ABSTRACT
Title of Dissertation: PREDICTORS OF DELAYED COLLEGE ENROLLMENT 
AND THE IMPACT OF SOCIOECONOMIC STATUS.
Heather T. Rowan, Doctor of Philosophy, 2005
Dissertation Directed By: Laura W. Perna, Ph.D.
Department of Education Policy and Leadership
This study contributes to our understanding of the decision of students to delay 
enrollment in college by building on the work of Hearn (1992).  This study uses 
constructs from models of traditional college student enrollment (Cabrera & LaNasa, 
2001; Perna, 2000; St. John, 2003), and data from the NELS:88/2000 longitudinal study 
of 1992 high school graduates.  Descriptive analyses are used to determine the 
appropriate break point between immediate and delayed enrollment, as well as compare 
differences in student characteristics based on timing of enrollment.   Multinomial logit 
regression is used to determine the predictors of delaying college enrollment rather than 
enrolling in college immediately after high school or not enrolling, and to explore if 
socioeconomic status is related to timing of college enrollment after controlling for other 
variables.  
At least six conclusions may be drawn from this study.  First, the appropriate 
break point between immediate and delayed enrollment is the end of the calendar year 
that a student graduated from high school.  This study also reveals the value of looking at 
a delayed group in comparison to immediate enrollment and no enrollment, since there 
are differences among all three groups for measures such as background characteristics, 
academic preparation and achievement, and social and cultural capital.  Third, a
combined model of college access based on traditional students (Cabrera & La Nasa, 
2001; Perna, 2000; St. John, 2003) is appropriate for examining the predictors of both 
immediate and delayed enrollment in college relative to no enrollment.  Fourth, measures 
of social and cultural capital are related to the timing of college enrollment but seem to be 
relatively more important in the decision to enroll immediately than in the decision to 
delay enrollment.  Fifth, graduates who delay enrollment average fewer resources and 
weaker preparation than graduates who enroll immediately, but average more resources 
and better preparation than graduates who do not enroll.  Finally, even after controlling 
for other variables, socioeconomic status is related to timing of college enrollment.  
Students who enroll immediately as well as those who delay enrollment have higher 
socioeconomic status than those who do not enroll, with those who enroll immediately 
having higher socioeconomic status than those who delay enrollment.
PREDICTORS OF DELAYED COLLEGE ENROLLMENT AND THE IMPACT OF 
SOCIOECONOMIC STATUS
by
Heather Terese Rowan
Dissertation submitted to the Faculty of the Graduate School of the
University of Maryland, College Park in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
2005
Advisory Committee:
Professor Laura W. Perna, Chair
Dr. Jacqueline E. King
Professor Susan Komives
Professor Jeffrey F. Milem
Dr. Nancy S. Shapiro
?Copyright by
Heather Terese Rowan
2005
ii
DEDICATION
To my parents, Patrick and Linda Rowan, who have supported me in all of my 
educational endeavors.
iii
ACKNOWLEDGEMENTS
There are so many people who helped to make the completion of this dissertation 
possible.  First, I would like to thank Laura Perna, my mentor, chair, and advisor.  Her 
role modeling, quick and meaningful feedback, and support helped me to push my 
thinking and make this dissertation a much better finished product.  I would also like to 
thank Jeff Milem for his advising, support and inclusion into many great research 
opportunities during the last four years.  I would also like to thank Susan Komives, Jaci 
King, and Nancy Shapiro, who took time out of their busy schedules to be insightful 
members of my committee.  
Next, I would not have made it through this program without the friendship, 
support, and assistance of the members of my cohort.  We have come a long way from 
those first days when we didn?t think we would ever make it to this point!  Kathleen, 
Michelle, Michele, and Courtney, I am thankful that you were the women I started with 
and that we were able to share so many of life?s experiences together over these last few 
years.  Michelle, I would especially like to thank you for all of your editing help and all 
of the daily phone calls over the past few months.
I have also been very lucky during the past four years to have great colleagues, 
who encouraged me and supported me.  To my University of Maryland colleagues in 
Office of Judicial Programs and in the Lumina Office, and my colleagues and friends at 
Mount St. Mary?s College, I thank you for all of your support and laughter in helping me 
to achieve this.  
iv
I would also like to thank my parents, sister, and all of my friends, who were so 
supportive during this process, even if they did not always understand what I was doing.  
Your patience, understanding and encouragement were so valuable to my education, and 
I am so grateful to have you all be a special part of this.  
Lastly, I would like to thank Mark for all of his love, patience, and support during 
my dissertation process.  Thank you for believing in me and encouraging me on the days 
when I wanted to give up, and for wiping away the many tears that went into this 
dissertation.  
v
TABLE OF CONTENTS
DEDICATION.................................................................................................................... ii
ACKNOWLEDGEMENTS...............................................................................................iii
LIST OF TABLES............................................................................................................. xi
LIST OF FIGURES .......................................................................................................... xii
INTRODUCTION .............................................................................................................. 1
Introduction..................................................................................................................... 1
Purpose............................................................................................................................ 5
Theoretical and Conceptual Framework......................................................................... 6
Research Method .......................................................................................................... 10
Significance of the Study.............................................................................................. 12
Limitations and Delimitations of the Study.................................................................. 14
LITERATURE REVIEW ................................................................................................. 16
Introduction................................................................................................................... 17
Delayed College Enrollment......................................................................................... 18
Theoretical and Conceptual Approaches to Traditional College Enrollment............... 22
Econometric Approaches.......................................................................................... 23
Academic Preparation and Achievement.............................................................. 25
College Cost.......................................................................................................... 26
Forgone Earnings.................................................................................................. 29
Limitations of the Econometric Model................................................................. 29
Summary of Econometric Approaches ................................................................. 30
Sociological Approaches .......................................................................................... 31
vi
Habitus .................................................................................................................. 33
Cultural Capital..................................................................................................... 35
Findings on Cultural Capital............................................................................. 38
Social Capital........................................................................................................ 40
Findings on Social Capital................................................................................ 47
Summary of Sociological Approaches.................................................................. 48
Combined Approaches.............................................................................................. 49
Combined Models for Low-Socioeconomic Status Students ............................... 51
Summary....................................................................................................................... 54
RESEARCH DESIGN AND METHODOLOGY ............................................................ 55
Introduction................................................................................................................... 56
Data............................................................................................................................... 57
Instrumentation ............................................................................................................. 61
Statistical Analysis........................................................................................................ 62
Conceptual Model and Variable Description................................................................ 63
Enrollment................................................................................................................. 65
Background Characteristics ...................................................................................... 66
Socioeconomic Status ........................................................................................... 66
Race/Ethnicity....................................................................................................... 69
Gender................................................................................................................... 69
Academic Preparation............................................................................................... 70
Academic Achievement........................................................................................ 70
Coursework........................................................................................................... 71
vii
College Costs ............................................................................................................ 71
Importance of Costs and Financial Aid ................................................................ 72
Tuition................................................................................................................... 72
Social Capital............................................................................................................ 73
Number of Financial Aid Contacts ....................................................................... 74
Parental Involvement in Student?s Education....................................................... 75
Closed Networks................................................................................................... 76
Student-Teacher Relations.................................................................................... 77
High School Based Support.................................................................................. 77
High School Control ............................................................................................. 78
Cultural Capital......................................................................................................... 80
Parental Expectations............................................................................................ 80
Parental Involvement ............................................................................................ 81
Educational Materials in the Home....................................................................... 82
Peer Encouragement ............................................................................................. 82
Involvement in Cultural Activities........................................................................ 83
Limitations of the Study................................................................................................ 87
RESULTS ......................................................................................................................... 93
Introduction................................................................................................................... 93
Research Question One: Timing of College Enrollment .............................................. 94
Research Question Two: Differences in High School Characteristics by Time of 
College Enrollment ....................................................................................................... 97
Research Question 3: Characteristics of Graduates Who Delay Enrollment.............. 104
viii
Research Question 4: Predictors of Timing of College Enrollment ........................... 114
Research Question 5: Socioeconomic Status as a Predictor of Timing of College 
Enrollment................................................................................................................... 119
DISCUSSION, CONCLUSIONS, IMPLICATIONS, AND RECOMMENDATIONS 
FOR FUTURE RESEARCH .......................................................................................... 124
Introduction................................................................................................................. 124
Discussion................................................................................................................... 125
Research Question One: Timing of College Enrollment ........................................ 125
Research Question Two: Variation in High School Characteristics Based on Time of 
College Enrollment ................................................................................................. 126
Research Question Three: Characteristics of Graduates Who Delay Enrollment .. 127
Research Question Four: Predictors of Timing of College Enrollment.................. 128
Research Question Five: Socioeconomic Status as a Predictor of Timing of College 
Enrollment............................................................................................................... 133
Conclusions................................................................................................................. 134
Implications................................................................................................................. 137
Recommendations for Future Research...................................................................... 142
APPENDIX..................................................................................................................... 146
Appendix A.  Comparison of 1992 high school graduates who first enrolled in fall 
1992 and spring 1993 by selected student characteristics .......................................... 146
Appendix B.  Comparison of 1992 high school graduates who first enrolled in college 
in spring 1993 versus academic year 1993-1994 enrollment by selected student 
characteristics.............................................................................................................. 148
ix
Appendix C.  Comparison of 1992 high school graduates who first enrolled in college 
in 1995 and 1996 calendar years by selected student characteristics. ........................ 150
Appendix D. Comparison of 1992 high school graduates who first enrolled in college 
in 1996 and 1997 calendar years by selected student characteristics. ........................ 152
Appendix E. Comparison of 1992 high school graduates who first enrolled in college 
in 1998, 1999, and 2000 calendar years by selected student characteristics .............. 154
Appendix F.  Comparison of 1992 high school graduates who first enrolled in college 
in 1992, 1993, and 1994 calendar years by selected student characteristics .............. 156
Appendix G.  Odds-ratios, co-efficients, and standard errors based on time of 
enrollment, with no enrollment as the baseline group ................................................ 159
References....................................................................................................................... 161
x
LIST OF TABLES
Table 3.1.  Percentage of socioeconomic status composite components by 
socioeconomic status quartile????????????????????67
Table 3.2. Components of the Factor Composite:  Parental Involvement in Student?s 
Education ???????????????????????????..76
Table 3.3. Components of the Factor Composite:  Parental Involvement with a Child?s
School ????????????????????????????...81
Table 3.4. Components of the Factor Composite:  Peer Encouragement ??????.83
Table 3.5.  Number and percentage of cases that are missing for each of the variables 
in the analyses from the final sample .????????????????...88
Table  3.6.  Percentage of cases with missing data for selected characteristics of 1992
high school graduates ???????????????????????89
Table 4.1.  Number and percentage of 1992 high school graduates who first enrolled 
in college in each of nine calendar years after graduating from high school.?...95
Table 4.2.  Distribution of 1992 high school graduates by timing of college 
enrollment ???????????????????????????.97
Table 4.3. Percentage of 1992 high school graduates who enroll in college by 2000 by
school type and status??????????????????????... 98
Table 4.4  Characteristics of 1992 high school graduates by enrollment timing ???100
Table 4.5.  Characteristics of 1992 high school graduates who delayed first enrollment 
in college by year of enrollment.??????????????????.107
Table 4.6.  Coefficients, standard errors, and odds-ratios for predictors of time of
enrollment among 1992 high school graduates ????????????..118
xi
Table 4.7.  Odds-ratios for predictors of enrollment timing of 1992 high school 
graduates ?.??????????????????????????.121 
xii
LIST OF FIGURES
Figure 1. Conceptual Model of Traditional College Enrollment??????????7
Figure 2. Variables in Conceptual Model of Timing of College Enrollment ????..64
1
CHAPTER 1
INTRODUCTION
Introduction
College enrollment rates vary systematically based on income and socioeconomic 
status, with lower enrollment rates for lower income students and students with low-
socioeconomic status. Fitzgerald and Delaney (2002) stated that the size of the gap in 
college enrollment between low- and high-income families was the same in 1970 and 
1997, 32 percentage points, even though there had been an increase in the college 
enrollment rates for families of all income levels.  Cabrera and LaNasa (2001) explored
college enrollment and the stages leading up to it-being college qualified, completing 
high school, and applying to a four-year college.  They found, after controlling for 
relevant college-choice variables in their study using the NELS data , a 26.4 percentage 
point gap in college application rates between low-socioeconomic status and high-
socioeconomic status students.  
Research (Perna, 2000) shows that students from low-income families are less 
likely to enroll in college even after controlling for other variables.  Perna (2000), in her 
examination of college enrollment among 1992 high school graduates of different 
racial/ethnic groups, included family income as a variable in her conceptual model.  In 
her multivariate analyses, she found that students with a higher family income had a 
significantly greater probability than students with a lower family income of enrolling in 
college in the fall after graduation from high school.  She also found that, when each 
group was examined separately, that income was not  a significant predictor of enrollment 
2
for African American or Hispanic students after controlling for other variables, but was 
still a significant predictor of college enrollment for White students.  
While research consistently shows lower rates of enrolling in college within one 
or two years of high school graduation for students with lower income and low-
socioeconomic status than for other students, little is known about what happens to low-
income students, or any student, who does not enroll in college immediately after high 
school.  The vast majority of research (Cabrera & La Nasa, 2001; Hossler, Braxton, & 
Coopersmith, 1989; Hossler & Gallagher, 1987; Manski & Wise, 1983; Perna, 2000; St. 
John, 2003) examines the college enrollment of traditional aged students immediately 
after high school, i.e., within two years of high school graduation.  While such work 
sheds light on the variables that affect traditional college enrollment, little is known about 
the decision-making process of students who may choose to enroll in college at a later 
date.  
A review of trends in enrollment suggests that growing numbers of individuals 
who do not transition immediately from high school to college may be enrolling in 
college at some later point in time. About 39 percent of all students participating in any 
type of postsecondary education were over the age of 25 in 1999, up from 28 percent in 
1970 (Choy, 2002). While these data may also reflect an increase in part-time and post-
baccalaureate enrollment, a share of this growth in older student enrollment may be 
attributable to an increase in the number of students who are enrolling in college several 
years after graduating from high school.
Although the growing share of older students implies growth in delayed 
enrollment, there is limited research about the characteristics of students who delay 
3
college enrollment, or how the predictors of delaying enrollment compare to the 
predictors of immediate enrollment or no enrollment.  Since literature on traditional 
student access has shown that low-socioeconomic status students are less likely than 
high-socioeconomic status students to attend college immediately after high school
(Cabrera & La Nasa, 2001; St. John, 2003) low-socioeconomic status students may 
represent a greater portion of the college population that delays enrollment to a later time.
Examining this nontraditional form of enrollment may inform practitioners and policy 
makers of avenues where the socioeconomic gap in enrollment can be closed, and lead to 
further exploration of the persistence of this group.  More generally, understanding the 
characteristics of students who delay entry and the predictors of the decision to delay 
enrollment may suggest ways to better assist all students in this group.
Some research identifies the characteristics of students who delay enrolling in
college immediately after graduating from high school (Aslanian, 2001; Choy, 2002; 
Choy & Bobbit, 2000; Cook & King, 2004; Hearn, 1992; Horn & Carroll, 1996), and 
supports the notion that low-socioeconomic status students are overrepresented in this
group.  Choy and Bobbit (2000) found, through their analysis of data from the 1995-1996 
National Postsecondary Student Aid Study, that 40%  of undergraduate students aged 24 
to 29 who enrolled in college were classified as low-income.
With the exception of Hearn (1992), few researchers have examined the variables 
that influence the decision of a student to delay entry into college after high school 
graduation.  Hearn explored the predictors of each of three non-traditional college 
enrollment behaviors: delaying enrollment by at least one year but not more than two 
years after graduating from high school, enrolling part-time, and enrolling in a non-
4
degree program.  He defined delayed enrollment as a break of between one and two years 
after graduating from high school partly because his dataset included data only up to two 
years after high school graduation.  Using the High School and Beyond dataset and 
logistic regression, he found that 1980 high school graduates who delayed enrollment
were disproportionately students from low-socioeconomic backgrounds and low-levels of 
academic preparation for college after controlling for other variables.  
Hearn?s (1992) study sheds light on the effects of such variables as 
socioeconomic status, educational aspirations, and academic background on the decision 
to enroll in college more than one year but less than two years after graduating from high 
school.  But, little is known about the predictors of the decision to enroll in college more 
than two years after graduating from high school.  Moreover, Hearn?s study used data 
describing students who graduated from high school in 1980.  The generalizability of his 
findings to later cohorts of students is unclear, given changes in the economic, social and 
political landscape since 1980.  Also, models of college student enrollment have 
advanced since the publication of Hearn?s study.  Attention to such concepts as social and 
cultural capital by researchers such as Perna (2000), McDonough (1997) and Cabrera and 
LaNasa (2001) has improved the understanding of the decision to enroll in college 
directly from high school, suggesting that these concepts may help explain the decision of 
students to delay enrollment as well.  
This introductory chapter first presents an overview of the study and the 
theoretical framework that supports it.  Next the problem statement, an overview of the 
research model, and research questions are presented.  The chapter also describes the 
expected implications and significance, as well as the limitations, of the study.
5
Purpose
The proposed study builds on the work of Hearn (1992) by looking at a more 
recent group of students, expanding the time after high school when a student may decide 
to enroll in college, and utilizing a more comprehensive conceptual model of the college 
enrollment decision.  This study also determines the most appropriate definition of 
delayed enrollment by examining the percentages of students who enroll in college each 
year after graduating from high school and comparing the characteristics of students who 
enroll in each year.  Using a conceptual model that is based on what is known about
decisions to enroll in college immediately after high school, this study examines the 
variables that differentiate a student?s decision to enroll in a two- or four-year college
immediately after high school graduation, delay first college enrollment, and not enroll in 
college within eight years of high school graduation.  The study also explores the 
relationship between socioeconomic status and patterns of college enrollment to 
determine if the socioeconomic status gap that is present in traditional enrollment is also 
present in delayed enrollment.  
Specifically, this study addresses the following five research questions:  
1. What percent of high school graduates enroll in college in each of eight years 
after graduation from high school?  What is the appropriate break point 
between immediate and delayed enrollment?
2. How do the characteristics of high school graduates who delay entry into a 
two- or four- year college degree program compare to the characteristics of 
high school graduates who enroll immediately after high school and the 
6
characteristics of high school graduates who do not enroll within eight years 
of high school graduation?
3. For high school graduates who delay enrollment, how does the timing of first 
enrollment vary by student characteristics?
4. What are the predictors of delaying entry into a two-  or four-year college 
degree program for up to eight years rather than enrolling immediately after
high school or not enrolling within eight years?
5. After controlling for other variables, is socioeconomic status related to the 
decision to delay enrollment in college?
Theoretical and Conceptual Framework
Figure 1 shows the conceptual model that guides this examination of delayed 
college enrollment.  This study tests a conceptual model that is developed based on what 
is known from prior research about ?traditional? college enrollment (i.e., enrollment in 
college immediately after graduation from high school), but especially the work of
Cabrera and LaNasa (2001) and Perna (2000).  Cabrera and LaNasa focused on 
understanding college enrollment processes of low-socioeconomic status students, while
Perna focused on understanding differences in college enrollment by race/ethnicity.  Both 
focused on the enrollment of high school graduates no more than two years after high 
school graduation.  Both also developed and tested conceptual models that incorporated 
aspects of human capital, social capital, and cultural capital theoretical frameworks.  A 
review and synthesis of prior research suggests that measures of background 
characteristics (e.g., socioeconomic status, race/ethnicity, gender), human capital (e.g., 
academic preparation and achievement), social capital (e.g., parental involvement in 
7
education), cultural capital (e.g., parental educational attainment), and financial 
resources (e.g., family income, tuition, and financial aid) all influence the decision of 
traditional age high school graduates to enroll in college within two years of graduating 
from high school.  
Figure 1. Conceptual Model of Traditional College Enrollment
The economic theory of human capital predicts that, as an individual invests in 
his/her human capital, he/she will realize higher returns in the form of income over the 
course of a lifetime (Becker, 1993).  This traditional econometric approach assumes that 
students make a decision to enroll in college based on the weighting and ranking of all 
alternatives relative to their preferences.  Enrollment in a two- or four-year degree
program is an investment in human capital that can generate returns, but that has costs 
associated with it.  Although the major economic return of higher education is higher 
lifetime earnings, participants also realize other benefits, including lower rates of 
unemployment and higher occupational status (Becker, 1993).
Timing of 
enrollment in a 2-
or 4-year college 
degree program
Background 
Characteristics
Social Capital
Cultural Capital
Financial Resources
Academic Preparation and  
Achievement
8
Students who enroll in college immediately after high school are more likely to 
receive a greater economic return over time than those who delay enrollment because 
they have more years over which to realize the return on their investment. This earnings 
profile may explain, at least in part, why college en rollment research generally focuses on 
early, rather than delayed, enrollment.  Nonetheless, an individual who  begins an 
investment in higher education eight years after high school graduation may still have 
more than 30 years in the workforce to realize the earnings premium. Even a delay in 
college enrollment of up to eight years after high school graduation may generate 
substantial economic, social, and psychological benefits for the individual participant, as 
well as for society at large.
Costs of enrolling in higher education can be direct, such as tuition, fees, and 
books.  The direct cost of college, aspects of which are typically included in traditional 
econometric models of college enrollment, is an important component in a student?s 
college enrollment decision.   Paulsen and St. John (2002) found that the cost of attending 
college and financial aid given to lower college costs were significant factors in a low-
income student?s college-choice decision.  They found through their multivariate 
analyses of 1987 National Postsecondary Student Aid Study (NPSAS) data that a 
majority of low-income students chose a college based on low cost and financial aid.  St. 
John (2003) posited that ?students respond to prices and subsidies based on their situated 
circumstances? (p. 187) particularly their family income and academic ability.  Heller 
(1997) also found, based on his review and synthesis of prior research, that the college 
enrollment of lower-income students tended to be affected to a greater degree by 
increases in tuition and decreases in aid than the college enrollment of students from 
9
families with higher incomes.  As a result of these findings, academic preparation and 
achievement, college costs, and financial aid are included in the conceptual model for this 
study.
I incorporated social capital and cultural capital into the conceptual model for 
this research, as suggested by Perna (2000), Paulsen (2001), and Hurtado, Inkelas, 
Briggs, and Rhee (1997), in order to test a more complete model of student decision 
making.  Perna (2000) showed that adding social and cultural capital to a traditional 
econometric model improved the explanatory power of the model, thereby lending more 
insight into the predictors of college enrollment.  
While human capital is an investment in education, training, health or mobility, 
social capital is an investment in relationships that helps the individual participating in 
the relationship or network to gain knowledge about resources available (Lin, 2001a).
An individual accumulates social capital as he or she participates in a network and from 
that network gains access to information or other resources that the person can use to his 
or her advantage.  Coleman?s (1988) approach to social capital focuses on a person?s 
connection to a social structure such as a family, community, or school, and the 
information and resources that one can gain from participation in that structure.  More 
specifically, he focuses on the transmission of social capital both within and outside of 
the family.  It is from the structure of the relationships between and among actors that 
information is gained and then can be converted into social capital.  The resources gained 
from participation in the network make possible a particular achievement such as 
enrollment in college.  Social capital in this study is measured by parental involvement in 
10
the student?s education, number of financial aid contacts, student-teacher relations, high 
school based support, and high school control.
Cultural capital is the passing on of values and norms from one generation to the 
next, or ?social reproduction? (Bourdieu & Passeron, 1977).  This symbolic good is held 
by all socioeconomic groups, with middle and upper class groups having the most 
economically valued forms of cultural capital (McDonough, 1997).  Those who have
more valued forms of cultural capital (i.e., members of the dominant class) may work to 
exclude others who do not have the same forms of cultural capital (Bourdieu, 1986; 
Lamont & Lareau, 1988).   The type of cultural capital that a student possesses can affect 
student enrollment choices, which are often influenced by cultures and values (Paulsen & 
St. John, 2002). Cultural capital in this study is measured by parental expectations, peer 
encouragement, parental involvement with the school, educational materials in the home, 
and participation in art, music, or dance classes.
Research Method
This study draws on data from the National Education Longitudinal Study: 1988-
2000 (NELS) to examine the research  questions.  Sponsored by the National Center for 
Educational Statistics (NCES) of the United States Department of Education, the NELS
tracked a group of students periodically from the 8
th
 grade through eight years out of high 
school, to study the transitions that students experience over time from middle school 
through high school and their entrance into the workforce and/or postsecondary 
education. I utilize information from the second (1992), third (1994), and fourth (2000) 
follow-ups to gain information about the end of their 12
th
 grade experience through eight 
11
years out of high school.  The sample is limited to students who graduated from high 
school on time in 1992, since college enrollment is usually limited to students who have 
completed high school.
The NELS is an appropriate dataset for this study for several reasons.  This 
dataset provides a large number of cases (analytic sample size= 8,567).  The dataset also 
has high response rates, ranging from 94.0% for the third follow-up (1994) to 77.6% for 
the fourth follow-up (Curtin, Ingels, Wu, & Heuer, 2002).  The dataset includes many 
variables that I can utilize to address my research questions.  Finally, the dataset is 
representative of the population of 1992 high school graduates nationwide when 
weighted by the normalized panel weight (F4F2PNWT).
Descriptive and multinomial logit regression analyses are used to address the 
research questions.  Descriptive analyses are used to address the first, second, and third
research questions, which require describing and comparing the characteristics of 
graduates. I use the results of the first research question to determine the number of years 
that appropriately differentiate between ?immediate? and ?delayed? enrollment. For 
research questions four and five, multinomial logit regression is used to examine the 
differences in the decision of high school graduates to delay enrollment rather than enroll 
in college immediately after graduation fro m high school or not enroll within eight years.  
Multinomial logit models are a type of loglinear model used for dependent 
variables that are categorical and have more than two responses (SPSS, 1999) .  The 
dependent variable, time of enrollment, has three categories: enroll immediately after 
graduating from high school, delay enrollment, or not enroll within eight years.  The 
multinomial logit model estimates the log-odds of one outcome occurring (e.g., no 
12
enrollment) relative to the baseline category (SPSS, 1999).  For this study the baseline 
category is delaying enrollment in college after graduating from high school.   
Significance of the Study
Existing research on college enrollment is largely limited to traditional college 
enrollment patterns (i.e., enrollment into college immediately after graduation from high 
school).  Little is known about the applicability of existing conceptual models of college 
enrollment to non-traditional college enrollment patterns such as delayed enrollment.  
Moreover, as Adelman (2002) stated based on his review of the literature, little is known 
about the initial college enrollment of socioeconomically disadvantaged students, with 
few exceptions (e.g., Cabrera & La Nasa, 2001; Hearn, 1992).  Paulsen and St. John
(2002) recommended, based on their research on the student-choice construct, the 
exploration of different patterns of student choice, such as differences in timing of 
enrollment, in order to refine college enrollment theory for different groups.
The research contributes to theory by testing the applicability of conceptual 
models that have been developed and tested on traditional college enrollment on the 
decision to delay college enrollment.  The findings may suggest ways in which models of 
college enrollment may need to be modified to understand this behavior.  This study may
help inform other researchers of the applicability of a model that incorporates measures 
of human, cultural, and social capital for examining the decision of students to delay 
entry into college after graduation from high school. 
This study also has implications for policy and practice.  Results of this study
shed light on how the characteristics of students who delay entry into college differ from 
the characteristics of students who enroll immediately after graduation from high school.     
13
Providing a more current description of the characteristics of graduates who delay entry 
into college may shed additional light on the characteristics and nature of the 
socioeconomic status stratification of higher education opportunity. Hearn (1992) found, 
using a sample of students that is now more than 20 years old, that students who delay 
entry are less prepared academically and of lower socioeconomic status students than 
students who do not delay college enrollment.
By exploring differences in the predictors of delayed versus immediate 
enrollment, the results of this study may suggest modifications to and improvements in 
policies and practices that promote enrollment in college at any point within eight years 
of high school graduation.  Such policies and programs may be particularly important for 
students who, for whatever reason, have less of the resources that are required to enroll in 
college immediately after graduation from high school.  
Gaps in enrollment by socioeconomic status and race as documented by Cabrera 
and LaNasa (2001) and Perna (2000) suggest persisting inequities in the opportunity to 
access the benefits of higher education.   Cabrera and LaNasa (2001) found that low-
socioeconomic status students were less likely to enroll in college due to a lack of 
academic qualifications and a lower rate of college applications.  Perna (2000) found that 
if students from underrepresented racial/ethnic groups had the same levels of academic 
ability, finances, and social and cultural capital as Whites, they would be as likely to 
enroll in college as White students.  But, she found that many African American and 
Hispanic students did not have access to these resources thereby perpetuating the 
observed racial/ethnic enrollment gap.  This study shows the extent to which 
socioeconomic status gaps in college enrollment persist even when a longer-term view of 
14
college enrollment is taken.  The results may provide additional insight i nto ways to 
equalize educational attainment across socioeconomic status by recognizing that 
enrollment even two to eight years after high school graduation has potential benefits to 
individuals and society.  
Promoting delayed college enrollment has benefits for both society and
individuals.  Society benefits from increased productivity, increased tax revenue, and less 
drain on the social welfare system.  Individuals with a college education benefit not only 
in terms of higher returns but also in terms of cognitive learning, emotional and moral 
development, and good health (Bowen, 1997).  Bowen argued that the most important 
benefit is intergenerational.  When a parent attains a college degree, his/her children are 
more likely to enroll in college in the future (Bowen, 1997).   
Limitations and Delimitations of the Study
This study has several limitations.  Since the study involves secondary data 
analysis, the study is limited to the use of the variables that are available in the NELS.  
Since social capital and cultural capital are complex constructs, it is difficult to find
appropriate measures of these constructs for the conceptual model in this study.  Proxies 
of social capital include parental involvement in a student?s education, number of 
financial aid contacts, student-teacher relations, high school control, and high school 
based support. Parental expectations and involvement, peer encouragement, educational 
materials in the home, and visits to museums/participation in the arts act as proxies of 
cultural capital for this study. While these proxies are appropriate, many of them (e.g.,
parental involvement and financial aid contacts) measure the quantity of the aspect of 
capital, not its quality.  High school control and the percent of students in the school 
15
receiving free or reduced price lunch is used as a proxy for resources available at the 
school attended, but again, more specific measures (e.g., quality of counseling for 
college), if available, would be preferred.
Predictors not included in the model which could be helpful to the study include a 
student?s perception of the benefits of higher education.  The NELS dataset does not have 
variables that are adequate proxies for student?s perceived benefits of college or forgone 
earnings.  
Reflecting the construction of the dataset, this study examines college enrollment 
only up to eight years after high school graduation.  Data from more than eight years after 
high school graduation may provide greater information related to college enrollment 
patterns, if it were available.  Nonetheless, the eight year time period may be an 
appropriate cut-off given the declining benefits of higher education to older individuals. 
Another limitation is that the model does not account for changes in economic 
circumstances that occurred after 1992 which may play a role in the decision to enter 
college two to eight years after high school graduation.  In the early 1990s when students 
in the sample graduated from high school, the economic picture was less stable than in 
more prosperous times of the late 1990s.  These economic changes may have encouraged 
graduates to enroll immediately into college, but deterred individuals who might have 
otherwise enrolled at a later point to stay in the labor market. 
Even with these limitations, the study is worth pursuing due to the likely 
contributions to theory, policy and practice.  The strengths of the research design, 
including the size of the sample, the longitudinal nature of the NELS dataset, and the 
number of variables that are available to represent complex constructs in the analyses 
16
also support the worth of the study.  The study tests the applicability of a conceptual 
model of traditional college enrollment to the decision to delay college enrollment.  The 
study also contributes to our understanding of the nature of socioeconomic status gaps in 
educational opportunity by examining socioeconomic gaps in delayed enrollment of 1992 
high school graduates.  
The following chapter explores the literature that informs this study.  The third 
chapter describes the methodology for the study.  Chapter four discusses findings of the 
study, and chapter five includes conclusions from the study.
17
CHAPTER 2
LITERATURE REVIEW
Introduction
This study uses descriptive and multinomial logistic regression to examine the 
predictors of a student?s decision to enroll in a two- or four-year college degree program 
immediately after graduation from high school, to delay enrollment for up to eight years 
after graduation, or to not enroll in college within eight years of high school graduation.  
The study also explores how time of entry in a college degree program varies based on 
graduates? socioeconomic status.  The following literature review provides background 
and support for this study.  
This study examines differences between graduates who enroll in college 
immediately after high school and those who delay as part of the third or final phase of 
Hossler and Gallagher?s (1987) model of the college enrollment process.  Hossler and 
Gallagher developed a three stage model of college choice based on a review of literature 
on the subject.  In their proposed model, developed for the immediate transition from 
high school to college, students gain understanding from multiple sources over time as to 
their higher education options.  In the first stage students become predisposed to continue 
their education after high school.  Hossler and Gallagher described the second stage as 
the search phase where a student gathers information about attending a postsecondary 
institution, including the characteristics and attributes of particular colleges and 
universities.  In the third phase of the choice process students decide whether and where 
to attend.  In their comprehensive review and synthesis of research, Hossler, Braxton, and 
Coopersmith (1989) further refined this model, labeling the third phase as ?choice.?  
18
Hossler, Schmidt, and Vesper (1999) also further refined the third phase of the model, 
making choices, to emphasize ?information processing.? Research (Cabrera & La Nasa, 
2001; Hurtado et al., 1997) suggests that this three phase model may be useful for 
understanding the characteristics of students who enroll in college directly after high 
school.  The applicability of the model to students who delay college enrollment after 
high school is not clear.  Regardless, this research focuses on the third phase of the 
process, choosing to enroll in college.
The literature review begins by describing what is known about the decision to 
delay college enrollment from the limited amount of research on this topic.  Because few 
researchers have examined delayed college enrollment, the chapter then reviews and 
analyzes models of immediate college enrollment, with particular attention to research 
that focuses on the college enrollment of low-socioeconomic status and low-income 
students.  I show how quantitative models for the immediate transition from high school 
to college have progressed from purely econometric, to models that include measures of 
sociological constructs, namely habitus, cultural capital, and social capital, to further 
understand the immediate college enrollment of high school students.  The chapter 
concludes by summarizing what is known and not known about the decision to delay 
entry into college after graduation from high school and describing the ways in which this 
study addresses the knowledge gap.  
Delayed College Enrollment
While much research has examined the decision of students to enroll in college, 
almost all of this research examines only the immediate transition to college from high 
19
school.  Entering the key words ?nontraditional students?, ?enrollment? , and ?delay? into 
the ERIC, Education Abstracts, Digital Dissertations, and Social Science Citation Index 
search engines generated over 200 journal articles, conference presentations, and 
dissertations published since 1985.  Only one of these ?hits? examined predictors of 
delayed college enrollment: Hearn (1992).  The following section describes the method, 
results, and limitations of his study.  
Hearn (1992) explored the predictors of delaying enrollment in college as a part 
of his examination of non-traditional college enrollment.  Using data from the 1982 
follow-up of the High School and Beyond (HS&B) longitudinal study of 1980 high 
school graduates, Hearn examined three aspects of non-traditional enrollment: enrolling 
in college part-time versus full-time; enrolling in a non-degree program rather than in a 
degree program at a two- or four-year institution; and delaying enrollment in college 
rather than enrolling in a two-year or a four-year institution within one year of graduation 
from high school. In his descriptive analyses, Hearn (1992) found that 52% of 1980 high 
school graduates enrolled in college within one year of graduating from high school, 10% 
enrolled between one and two years after graduating from high school, and 38% had not 
enrolled within two years.  Descriptive analyses also showed that higher percentages of 
students who were Black, from a larger family, of a lower socioeconomic status, and of a 
lower academic track in high school, and who had lower high school grades, test scores, 
and lower educational aspirations had delayed entry into college than had enrolled in 
college immediately after high school.  
Using logistic regression and limiting the analyses to students who enrolled in 
college within two years of high school graduation, Hearn (1992) examined the 
20
predictors of enrolling in college within one year of high school graduation rather than 
enrolling in college between one and two years after graduating from high school.  He 
found that some of the observed differences in time of enrollment disappeared after 
controlling for background characteristics, academic characteristics, and aspirations.  The 
control variables were entered into the model in three blocks.  The first block contained
such background variables as race/ethnicity, gender, parental socioeconomic status, and 
family size.  The second block included academic factors such as academic track in high 
school, high school grades, and test scores.  The final block included educational 
aspirations.   
The logistic analyses showed that students who delayed enrollment were less 
likely than those who enrolled in college immediately after high school to be 
academically prepared for college.  With an index of concordant-pair predictions at 68%, 
the analysis showed that, compared with those students who enrolled immediately after 
high school graduation, those who delayed enrollment were 27% less likely to be female, 
20% less likely to have been a part of an academic (college preparatory) curriculum, 26% 
less likely to have grades that were mostly A?s and B?s, and 17% less likely to aspire to 
attain a college or graduate degree.  Race and parental socioeconomic status were found 
to be significant, when only background and academic factors were being considered, 
with Black students and low-socioeconomic status students more likely than White and 
high socioeconomic status students, respectively, to delay enrollment.  When a measure 
of educational aspirations was added to the model, neither race nor socioeconomic status 
were significant predictors of delaying enrollment, suggesting that educational aspirations
21
mediated the direct effects of race and socioeconomic status on delaying enrollment 
(Hearn, 1992).
Hearn?s (1992) study illustrated differences in the predictors of immediate and 
delayed enrollment in college after high school.  His study also showed that the 
predictors of the decision to enroll in college vary based on the timing of the enrollment
and that the characteristics of students who delayed enrollment after high school 
graduation by between one and two years were different from the characteristics of 
students who enrolled within one year.  
Despite the contribution of Hearn (1992) to a virtually unresearched topic, his 
study also suggests several persisting knowledge gaps.  First, he examined the college 
enrollment decision of 1980 high school graduates.  The applicability of analyses 
describing high school graduates in 1980 to more recent cohorts is unclear given changes 
in economic, societal and political contexts in the United States. 
Second, Hearn (1992) defined ?delayed? college enrollment as enrolling in 
college within a limited period of graduating from high school (one to two years) rather 
than enrolling after a greater period of time.  While his study illustrated differences 
between these two groups, his study did not explore predictors of delaying college 
enrollment beyond two years of graduation from high school.  Although no available data 
describe trends in delayed enrollment per se, growth in the enrollment of older students 
suggests ?delayed? enrollment becoming more prevalent.  About 39 percent of all 
undergraduate student participating in any type of postsecondary education were over the 
age of 25 in 1999, compared with 28 percent in 1970 (Choy, 2002).     
22
Third, while Hearn?s (1992) conceptual model included measures of background 
characteristics, academic ability, and aspirations, a review of the body of research on 
traditional college enrollment that has been conducted since 1992 suggests that other 
variables also influence college enrollment decisions.  Other researchers (Cabrera & La 
Nasa, 2001; McDonough, 1997; Perna, 2000; St. John, 2003) have found that social 
capital, cultural capital, college cost, and financial aid also impact the immediate 
enrollment into college of high school graduates.  
Theoretical and Conceptual Approaches to Traditional College Enrollment
The bulk of the college enrollment literature has developed and tested models that 
defined college enrollment as enrollment no more than two years after graduating from 
high school.  Quantitative studies on this topic have been shaped by two major theoretical 
perspectives, the econometric approach (Kane, 1994; Manski & Wise, 1983) and the 
sociological status attainment approach (Alwin, Hauser, & Sewell, 1973; Hearn, 1991).  
The most recent quantitative researchers used models that relied on a combination of the 
two approaches (Cabrera & La Nasa, 2001; Perna, 2000; St. John, 2003) that is informed, 
at least in part, by findings from qualitative sociological research (McDonough, 1997).  
This section describes the primary theoretical approaches used to examine college 
enrollment.  It starts with a description of the econometric model pioneered by Manski 
and Wise (1983) and describes components of econometric models, namely academic 
achievement, college costs, and college benefits.  Then, sociological approaches focusing 
on habitus, social capital and cultural capital are examined.  Finally, combined models, 
23
which include components from both the econometric and sociological approaches, are
discussed.
Econometric Approaches
Econometric approaches to college enrollment are based on the economic theory 
of human capital, which predicts that as an individual invests in human capital, he/she is 
more likely to recognize higher returns (Becker, 1993).  An individual invests in human 
capital through education, training and medical care.  Becker (1993) stated that the most 
important investments in human capital are education and training because participation 
in these activities increases knowledge, skills, and analytical ability and in turn causes 
increased productivity, which is rewarded by greater earnings.
Human capital theory assumes that an individual decides to invest in human 
capital based on the ranking and weighting of all options according to preferences 
(Becker, 1993).  Hossler, Braxton, and Coopersmith (1989), in their review of research 
on college student enrollment, noted that econometric approaches assume that students 
will enroll in college if the net benefits of college are greater than the net benefits of all 
other alternatives.  Manski and Wise (1983) found that students decided whether and
where to attend college based on their skills and abilities balanced with their willingness 
and ability to pay for their education.  
One of the early models of college enrollment was that of Manski and Wise 
(1983).  They used econometric modeling, assuming that the decision to enroll in college 
is based on an assessment of costs and benefits of all alternatives, to explore student 
behavior affecting postsecondary school choices.  Their model posited that five 
24
components impacted the choices that a student made after high school graduation: 
academic aptitude, family income, cost and aid, high school quality, and labor market 
conditions.  They believed that these components influenced various steps in the college 
choice process: applying to college, being admitted, receiving financial aid, and selecting 
institutions with particular characteristics.  
Manski and Wise (1983) tested their model using the National Longitudinal Study 
(NLS) of 1972 high school graduates.  They found that an individual?s decision to apply 
to college within four years of graduating from high school was an important predictor of 
enrollment and was impacted by the student?s class rank, SAT scores relative to other 
students at the institution, and the level of education that their parents had attained.  
Kane (1994) also proposed an econometric model of college enrollment in his 
examination of Black and White students.  His model assumed that five components 
influenced the choice to attend college: pre-college preparation, gender, high school 
characteristics, high school location, family background, and tuition and financial aid.   
Kane (1994) tested his model using High School and Beyond data of 1980 high school 
graduates.  He found that the most important predictor of enrolling at a four-year college 
within two years of graduating from high school was standardized test scores, with high 
school grades, family income, and parental education also being important predictors of 
enrollment.  He also found that rising tuition prices and/or a decrease in available 
financial aid produced a disproportionate affect for low-income students, in that they 
were then less likely to enroll in college, due to cost sensitivity.  
The next sections will describe what is known from research about the primary 
components in the econometric models by Manski and Wise (1983) and Kane (1994), 
25
(academic preparation and achievement, and college cost and aid) as well as the 
limitations of the econometric model. 
Academic Preparation and Achievement
Academic preparation and achievement are key components of the econometric 
model of college enrollment, as demonstrated by Manski and Wise (1983), Kane (1994), 
and other researchers in their models of college student enrollment such as Perna (2000), 
Cabrera and La Nasa (2001) and St. John (2003).  Perna (2004) demonstrated through her 
review and synthesis of prior literature that academic preparation, defined as the quality 
and quantity of high school coursework, is essential to college enrollment, with it being a 
primary predictor even after controlling for other variables.  Adelman (2002), through his 
work with the NELS dataset , also found that academic preparation, defined as a 
composite measure of academic intensity of high school curriculum, standardized test 
scores, and high school class rank, was a strong positive predictor of college enrollment.  
He argued that academic preparation in high school helped motivate students to look 
towards a college education.  
Cabrera and LaNasa (2001), in their study of low-socioeconomic status student 
college enrollment, argued that academic preparation is required to be academically 
qualified to attend college, a key step in the college enrollment process.  Their college 
qualification index incorporated the following measures of academic preparation and 
achievement: high school grade point average, class rank, aptitude test scores, and 
SAT/ACT scores.  Even after controlling for background characteristics, social capital, 
and aspirations, whether the students were college qualified made a significant positive 
26
impact on whether the students would apply to a four-year college, which in turn 
impacted their enrollment. 
Drawing on work by Adelman, St. John (2003) also measured academic 
preparation by using the ?college qualification index?.  St. John split students into three 
groups: not college qualified, minimally/somewhat college qualified, and highly/very 
highly qualified.  His descriptive analyses showed that only 53% of low -income students, 
while 86% and 68% of high- and middle-income students respectively, were college 
qualified.  Of those low-income students who were college qualified, 52% went on to 
attend a four-year college, compared to 83% of the high-income group.  
In her study of differences in college enrollment decisions based on race, Perna 
(2000) measured academic ability and achievement using test scores and whether a 
student participated in an academic curricular program.  She found that both test scores 
and curricular track were significant positive predictors of the probability of a student 
enrolling in a four-year institution within two years of high school graduation regardless 
of the race or ethnicity of the student.  In a study of the relationship between parental 
involvement and college enrollment also using the NELS, Perna and Titus (in press)
found that academic preparation was a positive predictor of college enrollment for 
students.  Their proxy for academic preparation was intensity of college courses, which 
they measured by the highest level of math courses that a student had completed. 
College Cost
Econometric models assume that college cost and financial aid have an important 
impact on student college enrollment.  A review of research shows that college cost and 
27
financial aid impact enrollment in that real and perceived unmet need influences a 
student?s decision to enroll in college.
Kane (1994) explored college cost as one of the components of his study of 1980 
high school graduates.  He found that high tuition states had lower college enrollment 
rates.  He also found, through his use of the 1980 High School and Beyond study, that 
every ?$1,000 increase in tuition is associated with a ten percentage point decline in 
enrollment rates among youth from below median family socioeconomic status and a 4.4 
percentage point decline among youth above the median? (p. 24).  Manski and Wise 
(1983) also found in their descriptive analyses of the National Longitudinal Study of 
1972 that, as the cost of an institution increased, students were less likely to choose that 
institution.
In his reexamination of descriptive analyses of NELS data from Access to Higher 
Postsecondary Education for the 1992 High School Graduates by the National Center for 
Education Statistics (NCES), St. John (2003) found that finances impacted a student?s 
ability to attend college.  Parent and student concerns about paying for college negatively 
impacted enrollment, as did family finances and unmet need.  College costs and the 
amount of financial aid awarded especially impacted the enrollment of low-income
students.  St. John found that, compared to higher income students, smaller percentages 
of low-income students enrolled in college, and when they did enroll, were less likely to 
enroll in a four-year institution, partly because of unmet need, concern of paying for 
college, and family finances, which tended to be more of a challenge for low-income 
students and their families than students with higher incomes.  
28
Heller (1997), in his update to a prior review and synthesis by Leslie and 
Brinkman on student price response to higher education, concluded that there is a link 
between income and sensitivity to tuition and financial aid.  Heller (1997) specifically 
found that ?every $100 increase in tuition results in a drop in enrollments of .5 to 1 
percentage point across all types of institutions? (p. 650).  The increase in tuition is an 
even larger problem for low-income students as their enrollment is more sensitive than
the enrollment of high-income students to an increase in the cost of college, whether the 
increase in cost is in the form of a rise in tuition, a drop in available financial aid, or a 
combination of the two.    
Students? perception of college cost affects their college choice as well.  Paulsen 
and St. John (2002) used the financial nexus model in their study of the effects of 
perceptions and expectations about costs of college.  The financial nexus model 
examined the link between students? perceptions of the costs of college at time of 
enrollment compared to actual costs of college (e.g., tuition, aid, cost of living), and the 
relation of cost to choice.  In their study, they analyzed the relationships between social 
class and cost sensitivity by income using 1987 NPSAS data.  They found that low-
income students often made their college choice based on the tuition of the institution and 
the financial aid offer.   Through descriptive analyses they found that 64% of low-income 
students chose an institution based on low tuition, student aid, or both of these variables, 
while less than 50% off upper-middle income students chose an institution based on these 
variables.  
29
Forgone Earnings
Another direct cost of schooling is the forgone earnings that a student could earn 
as a full-time employee, rather than attending college on a full-time basis.  Many students 
attending college full-time work part-time, thereby decreasing, but generally not 
eliminating, the costs of forgone earnings.  The difference between the salary that 
students could be earning as a full-time employee and the part-time salary that students 
earn as a full-time student is the forgone earnings cost (Becker, 1993).   For some, 
especially low-socioeconomic status students, forgone earnings alone limit participation
in higher education (Kane, 1999).  Kane found, through his analysis of NCES data, that if 
a male high school graduate under the age of 24 was to work full time his average salary 
would be $16,900.  If this student was enrolled in college full-time for nine months a year 
without working, the student would lose $12,675.  This cost was greater than the cost of 
tuition at the time.  If the student has other family responsibilities, the forgone earnings 
may cause the student not to enroll. 
Limitations of the Econometric Model
While the econometric approach helps researchers assess some of the predictors 
of college enrollment, there are limitations to this approach.  One limitation to the 
econometric approach is the assumption that individuals act rationally with perfect 
information about the costs and benefits of all alternatives and have adequate resources 
with which to make a decision.  Paulsen (2001) and St. John (2003) commented that 
lower income students may not have accurate information about the monetary and 
nonmonetary returns on their investment in higher education.  Moreover, the costs and 
30
benefits of participating in higher education may vary greatly across individuals due to 
variables such as socioeconomic status, background, access to information, and ability 
(Bowen, 1997; Paulsen, 2001).  
Another limitation of this approach is that a person?s personal preference about 
the desire to participate in college is often not measured.  If the person has a negative 
attitude about college and studying, does not have an interest in the types of careers that 
college educated individuals have, and his/her parents do not expect the student to attend 
college, the individual may not be interested in continuing their education beyond high 
school, and be satisfied with the returns that he or she will receive (Becker, 1993).  
Nonetheless, these preferences are difficult to control for when testing human capital 
theory (Douglass, 1997).
Summary of Econometric Approaches
In summary, econometric models assume that students decide to enroll in college 
after assessing all costs and benefits of attendance.  The models put forth by Manski and 
Wise (1983) and Kane (1994) show that compo nents of the decision process include 
academic preparation, cost and aid, and background characteristics.  While these models 
identify important predictors of enrollment, there are weaknesses of the model, such as 
the assumption that every student has perfect information with which to make their 
decision, as well as that all students have similar preferences and tastes for college.  
Sociological constructs such as habitus, social capital, and cultural capital provide 
another perspective to college enrollment, which is the focus of the next section.
31
Sociological Approaches
Sociological status attainment models focus on the ways in which socioeconomic 
characteristics contribute to a student?s educational and occupational attainment.  The 
sociological status attainment models began with those developed by University of 
Wisconsin researchers (Alwin et al., 1973).  The Alwin et. al. model focused on the 
effects of socioeconomic status, ability, and high school experiences on educational 
attainment.       
Hearn (1991) drew on a sociological status attainment approach to examine 
students? choice of college to attend.  His model focused on the effects of such 
background socioeconomic characteristics as parental income, parental educational 
attainment, family size, race, and gender on choice, defined as selectivity of the 
institution and institutional expenditures per student.  The model also included academic 
characteristics such as standardized test scores, high school grades, curricular track and 
high-school activities (e.g., participation in student government, journalism, debate club), 
as well as postsecondary educational aspirations.  He tested this model with data from the 
first follow-up (1982) of the High School and Beyond longitudinal study of 1980 high 
school seniors and limited the analyses to students who attended college within one year 
of graduation.  In his descriptive analyses he found that low-socioeconomic status 
students and Black students were more likely than higher socioeconomic status students 
and White students to attend institutions that were less selective (i.e., lower average 
institutional average SAT scores) and that had lower costs.  In his multivariate analyses, 
Hearn (1991) found that the strongest predictors of institutional selection were academic 
characteristics, namely grades, test scores, and academic track.  Educational aspirations,
32
socioeconomic status, and race were also significant predictors of institutional selection.  
Specifically, even after controlling for other variables, Black students and low-
socioeconomic students were less likely than other students to attend selective and higher 
cost institutions.     
McDonough (1997) identified and explained the role of a wider range of 
sociological constructs in her qualitative case study of the college choice process.  As a 
qualitative study, her work also generated and contributed to theory underlying college 
enrollment.  The theoretical framework that she developed based on her study was that 
students? college choices are affected by their cultural capital and bounded rationalit y, 
which are influenced by their habitus .  She defined cultural capital as a symbolic good 
held by all socioeconomic groups, with those of higher socioeconomic classes possessing 
forms of this capital that are more economically valued.  Her theoretical framework 
posited that the types of cultural capital that a student has will affect the type of college 
that a student will choose to attend.  Drawing on the work of Pierre Bourdieu, she defined 
habitus as ?a deeply internalized, permanent, system of outlooks, experiences, and beliefs 
about the social world that an individual gets from his/her immediate environment? 
(McDonough, 1997, p. 9).  She hypothesized that habitus also impacted the student?s
college choice.  McDonough (1997) defined bounded rationality as the idea that a person 
makes decisions with available information.  She found that students made what can be
considered rational college choices based on the information that was readily available to 
them at the time of the decision.  
In her case study of the college choices of 12 White high school girls from four 
different high schools, McDonough (1997) also found that a student's habitus affected the 
33
level and type of college education that a student intended to acquire.   The study 
documented that cultural capital acquired by students from private schools often gave 
them the advantage necessary by supplementing the capital that they already had in order 
to maximize their educational choices.  She also discussed the roles of organizational 
habitus, which describes the effects of the organizational contexts and status cultures in 
high schools, on the college-choice process, with the impact of the school on the decision 
making process, and the consistency of views across the domains of schools, family and 
friends.  Her study showed that cultural capital and habitus are critical constructs to 
consider when examining college enrollment.  The next sections further examine these 
sociological concepts as well as social capital in order to better understand the predictors 
of college enrollment.  
Habitus
Habitus, as described by Horvat (2001) in her chapter focusing on equity and 
access in higher education and the contribution of Pierre Bourdieu to theory, is ?rooted in 
status group based familial history, and serves to generate and bound the range of 
possible action in social settings? (p. 209).  Habitus is a series of perceptions that are 
unconsciously collected by individuals and internalized to provide an understanding of 
rules of social interaction to determine a place in the world.  Individual action is shaped 
by a person?s habitus.  A person develops possibilities for action from their habitus 
(Horvat, 2001).    Horvat (2001; 2003) found in her longitudinal qualitative study of the 
college-choice process of 14 low- and high-income Black girls at three high schools that 
the habitus of a student affected their educational choices by the implicit and explicit 
34
expectations for college attendance that family and friends had.  These expectations 
influenced where a student found it possible to go to college.
One example of habitus is from Horvat?s (2001) study of how habitus operated 
for one high school female.  Many students at the high school she attended did not go on 
to college.  But, the girl?s mother and siblings attended college.  Their educational
attainment influenced her, establishing college as within the student?s realm of 
possibility.  Many of this girl?s peers did not have the same experience.  Consequently, 
few went on to college.
 McDonough (1997) also found that students? decisions to attend college were
impacted by their habitus.  The students in her study who attended the private preparatory 
high school internalized the belief that they would go to an elite college, and in turn 
considered only a narrow subset of institutions that fit this category.  Attending an elite 
institution was understood to be within their realm of possibility.  In contrast, at a public 
high school with fewer guidance resources, students did not perceive that they could go 
to any college they wanted, based on the messages of peers, family, and school personnel. 
As a result, students at this school made more conservative college choices, even when
they had similar ability characteristics as students at the private school.
Habitus addresses an assumption of the econometric model of college enrollment 
that a person makes a decision with perfect and complete information and considers all 
possible alternatives.  Horvat (2001) and McDonough (1997) both demonstrate that 
students do not have perfect information with which to make decisions about college 
choice and that they consider only a small range of options. Students consider only 
alternatives that are consistent with their habitus, or worldviews.
35
Horvat (2001) argued that habitus enables a person to activate his/her capital.  If 
students do not perceive they can go to college, they will not enact their capital in order 
to do so.  McDonough (1997) also found that a student will not activate cultural capital to 
help get to college if the student does not have the habitus to support it. 
Cultural Capital
Pierre Bourdieu is the leading scholar of cultural capital.   Horvat (2001)
described the Bourdieuian construct of cultural capital ?as the cultural resources such as 
high status cultural knowledge about art or music that can be legitimate power for an 
individual? (p. 211).  Bourdieu and Passeron (1977) argued that the lack of this 
knowledge leads to exclusion from high status groups, thereby exerting a strong form of 
power.  The power wielded through cultural capital ?is a power to shape other peoples' 
lives through exclusion and symbolic imposition? (p. 18).   While all socioeconomic 
groups have some form of this symbolic good, middle and upper class groups tend to 
define and have more economically valued forms of this good, and are able to invest in it 
and maintain it (McDonough, 1997). 
Cultural capital can be mobilized to enhance human capital.  Individuals benefit 
the most from the acquisition of cultural capital when they utilize it and invest it for 
things that are socially valued, like academic achievement and enrolling in college.  
Cultural capital is primarily transmitted by the values and activities of parents to their 
children through blatant and subtle messages about what is valued and important (e.g., 
enrolling in college) (Bourdieu, 1986; McDonough, 1997).  
36
Bourdieu (1986) described three states of cultural capital: embodied, objectified, 
and institutionalized.  The embodied state of accumulating cultural capital is investing 
time to build culture within oneself.  In order to accumulate capital in this state a person 
must invest time to assimilate, or improve himself or herself in order to embody the 
desired culture.  This is typically done unconsciously, as children often ?inherit? their 
parents? cultural capital, which happens in all families regardless of income.
The objectified state of cultural capital is in a person?s exposure to material goods 
such as books, computers, or the media (Bourdieu, 1986).  Examples of this state include 
exposure to the arts or to a diverse society.  The objectified state of cultural capital is 
connected to the embodied state, as the embodied state determines which materials are 
the most valued in a specific culture.  
The institutionalized state of cultural capital ?confers on its holder a conventional, 
constant, legally guaranteed value with respect to culture? (Bourdieu, 1986, p. 50).  This 
acquisition of a formal credential, such as a college degree, gives an individual a formal 
means of capital that the person can then convert into economic capital and use to his or 
her advantage.  An example of this would be academic qualifications, especially from a 
selective institution, which may expose a person to a different level of culture than 
previously experienced, and serves as a signal of status, which in turn increases the 
amount of desirable cultural capital.  
Drawing on the work of Bourdieu and Passeron and Weber, Lamont and Lareau 
(1988) developed a theoretical model, based on previous literature, of how individuals 
can be excluded from access to sought after types of cultural capital.  Lamont and Lareau
defined cultural capital as institutionalized values, attitudes, preferences and formal 
37
knowledge that are used for cultural and social exclusion from high status groups.  They 
posited that there were four ways in which cultural capital can affect outcomes like 
college enrollment: self-selection, overselection, relegation, and direct exclusion.  Self-
selection, overselection, and relegation are all based on Bourdieuian concepts, while 
direct exclusion is a concept based on the work of Weber.  
Lamont and Lareau (1988) described self-selection as where an individual 
estimates their chance of success based on their current knowledge and in turn adjusts 
aspirations accordingly.  Applying this framework to college enrollment, low-income 
youth often self-eliminate themselves from attending college because of the perceived 
low chance of succeeding in college and/or their low academic ability.  If they decide to 
attend college, low-income students must expend more effort because they possess less-
valued cultural resources.  Low-income students often have less-valued cultural 
resources, but are often in competition for the same spaces in a college class with 
students who have access to more plentiful resources.  These students are expected to 
perform to the same level as those with greater resources, and so need to perform more 
than others.  Relegation may occur if low-income students do not possess desirable forms 
of cultural capital and thus are relegated to less desirable positions in educational 
institutions (e.g., community colleges).   As a result they do not receive as a great a return 
on their educational investment.  While self-selection, overselection, and relegation are 
indirect routes of exclusion, direct exclusion may occur if low-income individuals are 
locked out of an opportunity by those of higher classes because they do not have the 
requisite capital to participate.  An example of this may be that if a student has not taken 
the appropriate high school courses, they may not be qualified to attend college.   
38
Findings on Cultural Capital
Parents, schools, and peers all transmit cultural capital to students.  Research has 
shown that cultural capital has an effect on college enrollment.  McDonough (1997), 
Gandara (2002), and Lareau (1987) all found that cultural capital, operationalized as 
parental involvement and support, positively influenced the educational progress of their 
children. McDonough?s qualitative study of high school girls supported the premise that 
the more desirable cultural capital, operationalized as support and expectations from 
parents, that a student possessed, the more likely she was to attend a selective college or 
university.  She found that parents were a key transmitter of cultural capital by 
communicating with their children about the value and process for attaining a college 
degree as well as informing them about the importance of that degree for future success.  
The student?s peer group also influenced this decision making process, as suggested by 
the finding that peers often applied to similar schools.  
Gandara (2002) also found parental support to be an important component of 
cultural capital that influenced college enrollment.  She stated, based on her review of 
literature, that an aspect of cultural capital surrounding a child?s education was parental 
knowledge of resources that may benefit one?s children (e.g., school curricula, financial 
aid information).  A parent?s knowledge of these resources can help promote college 
attendance.  Differences in types of cultural capital may also restrict the extent to which 
low-income parents engage in discussions about their child?s education.  She stated that 
low-income and minority parents often lack the cultural capital necessary in order to 
access the networks and resources that can benefit their student.  
39
Lareau (1987), in her qualitative study of parental participation in elementary 
schools, also found evidence of how cultural capital affected parents? relationships with 
their child?s school.  Her study, which involved participant-observation and interviews at
two first-grade classrooms in two community schools, one working-class and one middle-
class, explored the quantity and quality of parental involvement.  Her analyses show a 
difference in parental involvement in their student?s education between working class and 
middle class parents, that the difference in involvement reflected differences in cultural, 
social, and economic resources, and that the resources of the middle class were more 
highly valued by those at the school.  Parents at the middle-class schools spent more time 
volunteering in their child?s classroom and attending parent/teacher conferences than 
parents at the working class school.  Parents at middle-class schools activated their 
capital in order to participate in their child?s education.  Parents in the working class 
school also rarely contacted the school with academic concerns and when they did 
interact with school personnel they were very uncomfortable.  While all parents believed 
that their child?s education was extremely important, children of working class parents 
did not benefit in the same way due to their lower participation in the school.  In contrast 
to parents in the working class school, parents in the middle-class school were able to 
request additional educational resources, monitor teacher behavior, and consult with other 
parents and teachers about the educational experience of their child.  
DiMaggio and Mohr (1985) studied the effects of cultural capital on college 
attendance among students who graduated from high school in 1961, and were 
resurveyed ten years later.  Cultural capital was measured as high cultural attitudes, 
activities, and information (e.g., attending symphony concerts, having a cultivated self-
40
image, and literature reading).  They found that cultural capital was a significant positive 
predictor of enrolling in college even after controlling for background characteristics and 
ability.  
Perna and Titus (in press) included measures of cultural capital in their study of 
the relationship between social capital and college enrollment using the NELS.  Proxies 
of cultural capital included level of parent education, parental educational expectations, 
primary language spoken at home, and participation in cultural classes.  After controlling 
for background characteristics, economic capital, human capital, and social capital, level 
of parental education and parental expectations for the child?s educational attainment
were significant positive predictors of enrolling in a four-year college in the fall after 
graduating from high school. 
Social Capital
Social capital is another important component of sociological models of college 
access. Portes (1998), after his review of writings on social capital, defined social capital
as the ability of individuals to secure benefits through participation in a network or 
specific social structure.  Portes found that there are often contradictions in processes 
defined as social capital by different authors, but  concluded that there is a common 
growing understanding in the literature regarding a definition of social capital.  This 
section identifies the prominent social capital theorists, Lin, Coleman, and Bourdieu, and 
highlights the similarities and differences in their approaches. 
Coleman?s (1988) approach to social capital focuses on a person?s connection to a 
social structure such as a family, community, or school, and the information and 
resources that one can gain from participation in that structure.  More specifically, he 
41
focuses on the transmission of social capital both within and outside of the family.  It is 
from the structure of the relationships between and among actors that information is 
gained and then can be converted into social capital.  The connections make a particular 
achievement, such as enrollment in college, possible where it would not have been 
without the resources gained from participation in the network.  Actors with resources or 
knowledge facilitate the transfer of information that others can use to their future benefit
to produce different system-level behavior or individual outcomes.   Social capital allows 
an individual to invest in human and other forms of capital, and as a result there will most 
likely be an increase in human capital for the future generations of the person who
originally accessed social capital.
Coleman (1988) discussed that all social relations and social structures promote 
the transmission of some form of social capital.  He believed that a closed network, 
where all actors are connected in some way, can create strong norms within a social 
structure and can be a powerful form of social capital.  This closure to the social structure 
motivates trustworthiness that in turn creates obligations and expectations as well as the 
ability to have effective norms within the group.  
Coleman (1988) argued that social capital within a family is a benefit for a child?s 
development.  This social capital comes from time and energy that parents focus on their 
children to help their development.  The attention from, and physical presence of, the 
parent facilitates the passing on of the adult?s human capital.  McNeal (1999) used the 
NELS to study the effect of parental involvement as social capital on science 
achievement, truancy, and dropping out of high school.  He found that parental-student 
42
involvement, parental involvement at the school, and parental monitoring of the child?s 
behavior were all significant positive predictors of science achievement and truancy.  
Despite the benefits, some parents may not have access to the social capital
through involvement in the education that sup ports their children.  Coleman (1988) and 
others (e.g., Croninger & Lee, 2001) argue that social capital is not gained solely from 
the family but also from the school and the community.  Research supports the 
hypothesis that schools and communities are sources of social capital.  In both the 
schools and the community, a student can be a part of a network and in turn receive 
information that can be turned into social capital.  Coleman found evidence of network 
promotion of social capital by schools in his study of high school drop-out rates.  Using 
High School and Beyond data he found that students who attended Catholic high schools 
had lower drop out rates, a finding he attributed to the closed network of the school.  This 
network had stronger ties and in turn fostered greater community and family influence on 
students? high school attendance decisions. 
Using the NELS, Croninger and Lee (2001) found that social capital outside of 
the family reduced the likelihood of dropping out of high school between 10th and 12th 
grade after controlling for gender, academic achievement and academic behaviors.  Their 
measures of social capital included students' beliefs about how much their 10th grade 
teachers supported their efforts to succeed in school and teacher reports about whether 
individual 10th grade students received guidance from teachers about school and personal 
issues.  Results of the study showed that dropouts had less of the two forms of social 
capital than high school graduates, and that these two measures of social capital increased
43
the probability that a student completed high school after controlling for background 
characteristics, social risk factors, and academic background.   
There are multiple criticisms to the work of Coleman. Morrow (1999) stated that 
Coleman?s work does not consider the socioeconomic history of students and is gender-
blind.  She wrote that students who come from disadvantaged backgrounds may not have 
access to rewards for school achievement, and in turn may act reasonably based on their 
circumstances by not continuing with school.  She also posited that Coleman did not take 
into account the differences in social networks between men and women, only noting that 
women?s employment negatively impacted their children and the community.  
Stanton-Salazar (1997) also disagreed with Coleman?s notion that institutional 
agents such as teachers, counselors, and school peers can transmit resources and 
opportunities to disadvantaged students if they are given the necessary time and funding.  
He posited that institutional agents are often unable to transmit resources in the schools of 
minority students because of structural characteristics (e.g., high teacher/student ratio, 
fewer college preparation courses and programs).  He found, through his review and 
analysis of literature, that low-status children and youth tended to accumulate low 
amounts of social capital because of a lack of institutional support.  This lack of support 
made students? participation in school more challenging due to their perceived value of 
not being able to achieve as much based on class, ethnicity, and gender.  
 Stanton-Salazar (1997) defined social capital as ?the degree and quality of 
middle-class forms of social support inherent in a young person?s interpersonal network? 
(p. 5).  This definition supported the notion that some types of social capital are more 
valued by society than others.  Stanton-Salazar found that sources of valued social 
44
support were most likely to include extended family, school, peers and other community 
organizations.
Lin (2001a) and Burt (2001) disagreed with Coleman?s need for a closed social 
structure.  Lin (2001a), based on the work of Granovetter, argued that closed social 
structures were too narrowly focused, and believed in the advantages of more open 
networks with many connections.  She argued that open networks gave greater
opportunity for mobility between groups which was especially beneficial for those from 
lower socioeconomic backgrounds.  Lin also argued that closed networks are more 
beneficial for members of upper classes, individuals who already have access to 
resources within their network.   
Burt (2001) also criticized Coleman?s emphasis on closed networks for several 
reasons.  First, he posited that not dropping out of school was a weak performance 
criterion for estimating the effects of social capital.  Second, he stated that it is difficult to 
ascertain if it was the social capital of the parents or the students that affected dropping 
out of school.  Burt argued, using a review of previous research, that open networks with 
the existence of structural holes were preferred to closed networks.  Structural holes are 
created when individuals stay to themselves and only focus on the activities of their own 
network. Social capital is accessed through the brokerage of non-connected segments.
Brokering connections between groups is one way to add value to a network, by 
enlarging the group and as a result having a greater number of resources to share 
information.  Individuals who benefit the most from holes are those individuals who are 
able to bridge holes in the network, such as a low-income student who is able to 
45
participate in a college preparation program that results in access to resources that he or 
she would not normally receive in their school setting.  
In a different conceptualization, Bourdieu (1986) described social capital as 
resources, actual or potential, that are accessible due to membership as a part of a group.  
The amount of social capital possessed by an individual depends on the size and strength 
of the resources that the members of the network possess.  The purpose of this network of 
relationships is to continue to reproduce social relationships that may be needed 
immediately or in the future to gain information or resources (Bourdieu, 1986).  
Lin (2001a; 2001b) has closely aligned herself with the work of Bourdieu.  Lin 
(2001b) described social capital theory as how ?resources are embedded in one?s social 
network and how access to and use of such resources benefit the individual?s actions? (p. 
55).  These embedded resources produce profits in the form of information that flows 
easily to members of the network, and that can be used to their advantage.  
To better illustrate the concept of social capital, Lin (2001a) developed a model of 
a theory of social capital.  This model is made up of three stages, inequality, 
capitalization, and effects, and centers around access to the quality and quantity of 
embedded resources.  The first block of the model, inequality, looks at the precursors of 
social capital, such as access to technology, group participation, and the economy, as well 
as resources that are available to the individual.  Also in this block is an individual?s 
position in a social structure or community which can help or hinder an individual?s 
ability to invest in social capital.  The expectation is that the more available that resources 
are, the more likely that the resources will be mobilized.  
46
The second block in Lin?s model of a theory of social capital is the capitalization 
of resources.  This block examines where the resources are located and if they are 
accessible to an individual.  If the resources are available then the individual can choose 
to utilize them.  Inequality is again highlighted at this point in the model because those 
with lower collective assets are less likely to be able to mobilize capital (Lin, 2001a).  
The third block of the model, or the outcome, is the return after the mobilization 
of resources.  Lin (2001a) described two types of outcomes of social capital, instrumental 
and expressive.  An instrumental outcome occurs when a person gains additional 
resources not previously possessed and uses these resources to make progress. A low-
socioeconomic status high school student who enrolls in college as a result of gaining 
resources about enrolling in college from his/her teacher or another adult would be an 
instrumental outcome, because that student is able to make gains that family or 
community members before him or her were not able to achieve.  This idea has been 
criticized by Stanton-Salazar (1997) who argued that structural barriers limit the transfer 
of such resources from teachers to low-income students.  
Lin (2001a) defined expressive action as one which a person does not noticeably 
gain, but maintains the resources for their class or group.  An example of this would be 
the college enrollment of a student from an upper socioeconomic status background since
going to college is a norm for members of the upper socioeconomic status network.  Thus 
there is not a noticeable gain in the outcome of that student in comparison to others in the 
individual?s family or community.  
47
Findings on Social Capital
Some research examines the role of social capital on college enrollment.  
Researchers (Cabrera & La Nasa, 2001; Perna, 2000; Perna & Titus, in press; Plank & 
Jordan, 2001) have utilized such measures of social capital in their studies of college 
enrollment as parental involvement in their child?s education, assistance from school 
personnel, and amount of information received about financial aid.  These researchers all 
found that social capital was a positive predictor of college enrollment.  
Perna and Titus (in press) utilized data from the NELS to explore the  relationship 
between social capital, defined as parental involvement, and college enrollment among 
1992 high school graduates.  Their study found that parent involvement, measured by the 
frequency of discussing education-related issues with their student, the frequency of 
contacting the school about volunteering, and the frequency of contacting the school 
about academic matters, was a positive predictor of enrolling in a four-year college in the 
fall after graduating from high school after controlling for measures of economic capital, 
human capital, and cultural capital.   
Perna (2000) also included measures of social capital in her study of college 
enrollment among different racial/ethnic groups.  Using measures of parental 
involvement and high school support in her analyses of NELS data, she found that 
parental involvement and encouragement from school personnel were significant positive 
predictors of enrolling in a four-year college in the fall after high school for 1992 high 
school graduates even after controlling for background characteristics, costs and benefits, 
and ability.    
48
Cabrera and LaNasa (2001) included social capital in their research describing the 
path to college of low-socioeconomic status students.  Cabrera and LaNasa measured this 
sociological concept using indicators of parental involvement in their student?s education, 
information sources on financial aid, and high-school based support.  Using data from the 
NELS, they found that all three measures were significant positive predictors of applying 
to college for high school students in all socioeconomic status groups after controlling for 
background characteristics, academic ability, and personal educational aspirations.
Also using the NELS, Plank and Jordan (2001) included measures of social 
capital in their study of the effects of information, guidance, and actions on 
postsecondary destinations.  To test the impact of social capital on their four category 
dependent variable of college enrollment (enrollment in a four-year institution, full-time 
enrollment at a two-year institution, part-time enrollment at a two-year institution, or no 
enrollment) within two years of graduating from high school, they utilized multiple
variables as proxies of social capital: parental involvement with the student and the 
school, SAT test preparation and completion, information on financial aid resources, and 
support from school personnel.  Parent-student involvement, test taking, financial aid 
information, and school guidance were all positive predictors of both enrolling in a four-
year institution and enrolling at a two-year institution full- or part-time rather than not 
enrolling, even after controlling for background factors and academic achievement.  
Summary of Sociological Approaches
Sociological approaches to college enrollment focus on the effects of 
socioeconomic characteristics of an individual?s college enrollment.  Components of 
49
sociological models put forth by Hearn (1991) and others examined family background 
characteristics, including family income, family size, parents? education, and race.  In
qualitative studies, McDonough (1997) and Horvat (2001) illustrated the contribution of 
the sociological constructs of cultural capital and habitus to understanding college 
enrollment.  Such quantitative analyses as Cabrera and LaNasa (2001), Perna (2000), 
Perna and Titus (in press), and Plank and Jordan (2001) illustrated the contribution of 
social capital to college enrollment.  
Despite the contribution of this approach, sociological models lack the attention to 
the costs and benefits of enrollment, a focus of the econometric models.  Components of 
both econometric and sociological models inform our understanding of college 
enrollment behaviors, suggesting that a combi nation of these models is most effective. 
Combined Approaches
Among recent work are studies (Cabrera & La Nasa, 2001; Hossler & Gallagher, 
1987; Perna, 2000; Plank & Jordan, 2001; St. John, 2003) that draw on econometric and 
sociological models as well as recent qualitative studies to explain college enrollment.
Cabrera and LaNasa (2001), Perna (2000), and St. John (2003) include measures of 
academic achievement and college costs, which are parts of econometric models, but also 
include other forms of capital such as social capital and cultural capital to better measure 
the role of preferences, tastes, and values for higher education, factors that are 
acknowledged but generally not measured in traditional econometric models.   Cabrera 
and LaNasa (2001), Plank and Jordan (2001), and St. John (2003) utilized the combined 
approach to further study the college enrollment of low-socioeconomic status students.
50
Perna (2000), in her study of differences in college enrollment decisions based 
on race/ethnicity, tested a conceptual model that expanded a traditional econometric 
model of college enrollment by including measures of cultural capital and social capital.  
Her model hypothesized that the decision to enroll in college was based on the cost of 
college, perceived future benefits, family income, academic ability, and social and 
cultural capital.  Obscuring the difference between the two types of capital, Perna 
measured social and cultural capital by personal and parental educational aspirations, 
parental involvement, encouragement from peers and school personnel, assistance with 
college requirements, and use of test preparation tools.
Perna (2000) tested her model utilizing data from the NELS.  She found that 
family income was a positive predictor of enrolling in a four-year college within two 
years of high school graduation, both in the overall sample and when each of three 
racial/ethnic groups (Black, Hispanic, Asian American) was examined individually, after 
controlling for measures of direct costs and expected benefits of college, and academic 
ability.  When social capital and cultural capital were also controlled, income was no 
longer a significant predictor of enrollment for African American or Hispanic students, 
but was still a significant predictor of college enrollment for White students.  This finding 
suggests that high school graduates with lower family income are less likely to have the 
types of social and cultural capital that promote college enrollment.    Perna found that 
social capital and cultural capital impact the college enrollment decisions for all three 
racial/ethnic groups, but play a relatively bigger role in the decision to enroll for African 
Americans and Hispanics than Whites.   
51
Combined Models for Low-Socioeconomic Status Students
Combined models of college enrollment are especially beneficial when exploring 
the enrollment patterns of low-socioeconomic status students.  Integrating aspects of 
sociological approaches with econometric approaches helps to present a more holistic 
model of the student?s experience that informs their college enrollment choice.  Cabrera 
and LaNasa (2001) incorporated elements of both an econometric and a sociological 
approach in their study of low-socioeconomic status student enrollment.  Utilizing data 
from the NELS, their model included measures of socioeconomic status, academic 
preparation, personal educational aspirations, and components of social and cultural 
capital such as high school support, parental involvement, and parental expectations.  
Their model assumed that in order for a student to enroll in college, he/she needed to first 
be academically qualified to attend, then graduate from high school, and then actually 
apply to a four-year institution.  
Cabrera and LaNasa (2001) found a gap in college enrollment based on 
socioeconomic status, with lower percentages of low socioeconomic status students than 
high socioeconomic status students enrolling in college within two years of graduating 
from high school after controlling for other variables. They found variables that 
encouraged students to enroll in college included: having parents involved in their 
student?s education, prior academic performance, resources available to students about 
the college process and personal aspirations.  While application rates were similar for 
low-socioeconomic status students and higher socioeconomic status students after 
controlling for other variables, college attendance rates continued to be lower for students 
from low-socioeconomic status families than for students from high socioeconomic 
52
families net of other variables.  Cabrera and LaNasa hypothesized that this unexplained 
gap may be attributed to differences between low- and high-socioeconomic status 
students in ability to pay for college, quality of information about the college process, and 
other components of cultural capital that may not have been adequately measured in the 
study.  
St. John (2003) developed the balanced access model after seeing an inequality in 
college access over the last twenty years that he believed was related to more than just 
academic preparation.  He used this combined model to focus on the link between family 
finances and college enrollment and to examine his hypothesis that financial, academic 
and social factors impact college enrollment.  The academic and social factors that St. 
John (2003) included in his model were similar to those in other models (e.g., Cabrera & 
La Nasa, 2001; Perna, 2000) already described.  Social factors included  family income 
and background, and student plans and aspirations, while academic factors included 
becoming academically prepared, taking college entrance exams, and applying and being 
admitted to college.  He identified two types of financial factors, both the perceived 
unmet need as well as the actual unmet need, where unmet need is defined as the 
difference between college costs and financial aid.  St. John (2003) argued that this 
combination of academic readiness and ability to pay, coupled with the influence of 
financial aid, impacts enrollment in college.  
St. John (2003) tested the balanced access model using NELS data and found that 
financial aid was central to promoting college access since college cost and the ability to
pay influenced college enrollment.  He found that the correlation between academic 
preparation and enrollment was stronger for high- income students than for low-income 
53
students.  He concluded that, in order for a student to access college, the student must 
have the appropriate academic qualifications as well as the ability to pay for initial and 
continuous enrollment in college, either on their own or with governmental or 
institutional support.  
Plank and Jordan (2001) also tested a combined model of college enrollment 
using data from the NELS that measured the effects of information, guidance, and actions 
on college enrollment within two years of graduating from high school of low-
socioeconomic status students.  Their conceptual model, which utilized components of 
models developed by Hanson, Litten, and Hossler and Gallagher, included background 
characteristics, high school influences, media, social capital, college characteristics (e.g.,
costs) and public policy (e.g., financial aid).  Their four-category dependent variable 
measured enrollment in a four-year institution, full-time enrollment at a two-year 
institution, part-time enrollment at a two-year institution, or no enrollment.  They found 
through their multivariate analyses that low-socioeconomic status students were 
significantly less likely to enroll in a two-or four year college within two years of high 
school graduation than students of higher socioeconomic status.  They also found that 
ability and social capital (measured by parental involvement, test preparation, and
information about financial aid) were significantly positive predictors of enrollment in 
college.
In summary, combined models of college enrollment include components from 
both econometric and sociological approaches.  Research using this approach shows that 
immediate college enrollment is impacted by a student?s academic preparation, college 
costs, and amount and type of cultural and social capital (Cabrera & La Nasa, 2001; 
54
Perna, 2000; Plank & Jordan, 2001; St. John, 2003).  Because of these strengths, this 
study also draws on a combined model of college enrollment. 
Summary
A review and synthesis of prior research demonstrates a lack of research on 
delayed enrollment, with only one study published since 1985 (i.e., Hearn, 1992)
examining this topic.  Hearn (1992) explored background characteristics, academic 
achievement, and aspirations as predictors of delaying college enrollment.  He found that 
1980 high school graduates who delayed enrollment in college for one to two years after 
graduating from high school were more likely than those who attended college within one 
year to be male and to have lower aspirations and achievement.  While Hearn sheds light 
on the decision to delay college enrollment, more information about this behavior is 
needed using a more recent cohort of high school students, a more comprehensive 
conceptual model, and a longer time period to measure ?delay.?
      Research focusing on immediate college enrollment draws on econometric, 
sociological, and combined econometric and sociological models.  Econometric 
approaches focus on the academic preparation of students along with the costs and 
benefits of attending college.  This approach, utilized by Manski and Wise (1983) and 
Kane (1994) showed that academic preparation, tuition and aid, and family background 
impacts the decision to enroll in college within four and six years of high school 
graduation.  These models do not take into account the lack of information that a student 
may have about college and its benefits or other sociological concepts.  Sociological 
approaches, such as those utilized by Hearn (1991) and McDonough (1997), explore how 
55
socioeconomic characteristics and concepts such as social capital, cultural capital, and 
habitus impact a student?s college enrollment decision, but do not take into account the 
components of the cost/benefit analysis found in the econometric models.  Combined 
models, such as the work of Cabrera and LaNasa (2001), Perna (2000), and St. John 
(2003), draw on econometric and sociological approaches to test a more complete model 
and find that measures of constructs such as background characteristics, academic 
preparation, finances, social capital, and cultural capital all play a role in a student?s 
decision to enroll in college immediately after graduating from high school.  
Building on prior research, this study utilizes a combined model to examine the 
decision of students to enroll in college within eight years of graduating from high 
school.  This model includes measures of background and achievement explored by 
Hearn (1992), as well as college costs, social capital, and cultural capital.  This study 
tests the applicability of the model using a more recent cohort of students (i.e., 1992 high 
school graduates).  While Hearn (1992) defined ?delayed enrollment? as enrollment 
within one to two years of high school graduation, this study uses a broader definition of 
delay, up to eight years out of high school.  This study also explores the relationship 
between socioeconomic status and college enrollment since research consistently shows a 
relationship between enrollment and socioeconomic status even after controlling for other 
variables.  The following chapter explains the components of the conceptual model in 
greater detail as well as the methodology for the study.
56
CHAPTER 3
RESEARCH DESIGN AND METHODOLOGY
Introduction
This study builds on the work of Hearn (1992) by looking at a more recent cohort
of high school graduates, utilizing a more complete conceptual model, and expanding the 
time after high school when a graduate may decide to enroll in college.  The appropriate 
definition of delayed enrollment is generated by examining the characteristics of students 
who enroll in college each year after graduating from high school.  Using a conceptual 
model that is based on what is known about traditional student enrollment, this study 
utilizes multinomial logit regression to examine the variables that differentiate a student?s 
decision to enroll in a two- or four-year college immediately after high school graduation, 
delay first college enrollment, and not enroll in college within eight years of graduating 
from high school.  The study also explores the relationship between socioeconomic status 
and timing of college enrollment to determine if the socioeconomic status gap that is 
present in immediate enrollment is also present in delayed enrollment.  
Specifically, this study addresses the following five research questions:  
1. What percent of high school graduates enroll in college in each of eight years 
after graduation from high school?  What is the appropriate break point 
between immediate and delayed enrollment?
2. How do the characteristics of high school graduates who delay entry into a 
two- or four- year college degree program compare to the characteristics of 
high school graduates who enroll immediately after high school and the 
57
characteristics of high school graduates who do not enroll within eight years 
of high school graduation?
3. For high school graduates who delay enrollment, how does the timing of first 
enrollment vary by student characteristics?
4. What are the predictors of delaying entry into a two-  or four-year college 
degree program for up to eight years rather than enrolling immediately after 
high school or not enrolling within eight years?
5. After controlling for other variables, is socioeconomic status related to the 
decision to delay enrollment in college?
The data to be analyzed for this research study are from the National Education 
Longitudinal Study: 1988-2000 (NELS).  This chapter describes the dataset and the 
sample, the statistical analyses of the data, the dependent and independent variables in the 
analyses, and the limitations and delimitations of the study.  
Data
This study uses data from the second (1992), third (1994) and fourth (2000) 
follow-up to the National Education Longitudinal Study of 1988 (NELS) eighth graders.  
The NELS is a product of the National Center for Education Statistics  (NCES), which is a 
part of the United States Department of Education.  The purpose of the NELS was to 
track the transitions that students experienced over time from middle school through their 
entrance into the workforce and/or postsecondary education.  NELS includes a base year 
survey that was conducted with students who were in the 8
th
 grade in 1988.  Researchers 
attempted to follow-up with students in the initial sample four times, in 1990, 1992, 1994, 
58
and 2000, eight years after most of the students had graduated from high school (Curtin et 
al., 2002).  
NCES selected participants for the initial base year data collection period utilizing 
a two-stage stratified sample design.  Approximately 39,000 schools nationally had 
eighth grade classes, and researchers first used stratified sampling to select schools to 
participate.  To ensure adequate representation, private schools were oversampled.  
Schools were stratified first by school type and geographic region.  After that, schools 
were further stratified by urban, rural, and suburban location, as well as minority 
classification of the school, and then sorted by school size.  Schools were then randomly 
selected from within the stratified groups in order to have adequate representation of all 
types of schools.  From the population, 1,734 schools were selected, and 1,057 schools 
participated in the base year data collection (817 public schools, 104 Catholic schools, 
and 136 private schools) (Curtin et al., 2002).  
After schools were selected, approximately 26 eighth grade students were selected 
from each school to be a part of the sample.  Students were excluded from the sample if 
the school did not deem them eligible to participate due to a disability or language 
barrier.  Asian and Hispanic students were oversampled to make sure that there were 
adequate numbers of students of these groups in the sample.  In total, 24,599 eighth grade 
students were selected to participate in the 1988 base year survey (Curtin et al., 2002).  
The students who were part of the base year survey were resurveyed in each of 
four follow-up data collection periods.  Before each follow-up, NCES contractors tracked 
the location of participants and sent letters letting them know that they would be 
contacted and asked for their assistance in updating contact information.  Most 
59
participants completed the third and fourth survey over the phone with the NCES 
contractors using computer-assisted survey technology.  For individuals who were unable 
to participate in the survey over the phone, NCES contractors held in-person interviews 
using a similar computer assisted technology system.  High school and postsecondary 
transcripts were also collected as a part of the fourth collection procedure in 2000 (Curtin 
et al., 2002).  
Since not all students were located for subsequent follow-up surveys, the sample 
was freshened for the first (1990) and second (1992) follow-ups.  The third (1994) and 
fourth (2000) follow-ups utilized current participants (i.e., base year and freshened) with 
no additional freshening.  The researchers attempted to contact 15,257 participants for the 
fourth follow-up, with 77.6 percent responding, for a total of 11,914 cases (Curtin et al., 
2002). 
The analytic sample is limited to participants who were part of the second (1992), 
third (1994), and fourth (2000) waves of data collection, and who graduated from high 
school on schedule in spring 1992.  The sample is limited to high school graduates since 
a high school diploma is a requirement for college enrollment and I also wanted to have a 
consistent measure of high school completion.  I weighted the analytic sample using the 
normalized panel weight, before conducting the analysis.  Since this study explores 
students in the second through fourth follow-ups, the F4F2PNWT is used (Curtin et al., 
2002).  Weighting acounts for nonrandom selection, sampling error, and nonresponse.  
Normalizing the weight, which equates the number of cases to the actual size of the 
sample, reduces the impact of large sample sizes on statistical tests.
60
Design effects of the sample design are taken into account using a rigorous 
threshold of statistical significance for all statistical tests (p<.001). Specifically, the 
NELS sample design includes the clustering of students within schools.  If design effects 
are not taken into account there may be an increased likelihood of committing a Type I 
error (Thomas & Heck, 2001).   
The NELS is an appropriate dataset to utilize for my research for several reasons.  
First, the dataset is longitudinal in nature with a great deal of information available 
through the large number of variables that have been collected.  I can utilize data from 
different points in time (e.g., grades, parental involvement, and other experiences during 
high school enrollment) to help discern differences in the predictors of the timing of 
enrollment in a two-  or four-year college degree program.  Second, the timing of the 4
th
follow-up, eight years after high school graduation, allows an examination not only of
immediate college enrollment (i.e., enrollment right after high school) as in most college 
access studies, but also of delayed college enrollment, an outcome less commonly 
examined.  Third, the NELS is also an appropriate dataset because the sample is large in 
size (n=11,914) of those individuals who completed all three waves of the study that I am 
using.  
The fourth advantage to using the NELS is its  high response rate.  Response rates 
for the second and third follow-ups exceeded 90%.  There was a 92.5% completion rate 
for the second follow-up (1992), and a 94.0% completion rate for the third follow-up 
(1994) with 14,915 cases.   There are 12,144 cases for the fourth follow-up (2000), for a 
completion rate of 77.6% (Curtin et al., 2002).  When weighted the data are 
representative of the national population of 1992 high school graduates in 2000.  
61
Instrumentation
The NELS data were collected by NCES researchers utilizing multiple methods in 
1992 (second follow up), 1994 (third follow-up), and 2000 (fourth follow-up).  For the 
base-year survey through the second follow-up survey, student surveys and achievement 
tests were administered through the schools.  NELS project staff administered t hese paper 
and pencil instruments to students.  Teachers, principals, and parents also completed 
paper and pencil surveys (Curtin et al., 2002).  
In the second (1992) follow-up, students completed two separate types of 
instruments.  First students completed a 60-minute questionnaire asking about their
background, self-perceptions, aspirations, school experiences, and home environment.  
Students were also given a battery of cognitive tests on subjects including reading, math, 
science, and social studies.  The school administrator survey was between 40 and 60 
minutes in length and focused on school characteristics and environment.  The teacher 
instruments asked teachers to answer about specific students in relation to their school 
environment (Curtin et al., 2002).
The third follow-up was designed to capture the experiences of the cohort two 
years out of high school.  The questions focused on academic achievement, 
postsecondary access, work experiences, perceptions about school and job, and family 
structure and environment.  The third follow-up instrument, administered by computer 
assisted phone interviews, had approximately 64 items with some questions including 
sub-items (Curtin et al., 2002).  
62
The fourth follow-up surveyed students who were on average eight years out of 
high school.  Participants were asked a battery of questions in 10 different areas including 
primary activities, such as current education and employment status, educational 
completion, employment and training, current income and expenditures, as well as an 
update of other characteristics such as marital status, number of dependents, and 
community integration (Curtin et al., 2002).     
Statistical Analysis
I use descriptive analyses such as chi-square tests, and ANOVA to address the 
first three research questions.  The first question is: What percent of high school 
graduates enroll in college in each of eight years after graduation from high school?  
What is the appropriate break point between immediate and delayed enrollment?  The 
second question is: How do the characteristics of high school graduates who delay entry 
into a two- or four-year college degree program compare to the characteristics of high 
school graduates who enroll immediately after high school and the characteristics of high 
school graduates who do not enroll within eight years of high school graduation. The 
third question is: For high school graduates who delay enrollment, how does timing of 
first enrollment vary by student characteristics?  These analyses describe the 
characteristics of students in the sample and examine observed college enrollment rates.  
Chi-square tests are used to describe categorical independent variables.  ANOVA is used 
to examine differences in continuous independent variables across the three categories of 
the dependent variable: immediate enrollment in college, delayed enrollment in college, 
and no enrollment in college. 
63
The appropriate multivariate technique to address the fourth and fifth research 
questions is multinomial logit regression, because the dependent variable, enrollment by 
2000, has three categories.  This method estimates the log-odds of one outcome relative 
to a baseline category, which in this case is delayed college enrollment after high school.  
The logit coefficients that result from the analyses may be interpreted as the change in 
log odds associated with a one-unit change in the independent variable (SPSS, 1999). 
Odds-ratios are also utilized to show the change in odds of choosing a particular 
college enrollment status over the baseline category, delayed enrollment, that are
associated with a one-unit change in each independent, or explanatory, variable.  An odds 
ratio greater than one represents an increase in the likelihood of immediate enrollment or 
no enrollment, whereas an odds-ratio of less than one represents a decrease in the 
likelihood of immediate enrollment or no enrollment (Perna, 2001).    
The variables are entered into the analyses in five conceptually related blocks.  
The first block includes background characteristics, while the other blocks are financial 
resources, academic preparation and achievement, social capital, and cultural capital.  
Adding the blocks in this order shows how the addition of different groups of variables 
impacts the observed relationship between socioeconomic status and enrollment timing.
Conceptual Model and Variable Description
Using a conceptual model of traditional student enrollment, this study compares
predictors of delaying enrollment in college after high school with predictors of enrolling 
immediately and not enrolling.  A review of the literature on college enrollment suggests 
that an individual?s background characteristics, financial resources, academic preparation 
64
and achievement, social capital, and cultural capital influence a student?s decision to 
enroll in a two- or four-year college degree program.  Figure 2 shows the variables that 
measure these components of the conceptual model.
Figure 2. Variables in Conceptual Model of Timing of College Enrollment
Timing of 
enrollment in a 2-
or 4- year college 
degree program
Background Characteristics
*Socioeconomic Status 
*Race/Ethnicity
*Gender
Social Capital
*Number of financial aid 
contacts
*Parental involvement in 
student?s education
*Closed Networks
*Student-teacher relations
*HS based support
*HS control
*School participation in free 
and reduced lunch program
Cultural Capital
*Parental expectations
*Peer encouragement
*Parental involvement with 
school
*Educational materials in the 
home
* Participation art, music, or 
dance classes
Financial Resources
*Importance of college costs and 
financial aid
*Tuition
*Unemployment Rate
Academic Preparation and 
Achievement
*High school achievement
*Coursework
65
Most of the independent variables are measured using data from the 1992 follow-
up survey when students were seniors in high school.  Background characteristics are 
measured by socioeconomic status, race, and gender.  Financial resources are measured 
by the perceived importance of college costs and financial aid, as well as tuition and state 
unemployment rate.  Academic preparation is measured by high school achievement and 
coursework.  Social capital is measured by proxies for information about financial aid,
closed networks, parental involvement with the student?s education, teacher-student 
relations, high school based support, high school control, and school participation in free 
and reduced lunch program.  Cultural capital is measured by parental expectations, peer 
encouragement, parental involvement with the school, amount of educational materials in 
the home, and participation in art, music, or dance classes.  
Enrollment
The dependent variable has three categories: enroll immediately after graduating 
from high school, delay entry for up to eight years after graduating from high school, or 
not enroll within eight years of high school graduation.  Enrollment is defined as 
enrollment in either a two-or four-year college or university.  The variable is constructed 
based on the month and year of first attendance.  I determine the cut-off point between 
immediate enrollment and delayed enrollment, based on the descriptive analyses used to 
address the first research question.  I delayed this decision because some researchers have 
defined immediate enrollment as within two years of graduation from high school 
(Cabrera & La Nasa, 2001), or one year of high school graduation (Cabrera & La Nasa, 
2001; Hearn, 1992), while others have defined immediate enrollment as enrollment by
66
the fall after graduating from high school (Perna, 2000).  No prior study has 
systematically examined the most appropriate definition of ?delayed enrollment.?
Background Characteristics
Three variables measure background characteristics: socioeconomic status, 
race/ethnicity, and gender.
Socioeconomic Status
Socioeconomic status in the 12
th
 grade is a continuous, composite measure 
created by NCES based on the educational levels of the mother and the father, the 
occupational status of the mother and the father, and the family income in 1987 (Curtin et 
al., 2002).  I am using the standardized measure of socioeconomic status. Table 3.1
shows the relationship between this composite and three of its components: father?s 
educational attainment, mother?s educational attainment, and total family income in 
1987. A clear pattern emerges between socioeconomic status quartile, income, and 
education, even though low-socioeconomic status does not necessarily equal low-income.  
Eighty-eight percent of individuals in the lowest socioeconomic status quartile have a 
father with no more than a high school education, while 77% of those student?s in the 
highest socioeconomic status quartile have a father with at least some graduate education.  
The mother?s highest level of education also increased by socioeconomic status quartile.  
The mothers of high school graduates in the first socioeconomic status quartile primarily 
had a high school degree or less (89%), while 62% of mothers in the highest quartile had 
at least a graduate degree.  As for family income, over 50% of high school graduates in 
67
the lowest quartile had a total family income of less than $15,000, while 62% of those in 
the highest quartile made at least $50,000.
Table 3.1.  Percentage of socioeconomic status composite components by socioeconomic 
status quartile
Socioeconomic Status
Variable Quartile 
1 (Low)
Quartile 
2
Quartile 
3
Quartile 
4 (High)
Total
Father?s Education*** 100.0 100.0 100.0 100.0 100.0
Not finish HS 46.8 16.2 4.8 1.3 12.7
HS Graduate 41.3 53.6 37.4 7.3 31.6
Junior College 5.9 15.2 21.0 6.9 12.5
College LT 4 Yrs 2.7 7.5 14.5 8.0 8.8
Graduate College 2.6 6.2 16.3 34.7 18.1
Master?s Degree 0.5 0.9 4.7 25.1 10.1
Ph.D, M.D., etc. 0.2 0.5 1.2 16.7 6.1
Mother?s Education*** 100.0 100.0 100.0 100.0 100.0
Not finish HS 42.9 12.3 4.5 0.7 11.2
HS Graduate 46.2 58.6 44.2 13.7 38.1
Junior College 5.3 15.2 19.7 10.7 13.4
College LT 4 Yrs 3.7 7.2 14.3 12.6 10.3
Graduate College 1.5 5.2 12.5 34.9 16.3
Master?s Degree 0.1 0.8 3.5 22.3 8.5
Ph.D, M.D., etc. 0.2 0.6 1.2 5.1 2.2
Total Family Income in 1987*** 100.0 100.0 100.0 100.0 100.0
Less than $10,000 32.2 5.6 0.9 0.1 6.8
$10,000-$24,999 47.9 38.6 18.3 3.4 23.1
$25,000-$49,999 19.3 49.5 61.5 34.9 43.3
$50,000 or more 0.6 6.3 19.4 61.6 26.9
Source: Analyses of NELS:92/2000
***p<.001
Note: Data are weighted by normalized F4F2 panel weight
The use of socioeconomic status rather than separate measures of family income 
and parental education in examination of college enrollment has been criticized by some 
researchers (e.g., Paulsen & St. John, 2002).  Paulsen and St. John argue that 
socioeconomic status alone is not a complex enough measure to cover social class 
adequately.  In a differing opinion, Adelman (2002) recommended the use of 
68
socioeconomic status over parent income with NCES data that were collected before 
1996 because family income is self-reported by students and thus it is difficult to know if 
the data are reliable.  Also, since the NELS does not account for family size, it is dif ficult 
to correctly define ?low-income.?  He also recommended using socioeconomic status 
because, as a composite measure with multiple components, the measure is more reliable
than any of its components alone, and because a smaller number of cases are missing for 
socioeconomic status components than for the family income variable. Cabrera and 
LaNasa (2001) and Perna and Titus (2004) make similar arguments to justify the use of 
socioeconomic status in their work.  Socioeconomic status is an appropriate measure for 
this study, especially since the conceptual model for this study includes other measures of 
social and cultural capital.  
Cabrera and LaNasa (2001) utilized the 1988 version of the socioeconomic status 
composite variable in their examination of college enrollment for low-socioeconomic 
status traditional-age students.  Cabrera and LaNasa found differences in rates of college 
enrollment based on socioeconomic status, with low-socioeconomic status students less 
likely to enroll in a four- year college within one year of high school graduation after 
controlling for academic preparation, parental and individual aspirations, and financial 
aid information.  Using data from the High School and Beyond (HS&B) survey, a 
precursor to the NELS, Hearn (1992) utilized a similar measure, parental socioeconomic 
status, in his study of nontraditional college enrollment, and found that low -
socioeconomic status students were more likely than upper socioeconomic status students 
to ?delay? enrollment (i.e., enroll between one and two years after high school graduation
69
rather than enroll within one year) after controlling for background and academic 
characteristics such as high school academic track, high school grades, and test scores.
Race/Ethnicity
The four categories of race/ethnicity are White, Hispanic, Black, and Asian 
American.  White is the reference group.  Native Americans/Alaska Natives are excluded 
from the analyses due to their small sample size (n=78), which is 0.9% of the unweighted 
sample.
Hurtado et. al. (1997) found differences in college application behaviors, while 
Perna (2000) and Cabrera and LaNasa (2001) found some differences in the probability 
of college enrollment based on race/ethnicity after controlling for other variables.  Hearn 
(1992), when comparing delayed and immediate attendance, found no significant 
difference in the enrollment of Blacks and Whites after controlling for other variables.  In 
contrast, Perna (2000) found, after controlling for measures of social and cultural capital, 
family income, and academic achievement, that Blacks were 11% more likely than 
Whites to enroll in a four-year college within the semester after graduating from high 
school.  She also found that there was no difference between enrollment rates of 
Hispanics and Whites after controlling for other variables.  Other research (e.g., Cabrera 
& La Nasa, 2001) also shows higher enrollment rates for Black than White students when 
controlling for other variables.  
Gender
Gender is measured by a dichotomous variable, with men serving as the reference 
group.  Prior research (Hearn, 1992; Manski & Wise, 1983; Perna, 2000) consistently 
70
shows that women are more likely than men to enroll in college after controlling for other 
variables.  Hearn (1992) also found, after controlling for other variables, that students 
who delayed enrollment in college beyond one year but no more than two years after high 
school tended to be male, while women were more likely than men to enroll in college 
immediately after high school graduation.  
Academic Preparation and Achievement
Two variables are utilized to measure academic preparation: academic 
achievement and coursework.
Academic Achievement
Academic achievement is measured by the standardized composite reading and 
math IRT score in the 12
th
 grade.  This measure has less missing data than other possible 
measures of academic achievement, such as high school grade point average and college 
admission test scores.   Perna (2000) and Perna and Titus (in press) utilized this measure
and found that test scores were positively related to the probability of enrolling in a four-
year college net of other variables.  Also using the NELS, Plank and Jordan (2001), in 
their exploration of the effects of information and guidance on postsecondary 
destinations, utilized the 1990 composite reading and math IRT score (i.e., the test score 
from the first follow-up when students were in the tenth grade).  They found that 
academic achievement was a positive predictor of enrolling in a four-year institution net 
of controls for background characteristics, and social capital .  Croninger and Lee (2001)
utilized a similar measure, but from the 8
th
 grade, in exploring high school drop-out rates.  
71
They found that this measure of academic achievement was a significant negative 
predictor of high school drop-out rates after controlling for background characteristics 
and social capital. 
Coursework
The rigor of coursework completed is measured by a series of dichotomous 
variables that reflect the highest math course that a student took in high school.  These 
variables were computed using data on math courses from student high school transcripts.  
This measure has five categories: advanced math, algebra 2, algebra 1, other or no math, 
and missing or unknown math.  Other or no math is the reference group.
Adelman (2002) recommended rigor of coursework as a more reliable measure of 
academic preparation than curricular track.  Perna and Titus (2004; in press) utilized the 
highest level of math coursework taken in high school as a measure of academic 
preparation.  They found that taking at least algebra 1 was a significant positive predictor 
of college enrollment after controlling for such variables as background characteristics, 
parental involvement, and school characteristics.
Financial Resources
Three variables are utilized to measure college cost: perceived importance of 
college cost and financial aid, actual tuition in home state, and state unemployment rate.
72
Importance of Costs and Financial Aid
The importance of college costs to a student is measured by a composite variable 
comprised of two items from the second (1992) follow-up:  the perceived importance of 
college expenses and the importance of financial aid.  Both of these variables are on a 
scale of one to three with one being not important to three being very important.  These 
two variables were summed with a new range from one to six.  The variable was then 
recoded to create the new measure with three categories, from little importance (1) to 
very important (3).  Little importance is the reference category.
Tuition
The cost of college is measured by the average in-state tuition and fees at two-
year public colleges and universities in the student?s home state in 1992, when the 
student graduated from high school.  Tuition data are not included in the dataset, but are 
added using data from the Digest of Education Statistics (NCES, 1993).  The data are 
weighted by the number of full-time equivalent undergraduates, but were not adjusted to 
reflect student residency.  Arkansas, South Dakota, and the District of Columbia did not 
have two-year institutions listed, so the average in-state tuition and fees at four-year 
institutions was used, since this would most likely be the least expensive option for 
graduates in those states. Kane (1994), Perna (2000), and Schwartz (1985) utilized a 
similar variable and found that higher tuition is negatively related to college enrollment, 
regardless of other variables.
73
Unemployment Rates
Manski and Wise (1983) utilized a measure of forgone earnings to test their 
econometric model of college enrollment.  This measure was the expected monthly 
earnings of a student based on his or her race, gender, and ability level.  They found that 
as expected forgone earnings increased, enrolling in college became a less likely 
decision.  While actual forgone earnings would be the ideal measure, when the sample of 
those graduates not enrolled by 1994 was broken down by race and ability level, there 
were too few cases in most of the categories (i.e., less than 50 in each group except for 
Whites) to determine reliable estimates of forgone earnings. 
Instead of a measure of earnings, I utilize June 1992 state unemployment rates, 
obtained from the Bureau of Labor Statistics (1992 Unemployment Rates, 2004), in order 
to provide a measure of labor market conditions.  Kane (1999) found that the higher 
unemployment rates, the more likely that a student was to enroll in college.  
Social Capital
Seven variables are utilized to measure social capital: number of financial aid 
contacts that a student had in the 12
th
 grade, parental involvement in a student?s 
education, existence of closed networks, student-teacher relations, high school based 
support, high school control, and the percentage of students in a particular school 
participating in the free and reduced price lunch program.  
74
Number of Financial Aid Contacts
A composite variable, information sources on financial aid, is included as one
measure of social capital.  The composite is created utilizing five variables from the 1992 
follow-up.  The variables, which all asked for yes/no responses, are: whether the student 
read information from the school on financial aid, whether the student talked to the 
school representative about financial aid, whether the student talked to a teacher/school 
counselor about financial aid, whether the student read Department of Education
information on financial aid, and whether the student talked to an adult about financial 
aid.  The composite variable is created by adding the number of financial aid contacts .
The composite variable has five categories: no financial aid contacts, one to two financial 
aid contacts, three to four financial aid contacts, and five or more financial aid contacts.
No financial aid contacts is the reference group.
Cabrera and LaNasa (2001), in their study using the same dataset, found that a 
composite of these items was reliable (
=.73) and that the composite was a positive 
predictor of applying to a four-year college for students graduating from high school in 
1992.  Plank and Jordan (2001) also utilized a similar variable and found that the number 
of information sources that a student utilized was a positive predictor of enrolling at a 
four-year institution, rather than not enrolling or enrolling part-time at a two-year 
institution, after controlling for background characteristics, parental involvement, and 
college search activities.
75
Parental Involvement in Student?s Education
Parental involvement in their student?s education, is a measure reflecting social 
capital within the family.  This measure utilizes information from six variables in the 
1992 follow-up.  Exploratory factor analysis is utilized to create this measure and the
results are provided in Table 3.2.  Factor analysis allows for identification of an 
unobservable variable that is common to other variables (Pedhazur & Schmelkin, 1991).  
The variables that are considered in the factor analysis are the frequency that the student
and parent: discussed school courses, discussed school activities, discussed things studied 
in class, discussed grades, discussed preparation for the ACT/SAT test, and discussed 
going to college.  Each of these items is measured on a Likert scale with 1=never to 
3=often.  A factor composite of these variables was used in a study employing the same 
dataset by Perna (2000) and the items together created a highly reliable factor (
 =.83).  
Perna (2000) found that, after controlling for race and sex, costs and benefits, ability and 
other measures of social and cultural capital, this measure of parental involvement was 
positively related to the likelihood that White students would enroll in college 
immediately after high school, but was not related to the likelihood of enrolling for 
students of other racial and ethnic groups.  Cabrera and LaNasa (2001) also utilized a 
similar composite measure and found that students with higher levels of parental 
involvement were more likely to apply to a four- year college, a necessary precursor to 
enrollment, after controlling for socioeconomic status, race, gender, aspirations, and other 
social and cultural capital measures.  
76
Table 3.2. Components of the Factor Composite:  Parental Involvement in Student?s 
Education
Factor Components Loadings
Discussed school courses with parent .770
Discussed school activities with parent .769
Discussed things studied in class with parent .746
Discussed grades with parent .741
Discussed preparation for the ACT/SAT test with parent .696
Discussed going to college with parent .680
Alpha reliability coefficient .828
Source: Analyses of NELS:92/2000
Note: Data are weighted by normalized F4F2 panel weight
Closed Networks
The number of other parents to whom a parent talks is a measure of a closed 
network, another aspect of social capital.  This categorical measure reflects whether a 
parent reports talking to no other parent, one or two parents, three to five parents, six to 
ten parents, eleven to twenty parents, or more than twenty parents.  More than twenty 
parents is the reference category.  Perna and Titus (in press) utilized this measure in their 
study of parental involvement and college enrollment. While Perna and Titus did not 
find this to be a significant predictor of college enrollment after controlling for other 
variables, Coleman (1988) argues that such a measure is an important dimension of social 
capital.  He stated that social capital is formed through social structures and relationships.  
He argued that a key way that social capital may be formed was through a closed 
network, where all actors are connected in some way.  He believed that a closed network
could create strong norms within a social structure.  These closed social structures may 
serve as a powerful form of social capital by motivating trustworthiness that in turn 
creates obligations and expectations as well as the ability to have effective group norms.  
77
Student-Teacher Relations
A composite variable, student-teacher relations, is computed from three variables 
from the 1992 follow-up survey that reflect the extent to which students agree or disagree 
that teachers: are interested in them, are good at teaching, and help them with their 
homework.  Two of the variables: teachers are interested in students and are good at 
teaching are recoded into dichotomous variables so that all variables are on the same 
metric.  The variables were then summed and then recoded to create the new variable.  
The metric for this variable is on a 1 to 4 scale, with 1 labeled as strongly agree and 4 
labeled as strongly disagree.  Strongly disagree is the reference category.
Croninger and Lee (2001), who used the NELS in their study of dropping out of 
high school, employed a similar measure of social capital and found that it was a 
significant negative predictor of dropping out of high school.  The reliability of the factor 
was high (
 =.79).  Croninger and Lee argued that social networks developed with 
teachers may impact a student?s educational progress, especially if other forms of social 
capital are not present in their lives.     
High School Based Support
Croninger and Lee (2001) and Perna and Titus (in press) argued that school 
programs may have a large impact on the social capital of the students and may 
supplement resources from family and the community.   In this study, three variables are 
combined into a dichotomous variable to measure high school based support: if a student 
received help from high school personnel with college application procedures, if a 
78
student received help from high school personnel with financial aid application 
procedures, and if a student received assistance from high school personnel in writing 
college application essays. These items are yes/no items.  If a student answered ?yes? to 
at least one of the items, then it was classified as a ?yes? for the composite.  No is the 
reference category.
Cabrera and LaNasa (2001) utilized the three separate measures of high school 
support to study the college choice of low-socioeconomic status students and found that 
two of the measures, high school assistance in completing college applications and high 
school assistance in completing college essays were significant positive predictors of 
applying to a four-year college among the overall sample.  None of the three were related 
to the probability of applying among low-socioeconomic status students.  Perna (2000)
measured high school support in her study by combining the three items into a 
dichotomous variable as to whether or not a student received any of these three forms of 
assistance (yes/no).  She found that receiving no help from high school personnel with 
college requirements was a significant negative predictor of college enrollment after 
controlling for other variables. 
High School Control
High school control is also a measure of social capital used in this study.  The 
variable has two categories, public and private.  Public is the reference category.
McDonough (1997) found that students who attend private schools may have higher 
expectations of attending college and have more capital available from personnel at these 
schools.  She posited greater availability of networks and resources that promote college 
79
enrollment at private than public schools.  Perna (2000) utilized this variable in studying 
immediate enrollment into college of 1992 high school graduates and found that school 
control was unrelated to college enrollment after controlling for other variables.  
Coleman (1988) found, using the High School and Beyond dataset, that the type of high
school that a student attends affected high school persistence, a necessary step toward
college enrollment.
Free and Reduced Price Lunch
The final measure of social capital used in this study is the percentage of students 
at a school who participate in the free and reduced lunch program.  This variable was 
negatively skewed so it was recoded into quartiles. The fourth quartile is the reference 
category.  The percent of students receiving free and reduced price lunch is a proxy for 
the resources available at a student?s school.  The rationale is that, the more students who 
participated in this program, the less likely that other students and parents at the school
had adequate resources to promote college enrollment.  A similar measure was utilized by 
Hamrick and Stage (1998) to identify students from resource-poor schools.  Hamrick and 
Stage stated that at resource-poor schools, such as those with high percentages of students 
in the free and reduced price lunch program, the less time that school personnel spent on 
promotion of college enrollment.  At resource-poor schools, school personnel were more 
likely to be spending time with students with academic or behavioral problems.  Hamrick 
and Stage found that, for students from schools with high participation in free and 
reduced lunch programs, gender, socioeconomic status and parental expectations were all 
predictors of college predisposition.
80
Cultural Capital
Five variables are utilized to measure cultural capital: parental expectations, peer 
encouragement, parental involvement with the child?s school, educational materials in the 
home, and participation in art, music, or dance classes.
Parental Expectations
The highest expectation for the student?s educational attainment of the student?s 
mother is included as a measure of cultural capital because it illustrates the value of a 
college education to a parent.  While Cabrera and LaNasa (2001) chose the parent with 
the highest expectations, I utilize the expectations of the mother only, as Perna (2000)
did, in order to make the implications of the findings clearer. The categories for this 
variable are (1) no postsecondary expectations, (2) expected to attend a two-year 
academic or technical college, (3) expected to complete a bachelor?s degree, or (4) 
expected to attain a graduate or professional degree. The categorical variable is coded 
into a series of dummy variables with expected graduate education as the reference 
category.
Cabrera and LaNasa (2001) found that there was a greater probability of a student 
applying to a four-year college if at least one parent expected a bachelor?s degree or 
higher from their child.  Perna (2000) also found that a mother?s educational expectation
had a significant positive effect on college enrollment of a high school graduate net of 
other variables.  
81
Parental Involvement with a Child?s School
The involvement of parents with the student?s school is also considered as a 
measure of cultural capital.   The factor composite is comprised of four variables 
measuring the frequency with which the parent contacted the school about their student?s 
academic performance, academic program, plans after high school, and college 
preparation course selection.  Each of these variables is coded on a one to four scale, with 
one being never to four being more than four times.  Table 3.3 shows that the alpha 
reliability coefficient for the variable is .802.
Table 3.3. Components of the Factor Composite:  Parental Involvement with a Child?s 
School
Factor Components Loadings
Contacted School About Academic Program .827
Contacted School About College Preparation Course Selection .825
Contacted the School About Teen?s Plans After High School .820
Contacted School About Academic Performance .709
Alpha reliability coefficient .802
Source: Analyses of NELS:92/2000
Note: Data are weighted by normalized F4F2 panel weight
Lareau (1987) and Lareau and Horvat (1999) both consider parental involvement 
in the schools to be measure of cultural capital.  Parents are activating their capital in 
order to participate in their child?s education, which is a cultural norm in middle and 
upper class families (Lareau, 1987).  Although both of their qualitative studies focused on 
elementary school children, both found that families who activated their cultural capital 
by interacting with the school, such as attending school events and having regular 
individual contact with the classroom teacher had children who were more likely to 
succeed in school.   
82
Educational Materials in the Home
Educational materials in the home is another proxy of cultural capital.  A person?s 
exposure to material goods such as books or computers can be classified as the 
objectified state of cultural capital (Bourdieu, 1986).  Three survey items are utilized to 
construct the variable: students have a daily newspaper, students have a home computer, 
and students have more than fifty books.  The new variable reflects whether a student has 
any of these three items (yes/no).   No is the reference category.
Peer Encouragement
In order to capture peer attitudes, a composite variable, peer encouragement, is
created using factor analysis and three variables from the 1992 follow-up survey.  The 
variables that are considered in the exploratory factor analysis are how important it is for 
friends to get good grades, study, and finish high school.  The metric for these variables 
is on a one to three scale, with one being not important to three being very important.  
Table 3.4 shows three variables load on one factor with an alpha reliability coefficient of 
.734.
Perna (2000) found that a similar composite was not significantly related to 
college enrollment among White, Black, or Hispanic students after controlling for other 
variables.  Nonetheless, McDonough (1997) found that peers can be a strong transmitter 
of cultural capital through the values that they place on education, which in turn influence
other students. She found a seamless environment for high socioeconomic status girls, in
83
the expectation that a student would attend college was present from friends, family, and 
the school.  
Table 3.4. Components of the Factor Composite:  Peer Encouragement
Factor Components Loadings
Among friends, how important is it to get good grades .871
Among friends, how important is it to study .821
Among friends, how important is it to finish high school .735
Alpha reliability coefficient .734
Source: Analyses of NELS:92/2000
Note: Data are weighted by normalized F4F2 panel weight
Participation in Cultural Activities
Downey and Powell (1993) recommended the use of a variable taking cultural 
classes (i.e., art, music, or dance) at least one time a week, as a proxy for cultural capital.  
Perna and Titus (in press) utilized a measure of participation in cultural classes such as 
those mentioned above in their study of the impact of parental involvement or racial 
group differences in college attendance.   The variable from the NELS reflects  whether a 
student has taken music, art, or dance classes.  Response categories are (1) never/rarely, 
(2) less than one time per week, (3) one to two times a week, (4) almost every day.
Never is the reference category.  While Perna and Titus (in press) did not find this 
variable to be a significant predictor of college enrollment , Horvat (2001) argues that 
such a measure is an important dimension of cultural capital.
Post-Graduation Characteristics
In order to provide a more complete picture of graduates who delay enrollment, 
additional post-high school characteristics are included for use in the descriptive 
84
analyses.  Seven variables are utilized to measure post-high school characteristics for 
those students who delay enrollment in college: married before enrollment, children 
before enrollment, working full- or part-time in 1993, occupation in 1993, expected 
occupation in 2000, earnings in 1993, and expected income at age 30.  These variables 
were chosen based on the assumption that, the further a student was from their high 
school experience, the greater chance that a student?s decision to enroll in college was 
influenced by post-high school graduation activities and experiences.  
Married before enrollment
The measure for whether a student was married before he or she enrolled in 
college is based on date of first marriage.  This variable was recoded into categories by 
calendar year.  This date was then compared to year of enrollment to create a 
dichotomous variable, married before enrollment-yes or no.  Teachman and Polonko 
(1988) found in their longitudinal study of data from the 1972 NLS that marriage reduced 
the probability of college enrollment.  
Children before enrollment
The measure for whether a student had a child before enrollment in college is 
based on date of first child.  The variable was recoded into categories by calendar year.  
The year of first child was then compared to year of enrollment to form a dichotomous 
variable, child before enrollment-yes or no.  Having children was found to reduce the 
probability of college enrollment in their longitudinal study of data from the 1972 NLS 
(Teachman & Polonko, 1988).
85
Employment Status in 1993
Work experience immediately after high school graduation is measured by a 
dichotomous variable that asks whether a student is working full-time or part-time. If a 
student works part-time while enrolled in school, the decreased earnings compared to 
someone who works full-time is a direct cost of college enrollment.  The student may 
forgo some of the earnings that they would make as a full-time employee if he or she was 
not enrolled in school.  Low-socioeconomic status students may not have the option to 
only work part-time and the forgone earnings alone may keep an individual from 
participating in higher education (Kane, 1999).  Also, students may choose not to forgo 
earnings, and continue to work full-time while going to school either part-time or full-
time, or delay time of enrollment in order to save money for college.
Occupation in 1993
An additional measure of employment status immediately after high school 
graduation is occupation in 1993.  This 30 category variable was recoded into 15 
categories.  Occupations were grouped together based on type of position (e.g., clerical-
secretarial, clerical-financial, clerical-other).  The 15 categories are: clerical, craftsman, 
farmer, laborer, manager, military, arts or entertainment, professional, owner, service, 
sales, education, technical, and not working.  
86
Expected occupation in 2000
The variable, expected occupation in 2000, was asked as a part of the third 
follow-up in 1994.  Participants were asked to project their occupation six years in the 
future.  The categories of this variable were the same as occupation in 1993, so the same 
recoding process was utilized.  This variable may suggest whether the student has an 
interest in a career that requires a college education.  These future occupation goals 
influence the decision to attend college in the econometric model (Becker, 1993).  
Income in 1993
The continuous standardized measure, total earnings from jobs January to 
December 1993 is utilized to measure 1993 income.  This measure is standardized for use 
in these analyses. If a graduate chooses to attend college, he or she may forgo some of 
his or her earnings in order to attend school (Becker, 1993).  He or she may not be willing 
to forgo earnings in order to enroll in college, or may choose to delay enrollment.  Also, 
the higher a graduate?s income, the less likely he or she may be to see a need to enroll in 
college.
Expected income at age 30
The continuous measure, expected income at age 30 is a measure that was 
collected from participants in 1994, when the average age of the participants was 20.    
This measure is standardized for use in these analyses.  Future income expectations may 
influence an individual to enroll in college due to the increased earnings potential of an 
individual with a college degree (Becker, 1993).  
87
Limitations of the Study
This study has at least five limitations.   The first limitation is that data are
missing for many independent variables.  Table 3.5 shows that a few variables such as 
academic achievement (13.3%), closed networks (16.0%), parental expectations (13.7%) 
and parental involvement with the school (13.9%) have relatively large amounts of 
missing data. Listwise deletion reduces the number of cases by 48.5%, resulting in an 
analytic sample of 4,415 cases.
Table 3.5.  Number and percentage of cases that are missing for each of the variables in 
the analyses from the final sample (n=8,567)
Variable Number
Complete
Number
Missing
Percent
Missing
Socioeconomic Status 8476 91 1.3
Race 8552 15 0.2
Gender 8567 0 0.0
Academic Achievement 7424 1143 13.3
Coursework 8407 160 1.9
College Costs 7899 668 7.8
Tuition 8546 21 0.2
Unemployment 8546 21 0.2
Number of Financial Aid Contacts 7889 679 7.9
Parental Involvement in Student?s Education 7510 1057 12.3
Closed Networks 7197 1370 16.0
Student Teacher Relations 8342 225 2.6
High School Based Support 7961 606 7.1
High School Control 8546 2121 0.2
Free and Reduced Price Lunch Quartile 8290 277 3.2
Parental Expectations 7397 1170 13.7
Parental Involvement with School 7379 1188 13.9
Educational Materials 8267 301 3.5
Peer Encouragement 8012 555 6.5
Involvement in Cultural Activities 8220 347 4.0
Number of cases with data for all variables 4,415 4,152 48.5
Source: Analyses of NELS:92/2000
Note: Data are weighted by normalized F4F2 panel weight
88
Moreover, missing data analysis showed that the data are not randomly missing 
(Table 3.6).  Students who delayed enrollment (55.0%) and students who did not enroll 
(66.5%) in college were more likely to have missing data on one or more variables than 
students who enrolled in college immediately (35.0%).  Higher percentages of Hispanic 
(65.0%) and Black (63.6%) students were missing data on one or more variables than 
Asian (45.3%) or White (44.6%) students.  Missing data were more common among men 
(50.9%) than women (46.4%).  A higher percentage of students from the lowest 
socioeconomic status quartile (62.7%) than of students in the second, third, and fourth 
quartiles (51.6%, 47.2, and 37.8 respectively) had missing data on one or more variables.  
Missing data were also more common among students in the lowest academic 
achievement quartile (67.7%) than among students in the second, third, and fourth 
quartiles (46.9%, 37.7%, and 24.2% respectively).  
Because of the magnitude and non-randomness of the missing data, I considered
potential strategies for minimizing missing data.  Other researchers utilizing the NELS
have described missing data as a limitation in their studies.  Croninger and Lee (2001)
dealt with their missing data by excluding cases that were missing data on key variables.  
This substantially decreased the sample size for their study.  For my study missing data 
analysis revealed that listwise deletion of missing data would result in a large reduction 
of number of cases in the sample from 8,567 to 4,415.  Pairwise deletion is another way 
to deal with missing data, but this is only recommended for randomly missing data 
(Cohen & Cohen, 1983).  This procedure is not adopted in this study procedure due to the 
non-randomness of the missing data, as described above.  Perna (2000), in order to 
maintain the size of the sample, imputed missing data for some continuous variables 
89
based on the average value for cases with the data for the variable by race and 
socioeconomic status quartile. 
Table  3.6.  Percentage of cases with missing data for selected characteristics of 1992 
high school graduates
Characteristics % Missing
Enrollment
     Enrolled Immediately 35.0
     Delayed Enrollment 55.0
     No Enrollment 66.5
Race
     Asian 45.3
     Hispanic 65.0
     Black 63.6
     White 44.6
Gender
     Men 50.9
     Women 46.4
Socioeconomic Status Quartile
     1
st
62.7
     2
nd
51.6
     3
rd
47.2
     4
th
37.8
Academic Achievement Quartile
     1
st
67.7
     2
nd
46.9
     3
rd
37.7
     4
th
24.2
Source: Analyses of NELS:92/2000
Note: Data are weighted by normalized F4F2 panel weight
***p<.001
In order to maintain the size of my sample, I imputed missing data for the three 
continuous variables with at least ten percent missing data: academic achievement, 
parental involvement with the school, and parental involvement in a student?s education.  
Imputation was based on the average value for students of the same race/ethnicity and 
gender. I was not able to include socioeconomic status in the imputation procedure
because there were too few cases in many of the socioeconomic quartile, race/ethnicity,
and gender cells for the average to be a reliable measure.  Nonetheless, this process may 
90
cause an underestimation of standard errors thereby causing the coefficient for imputed 
variables to be falsely statistically significant.  For categorical variables, math 
coursework, closed networks, and parental expectations, I added a dummy variable to 
indicate whether the data were missing (Cohen & Cohen, 1983).  The implications of the 
missing data techniques will be kept in mind when interpreting the results of the analyses.
Since I am conducting this study using secondary data analysis, a second 
limitation is that I must use variables that have already been collected.  Since complex 
constructs such as social and cultural capital are utilized in the conceptual model, it is 
difficult to find appropriate proxies.  Proxies of social capital in the analyses include 
parental involvement in a student?s education, number of financial aid contacts, student-
teacher relations, high school control and high school based support.  Parental 
expectations and involvement, peer encouragement, educational materials in the home, 
and visits to museums/ participation in the arts, act as proxies of cultural capital for this
study.  While these proxies appear to be adequate, many of the measures such as parental 
involvement in both home and school and type of financial aid contacts measure the 
quantity of interaction, not the nature of the interaction or its quality.  These measures 
cannot identify the type of resource accessed through involvement or the quality of the 
resources available through networks.  High school control is also used as a proxy for 
what resources a school has, but again, a more adequate proxy, would be recommended.  
The third limitation of the study is that predictors not included in the model which 
could be helpful to the study include a student?s perception of the benefits of higher 
education.  The NELS does not have variables as part of the dataset that are adequate 
proxies for benefits of college.  
91
Reflecting the construction of the dataset, college enrollment only up to eight 
years after high school graduation is examined, which is a fourth limitation.  Data from 
more than eight years after high school graduation may provide greater information 
related to college enrollment patterns, if it were available.  Nonetheless, the eight year 
time period may be an appropriate cut-off given the declining benefits of higher 
education to older individuals (Becker, 1993). 
A fifth limitation is that, in the early 1990?s, when students in the sample 
graduated from high school, the economy was different than in the late 1990s.  These 
economic changes may have encouraged students to enroll immediately into college, but 
encouraged individuals who might have otherwise enrolled at a later point to stay in the 
labor market.  The model does not include measures of changes in economic forces that 
occurred after 1992 which may play a role in the decision to enter college two to eight 
years after high school graduation. 
A final limitation is that I have only included those students who graduated from 
high school on time in 1992.  I made this decision because students need to graduate in 
order to qualify for college, and because I wanted to examine students who completed 
high school in a similar manner.  This decision excludes some students who may have 
completed high school later, or who completed a GED in lieu of high school graduation.
Even with these limitations, the study is worth pursuing due to the likely 
contribution to theory, policy and practice.  The strengths of the research design, 
including the size of the sample, the longitudinal nature of the NELS dataset, and the 
number of variables that are available to represent complex constructs in the analyses 
also support the worth of the study.  The study tests the applicability of a conceptual 
92
model of traditional college enrollment to the decision to delay college enrollment.  The 
study also contributes to our understanding of the nature of socioeconomic status gaps in 
educational opportunity by examining socioeconomic gaps in delayed enrollment of 1992 
high school graduates.  
93
CHAPTER 4
RESULTS
Introduction
This study uses descriptive and multivariate analyses to examine college 
enrollment patterns of 1992 high school graduates within eight years of graduation.  The 
study determines the most appropriate definition of delayed enrollment based on 
differences in student characteristics at year of enrollment.  The relationship between 
socioeconomic status and timing of college enrollment is also explored to determine if 
the socioeconomic status gap that is present in immediate enrollment is also present in 
delayed enrollment.  
The chapter addresses the first research question by presenting the percentage of 
1992 high school graduates who enrolled in college in each of eight calendar years after 
high school graduation.  Next, a break point between immediate and delayed enrollment 
is determined, by examining differences in time of enrollment for enrolling in 1992, 1993 
and 1994. Moving on to the second research question, the chapter presents descriptive 
statistics that compare the characteristics of three groups: high school graduates who 
enroll immediately in college, those who delay enrollment, and those who do not enroll in 
college within eight years of graduating from high school.  The third research question is 
addressed by presenting descriptive statistics that examine the characteristics of graduates 
who delay enrollment by each year of delayed enrollment.  To address the fourth and fifth 
research questions, the results of multivariate analyses are presented.  These results
examine the predictors of delaying college enrollment after graduating from high school 
94
rather than enrolling in college immediately or not enrolling.  The relationship between 
socioeconomic status and time of enrollment is also addressed through the multivariate 
analyses.
Research Question One: Timing of College Enrollment
The first research question explores the percent of students that enroll each year 
after graduation from high school.  This information helps to determine the break point 
between immediate and delayed college enrollment.  Table 4.1 shows the number and 
percentage of 1992 high school graduates in the sample who enrolled in college in each 
of eight years after high school graduation.  The majority of 1992 high school graduates 
(68%) first enrolled in college during the 1992 calendar year.  A small percentage of 
1992 high school graduates (1%) enrolled in college before 1992, possibly through dual 
enrollment programs or other special programs that allow high school students to take 
classes before earning a high school diploma.  Graduates continued to enroll in college in 
subsequent years, with about 9% of 1992 high school graduates enrolling in 1993 and 
between 0.1% and 2.4% enrolling each year between 1994 and 2000.  About 14% of 
1992 high school graduates had not enrolled in college within eight years of graduating 
from high school.
The initial analyses broke down years of enrollment by academic year rather than 
calendar year.  However, disaggregating the 1992-1993 academic year showed significant 
differences between graduates who first enrolled in college in 1992 and graduates  who 
first enrolled in spring 1993 in terms of race, socioeconomic status, achievement test 
scores, highest level of math coursework, parental involvement, high school control, and 
95
mother?s expectations (Appendix A).  In contrast, chi-square and ANOVA analyses 
revealed no significant differences between graduates who first enrolled in spring 1993 
and graduates who enrolled in the 1993-1994 academic year, except for race and high 
school based support (See Appendix B).  In other words, the analyses suggest that 1992 
high school graduates who enrolled in college in fall 1992 are different from high school 
graduates who first enrolled in spring 1993.  But, high school graduates who first enrolled 
in spring 1993 show few differences from high school graduates who first enrolled in the 
1993-1994 academic year.  Therefore, the analyses are based on calendar year rather than 
academic year of first enrollment. 
Table 4.1.  Number and percentage of 1992 high school graduates who first enrolled in 
college in each of nine calendar years after graduating from high school
Year Unweighted
Frequency
Weighted
Frequency
Weighted 
Percent
Pre 1992 87     99 1.2
1992 5,981 5,786 67.5
1993 747   793 9.3    
1994 205   203 2.4
1995 120   160 1.9
1996 78     69     0.8
1997 95   110     1.3
1998 80     84     1.0
1999 62     58     0.7
2000 15     11 0.1
Not Enrolled 1,113 1,196      14.0
Total 8,583 8,567 100.0
Source: Analyses of NELS:92/2000
Note: Data are weighted by normalized F4F2 panel weight
Table 4.1 shows that in six of the ten years, the unweighted number of graduates
who enrolled in college was less than 100: pre-1992, 1996, 1997, 1998, 1999, and 2000.  
In order to increase the reliability of the analyses, I collapsed the ten years into six: 1992
(includes those who enrolled pre-1992), 1993, 1994, 1995-1996, 1997, and 1998-2000.  
96
Chi-square and ANOVA analyses suggest that 1995 and 1996 should be considered 
together rather than 1996 and 1997 (Appendices C & D).  Chi-square and ANOVA 
analyses revealed no significant differences between graduates who first enrolled in 1995 
and graduates who first enrolled in 1996 except for race and the number of parents that 
other parents talk to (Appendix C), but significant differences between graduates who 
first enrolled in 1996 and graduates who first enrolled in 1997 for four variables: race, 
gender, parental involvement in a student?s education, and educational materials in the 
home (Appendix D).  Chi-square and ANOVA tests revealed no significant differences 
among graduates who first enrolled in 1998, 1999, or 2000, supporting the decision to 
combine these three categories (Appendix E).     
In order to determine the appropriate break point for immediate and delayed 
enrollment, I ran chi-square and ANOVA tests comparing 1992 and 1993 enrollment, and 
chi-square and ANOVA tests comparing 1993 and 1994 enrollment.  These results are 
displayed in Appendix F.  I found that there were significant differences between 
graduates who first enrolled in 1992 and graduates who first enrolled in 1993 for race, 
socioeconomic status, test scores, highest level of math, social capital characteristics 
including number of financial aid contacts, parental involvement with the student, 
number of parents that other parents talk to, positive student-teacher relations, high 
school based support, high school control, free and reduced price lunch quartile, and 
cultural capital characteristics, including mother?s educational expectations, peer 
encouragement, and educational materials in the home.  A comparison of 1993 and 1994 
enrollment revealed no significant differences (Appendix F) for any of the categorical or 
continuous variables examined.  Based on the examination of these three years, I define 
97
immediate enrollment as enrolled in college before the end of the 1992 calendar year, and 
delayed enrollment as enrolled in college between January 1993 and 2000.
Research Question Two: Differences in High School Characteristics by Time of College 
Enrollment
The second research question explores how the characteristics of graduates who 
delay enrollment in college compare to the characteristics of graduates who enroll 
immediately and those who do not enroll.  Table 4.2 shows the distribution of 1992 high 
school graduates by timing  of college enrollment.  While 68% of the students who 
graduated from high school in 1992 enrolled in college immediately (i.e., 1992 calendar 
year), 17% delayed enrollment (i.e., enrolled between 1993 and 2000), and 14% of 
graduates did not enroll in college by 2000.  
Table 4.2.  Distribution of 1992 high school graduates by timing of college enrollment
Year Weighted 
Frequency
Percent
Immediate Enrollment 5,885 68.4
Delayed Enrollment 1,486 17.3
No Enrollment 1,196 13.9
Total 8,567 100.0
Source: Analyses of NELS:92/2000
Note: Data are weighted by normalized F4F2 panel weight
In order to further understand the immediate and delayed groups, Table 4.3 shows 
the distribution of graduates who enroll by type of institution (i.e., public/private, 2-
year/4-year) as well as a proxy of full or part-time enrollment (i.e., if they are primarily a 
student or an employee).  The type of institution in which a graduate enrolls varies 
significantly by time of enrollment.  Higher percentages of graduates who enroll 
immediately attended four-year institutions (68%), while higher percentages of those who 
delayed enrollment attended two-year institutions (70%).  In terms of enrollment status, 
98
the majority of graduates view themselves primarily as students, although the percentage
is higher for those who enroll immediately (85%) than for those who delay enrollment 
(65%). 
Table 4.3.  Percentage of 1992 high school graduates who enroll in college by 2000 by 
school type and status
Student Characteristics Immediate Delayed Total
Type of Institution*** 100.0 100.0 100.0
   Private 2-year 1.1 2.7 1.4
   Public 2-year 30.7 67.4 37.0
   Private 4-year 23.1 7.4 20.4
   Public 4-year 45.1 22.5 41.2
Enrollment Status*** 100.0 100.0 100.0
   Primarily a student 84.9 65.2 81.3
   Primarily an employee 15.1 34.8 18.7
Source: Analyses of NELS:92/2000
Note: Data are weighted by normalized F4F2 panel weight
***p<.001
A comparison of the characteristics of 1992 high school graduates by type of 
enrollment shows si gnificant differences among the three categories for all characteristics 
except the importance of cost and aid and two-year tuition rates. In terms of background 
characteristics, Table 4.4 shows that timing of enrollment varies based on race/ethnicity, 
gender, and socioeconomic status.  Black graduates represent a higher percentage of high 
school graduates who delay enrollment (17%) than high school graduates who do not 
enroll (12%) and high school graduates who enroll immediately (9%).  Men make up a 
higher percentage of graduates who delay enrollment (54%) and do not enroll (60%) than 
of high school graduates who enroll immediately (48%).  Graduates who enrolled 
immediately in college had higher socioeconomic status (0.236) than those graduates who
delayed first enrollment (-0.323) or did not enroll in college by 2000 (-0.763).     
In terms of financial resources, neither importance of cost and aid nor two-year 
tuition vary by timing of enrollment.  Table 4.4 shows state unemployment rates vary 
99
among the three groups.  The standardized mean is about 8% of a standard deviation 
lower for those graduates who did not enroll (-0.108) than those who enrolled 
immediately (0.020).
In terms of academic preparation and achievement, graduates who delay 
enrollment or do not enroll have lower achievement test scores and lower levels of math 
than graduates who enroll immediately.  Table 4.4 shows that graduates who enrolled 
immediately in college averaged more than ? standard deviation higher on achievement 
test scores than graduates who delayed first enrollment and about one standard deviation 
higher on test scores than graduates who did not enroll in college by 2000.  Table 4.4 also 
shows that 37% of graduates who enrolled in college immediately after high school took 
advanced math compared to only 10% of those who delayed enrollment and 6% of those 
who did not enroll.  
With regard to social capital characteristics, there were statistically significant 
differences among the three enrollment groups for all characteristics: number of financial 
aid contacts, parental involvement with the student?s education, number of parents that 
other parents talk to, positive student-teacher relations, high school based support, high 
school control, and free and reduced price lunch quartile.  Table 4.4 shows that only 17% 
of the immediate enrollment group had zero financial aid contacts in the 12
th
 grade 
compared with 24% of those graduates who delayed enrollment and 33% of those 
graduates who did not enroll.  Parental involvement was also higher for graduates who 
enrolled immediately in college with average parental involvement scores nearly ? of a 
standard deviation lower for graduates who delayed enrollment and about 2/3 of a 
100
standard deviation lower for graduates who did not enroll than for graduates who enrolled 
immediately.  
Table 4.4.  Characteristics of 1992 high school graduates by enrollment timing
Student Characteristics Immediate
(1)
Delayed 
(2)
None
(3)
Total
Race*** 100.0 100.0 100.0 100.0
APA 5.0 3.6 1.4 4.2
Hispanic 8.0 10.6 9.0 8.6
Black 8.5 17.0 11.8 10.5
White 78.0 67.7 75.6 75.9
Amer. Indian/AK Native 0.6 1.1 2.1 0.9
Gender*** 100.0 100.0 100.0 100.0
Male 47.7 53.6 59.5 50.4
Female 52.3 46.4 40.5 49.6
Socioeconomic Status***
a, b, c
      Mean 0.236 -0.323 -0.763 0.002
      Standard deviation  0.950  0.947  0.788  0.999 
Importance of Cost and Aid 100.0 100.0 100.0 100.0
Little Importance 10.5 9.4 11.6 10.5
Some Importance 42.6 40.6 39.8 42.0
Very Important 46.9 50.0 48.6 47.6
Two-Year Tuition
      Mean 0.002 -0.033 0.062 0.004
      Standard deviation  1.018  0.961  0.939  0.997 
1992 Unemployment Rate***
b
      Mean 0.020 -0.013 -0.108 -0.003
      Standard deviation  1.027  0.916  0.956  0.999 
Reading and Math IRT Score ***
 a, b, c
      Mean  0.248 -0.389 -0.738 0.005
      Standard deviation  0.937  0.895  0.904  0.999 
Highest Level of Math*** 100.0 100.0 100.0 100.0
Advanced Math 36.6 10.1 5.6 27.6
Algebra 2 36.1 27.9 18.2 32.2
Algebra 1 22.8 48.8 48.6 30.9
     Missing or Unknown 1.0 4.2 3.3 1.9
Other or None 3.5 9.0 24.3 7.4
101
Table 4.4.  Characteristics of 1992 high school graduates by enrollment timing 
(continued)
Student Characteristics Immediate
(1)
Delayed 
(2)
None
(3)
Total
Financial Aid Contacts*** 100.0 100.0 100.0 100.0
None 16.5 24.1 32.9 19.7
One or Two 29.3 35.0 35.7 31.0
Three or Four 41.1 31.3 25.7 37.6
At least five 13.1 9.6 5.8 11.7
Parental Involve w/ Student***
 a, b, c
   Mean 0.178 -0.253 -0.574 0.003
      Standard deviation  0.943  1.003  0.987  0.999 
# of Parents Talk To*** 100.0 100.0 100.0 100.0
None 5.5 9.1 9.6 6.6
1 or 2 18.6 26.0 24.9 20.6
3 to 5 28.8 31.6 30.1 29.4
6 to 10 22.2 16.0 19.6 20.9
11-20 12.4 8.6 9.7 11.4
20+ 12.5 8.7 6.1 11.1
+ Student-Teacher Relations*** 100.0 100.0 100.0 100.0
Strongly Disagree 2.6 5.1 3.5 3.2
Disagree 7.3 11.5 11.6 8.6
Agree 21.0 24.1 27.7 22.5
Strongly Agree 69.1 59.3 57.2 65.8
High School Based Support*** 100.0 100.0 100.0 100.0
No 32.0 48.7 51.8 37.1
Yes 68.0 51.3 48.2 62.9
High School Control*** 100.0 100.0 100.0 100.0
Public 88.8 95.8 98.8 91.4
Private 11.2 4.2 1.2 8.6
Free and Reduced Lunch Quart*** 100.0 100.0 100.0 100.0
First 72.7 65.0 61.1 69.8
Second 19.9 22.9 25.7 21.2
Third 5.2 7.1 10.8 6.3
Fourth 2.2 5.0 2.4 2.7
102
Table 4.4.  Characteristics of 1992 high school graduates by enrollment timing 
(continued)
Student Characteristics Immediate
(1)
Delayed 
(2)
None
(3)
Total
Level of Educ Expect by Mother *** 100.0 100.0 100.0 100.0
No Postsecondary Expectations 2.8 6.8 25.1 6.0
  2-Year College 12.5 25.2 33.3 16.9
Bachelor?s 46.7 47.6 30.5 45.0
Graduate or Professional Degree 38.0 20.4 11.1 32.1
Parental Involve w/ School***
 b, c
      Mean 0.030 -0.100 -0.148 0.000
      Standard deviation  0.989  1.023  1.022  1.000 
Peer Encouragement***
a, b
      Mean 0.076 -0.165 -0.158 0.003
      Standard deviation  0.981  1.006  1.044  1.000 
Educational Materials in Home*** 100.0 100.0 100.0 100.0
No 2.2 7.3 4.9 3.5
Yes 97.8 92.7 95.1 96.5
Takes Music, Art or Dance*** 100.0 100.0 100.0 100.0
Never 84.1 87.1 89.3 85.3
Less than 1 time a week 4.5 3.7 3.4 4.2
1 to 2 times a week 8.4 6.6 3.1 7.4
Almost very day 3.0 2.7 4.3 3.1
Source: Analyses of NELS:92/2000
Note: Data are weighted by normalized F4F2 panel weight
***p<.001
a
 Denotes significance in comparison between Column 1 (Immediate Enrollment) and 
Column 2 (Delayed Enrollment), p< .001
b
 Denotes significance in comparison between Column 1 (Immediate Enrollment) and 
Column 3 (No Enrollment), p< .001
c
 Denotes significance in comparison between Column 2 (Delayed Enrollment) and 
Column 3 (No Enrollment), p< .001
Higher percentages of graduates whose parents spoke to two or fewer other 
parents delayed enrollment (35%) or did not enroll (35%) than enrolled immediately 
(24%).  A higher percentage of graduates who enrolled in college immediately agreed or 
strongly agreed that they had positive student teacher relations (90%) than of graduates 
103
who delayed enrollment (83%) or did not enroll (85%).  A higher percentage of graduates 
who enrolled immediately in college stated that they received support from their high 
school (68%) than graduates who delayed first enrollment (51%) or did not enroll in 
college (48%).  
Private school attendance was higher for those graduates who enrolled in college 
immediately (11%) than for graduates who delayed enrollment (4%) or did not enroll 
(1%).  Higher percentages of graduates who delayed enrollment (12%) or did not enroll 
(13%) than of graduates who enrolled immediately (7%) attended a school that was in the 
third or fourth quartile (high) of free and reduced price lunch program participation.
Cultural capital characteristics including mother?s educational expectations, 
parental involvement with the school, peer encouragement, educational materials in the 
home, and participation in art, music, or dance also vary significantly by type of 
enrollment.  A higher percentage of graduates who did not enroll than of graduates who 
delayed or enrolled immediately had mothers with no postsecondary expectations for 
their children .  Table 4.4 shows that 25% of those who did not enroll had a mother 
without postsecondary expectations compared with 7% of those who delayed enrollment 
and 3% of those who enrolled immediately.  In contrast, more than one-third (38%) of 
those who enrolled immediately had a mother who expected a graduate or professional 
degree versus only 20% of those who delayed enrollment and 11% of those who enrolled 
immediately.   
In terms of parental involvement with the school, the standardized mean is lower 
for those graduates who did not enroll (-0.148) than those who enrolled immediately 
(0.030).  In terms of peer encouragement, the standardized mean is lower for those 
104
graduates who delayed enrollment (-0.165) and did not enroll (-0.158) than for those who 
enrolled immediately (0.076).  While high percentages of graduates in all groups had 
educational materials in the home, higher percentages of graduates who delayed 
enrollment (7%) or did not enroll (5%) did not have these educational materials in their 
home compared to those graduates who enrolled immediately (2%).  Finally, there are 
statistically significant differences in the percentages of graduates who took art, music, or 
dance classes but the differences are not substantive.  About 85% of all 1992 high school 
graduates never took a music, art, or dance class during high school.
Research Question 3: Characteristics of Graduates Who Delay Enrollment
The third research question examines the characteristics of graduates who delay 
enrollment (Table 4.5).  The characteristics include those already shown above as well as 
additional post-high school characteristics.  The analyses reveal differences in student 
characteristics by year of enrollment, but no clear patterns emerge in the nature of the 
differences.  For virtually all characteristics, 1997 is an outlier from other years.
The initial characteristics examined are background characteristics.  Table 4.5
shows that race/ethnicity does not vary by year of delayed enrollment.  Among high 
school graduates who delay, 17% are Black, 11% are Hispanic, 4% are Asian, 1% are 
Native American, and 68% are White.  In each year except 1997, at least half of enrollees 
were men.  Men represented only 38% of enrollees in 1997 but about 50% of enrollees in 
1993 and 1994, and more than 60% of enrollees in 1995-1996 and 1998-2000.   Table 4.5
also shows that graduates who enrolled after 1996 had about ? (1998-2000) to ? (1997) 
of a standard deviation lower socioeconomic status than those who enrolled in 1993.
105
In terms of financial resources, year of enrollment varied by the importance of 
cost and aid and state unemployment rates, but not by two-year tuition rates.  For those 
who delayed enrollment, the percentages reporting little importance of cost and aid in the 
12
th
 grade ranged from 7% to 9% in 1993, 1994, and 1998-2000, to 17% in 1995-1996 
and 1997 (Table 4.5).  Unemployment rates also vary by year of enrollment but with no 
pattern.  The standardized average 1992 unemployment rate was lowest for those 
enrolling in 1995-1996  (-0.290) and the highest for those who first enrolled in 1998-
2000 (0.177).    
In terms of academic achievement, average test scores were substantially below 
the mean for all 1992 high school graduates regardless of year of enrollment.  Average 
test scores were somewhat higher for 1992 high school graduates who first enrolled in 
1993 (-0.273) than for high school graduates who first enrolled between 1995-1996        
(-0.591) and 1997 (-0.670).  The percentage of graduates who completed advanced math 
varied by year of enrollment, ranging from 6% to 7% in 1994 and 1997 to 14% in 1998-
2000.
The following social capital characteristics vary by year of enrollment: number of 
financial aid contacts, parental involvement with the student?s education, the number of 
parents that other parents talked to, and free and reduced lunch quartile.  Table 4.5 shows 
that year of enrollment was unrelated to positive student-teacher relations, high school 
based support, and high school control.  The percentage of graduates with less than two 
financial aid contacts in the 12
th
 grade increased from 56% in 1993 to 72% in 1996 and 
then decreased to 53% in 1998-2000.  Average parental involvement with their student?s 
education was below the average for all 1992 high school graduates regardless of year of 
106
delayed enrollment.  Average parental involvement ranged from 15% of a standard 
deviation below the mean in 1993 and 18% of a standard deviation below the mean in 
1998-2000 to nearly 1 standard deviation below the mean in 1997.  The percentage of 
graduates whose parents talked to fewer than two other parents also varied by year of 
enrollment.  More than 40% of graduates who enrolled after 1997 had parents who spoke 
to fewer than two other parents compared with only about 1/3 of graduates who enrolled 
in 1993 and 1995-1996.  About 13% of graduates who enrolled between 1998 and 2000 
attended a high school that was in the fourth quartile (high) of free and reduced price 
lunch program participation compared to less than 5% of graduates who enrolled in 
college in 1993, 1995-1996, and 1997.
107
Table 4.5.  Characteristics of 1992 high school graduates who delayed first enrollment in college by year of enrollment
Characteristics 1993
(1)
1994
(2)
1995-1996
(3)
1997
(4)
1998-2000
(5)
Total
Race 100.0 100.0 100.0 100.0 100.0 100.0
APA 4.3 3.5 4.0 0.0 2.0 3.6
Hispanic 10.7 9.4 8.8 15.5 10.5 10.6
Black 16.9 12.4 20.7 10.9 22.4 17.0
White 67.1 73.3 64.8 73.6 63.8 67.7
Amer. Indian/AK Native 0.9 1.5 1.8 0.0 1.3 1.1
Gender*** 100.0 100.0 100.0 100.0 100.0 100.0
Male 51.7 49.8 65.5 38.2 61.4 53.6
Female 48.3 50.2 34.5 61.8 38.6 46.4
Socioeconomic Status***
c,f,g
      Mean -0.247 -0.326 -0.239 -0.806 -0.491 -0.323
      Standard deviation 0.959 1.054 0.831 0.930 0.782 0.947
Importance of Cost and Aid*** 100.0 100.0 100.0 100.0 100.0 100.0
Little Importance 7.1 8.8 16.5 16.5 8.9 9.5
Some Importance 44.7 30.8 29.3 56.5 37.0 40.5
Very Important 48.2 60.4 54.3 27.1 54.1 50.0
Two-Year Tuition
      Mean -0.040 -0.158 0.024 0.011 0.016 -0.037
      Standard deviation 0.958 1.012 0.873 1.095 0.948 0.963
Unemployment Rate***
b,e,h
      Mean 0.013 0.076 -0.290 -0.008 0.177 -0.010
      Standard deviation 0.885 0.881 0.933 0.954 0.992 0.917
108
Table 4.5.  Characteristics of 1992 high school graduates who delayed first enrollment in college by year of enrollment 
(continued)
Characteristics 1993
(1)
1994
(2)
1995-1996
(3)
1997
(4)
1998-2000
(5)
Total
Test Score ***
  b, c
      Mean -0.273 -0.372 -0.591 -0.670 -0.495 -0.389
      Standard deviation 0.855 0.923 0.955 0.962 0.808 0.895
Highest Math Coursework*** 100.0 100.0 100.0 100.0 100.0 100.0
Advanced Math 11.0 6.9 9.2 6.3 13.7 10.1
Algebra 2 28.4 26.6 27.5 28.8 26.1 27.8
Algebra 1 45.7 51.7 54.6 55.9 46.4 48.7
Other or None 7.4 13.8 7.9 8.1 13.7 9.1
      Missing or Unknown 7.4 1.0 0.9 0.9 0.0 4.3
# of Financial Aid Contacts*** 100.0 100.0 100.0 100.0 100.0 100.0
None 22.0 20.2 25.7 37.5 29.5 24.1
One or Two 33.6 39.9 46.5 29.5 23.7 35.0
Three or Four 35.0 30.6 22.5 11.4 37.4 31.3
At least five 9.4 9.3 5.3 21.6 9.4 9.6
Parental Involvement w/ Student***
 c,f,g,i
      Mean -0.150 -0.257 -0.367 -0.922 -0.180 -0.257
      Standard deviation 0.972 0.958 0.977 0.932 1.114 1.004
# of Parents Talk To*** 100.0 100.0 100.0 100.0 100.0 100.0
None 6.9 8.0 11.8 20.5 9.2 9.0
1 or 2 26.0 29.4 19.2 26.0 32.8 26.0
3 to 5 28.5 33.1 45.3 24.7 26.1 31.6
6 to 10 18.8 10.4 13.3 12.3 16.8 16.1
11-20 9.7 9.8 4.4 8.2 9.2 8.6
20+ 10.1 9.2 5.9 8.2 5.9 8.7
109
Table 4.5.  Characteristics of 1992 high school graduates who delayed first enrollment in college by year of enrollment 
(continued)
Characteristics 1993
   (1)
1994
(2)
1995-1996
(3)
1997
(4)
1998-2000
(5)
Total
Positive Student-Teacher Relations 100.0 100.0 100.0 100.0 100.0 100.0
Strongly Disagree 4.9 7.0 7.5 0.9 2.7 5.1
Disagree 10.5 11.4 13.3 18.7 9.5 11.6
Agree 23.9 24.4 23.9 14.0 32.0 24.1
Strongly Agree 60.7 57.2 55.3 66.4 55.8 59.3
High School Based Support 100.0 100.0 100.0 100.0 100.0 100.0
No 46.2 55.9 55.3 39.3 49.6 48.7
Yes 53.8 44.1 44.7 60.7 50.4 51.3
High School Control 100.0 100.0 100.0 100.0 100.0 100.0
Public 94.1 98.5 96.1 99.1 98.0 95.8
Private 5.9 1.5 3.9 0.9 2.0 4.2
Free and Reduced Lunch Quartile*** 100.0 100.0 100.0 100.0 100.0 100.0
First 67.7 62.0 65.3 67.0 52.7 65.0
Second 22.6 26.0 25.8 13.8 24.0 23.0
Third 5.6 4.7 6.7 17.4 10.3 7.0
Fourth 4.2 7.3 2.2 1.8 13.0 5.0
Mother?s Educ Expectation *** 100.0 100.0 100.0 100.0 100.0 100.0
      None 4.7 8.5 9.3 15.9 8.6 6.8
      2-year institution 21.7 24.2 33.7 31.7 30.5 25.1
Bachelor?s 52.9 46.4 37.8 38.1 37.1 47.6
Graduate or Professional Degree 20.7 20.9 19.2 14.3 23.8 20.4
110
Table 4.5.  Characteristics of 1992 high school graduates who delayed first enrollment in college by year of enrollment 
(continued)
Characteristics 1993
(1)
1994
(2)
1995-1996
(3)
1997
(4)
1998-2000
(5)
Total
Parental Involvement w/ School
      Mean 0.017 -0.021 -0.176 -0.199 0.250 -0.010
      Standard deviation 0.980 0.979 0.986 0.910 1.316 1.022
Peer Encouragement
      Mean -0.120 -0.326 -0.149 -0.232 -0.192 -0.168
      Standard deviation 0.988 1.096 0.944 1.107 0.979 1.006
Educational Materials in Home*** 100.0 100.0 100.0 100.0 100.0 100.0
No 5.9 7.2 4.5 26.9 3.4 7.2
Yes 94.1 92.8 95.5 73.1 96.6 92.8
Takes Music, Art or Dance Classes 100.0 100.0 100.0 100.0 100.0 100.0
Never 86.3 88.8 88.8 86.8 86.1 87.0
Less than 1 time a week 3.7 4.3 4.5 3.8 2.8 3.8
1 to 2 times a week 7.3 4.3 2.2 9.4 10.4 6.6
Almost very day 2.7 2.7 4.5 0.0 0.7 2.6
Married Before Enrolled*** 100.0 100.0 100.0 100.0 100.0 100.0
Yes 6.1 10.0 9.3 25.7 8.1 8.8
No 93.9 90.0 90.7 74.3 91.9 91.2
Child Before Enrolled*** 100.0 100.0 100.0 100.0 100.0 100.0
Yes 4.0 4.0 5.7 14.7 5.6 5.2
No 96.0 96.0 94.3 85.3 94.4 94.8
111
Table 4.5.  Characteristics of 1992 high school graduates who delayed first enrollment in college by year of enrollment 
(continued)
Characteristics 1993
(1)
1994
(2)
1995-1996
(3)
1997
(4)
1998-2000
(5)
Total
1993 Total Earnings from Jobs***
a b c d
      Mean 0.106 0.454 0.542 0.604 0.618  0.313 
      Standard deviation 0.933 1.000 0.852 0.948 1.340 1.013
Income Expected at Age 30
      Mean -0.011 -0.119 0.060 -0.153 -0.094 -0.033
      Standard deviation 1.165 0.316 1.400 0.356 0.352 1.037
Working Full- or Part-Time 1993 100.0 100.0 100.0 100.0 100.0 100.0
Full-time 69.9 63.4 62.0 65.5 66.0 67.0
Part-time 30.1 36.6 38.0 34.5 34.0 33.0
Occupation 1993*** 100.0 100.0 100.0 100.0 100.0 100.0
Clerical 31.7 30.2 18.4 39.6 32.9 30.1
Craftsman 9.8 9.5 12.9 8.5 15.0 10.6
Farmer 2.4 2.1 1.0 0.0 2.1 1.9
Laborer 11.6 7.4 10.9 5.7 10.0 10.2
Manager 4.6 9.5 4.5 5.7 10.0 5.9
Military 2.1 6.9 11.9 6.6 5.7 5.0
      Art or Ent. Professional 1.0 1.1 0.5 0.0 2.9 1.1
Professional 1.8 0.0 1.0 0.0 0.7 1.1
Owner 0.1 0.0 0.0 2.8 0.7 0.4
Service 20.3 18.0 22.4 12.3 10.0 18.5
Sales 12.9 14.8 9.0 15.1 7.1 12.1
Education 0.1 0.0 0.0 0.0 0.7 0.2
Technical 1.5 0.5 7.5 3.8 2.1 2.6
Not Working 0.1 0.0 0.0 0.0 0.0 0.1
112
Table 4.5.  Characteristics of 1992 high school graduates who delayed first enrollment in college by year of enrollment 
(continued)
Characteristics 1993
(1)
1994
(2)
1995-1996
(3)
1997
(4)
1998-2000
(5)
Total
Expected Occupation at Age 30*** 100.0 100.0 100.0 100.0 100.0 100.0
Clerical 4.1 6.7 3.9 22.0 6.3 5.7
Craftsman 3.5 6.1 12.1 8.5 7.0 5.9
Farmer 2.2 1.7 1.5 1.2 1.4 1.9
      Homemaker 0.4 0.6 1.0 3.7 0.0 0.7
Laborer 0.0 0.0 1.0 2.4 0.7 0.4
Manager 7.8 4.4 6.3 4.9 4.9 6.7
Military 1.1 0.0 7.8 2.4 2.8 2.2
      Art or Ent. Professional 7.2 5.6 6.8 3.7 4.2 6.4
Professional 40.1 27.2 22.3 17.1 39.9 34.1
Owner 5.9 5.6 8.3 4.9 7.0 6.2
Service 7.9 12.8 18.9 13.4 11.9 11.0
Sales 0.5 2.8 1.0 1.2 1.4 1.0
Education 6.8 16.1 2.4 7.3 2.1 6.9
Technical 9.9 9.4 6.3 6.1 10.5 9.1
Not Working 2.6 1.1 0.5 1.2 0.0 1.7
Source: Analyses of NELS:92/2000
***p<.001
a
 Denotes significance in comparison between Column 1 (1993 Enrollment) and Column 2 (1994 Enrollment), p< .001
b
 Denotes significance in comparison between Column 1 (1993 Enrollment) and Column 3 (1995-1996 Enrollment), p< .001
c
 Denotes significance in comparison between Column 1 (1993 Enrollment) and Column 4 (1997 Enrollment), p< .001
d
 Denotes significance in comparison between Column 1 (1993 Enrollment) and Column 5 (1998-2000 Enrollment), p< .001
e
 Denotes significance in comparison between Column 2 (1994 Enrollment) and Column 3 (1995-1996 Enrollment), p< .001
f
 Denotes significance in comparison between Column 2 (1994 Enrollment) and Column 4 (1997 Enrollment), p< .001
g
 Denotes significance in comparison between Column 3 (1995-1996 Enrollment) and Column 4 (1997 Enrollment), p< .001
h
 Denotes significance in comparison between Column 3 (1995-1996 Enrollment) and Column 5 (1998-2000 Enrollment), p< .001
i
 Denotes significance in comparison between Column 4 (1997 Enrollment ) and Column 5 (1998-2000 Enrollment), p< .001
Notes: Data are weighted by normalized F4F2 panel weight.  Continuous variables are standardized
113
In terms of cultural capital, mother?s educational expectations and the presence of 
education materials in the home varied by year of enrollment (Table 4.5).  Year of enrollment 
did not vary by the following variables: parental involvement with the school, peer 
encouragement, and participation in music, art, or dance classes.  While a pattern did not 
emerge, the percentage of students whose mothers had no postsecondary expectations for 
them appears to be lower for those graduates who enrolled in 1993 (5%) than for graduates in 
other enrollment years (over 9%).  Over 93% of all graduates who delayed enrollment had 
educational materials in their home with the exception of those who enrolled in 1997 (73%).   
Table 4.5 also shows the post-high school characteristics of graduates who delay 
enrollment after high school.  In terms of family responsibilities, Table 4 shows that the 
percentages of graduates who were married and/or had a child before enrollment were stable 
and low with the exception of 1997.  Between 6% and 10% of the graduates who delayed 
enrollment in college were married at the time of enrollment, with the exception of those 
graduates who enrolled in 1997 (26%).  Between 4% and 6% of graduates had a child when 
they first enrolled, again with the exception of those graduates who enrolled in 1997 (15%).  
Table 4.5 shows that year of delayed enrollment did not vary by expected income at 
age 30 (collected in 1993), but by actual earnings in 1993.  Post hoc tests with a Bonferoni 
adjustment showed significant differences between 1993 earnings of graduates who first 
enrolled in 1993 and 1993 earnings of high school graduates who first enrolled in each of the 
other years.  Average 1993 earnings were between 1/3 and ? standard deviation lower for 
graduates who enrolled in 1993 than for graduates who enrolled in later years.  
About two-thirds of all graduates who delayed enrollment, regardless of when they 
first enrolled, were working full-time in 1993.  Exploring the occupations that members of 
114
the sample had in 1993 shows variation by year of enrollment.  Table 4.5 shows the most 
represented occupations for all years of delayed enrollment were clerical (30%), service
(19%), and sales (12%).  About 12% of graduates who enrolled in college in 1995-1996 were 
in the military in 1993, a higher percentage than other years of first enrollment.  
Approximately 6% to 7% of graduates who enrolled in 1994, 1997, and 1998-2000 and only 
2% of graduates who enrolled in 1993 reported being in the military as their 1993 
occupation.  
In the 1994 NELS follow-up graduates were asked their expected occupation at age 
30.  Table 4.5 shows that the most popular response was a professional position, reported by 
34% of all respondents who delayed enrollment.  The percent of high school graduates who 
reported in 1993 that they expected to hold a professional position (e.g., medical non-MD, 
engineering, physician, legal) at age 30 ranged from 17% of graduates who enrolled in 
college in 1997 to 40% of high school graduates who first enrolled in college in 1993 or 
between 1998 and 2000.   
Research Question 4: Predictors of Timing of College Enrollment
The fourth research question examines the predictors of enrolling immediately into 
college or not enrolling rather than delaying enrollment.   Multinomial logit analyses show 
that background characteristics such as race, gender, and socioeconomic status are significant 
predictors of enrollment, as are measures of academic preparation and achievement.  None of 
the financial factors are significant predictors of time of enrollment net of other variables.  
Some social capital characteristics are significant predictors of time of enrollment, and only 
two cultural capital characteristics, mother?s postsecondary expectations and peer 
encouragement are significant predictors of time of enrollment. 
115
Table 4.6 shows the odds-ratios for enrolling in college immediately after graduating 
from high school and for not enrolling in college by 2000 relative to delaying college 
enrollment after controlling for all components of the conceptual model.  The model includes 
the background characteristics of race, gender, and socioeconomic status.  Financial 
resources, academic preparation and achievement, social capital, and cultural capital are also 
components of the model.  Including measures of these constructs improves the explanatory 
power of the model but it is not easy to determine goodness of fit in the multinomial logit 
model.  The final model correctly classifies the timing of enrollment for 75% of the cases.   
With regard to research question four, the predictors of timing of enrollment, the 
results show that Asian, Hispanic, and Black high school graduates are less likely than White 
high school graduates to not enroll in college rather than delay enrollment in college when 
controlling for other background characteristics as well as measures of financial resources, 
academic achievement and preparation, social capital, and cultural capital (odds-ratios= 0.19, 
0.39, and 0.54 respectively).  Asian and Hispanic graduates are as likely as White graduates 
to enroll in college immediately after high school graduation after controlling for the above-
mentioned variables.  Black graduates are less likely than White graduates to enroll 
immediately rather than delay enrollment net of other variables (odds ratios= 0.56).  
Women are less likely than men to not enroll in college than to delay enrollment 
(odds-ratio=0.67), but women are as likely as men to enroll immediately after high school 
graduation as they are to delay enrollment after controlling for other variables.  
Socioeconomic status is also a positive predictor of immediate enrollment (odds-ratio=1.40) 
and a negative predictor of not enrolling (odds-ratio=0.57) compared to delaying enrollment, 
116
even after controlling for race, gender, financial resources, academic achievement and 
preparation, social capital, and cultural capital.  
Academic achievement and preparation are important predictors of enrollment timing 
even after controlling for measures of background characteristics, social capital, and cultural 
capital.  Table 4.6 shows that test scores are negatively related to the odds of not enrolling 
(odds-ratio=0.79) rather than delaying enrollment in college net of other variables.  Test 
scores are unrelated to the probability of immediate enrollment rather than delayed 
enrollment.  Table 4.6 shows that the odds of enrolling in college immediately after high 
school graduation rather than delaying enrollment increase with the level of math completed 
(odds-ratio for Algebra 2=2.00; odds-ratio for advanced math=3.41).  Table 4.6 also shows 
that high school graduates who completed at least algebra 2 are less likely than high school 
graduates with lower or no math to not enroll than they are to delay enrollment (odds-ratios= 
0.51 for Algebra 1 and Algebra 2).  The odds-ratio for advanced math on no enrollment is 
comparable in magnitude to the odds-ratios for algebra 1 and algebra 2 but it is not 
significant, likely because of the relatively small percentage of high school graduates who 
both took advanced math and delayed enrollment (Table 4.6).  
The analyses revealed that one measure of social capital is related to not enrolling in 
college rather than delaying enrollment: number of financial aid contacts.  The likelihood of 
not enrolling rather than delaying enrollment decreases only when a high school graduate has 
at least five financial aid contacts compared to zero contacts (odds ratio=0.32).  Parental 
involvement in a student?s education and high school support are also related to time of 
enrollment.  Table 4.6 shows that after controlling for other variables, parental involvement 
in a student?s education is a positive predictor of immediate enrollment rather than delayed
117
enrollment (odds-ratio=1.16).  Table 4.6 also shows that, after controlling for other variables, 
the likelihood of enrolling immediately rather than delaying enrollment is higher (odds-
ratio=1.63) for graduates who report high school support than for high school graduates who 
do not report this support.  The other measures of social capital, number of parents that other 
parents talk to, student-teacher relations, school type, and free and reduced lunch quartile, 
were unrelated to timing of enrollment net of other variables.     
Table 4.6 shows that two measures of cultural capital are related to timing of 
enrollment after controlling for other variables: mother?s expectations for postsecondary 
education and peer encouragement.  The odds of enrolling in college immediately after high 
school graduation rather than delaying enrollment are lower for those whose mothers expect 
them to attain no postsecondary education (odds ratio=0.50), complete two-years of college 
(odds ratio= 0.52), or earn a bachelor?s degree (odds ratio= 0.61) than for those whose 
mothers expect them to attend graduate school, net of other variables.  A graduate is 
substantially more likely to not enroll than to delay enrollment when a graduate?s mother has 
no postsecondary expectations rather than expectations for graduate school attendance (odds-
ratio=3.25).  Table 4.6 shows that, after controlling for other variables, peer encouragement 
is a positive predictor of immediate enrollment compared to delaying enrollment (odds-
ratio=1.78).  Parental involvement with the school, educational materials in the home, and 
participation in art, music, and dance are all unrelated to timing of college enrollment net of 
other variables.  
118
Table 4.6.  Coefficients, standard errors, and odds-ratios for predictors of time of enrollment among 
1992 high school graduates
Immediate Enrollment No Enrollment
Independent Variable B Stand 
Error
Exp(B) B Stand 
Error
Exp(B)
Race/ethnicity
     Asian 0.14 0.19 1.15 -1.69 0.50 0.19***
     Hispanic 0.09 0.15 1.09 -0.96 0.23 0.39***
     Black -0.58 0.12 0.56*** -0.63 0.18 0.54***
     White (ref.)
Gender
     Women 0.19 0.08 1.21 -0.41 0.11 0.67***
     Men (ref.)
SES 0.43 0.40 1.40*** -0.56 0.07 0.57***
Importance of Cost & Aid
     Very Important 0.07 0.13 1.07 0.05 0.18 1.05
     Some Importance 0.05 0.13 1.05 -0.04 0.18 0.97
     Little Importance (ref.)
Tuition -0.01 0.04 0.99 0.14 0.06 1.15
Unemployment 0.05 0.04 1.05 -0.20 0.05 0.98
Test Scores 0.15 0.04 1.16 -0.24 0.07 0.79***
Highest Level of Math
     Advanced Math 1.23 0.19 3.41*** -0.72 0.25 0.49
     Algebra 2 0.69 0.17 2.00*** -0.67 0.19 0.51***
Algebra 1 0.11 0.16 1.12 -0.67 0.17 0.51***
     Missing -1.21 0.27 0.30*** -1.52 0.38 0.22***
     None or other (ref.)
# of Financial Aid Contacts
     One or Two 0.26 0.10 1.30 -0.20 0.13 0.82
     Three or Four 0.21 0.11 1.24 -0.48 0.15 0.62
     At Least Five 0.22 0.15 1.24 -1.13 0.24 0.32***
None (ref.)
Par Involve with Stu Educ 0.15 0.04 1.16*** -0.02 0.06 0.98
# of Parents Talk To 
     Missing -0.55 0.16 0.58*** 0.37 0.24 1.44
     None -0.46 0.20 0.63 0.20 0.30 1.23
     1 or 2 -0.38 0.15 0.69 0.28 0.24 1.32
     3 to 5 -0.19 0.15 0.82 0.23 0.24 1.26
     6 to 10 0.10 0.16 1.11 0.02 0.25 1.02
     11-20 0.24 0.18 1.27 0.49 0.28 1.63
20+(ref.)
Student-Teacher Relations
     Strongly Agree -0.69 0.20 0.93 0.24 0.27 1.27
     Agree -0.29 0.20 0.75 0.20 0.28 1.22
     Disagree -0.27 0.22 0.77 0.09 0.30 1.09
Strongly Disagree (ref.)
High School Support
     Yes 0.49 0.08 1.63*** 0.02 0.11 1.02
No (ref.)
119
Table 4.6.  Coefficients, standard errors, and odds-ratios for predictors of time of enrollment 
among 1992 high school graduates (continued)
Immediate Enrollment No Enrollment
Independent Variable B Stand 
Error
Exp(B) B Stand 
Error
Exp(B)
School Type
     Private 0.49 0.16 1.64 -1.07 0.37 0.34
     Public (ref.)
Free and Reduced Lunch
     1
st
 Quartile (Low %) 0.38 0.20 1.47 0.99 0.37 2.70
     2
nd
  Quartile 0.42 0.21 1.52 0.47 0.18 2.03
     3
rd
 Quartile 0.54 0.24 1.72 0.12 0.17 3.13
4
th
 Quartile (ref.)
Mother?s PSE Expectations
     Missing -1.01 0.13 0.36*** 0.34 0.19 1.40
     None -0.70 0.20 0.50*** 1.18 0.24 3.25***
     Two-Year -0.65 0.12 0.52*** 0.47 0.18 1.60
     Bachelor?s -0.50 0.10 0.61*** 0.12 0.17 1.12
Grad School (ref.)
Parental Involve w/ School -0.02 0.04 0.98 -0.01 0.05 0.98
Peer Encouragement 0.16 0.04 1.78*** 0.12 0.05 1.13
Educ Materials in Home
     Yes 0.35 0.19 1.42 0.75 0.25 2.11
 No (ref.)
Took Art, Music, or Dance
     Almost daily 0.29 0.23 1.34 0.39 0.29 1.47
     1 to 2x Week -0.21 0.15 0.81 -0.39 0.25 0.68
     1x Wk -0.16 0.19 0.85 0.18 0.28 1.20
Never (ref.)
Number of Cases in Analyses 6,825
(change -2 log likelihood) 8,122
Pseudo R
2 
(Cox & Snell) 0.30
Percent Classified Correctly 75.4
Notes: Enrollment status is relative to delayed enrollment in college.
Data weighted by normalized F4F2 panel weight.
***p<.001
Source: Analyses of NELS data
Research Question 5: Socioeconomic Status as a Predictor of Timing of College Enrollment
The fifth and final research question examines the relationship between 
socioeconomic status and timing of enrollment in college.  The multinomial logistic 
regression analyses were conducted using no enrollment as the reference category. Using the 
blocked entry strategy suggested by the conceptual framework, the analyses show the change 
120
in coefficients for socioeconomic status when additional blocks were added to the model 
(Table 4.7).  The analyses reveal that, after controlling only for race/ethnicity and gender, 
1992 high school graduates with higher socioeconomic status are more likely to both 
immediately enroll in college and delay enrollment than to not enroll in college by 2000 
(columns 1 & 2).  Table 4.7 shows the magnitude of the positive effect of socioeconomic 
status is larger for immediate enrollment (odds ratio=3.68) than for delayed enrollment (odds 
ratio=1.97).  Adding measures of financial resources does not change the relationship 
between socioeconomic status and enrollment timing (columns 3 & 4).  Adding measures of
academic preparation and achievement somewhat reduces the magnitude of the positive 
effect of socioeconomic status on immediate enrollment from an odds-ratio of 3.6 to an odds-
ratio of 2.7 (columns 3 & 5).  Adding academic preparation and achievement to the model 
has little impact on the relationship between socioeconomic status and the likelihood of 
delayed enrollment rather than no enrollment.  Also controlling for measures of social capital 
(columns 7 & 8) and cultural capital (columns 9 & 10) has minimal effect on the relationship 
between socioeconomic status and enrollment timing.  Even after controlling for all other 
variables in the model, timing of enrollment continues to vary by socioeconomic status.  As 
socioeconomic status increases so does the likelihood of both enrolling immediately and 
delaying enrollment rather than not enrolling.  The magnitude of the positive effect of 
socioeconomic status continues to be larger for immediate enrollment (odds ratio= 2.45) than 
for delayed enrollment (odds ratio= 1.75) relative to no enrollment, even after controlling for 
other variables.  
121
Table 4.7.  Odds-ratios for predictors of enrollment timing of 1992 high school graduates (See Appendix G for Coefficients and Standard Errors 
for Complete Model)
Background 
Characteristics
Financial 
Resources
Academic Prep and 
Achievement
Social Capital Cultural Capital
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Independent Variable Immed Delay Immed Delay Immed Delay Immed Delay Immed Delay
Race/ethnicity
     Asian 4.68*** 3.68*** 4.08*** 3.04*** 4.38*** 3.85*** 4.68** 3.46*** 6.20*** 5.41***
     Hispanic 2.75*** 2.56*** 2.87*** 2.77*** 3.25*** 2.89*** 3.62*** 3.14*** 2.83*** 2.60***
     Black 1.28 2.24*** 1.23 2.21*** 1.88*** 2.73*** 1.45 2.37*** 1.04 1.88***
     White (ref.)
Gender
     Women 2.03*** 1.40*** 1.94*** 1.37*** 1.91*** 1.37*** 1.73*** 1.33 1.82*** 1.53***
     Men (ref.)
SES 3.68*** 1.97*** 3.62*** 1.97*** 2.66*** 1.81*** 2.52*** 1.76*** 2.45*** 1.75***
Importance of Cost & Aid
     Very Important 1.62*** 1.28 1.33 1.16 0.96 0.94 1.02 0.95
     Some Importance 1.34 1.24 1.19 1.16 1.06 1.05 1.08 1.04
     Little Importance (ref.)
Tuition 0.92 0.89 0.90 0.89 0.86*** 0.95 0.86 0.87
Unemployment 1.11 1.10 1.13 1.11 1.11 1.03 1.07 1.02
Test Scores 1.65*** 1.36*** 1.69*** 1.34*** 1.48*** 1.27***
Highest Level of Math Taken
     Missing 1.27 3.46*** 0.98 3.06*** 1.36 4.57***
     Advanced Math 10.00*** 2.28*** 6.22*** 1.83 7.00*** 2.05
     Algebra 2 5.34*** 2.61*** 4.09*** 2.12*** 3.89*** 1.95***
Algebra 1 2.11*** 2.15*** 2.12*** 2.03*** 2.19*** 1.96***
     None or other (ref.)
122
Table 4.7.  Odds-ratios for predictors of enrollment timing of 1992 high school graduates (continued)
Background 
Characteristics
Financial Resources Academic Prep and 
Achievement
Social Capital Cultural Capital
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Independent Variable Immed Delay Immed Delay Immed Delay Immed Delay Immed Delay
# of Financial Aid Contacts
     One or Two 1.60*** 1.28 1.58*** 1.22
     Three or Four 2.05*** 1.54 2.00*** 1.62
     At Least Five 3.91*** 2.84*** 3.82*** 3.08***
None (ref.)
Par Involve with Stu Educ 1.28*** 1.04 1.18 1.02
# of Parents Talk To 
     Missing 0.44*** 0.72 0.40*** 0.69
     None 0.57 0.86 0.51 0.82
     1 or 2 0.53 0.77 0.52 0.76
     3 to 5 0.70 0.80 0.65 0.79
     6 to 10 0.92 0.83 1.09 0.98
     11-20 0.79 0.58 0.78 0.61
20+ (ref.)
Student-Teacher Relations
     Strongly Agree 0.70 0.78 0.73 0.79
     Agree 0.58 0.80 0.62 0.82
     Disagree 0.70 0.91 0.70 0.92
Strongly Disagree (ref.)
High School Support
     Yes 1.61*** 0.97 1.59*** 0.98
No (ref.)
School Type
     Private 4.86*** 2.82 4.76*** 2.91
     Public (ref.)
123
Table 4.7.  Odds-ratios for predictors of enrollment timing of 1992 high school graduates (continued)
Background 
Characteristics
Financial Resources Academic Prep and 
Achievement
Social Capital Cultural Capital
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Independent Variable Immed Delay Immed Delay Immed Delay Immed Delay Immed Delay
Free and Reduced Lunch
     1
st
 Quartile (Low %) 0.68 0.48 0.54 0.37
     2
nd
  Quartile 0.87 0.59 0.75 0.49
     3
rd
 Quartile 0.48 0.34*** 0.55 0.32
4
th
 Quartile (ref.)
Mother?s PSE Expectations
     Missing 0.26*** 0.71
     None 0.15*** 0.31***
   Two-Year 0.32*** 0.63
     Bachelor?s 0.54*** 0.89
Grad School (ref.) 
Parental Involve w/ School 1.00 1.02
Peer Encouragement 1.04 0.88
Educ Materials in Home
     Yes 0.67 0.47
 No (ref.)
Took Art, Music, or Dance
     Almost daily 0.91 0.68
     1 to 2x Week 1.20 1.48
     1x Wk 0.71 0.84
Never (ref.) 
# of Cases in Analyses 8,472 7,794 7,770 7,365 6,825
(-2 log likelihood) 10,343 10,983 10,107 9,151 8,122
Pseudo R
2 
(Cox & Snell) 0.16 0.14 0.24 0.27 0.30
Percent Classified Correctly 69.6 71.8 73.2 74.5 75.4
Notes: Enrollment status is relative to not enrolling in college.  Continuous variables are standardized.  Data weighted by normalized F4F2 panel weight.  ***p<.001    Source: 
Analyses of NELS data
124
CHAPTER 5
DISCUSSION, CONCLUSIONS, IMPLICATIONS, AND RECOMMENDATIONS 
FOR FUTURE RESEARCH
Introduction
Using a combined conceptual model that is based on what is known about 
traditional student enrollment (Cabrera & La Nasa, 2001; Perna, 2000; St. John, 2003), 
and the National Education Longitudinal Study: 1992-2000 (NELS), this study builds on 
the work of Hearn (1992) to examine the variables that differentiate a student?s decision 
to enroll in a two- or four-year college immediately after high school graduation, delay 
first enrollment, or not enroll within eight years of graduating from high school.  The 
study also explores the relationship between socioeconomic status and timing of college 
enrollment to determine if the socioeconomic status gap that is present in immediate
enrollment is also present in delayed enrollment.  
Specifically, this study addresses the following five research questions:  
1. What percent of high school graduates enroll in college in each of eight years 
after graduation from high school?  What is the appropriate break point 
between immediate and delayed enrollment?
2. How do the characteristics of high school graduates who delay entry into a 
two- or four- year college degree program compare to the characteristics of 
high school graduates who enroll immediately after high school and the 
characteristics of high school graduates who do not enroll within eight years 
of high school graduation?
125
3. For high school graduates who delay enrollment, how does the timing of first 
enrollment vary by student characteristics?
4. What are the predictors of delaying entry into a two- or four-year college 
degree program for up to eight years rather than enrolling immediately after 
high school or not enrolling within eight years?
5. After controlling for other variables, is socioeconomic status related to the 
decision to delay enrollment in college?
The first section of this chapter discusses the findings of the study.  Next, 
conclusions that may be drawn from the results are presented.  Finally, the chapter 
discusses the implications of the study and suggests directions for future research.   
Discussion
Research Question One: Timing of College Enrollment
Previous research does not utilize a consistent definition of immediate versus 
delayed enrollment.  Hearn (1992) defined delayed enrollment as enrollment between one 
and two years of high school graduation.  Other researchers have defined immediate 
college enrollment as enrollment in the fall after high school graduation (Perna, 2000), or 
within two years after high school graduation (Cabrera & La Nasa, 2001).  Based on 
comparisons of the characteristics of students who enroll in 1992, 1993, and 1994, this 
study demonstrates that the end of the fall semester of the year of high school graduation 
is the appropriate break point between immediate and delayed enrollment.
While previous research (e.g., Hearn, 1992) has not examined delaying 
enrollment in college beyond two years after graduation from high school, the results of 
126
this study provide a more complete picture as to the timing of college enrollment within 
eight years of high school graduation.  The majority of 1992 graduates (69%) enrolled in 
college immediately after graduation from high school (i.e., during the 1992 calendar 
year), while only 17% delayed enrollment and 14% did not enroll within eight years of 
graduation from high school.  The results of this study suggest that 1992 high school 
graduates enrolled in college, immediately and after a delay, at a higher rate than 1980 
high school graduates.  Hearn found that 52% of high school graduates enrolled in 
college immediately after high school, while 10% delayed entry into college. 
Research Question Two: Differences in High School Characteristics Based on Time of 
College Enrollment
 Hearn?s (1992) descriptive analyses showed that higher percentages of graduates 
who were Black, and who had lower socioeconomic status, lower academic track, and 
lower test scores delayed entry into college than enrolled immediately after high school 
graduation.  Consistent with Hearn?s findings, this study shows that, compared to high 
school graduates who enrolled immediately, a higher percentage of high school graduates 
who delayed enrollment in college were Black, and had lower socioeconomic status and 
lower levels of academic preparation (as measured by test scores and highest level of 
math taken).  The descriptive analyses also show that higher percentages of graduates 
who delayed enrollment rather than enrolled immediately were men, had no financial aid 
contacts, had parents who spoke to two or fewer other parents, attended public schools, 
and had a mother with no postsecondary expectations.  Compared with high school 
graduates who enrolled immediately, higher percentages of graduates who delayed 
127
enrollment also had parents with lower levels of parental involvement with the student 
and the school.
  In comparing delayed enrollment and no enrollment, the descriptive analyses 
show that higher percentages of graduates who delayed enrollment than of graduates who 
did not enroll were Black and women, took either advanced math or algebra 2, had three 
or more financial aid contacts, and had a mother who expected a bachelor?s or graduate 
degree.  Compared with graduates who did not enroll, graduates who delayed enrollment 
also had higher socioeconomic status, test scores, and parental involvement with the 
student and the school.
When looking at these three groups a pattern emerges.  Graduates who delayed 
enrollment were in between graduates who enrolled immediately and graduates who did
not enroll.  For graduates who delayed enrollment, levels of socioeconomic status, test 
scores, average math coursework, mother?s educational expectations , and parental 
involvement were lower than for graduates who enrolled in high school immediately after 
high school, but higher than for graduates who did not enroll within eight years of 
graduation from high school.  
Research Question Three: Characteristics of Graduates Who Delay Enrollment
When the delayed enrollment group is broken out by year of enrollment, 
descriptive analyses show no clear patterns.  Those who enrolled in 1997 tended to be 
outliers for most of the characteristics, with no clear rationale as to why this occurs.  For
example, race, level of math taken, high school support, and high school control, had 
similar percentages for those graduates who enrolled between 1995 and 2000.  The 
128
distributions vary for variables such as gender, importance of cost and aid, and number of 
financial aid contacts.  All graduates in the delayed group had test scores below the mean.  
The importance of cost and aid and unemployment rates varied by year of enrollment, but 
a pattern of difference did not emerge.  As for post-high school characteristics, low 
percentages of graduates who delayed enrollment were married and had a child.  This is 
consistent with previous research (Teachman & Polonko, 1988), which found that people 
who were not married and did not have children were more likely to enroll in college.  
This study also shows that the majority of those who delayed enrollment worked full-
time after graduating from high school, with large percentages holding jobs in clerical, 
service, and sales fields.  Large portions of graduates who delayed college enrollment 
also aspired to hold a professional position by the age of 30.  
Research Question Four: Predictors of Timing of College Enrollment
Due to a lack of research on delayed enrollment, the conceptual model for this 
study is drawn primarily from combined models of college access developed for 
traditional college enrollment (Cabrera & La Nasa, 2001; Perna, 2000; St. John, 2003).  
These models focus on the decision to enroll in a four-year college immediately after 
high school graduation compared to not enrolling.  In this study, multinomial logit 
analyses examine the relationship between variables measuring student background 
characteristics, financial resources, academic preparation and achievement, social capital, 
and cultural capital and timing of college enrollment.  Timing of college enrollment has 
three categories: enroll in a two-year or four-year college immediately after high schoo l 
graduation in 1992, delay enrollment into a two-year or four-year college, and not enroll
129
in college by 2000.  The multinomial logit analyses show that background characteristics, 
academic preparation and achievement, and some measures of social capital and cultural 
capital were significant predictors of timing of enrollment.  
Race/ethnicity was related to enrollment timing even after controlling for other 
variables.   Asian and Hispanic graduates were as likely as Whites to enroll immediately 
in college as they were to delay enrollment, but less likely than White graduates to not 
enroll than they were to delay enrollment.  Black graduates were both less likely than 
Whites to enroll immediately and not enroll than they were to delay enrollment.  This is 
consistent with findings of Hearn (1992) who found that Black graduates were more 
likely than Whites to delay college enrollment than enroll immediately.  But, these 
findings are different than those of previous researchers comparing immediate enrollment 
to no enrollment.  Cabrera and LaNasa (2001) and Perna (2000) found that Hispanic 
graduates were as likely as White graduates to enroll immediately in a four-year college, 
while Black graduates were more likely than Whites to enroll immediately in a four-year 
college after controlling for other variables.  These different findings may be attributable 
to differences in the dependent variable.  This study considers enrollment in a two-year or 
four-year college or university, while Cabrera and LaNasa and Perna consider only 
enrollment in four-year institutions.  Differences in these findings may also suggest the 
importance of considering timing of enrollment when examining racial/ethnic group 
differences in college enrollment.
While Hearn (1992), in his comparison of immediate enrollment to delayed 
enrollment, found that men were more likely to delay enrollment than women, this study 
found that men and women were as likely to enroll immediately as to delay enrollment in 
130
college.  This study also found that women were more likely than men to delay 
enrollment than to not enroll after controlling for other variables.  These findings, that 
overall women were more likely to enroll in college than men, are consistent with 
previous research on traditional college enrollment.  Such research on immediate versus 
no enrollment at four-year colleges and universities generally shows that women are 
more likely than men to enroll in college immediately rather than not enroll after 
controlling for other variables (Manski & Wise, 1983; Perna, 2000).  
This study shows that graduates who delayed enrollment in college had lower 
socioeconomic status net of other variables than those who enrolled immediately, but 
higher socioeconomic status than graduates who did not enroll.  This relationship 
between enrollment timing and socioeconomic status is consistent with the work of Hearn 
(1992), who found that those who delayed enrollment tended to have lower 
socioeconomic status than those graduates who enrolled immediately.  Other researchers 
(Cabrera & La Nasa, 2001; Perna & Titus, 2004) showed that high school graduates with 
higher incomes and socioeconomic status were more likely to enroll in college 
immediately rather than not enroll within one or two years of high school graduation.  
Previous findings on the impact of financial resources on immediate versus no 
college enrollment highlighted the importance of college cost as a predictor (Kane, 1994; 
Manski & Wise, 1983; Paulsen & St. John, 2002; St. John, 2003).  Contrary to those 
findings, none of the measures of financial resources were significant predictors of 
enrollment timing.  
Measures of academic preparation and achievement were also significant 
predictors of enrollment timing.  The multinomial analyses showed that academic 
131
preparation and achievement were related to the timing of enrollment.  Students with 
above average test scores were less likely to not enroll than they were to delay 
enrollment.  Level of math was related to timing of enrollment as well.  Graduates who 
took at least algebra 2 were more likely than graduates with lower levels of math to enroll 
immediately than to delay, while graduates with at least algebra 1 were less likely to not 
enroll than to delay.  Hearn (1992) also found that academic preparation and achievement 
were positively related to immediate rather than delayed enrollment.  Specifically, Hearn 
found that graduates who took a college preparatory curriculum and/or had grades that 
were mostly A?s and B?s were more likely than other graduates to enroll immediately 
rather than delay enrollment.  Research on traditional college enrollment, which focused 
on immediate enrollment compared to no enrollment (e.g., Adelman, 2002; Cabrera & La 
Nasa, 2001; Perna, 2000; Perna & Titus, in press; Plank & Jordan, 2001), also showed
that academic preparation and achievement were significant positive predictors of 
immediate rather than no college enrollment for high school graduates.  
In terms of measures of social capital, parental involvement in a student?s 
education and high school based support were positive predictors of enrollment timing.  
After controlling for other variables, graduates with higher parental involvement in 
student?s education and greater high school support were more likely than others to enroll 
immediately than they were to delay enrollment.  This finding is in line with research 
surrounding traditional college enrollment, comparing immediate and no enrollment
(Cabrera & La Nasa, 2001; Perna, 2000), which found that graduates with higher levels 
of parental involvement were more likely to enroll in college immediately.  Perna (2000)
132
also found that the lack of high school support was a negative predictor of immediate 
college enrollment.
Financial aid contacts were also related to enrollment timing.  Graduates with at 
least five financial aid contacts were less likely to not enroll by 2000 than to delay 
enrollment, net of other variables.  Cabrera and LaNasa (2001) and Plank and Jordan 
(2001) found financial aid contacts to be a positive predictor of enrolling in college 
immediately after high school rather than not enrolling within one to two years.  
In terms of measures of cultural capital, only mother?s postsecondary expectations 
for her student and peer encouragement were related to timing of college enrollment, net 
of other variables.  Graduates whose mothers expected them to atain less than graduate 
school were less likely to immediately enroll than to delay enrollment, while graduates 
whose mothers did not expect them to enroll in postsecondary education were more likely 
to not enroll than to delay enrollment.  Research examining immediate college enrollment 
also showed that parental expectations were positively related to enrollment in four-year 
colleges (Cabrera & La Nasa, 2001; Perna, 2000).  
Peer encouragement was positively related to the likelihood of immediate college 
enrollment rather than delayed enrollment, but was unrelated to the likelihood of no 
enrollment rather than delayed enrollment.  This finding suggests that as a high school 
graduate moves away from the high school experience, high school peers are no longer 
important to the college enrollment decision.  Prior research on the role of peers is 
inconsistent.  In her study of immediate college enrollment, McDonough (1997) found 
peers to be an important component of the four-year college choice process.  In contrast, 
133
using data from NELS:92/94, Perna (2000) found peers to be unrelated to immediate 
enrollment in a four-year college.  
Research Question Five: Socioeconomic Status as a Predictor of Timing of College 
Enrollment
Like other research (Cabrera & La Nasa, 2001; Perna, 2000; Perna & Titus, 
2004), this study shows that socioeconomic status is an important predictor of timing of 
college enrollment, even after controlling for other variables.  Graduates with above 
average socioeconomic status were more likely than graduates with below average 
socioeconomic status to both enroll immediately in college and delay enrollment than to 
not enroll within eight years of high school graduation.  Socioeconomic status had a 
greater positive impact on the likelihood of immediate than delayed enrollment relative to 
no enrollment.  Controlling for other variables, especially measures of academic 
preparation and achievement, explained only a small portion of the observed 
socioeconomic status differences in enrollment timing.  Similarly, Hearn (1992) found 
that graduates who delayed enrollment tended to have lower socioeconomic status than 
those graduates who enrolled immediately.  Researchers who examined traditional 
college enrollment (Cabrera & La Nasa, 2001; Perna, 2000; Perna & Titus, 2004) also 
found that graduates with higher income and socioeconomic status were more likely to 
enroll immediately in college than to not enroll within one to two years of high school 
graduation.   
134
Conclusions
There are at least six conclusions from this study.  First, the appropriate break 
point between immediate and delayed enrollment is the end of the calendar year that a 
student graduates from high school.  This finding will be helpful for other researchers 
because, until this point, there has been no consistent definition of immediate enrollment, 
due at least in part of a general lack of research on those who delay enrollment.  Future 
research, policy and practice should adopt this definition of immediate versus delayed 
enrollment.  
Second, this study revealed the value of distinguishing among delayed enrollment,
immediate enrollment, and no enrollment.  This research shows that there are differences 
in background characteristics, academic preparation and achievement, social capital, and 
cultural capital based on time of enrollment.  Most existing research on college 
enrollment has been limited to traditional college enrollment patterns.  However, this 
study shows that a substantial minority of high school graduates (17%) delay enrollment 
into college, that students who delay enrollment differ in important respects from other 
students, and that the predictors of delayed enrollment are different than the predictors of 
immediate and no enrollment.
Third, graduates who delay enrollment average fewer resources and weaker 
preparation than graduates who enroll immediately, but average more resources and 
better preparation than graduates who do not enroll.  That is, compared with graduates 
who enrolled immediately, graduates who delayed enrollment had lower levels of 
socioeconomic status, math coursework, parental involvement in the school, high school 
support, peer encouragement, and mother?s expectation for their postsecondary 
135
attainment.  But, compared to graduates who did not enroll by 2000, graduates who 
delayed enrollment had higher socioeconomic status, test scores, and math coursework, 
higher numbers of financial aid contacts, and higher maternal expectations for
postsecondary attendance.  Thus, the results of this study suggest that graduates who 
decide to enroll in college more than six months after graduation from high school differ 
in important respects from graduates who enroll immediately after high school graduation 
and graduates who do not enroll within eight years of graduation from high school.  
Fourth, the combined model of college access for traditional students (Cabrera & 
La Nasa, 2001; Perna, 2000; St. John, 2003) is appropriate for examining the predictors 
of both immediate and delayed enrollment in college relative to no enrollment.  The 
analyses in this study show that measures of background characteristics, academic 
preparation and achievement, social capital, and cultural capital are all predictors of 
enrollment timing.  While research by others (Cabrera & La Nasa, 2001; Perna, 2000; 
Perna & Titus, in press; Plank & Jordan, 2001) on the immediate college enrollment of 
high school graduates showed the benefits of a model that includes concepts from 
economic human capital theory and cultural capital and social capital theories, this study 
shows the benefits of a conceptual model that draws on these theories for understanding 
delayed college enrollment.  Specifically, the multinomial logit analyses show that some 
level of academic, cultural and social resources is required not only to enroll immediately 
into college after graduation from high school (as demonstrated in prior research), but 
also to delay enrollment in college.
Fifth, the multinomial logit analyses show that measures of social and cultural 
capital are related to the timing of college enrollment but seem to be relatively more 
136
important to the decision to enroll immediately than to the decision to delay enrollment.  
This finding suggests that graduates who delay enrollment are less connected with the 
home environment from their youth than graduates who enroll immediately, thereby 
reducing the influence of high-school-based measures of cultural and social capital.  Post-
high school experiences (e.g., marriage, children, home ownership, occupation) may play 
a greater role in the decision to delay enrollment than high school characteristics and 
experiences.  
Finally, even after controlling for other variables, socioeconomic status is related 
to timing of college enrollment.  Consistent with other research, those with lower 
socioeconomic status are less likely than their higher socioeconomic status peers to gain 
access to college at any time.  St. John (2003) found that only 52% of college-qualified, 
low-income students enrolled in college within two years after high school, while 83% of 
high-income students enrolled immediately.  He concluded that family income and lack 
of financial aid influence college-going.  In this study, after controlling for other 
variables, socioeconomic status had a large positive impact on immediate enrollment into 
college, and a smaller but still positive impact on delayed enrollment into college, 
relative to not enrolling.  
The continued statistical significance of socioeconomic status even after 
controlling for other variables suggests that there are barriers to immediate enrollment 
and delayed enrollment based on socioeconomic status that are not explained by this 
study.  While the barriers are greater for immediate than delayed enrollment, the sources 
of the continued barriers to both types of enrollment are not measured by the variables in 
this model.  The lower effect of socioeconomic status on delayed than immediate 
137
enrollment suggests that individuals who delay may engage in some types of behaviors
after leaving high school (e.g., saving money for college, receiving additional guidance or 
direction) that helps mitigate at least some of the barriers to college enrollment that are 
related to socioeconomic status. 
Implications
This study has at least five implications for theory, policy and practice. First, the 
research contributes to theory by testing the applicability of a conceptual model of 
traditional college enrollment to the decision to delay college enrollment.  The findings 
suggest that measures of background characteristics and academic preparation and 
achievement are significant predictors of delaying enrollment.  This study also shows
that, while some measures of social and cultural capital used in traditional models may be 
relevant, more current proxies of these measures including post-high school 
characteristics may be necessary to understand enrollment timing.
Second, the findings from this study provide campus leaders and others at 
colleges and universities with additional information about the characteristics of students 
who delay enrollment.  Specifically, this study shows that high school graduates who 
delay college enrollment have lower socioeconomic status, less academic preparation, 
and lower parental involvement, mother?s expectations, and high school support for 
enrollment than students who enroll in college immediately after graduating from high 
school.  Consistent with research showing that students who delay enrollment are less 
likely than others to persist (Choy, 2002), these findings suggest that students who delay 
enrollment may face greater challenges than students who enroll immediately in terms of 
adjusting to and succeeding in college once enrolled.  In order to promote the academic 
138
success of students who delay enrollment, colleges and universities should consider the 
availability of programs to assist this population, including remedial education, tutoring, 
advising, and academic skills development.  Colleges and universities should also 
consider adapting orientation programs to address the needs of students who delay 
enrollment and with encouraging a peer support system of graduates who delayed 
enrollment.  Research shows that low-income adult students who receive such 
institutional support as advising, peer support programs, and childcare are more likely to 
complete their education (Matus-Grossman & Gooden, 2001).   
Third, the findings from this study suggest that there is a need for greater special 
attention by policy makers and administrators to delayed enrollment behavior, in order to 
be able to improve college access for those who do not enroll immediately, in order to 
increase college access at all levels.  Many policies and practices seem to focus almost 
exclusively on increasing immediate enrollment.  For example, the Indiana 21
st
 Century 
Scholars program provides scholarships only to those who enroll in college within two 
years of high school graduation.  But, this research shows that 17% of graduates delay 
enrollment into college, and that students who delay enrollment rather than enroll 
immediately differ in several respects.  That is, compared with graduates who enroll 
immediately, graduates who delay enrollment have lower levels of socioeconomic status, 
academic preparation, parental involvement, and mother?s expectations for their 
postsecondary attainment.  New policies and programs must focus on providing 
information about college at locations beyond the high school, at places such as social 
service agencies, job sites, shopping malls, and directed mailings.  College preparation 
and remedial programs for those students who have graduated from high school, and did 
139
not enroll in college immediately are also possible options to increase college-going for 
those who delay enrollment, since students who do not enroll in college immediately 
often have lower levels of participation, and lower levels of support for their college 
going. 
Fourth, the finding that graduates who are from low socioeconomic status families 
are more likely to delay enrollment or not enroll than enroll immediately has implications 
for policy and practice.  Controlling for other variables, especially measures of academic 
preparation and achievement, explains only a small portion of the observed 
socioeconomic status differences in enrollment timing.  Since Hearn?s (1992) study 
documenting socioeconomic status as a predictor of delaying enrollment in college,
college enrollment rates have increased for all groups, but the enrollment gap between 
low- and high- socioeconomic status students is still present.  The results of this study 
show that the likelihood of both enrolling immediately and delaying enrollment increase 
with socioeconomic status.  These results suggest the continued need for policies and 
practices that address the barriers to both types of enrollment for high school graduates 
with low socioeconomic status.  
One example of a state program that addresses the college access issue is the 
Indiana 21
st
 Century Scholars program.  Lower-income students in the state with a 2.0 
grade point average who receive a high school diploma and agree to be alcohol and drug 
free can participate in a college preparation program and receive a full tuition award to an 
Indiana public college or university or the comparable amount the attend a private 
institution within the state, if the student enrolls within two years of graduating from high 
140
school.  For the 1999-2000 year, over 5,554 scholarships were awarded (Twenty-first 
century scholars 1990-2001 progress report, 2002).  
Higher education institutions should also partner with schools and communities 
serving low-socioeconomic status graduates to sponsor outreach programs for students 
and parents that provide information about, and preparation for, college.  Programs like 
The College Opportunity and Career Help (C.O.A.C.H.) program is an example of a 
partnership that pairs Harvard students, faculty, and staff with Bosto n Public School 
students to provide information and assistance with the college admission process 
(College Opportunity and Career Help Program, 2005).  In order to benefit students, it is 
recommended that college preparation programs such as these be structured as long term 
programs that create a supportive peer group, involve parents, and focus on academic 
preparation (Gandara & Bial, 2001; Swail & Perna, 2002).  
Another example of a collaborative effort between an institution and the 
community in order to educate women who did not enroll in college immediately after 
high school is the Women in Community Development (WICD) program, established in 
1997 at Wellesley College.  This Boston-based collaborative program works within the 
confines of welfare reform to provide higher education access to low-income women.  
The goals of the program are to provide access to women so that they can achieve an 
undergraduate degree and increase their opportunities in the workforce.  Program 
components include peer support, financial assistance, academic guidance, job referral, 
and leadership training (Marx, 2002).  While this program has been successful at 
graduating students who were part of the program, this is a small program (less than 10 
students), so does not impact a large group of people.  A recommendation is for 
141
programs such as this one to be replicated in order to increase the number of people who 
will gain access to college.
Fifth, a potential explanation for the failure of a substantial minority of students to 
not enroll immediately in college after high school graduation is the ?disconnect? 
between K-12 and higher education (A shared agenda: A leadership challenge to improve 
college access and success., 2004).  The findings also suggest that students who are 
relatively less prepared and with relatively fewer resources are most affected by the 
disconnect.  From her work with The Bridge Project, Venezia (2003) recommended that 
there needs to be better alignment in the K-16 system between state assessment tests,
college placement tests, and coursework so that students are better able to understand 
what is needed to academically prepare for college.  
Since individuals with a college education benefit not only in terms of higher
personal economic returns but also in terms of cognitive learning, emotional and moral 
development, and good health, it is in the best interest of society to promote college 
enrollment at all ages (Bowen, 1997).  Bowen concluded that educational attainment also 
has generational benefits for the children of these individuals.  In other words, when a 
parent attains a college degree, his/her children are more likely to enroll in college in the 
future. Bowen also concluded that society benefits from individual participation in 
education through increased productivity, and a decreased need for public assistance and 
decreased unemployment.  Funding and implementing policies to promote enrollment 
and completion for graduates who enroll immediately as well as delay enrollment will 
result in numerous benefits for individuals and society.  Policymakers may be able to 
promote the use of resources for higher education, by highlighting the benefits of higher 
142
education in order to improve economic development, and to provide greater stability in 
society (IHEP, 2005).  
Recommendations for Future Research
The findings of this study add to a limited body of knowledge on delayed 
enrollment since graduates who delayed enrollment have not been studied since Hearn?s 
(1992) work.  This study demonstrates that the appropriate break point in differentiating
between immediate and delayed enrollment is the end of the calendar year that a student 
graduates from high school.  This study shows the importance of using a combined model 
of college enrollment which includes background characteristics, academic preparation 
and achievement, financial resources, social capital, and cultural capital to examine the 
predictors of delaying college enrollment.  While Hearn?s (1992) model focused 
primarily on background characteristics and academic preparation and achievement, this 
research also explored the constructs of financial resources, social capital, and cultural 
capital.  Measures of background characteristics, academic preparation and achievement, 
social capital, and cultural capital were related to timing of enrollment.
This study suggests at least four recommendations for future research.  First, 
future research should further explore at least three aspects of the predictors of 
enrollment timing.  This study found no relationship between financial factors and timing 
of enrollment in the multivariate analyses, but past research (Kane, 1994; Perna, 2000; 
Schwartz, 1985; St. John, 2003) found that higher tuition is negatively related to 
immediate college enrollment even after controlling for other variables.  The descriptive 
analyses show that, regardless of timing, approximately 90% of graduates found cost and 
143
aid to have at least some importance in their decision to enroll in college.  These results 
suggest that the importance of cost and aid were unrelated because they are a high 
priority for all three categories of enrollment.  Additional measures of financial resources, 
such as actual financial aid received, perceived cost and perceived availability of aid, 
impact of college cost increases, availability of tuition remission from the workplace, and 
actual cost of college attendance may provide more insight into the relationship between 
financial resources and enrollment timing. 
Future research should also examine the effects on enrollment timing of post-high 
school characteristics and experiences (e.g., marriage, children, occupation, income).  
The descriptive analyses revealed no clear patterns except that rates of marriage and 
childbirth are generally low for individuals who delay enrollment.  Further exploration of 
the relationship between post-high school characteristics and enrollment timing will 
likely enhance understanding of the predictors of delayed college enrollment.  
Further research should also develop and examine additional and more complete 
proxies for social capital and cultural capital. This study found measures of the home 
environment and high-school environment are relatively less important to the decision to 
delay enrollment than the decision to enroll immediately suggesting that post-high school 
measures of social and cultural capital, such as the influence of the world of work after 
high school, may be required to more completely understand enrollment timing.  Finding 
appropriate measures in the NELS dataset of these complex constructs is challenging.  
Many of the available proxies in NELS look at the quantity of interactions and not the 
quality, while other proxies measure ?valued? forms of capital that may not be seen as 
equally important by all groups (Stanton-Salazar, 1997). Future research should identify 
144
and explore additional measures of cultural and social capital, particularly measures for 
post-high school experiences and measures that go beyond quantity to indicate quality.
Second, future research should further explore delayed enrollment through 
qualitative research methods.  Research that includes interviews or focus groups of high 
school graduates who delay enrollment would allow researchers to explore predictors of 
enrollment including quality and specifics about interactions with parents, teachers, and 
school counselors, a limitation of this study.  A qualitative study would also allow for a 
broader study that may capture the importance of other predictors of timing of college 
enrollment that were not a part of the conceptual model that was tested.
Third, future research should study a more current group of high school graduates, 
and also include GED recipients.  This study examined students who graduated from high 
school over ten years ago, and there have been many changes in society since that point.  
The changing economy and world events such as September 11
th
 may limit the 
generalizability of findings from this study to later cohorts.  Characteristics of the
economy during the 1990s may have motivated some graduates to enroll in college in 
order to receive skills to get a better job with a higher rate of return, but changes in 
college costs may have deterred the immediate and delayed enrollment of other
graduates.  World events may have impacted the number of graduates entering the 
military rather than enrolling immediately in college, which may lead to larger numbers 
of graduates who delay college enrollment.
Finally, further research should further explore the enrollment characteristics of 
the delayed enrollment group.  While the characteristics of delayed enrollment (e.g., type 
of institution, full-time or part-time) were mentioned briefly, attention to these 
145
characteristics was not a primary focus of this dissertation.  These descriptive analyses 
show that a higher percentage of students who delayed enrollment than of students who 
enrolled immediately choose to attend a two-year public institution.  Nonetheless, the 
causes of this finding were not explored. More research is also needed about the 
experience of students who delay enrollment after they enroll, including their persistence 
to degree completion.  Such research will generate a more complete picture of this 
enrollment group and suggest additional approaches for promoting college access and 
success for all students.
146
APPENDIX
Appendix A.  Comparison of 1992 high school graduates who first enrolled in fall 1992 
and spring 1993 by selected student characteristics
Student Characteristics 1992 1993 Total
Race*** 100.0 100.0 100.0
     APA 5.0 5.4 5.1
     Hispanic 7.8 10.4 7.9
     Black 8.6 22.7 9.4
     White 78.0 61.5 77.1
     Amer. Indian/AK Native 0.6 0.0 0.5
Gender 100.0 100.0 100.0
     Male 47.8 43.5 47.6
     Female 52.2 56.5 52.4
Socioeconomic Status***
      Mean  0.247 -0.366 0.214
Reading and Math IRT Score ***
      Mean    0.260 -0.306 0.229
Highest Level of Math Coursework*** 100.0 100.0 100.0
     Algebra 1 22.5 42.9 23.6
     Algebra 2 36.2 26.2 35.7
     Advanced Math 37.1 8.9 35.6
     Missing, other or unknown math 4.2 22.0 5.1
Parental Involvement w/ Child***
      Mean 0.182 -0.166 0.172
Number of Parents Other Parents Talk To 100.0 100.0 100.0
     None 5.4 3.9 5.4
     1 or 2 18.6 25.8 18.9
     3 to 5 28.4 33.2 28.6
     6 to 10 22.4 21.0 22.3
     11-20 12.6 10.0 12.5
     20+ 12.6 6.1 12.3
Positive Student-Teacher Relations 100.0 100.0 100.0
     Strongly Disagree 2.6 3.3 2.6
     Disagree 7.3 10.7 7.5
     Agree 21.2 19.5 21.1
     Strongly Agree 68.9 66.4 68.8
147
Appendix A.  Comparison of 1992 high school graduates who first enrolled in fall 
1992 and spring 1993 by selected student characteristics (continued)
Student Characteristics 1992 1993 Total
High School Based Support 100.0 100.0 100.0
     No 31.7 36.1 31.9
     Yes 68.3 63.9 68.1
High School Control*** 100.0 100.0 100.0
     Public 88.5 96.5 88.9
     Private 11.5 3.5 11.1
Free and Reduced Lunch Quartile 100.0 100.0 100.0
     First 72.9 69.3 72.7
     Second 20.0 20.6 20.0
     Third 4.9 5.6 4.9
     Fourth 2.3 4.6 2.4
Mother?s Highest Level of Education Expected*** 100.0 100.0 100.0
     No postsecondary Expectations 2.8 4.7 2.9
     2-year academic or technical college 12.1 20.8 12.5
     Bachelor?s 46.8 52.7 47.1
    Graduate or Professional Degree 38.3 21.9 37.4
Peer Encouragement
      Mean 0.079 -0.009 0.075
Parental Involvement w/ School
      Mean 0.024 -0.073 0.020
Educational Materials in the Home 100.0 100.0 100.0
     No 2.3 3.6 2.3
     Yes 97.7 96.4 97.7
Takes Music, Art or Dance Classes 100.0 100.0 100.0
     Never 84.4 85.8 84.5
     Less than 1 time a week 4.4 5.3 4.4
     1 to 2 times a week 8.5 4.6 8.3
     Almost very day 2.7 4.3 2.8
Source: Analyses of NELS:92/2000
Note: Data are weighted by normalized F4F2 panel weight
***p<.001
148
Appendix B.  Comparison of 1992 high school graduates who first enrolled in college in 
spring 1993 versus academic year 1993-1994 enrollment by selected student 
characteristics
Student Characteristics Sp 1993 1993-
1994
Total
Race*** 100.0 100.0 100.0
     APA 5.4 3.8 4.4
     Hispanic 10.4 10.3 10.3
     Black 22.7 11.9 15.7
     White 61.5 72.6 68.6
     Amer. Indian/AK Native 0.0 1.4 0.9
Gender 100.0 100.0 100.0
     Male 43.5 54.5 50.6
     Female 56.5 45.5 49.4
Socioeconomic Status
      Mean -0.366 -0.190 -0.252
Reading and Math IRT Score 
      Mean -0.306 -0.255 -0.274
Highest Level of Math Coursework 100.0 100.0 100.0
     Algebra 1 42.9 48.3 46.4
     Algebra 2 26.2 29.1 28.1
     Advanced Math 8.9 11.5 10.6
     Missing, other, or unknown math 22.0 11.1 14.9
Parent Involvement w/ Child
      Mean -0.166 -0.256 -0.172
Number of Parents Other Parents Talk To 100.0 100.0 100.0
     None 3.9 8.3 6.9
     1 or 2 25.8 25.8 25.8
     3 to 5 33.2 27.6 29.4
     6 to 10 21.0 16.6 18.1
     11-20 10.0 9.4 9.6
     20+ 6.1 12.3 10.2
Positive Student-Teacher Relations 100.0 100.0 100.0
     Strongly Disagree 3.3 6.4 5.3
     Disagree 10.7 10.9 10.9
     Agree 19.5 26.1 23.8
     Strongly Agree 66.4 56.5 60.0
149
Appendix B.  Comparison of 1992 high school graduates who first enrolled in college 
in spring 1993 versus academic year 1993-1994 by selected student characteristics
(continued)
Student Characteristics Sp 1993 1993-1994 Total
High School Based Support*** 100.0 100.0 100.0
     No 36.1 52.6 46.7
     Yes 63.9 47.4 53.3
High School Control
  Public 96.5 93.5 94.6
     Private 3.5 6.5 5.4
Free and Reduced Lunch Quartile 100.0 100.0 100.0
     First 69.3 66.7 67.6
     Second 20.6 23.8 22.7
     Third 5.6 5.9 5.8
     Fourth 4.6 3.6 3.9
Mother?s Highest Level of Education Expected 100.0 100.0 100.0
     No postsecondary Expectations 4.7 5.4 5.1
     2-year academic or technical college 20.8 22.3 21.8
     Bachelor?s 52.7 51.4 51.8
     Graduate or Professional Degree 21.9 20.9 21.2
Parental Involvement w/ School
      Mean -0.073 0.052 0.010
Peer Encouragement
      Mean -0.009 -0.199 -0.132
Educational Materials in the Home 100.0 100.0 100.0
No 3.6 7.1 5.8
Yes 96.4 92.9 94.2
Takes Music, Art or Dance Classes 100.0 100.0 100.0
Never 85.8 86.8 86.5
Less than 1 time a week 5.3 3.1 3.9
1 to 2 times a week 4.6 8.1 6.9
Almost very day 4.3 2.0 2.8
Source: Analyses of NELS:92/2000
***p<.001
Note: Data are weighted by normalized F4F2 panel weight
150
Appendix C.  Comparison of 1992 high school graduates who first enrolled in college in 
1995 and 1996 calendar years by selected student characteristics.
Student Characteristics 1995 1996 Total
Race*** 100.0 100.0 100.0
     APA 1.3 10.4 4.0
     Hispanic 7.5 11.9 8.8
     Black 26.9 6.0 20.7
     White 64.4 65.7 64.8
     Amer. Indian/AK Native 0.0 6.0 1.8
Gender 100.0 100.0 100.0
     Male 65.6 65.2 65.5
     Female 34.4 34.8 34.5
Socioeconomic Status
      Mean -0.131 -0.486 -0.239 
      Standard deviation  0.759  0.938  0.831 
Reading and Math IRT Score 
      Mean -0.631 -0.508 -0.591 
      Standard deviation  0.970  0.925  0.955 
Highest Level of Math Coursework 100.0 100.0 100.0
     Algebra 1 55.6 52.2 54.6
     Algebra 2 25.2 33.3 27.6
     Advanced Math 12.5 2.9 9.6
     Missing, other or unknown math 6.7 11.6 8.2
Parent Involvement w/ Child
      Mean -0.333 -0.428 -0.363 
      Standard deviation  1.032  0.848  0.976 
Number of Parents Other Parents Talk To*** 100.0 100.0 100.0
 None 14.1 7.3 12.3
     1 or 2 18.8 20.0 19.1
     3 to 5 52.3 25.5 45.1
     6 to 10 6.0 32.7 13.2
     11-20 3.4 7.3 4.4
     20+ 5.4 7.3 5.9
Positive Student-Teacher Relations 100.0 100.0 100.0
     Strongly Disagree 10.1 1.5 7.5
     Disagree 12.0 14.7 12.8
     Agree 21.5 30.9 24.3
     Strongly Agree 56.3 52.9 55.3
151
Appendix C.  Comparison of 1992 high school graduates who first enrolled in college in 
1995 and 1996 calendar years by selected student characteristics (continued)
Student Characteristics 1995 1996 Total
High School Based Support 100.0 100.0 100.0
     No 57.7 50.0 55.3
     Yes 42.3 50.0 44.7
High School Control 100.0 100.0 100.0
     Public 98.1 91.3 96.1
     Private 1.9 8.7 3.9
Free and Reduced Lunch Quartile 100.0 100.0 100.0
     First 65.4 64.7 65.2
     Second 26.4 23.5 25.6
     Third 6.3 7.4 6.6
     Fourth 1.9 4.4 2.6
Mother?s Highest Level of Education Expected 100.0 100.0 100.0
     No postsecondary Expectations 8.8 10.2 9.3
     2-year academic or technical college 30.1 40.7 33.7
     Bachelor?s 44.2 25.4 37.8
    Graduate or Professional Degree 16.8 23.7 19.2
Parental Involvement w/ School
      Mean -0.120 -0.321 -0.176 
      Standard deviation  1.084  0.663  0.988 
Peer Encouragement
      Mean -0.026 -0.428 -0.146 
      Standard deviation  0.844  1.101  0.944 
Educational Materials in the Home 100.0 100.0 100.0
     No 3.7 6.2 4.5
     Yes 96.3 93.8 95.5
Takes Music, Art or Dance Classes 100.0 100.0 100.0
     Never 91.7 83.1 89.2
     Less than 1 time a week 4.5 3.1 4.1
     1 to 2 times a week 0.6 6.2 2.3
     Almost very day 3.2 7.7 4.5
Source: Analyses of NELS:92/2000
***p<.001
Note: Data are weighted by normalized F4F2 panel weight
152
Appendix D. Comparison of 1992 high school graduates who first enrolled in college in 
1996 and 1997 calendar years by selected student characteristics.
Student Characteristics 1996 1997 Total
Race*** 100.0 100.0 100.0
     APA 10.4 0.0 4.0
     Hispanic 11.9 15.5 14.1
     Black 6.0 10.9 9.0
     White 65.7 73.6 70.6
     Amer. Indian/AK Native 6.0 0.0 2.3
Gender*** 100.0 100.0 100.0
     Male 65.2 38.2 48.6
     Female 34.8 61.8 51.4
Socioeconomic Status
      Mean -0.486 -0.806 -0.682 
      Standard deviation  0.938  0.930  0.943 
Reading and Math IRT Score 
      Mean -0.508 -0.670 -0.607 
      Standard deviation  0.925  0.962  0.948 
Highest Level of Math Coursework 100.0 100.0 100.0
     Algebra 1 52.2 56.4 54.7
     Algebra 2 33.3 29.1 30.7
     Advanced Math 2.9 6.4 5.0
     Missing, other or unknown math 8.7 8.1 9.6
Parental Involvement w/ Child***
      Mean -0.428 -0.918 -0.733 
      Standard deviation  0.848  0.931  0.929 
Number of Parents Other Parents Talk To 100.0 100.0 100.0
     None 7.3 20.5 14.8
     1 or 2 20.0 26.0 23.4
     3 to 5 25.5 24.7 25.0
     6 to 10 32.7 12.3 21.1
     11-20 7.3 8.2 7.8
     20+ 7.3 8.2 7.8
Positive Student-Teacher Relations 100.0 100.0 100.0
     Strongly Disagree 1.5 0.9 1.1
     Disagree 14.7 18.7 17.1
     Agree 30.9 14.0 20.6
     Strongly Agree 52.9 66.4 61.1
153
Appendix D.  Comparison of 1992 high school graduates who first enrolled in college in 
1996 versus 1997 by selected student characteristics. (continued)
Student Characteristics 1996 1997 Total
High School Based Support 100.0 100.0 100.0
     No 50.0 39.3 43.6
     Yes 50.0 60.7 56.4
High School Control 100.0 100.0 100.0
     Public 91.3 99.1 96.1
     Private 8.7 0.9 3.9
Free and Reduced Lunch Quartile 100.0 100.0 100.0
     First 64.7 67.0 66.1
     Second 23.5 13.8 17.5
     Third 7.4 17.4 13.6
     Fourth 4.4 1.8 2.8
Mother?s Highest Level of Education Expected 100.0 100.0 100.0
     No postsecondary Expectations 10.2 15.9 13.1
     2-year academic or technical college 40.7 31.7 36.1
     Bachelor?s 25.4 38.1 32.0
    Graduate or Professional Degree 23.7 14.3 18.9
Parental Involvement w/ School
      Mean -0.321 -0.199 -0.252 
      Standard deviation  0.663  0.911  0.812 
Peer Encouragement
      Mean -0.428 -0.229 -0.304 
      Standard deviation  1.101  1.107  1.106 
Educational Materials in the Home*** 100.0 100.0 100.0
     No 6.2 26.9 19.1
     Yes 93.8 73.1 80.9
Takes Music, Art or Dance Classes 100.0 100.0 100.0
     Never 83.1 86.8 85.4
     Less than 1 time a week 3.1 3.8 3.5
     1 to 2 times a week 6.2 9.4 8.2
     Almost very day 7.7 0.0 2.9
Source: Analyses of NELS:92/2000
***p<.001
Note: Data are weighted by normalized F4F2 panel weight
154
Appendix E. Comparison of 1992 high school graduates who first enrolled in college in 
1998, 1999, and 2000 calendar years by selected student characteristics
Student Characteristics 1998 1999 2000 Total
Race 100.0 100.0 100.0 100.0
     APA 1.2 3.4 0.0 2.0
     Hispanic 14.3 5.2 9.1 10.5
     Black 26.2 15.5 27.3 22.2
     White 57.1 74.1 63.6 64.1
     Amer. Indian/AK Native 1.2 1.7 0.0 1.3
Gender 100.0 100.0 100.0 100.0
     Male 67.9 51.7 63.6 61.4
     Female 32.1 48.3 36.4 38.6
Socioeconomic Status
Mean -0.364 -0.586 -0.976 -0.491 
      Standard deviation  0.791  0.722  0.827  0.782 
Reading and Math IRT Score 
      Mean -0.431 -0.623 -0.308 -0.495 
      Standard deviation  0.725  0.895  0.914  0.808 
Highest Level of Math Coursework 100.0 100.0 100.0 100.0
     Algebra 1 40.5 51.7 63.6 46.4
     Algebra 2 25.0 25.9 36.4 26.1
     Advanced Math 23.8 1.7 0.0 13.7
     Missing, other or unknown math 10.7 20.7 0.0 13.8
Parent Involvement w/ Child
      Mean -0.013 -0.382 -0.418 -0.176 
      Standard deviation  1.171  0.989  1.167  1.113 
Number of Parents Other Parents Talk To 100.0 100.0 100.0 100.0
     None 2.9 22.5 0.0 9.2
     1 or 2 42.9 17.5 22.2 32.8
     3 to 5 27.1 25.0 22.2 26.1
     6 to 10 15.7 17.5 22.2 16.8
     11-20 5.7 12.5 22.2 9.2
     20+ 5.7 5.0 11.1 5.9
Positive Student-Teacher Relations 100.0 100.0 100.0 100.0
     Strongly Disagree 2.5 0.0 10.0 2.1
     Disagree 8.6 10.9 10.0 9.6
   Agree 43.2 20.0 10.0 32.2
     Strongly Agree 45.7 69.1 70.0 56.2
155
Appendix E. Comparison of 1992 high school graduates who first enrolled in college 
in 1998, 1999, and 2000 calendar years by selected student characteristics (continued)
Student Characteristics 1998 1999 2000 Total
High School Based Support 100.0 100.0 100.0 100.0
     No 44.4 52.9 75.0 49.3
     Yes 55.6 47.1 25.0 50.7
High School Control 100.0 100.0 100.0 100.0
     Public 98.8 96.6 100.0 98.0
     Private 1.2 3.4 0.0 2.0
Free and Reduced Lunch Quartile 100.0 100.0 100.0 100.0
     First 46.9 61.1 60.0 53.1
     Second 18.5 31.5 20.0 23.4
     Third 13.6 7.4 0.0 10.3
     Fourth 21.0 0.0 20.0 13.1
Mother?s Highest Level of Education 
Expected
100.0 100.0 100.0 100.0
     No postsecondary Expectations 9.3 6.8 12.5 8.5
     2-year academic or technical college 31.5 29.5 37.5 31.1
     Bachelor?s 40.7 34.1 25.0 36.8
    Graduate or Professional Degree 18.5 29.5 25.0 23.6
Parental Involvement w/ School
      Mean  0.531 -0.152 -0.053  0.250 
      Standard deviation  1.525  0.855  0.746  1.137 
Peer Encouragement
      Mean -0.288 -0.108  0.193 -0.189 
      Standard deviation  0.929  1.064  0.858  0.979
Educational Materials in the Home 100.0 100.0 100.0 100.0
     No 2.4 1.9 18.2 3.4
     Yes 97.6 98.1 81.8 96.6
Takes Music, Art or Dance Classes 100.0 100.0 100.0 100.0
     Never 80.2 90.7 100.0 85.5
     Less than 1 time a week 4.9 0.0 0.0 2.8
     1 to 2 times a week 14.8 7.4 0.0 11.0
     Almost very day 0.0 1.9 0.0 0.7
Source: Analyses of NELS:92/2000
***p<.001 Note: Data are weighted by normalized F4F2 panel weight
156
Appendix F.  Comparison of 1992 high school graduates who first enrolled in college in 
1992, 1993, and 1994 calendar years by selected student characteristics
Student Characteristics 1992 1993 1994 Total
Race*** 
a
100.0 100.0 100.0 100.0
APA 5.0 4.3 3.5 4.2
Hispanic 8.0 10.7 9.4 8.6
Black 8.5 16.9 12.4 10.5
White 78.0 67.1 73.3 75.9
Amer. Indian/AK Native 0.6 0.9 1.5 0.9
Gender*** 100.0 100.0 100.0 100.0
Male 47.7 51.7 49.8 50.4
Female 52.3 48.3 50.2 49.6
SES***
 a
     Mean 0.236 -0.247 -0.326 0.002
     Standard deviation 0.950 0.959 1.054 0.999
Importance of Cost and Aid*** 100.0 100.0 100.0 100.0
Little Importance 10.5 7.1 8.8 10.5
Some Importance 42.6 44.7 30.8 42.0
     Very Important 46.9 48.2 60.4 47.6
Two-Year Tuition
      Mean 0.002 -0.366 -0.154 0.004
      Standard deviation 1.018 0.956 1.009 0.997
Unemployment Rate***
      Mean 0.020 0.010 0.072 -0.003
      Standard deviation 1.027 0.884 0.880 0.999
Test Score ***
 a
     Mean 0.248 -0.273 -0.372 0.005
      Standard deviation 0.937 0.855 0.923 0.999
Highest Math Coursework***
 a
100.0 100.0 100.0 100.0
Algebra 1 22.8 45.7 52.0 30.9
Algebra 2 36.1 28.4 26.6 32.2
Advanced Math 36.6 11.0 6.9 27.7
Missing or Unknown 1.0 7.4 1.0 1.9
Other or None 3.8 7.5 14.5 7.9
157
Appendix F.  Comparison of 1992 high school graduates who first enrolled in college 
in 1992, 1993, and 1994 calendar years by selected student characteristics
(continued)
Student Characteristics 1992 1993 1994 Total
# of Fin Aid Contacts***
a
100.0 100.0 100.0 100.0
None 16.5 22.0 20.2 19.7
One or Two 29.3 33.6 39.9 31.0
Three or Four 41.1 35.0 30.6 37.6
At least five 13.1 9.4 9.3 11.7
Parental Involvement w/ Stu dent***
a
      Mean 0.025 0.019 -0.011 -0.001
      Standard deviation 0.931 0.871 0.898 0.930
# of Parents Talk To***
 a
100.0 100.0 100.0 100.0
None 5.5 6.9 8.0 6.6
1 or 2 18.6 26.0 29.4 20.6
3 to 5 28.8 28.5 33.1 29.4
6 to 10 22.2 18.8 10.4 20.9
11-20 12.4 9.7 9.8 11.4
20+ 12.5 10.1 9.2 11.1
+ Student-Teacher Relations***
 a
100.0 100.0 100.0 100.0
Strongly Disagree 2.6 4.9 7.0 3.1
Disagree 7.3 10.5 11.4 8.6
Agree 21.0 23.9 24.4 22.5
Strongly Agree 69.1 60.7 57.2 65.8
HS Based Support***
 a
100.0 100.0 100.0 100.0
No 32.0 46.2 55.9 37.1
Yes 68.0 53.8 44.1 62.9
High School Control***
 a
100.0 100.0 100.0 100.0
Public 88.8 94.1 98.5 91.4
Private 11.2 5.9 1.5 8.6
Free & Reduced Lunch Quart***
 a
100.0 100.0 100.0 100.0
1
st
(Low Participation) 72.7 67.7 62.0 69.7
2
nd
19.9 22.6 26.0 21.2
3
rd
5.2 5.6 4.7 6.3
4
th
 (High Participation) 2.2 4.2 7.3 2.7
158
Appendix F.  Comparison of 1992 high school graduates who first enrolled in college 
in 1992, 1993, and 1994 calendar years by selected student characteristics
(continued)
Student Characteristics 1992 1993 1994 Total
Mother?s Educ Expectations***
 a
100.0 100.0 100.0 100.0
No Expectations 2.8 4.7 8.5 6.0
2-year Institution 12.5 21.7 24.2 16.9
Bachelor?s 46.7 52.9 46.4 45.0
Grad or Prof Degree 38.0 20.7 20.9 32.1
Parental Involvement w/ School***
      Mean 0.160 -0.125 -0.204 0.004
      Standard deviation 0.894 0.913 0.872 0.939
Peer Encouragement***
a
     Mean 0.076 -0.117 -0.323 0.172
     Standard Deviation 0.981 0.988 1.096 1.22
Educ Materials in the Home***
 a
100.0 100.0 100.0 100.0
No 2.2 5.9 7.2 3.5
Yes 97.8 94.1 92.8 96.5
Particpate in the Arts*** 100.0 100.0 100.0 100.0
Never 84.1 86.3 88.8 85.3
Less than 1 time a week 4.5 3.7 4.3 4.2
1 to 2 times a week 8.4 7.3 4.3 7.4
Almost very day 3.0 2.7 2.7 3.1
Source: Analyses of NELS:92/2000
***p<.001
a
 Denotes significance in comparison between Column 1 (Enrollment in 1992) and 
Column 2 (Enrollment in 1993), p< .001
Note: Data are weighted by normalized F4F2 panel weight
All continuous variables are standardized
159
Appendix G.  Odds-ratios, co-efficients, and standard errors based on time of enrollment, 
with no enrollment as the baseline group
Immediate Enrollment Delayed Enrollment
Independent Variable B Stand
Error
Exp(B) B Stand
Error
Exp(B)
Race/ethnicity
Asian 1.83 0.48 6.20*** 1.69 0.50 5.41***
Hispanic 1.04 0.22 2.83*** 0.96 0.23 2.60***
Black 0.04 0.17 1.04 0.63 0.18 1.89***
White (ref.)
Gender
Women 0.60 0.10 1.82*** 0.41 0.11 1.51***
Men (ref.)
SES 0.90 0.07 2.45*** 0.56 0.07 1.75***
Importance of Cost and Aid
Very Important 0.02 0.17 1.02 -0.05 0.18 0.95
Some Importance 0.08 0.17 1.08 0.04 0.18 1.04
Little Importance (ref.)
Tuition -0.15 0.05 0.86 -0.14 0.06 0.87
Unemployment 0.07 0.05 1.07 0.02 0.05 1.02
Test Scores 0.39 0.06 1.48*** 0.24 0.07 1.27***
Highest Level of Math
Adv Math 1.95 0.23 7.00*** 0.72 0.25 2.05
Algebra 2 1.36 0.18 3.90*** 0.67 0.19 1.95***
Algebra 1 0.78 0.16 2.19*** 0.67 0.17 1.96***
     Missing 0.31 0.40 1.36 1.52 0.38 4.57***
None or other (ref.)
# of Financial Aid contacts
One or Two 0.46 0.13 1.58*** 0.20 0.13 1.22
Three or Four 0.70 0.14 2.00*** 0.48 0.15 1.62
At Least Five 1.34 0.23 3.82*** 1.13 0.24 3.08***
  None (ref.)
Par Involve with Stu Educ 0.17 0.06 1.18 0.02 0.06 1.02
# of Parents Talk To
Missing -0.913 0.22 0.40*** -0.37 0.24 0.69
None -0.67 0.28 0.51 -0.20 0.30 0.82
One or Two -0.66 0.22 0.52 -0.28 0.24 0.76
Three to Five -0.42 0.22 0.65 -0.23 0.24 0.79
Six to Ten 0.08 0.23 1.09 -0.17 0.25 0.98
Eleven to Twenty -0.25 0.25 0.78 -0.49 0.28 0.61
20+ (ref.)
Student-Teacher Relations
Strongly Agree -0.31 0.25 0.73 -0.24 0.27 0.79
Agree -0.49 0.26 0.62 -0.20 0.28 0.82
Disagree -0.35 0.28 0.70*** -0.09 0.30 0.92
Strongly Disagree (ref.)
160
Appendix G.  Odds-ratios, co-efficients, and standard errors based on time of enrollment, 
with no enrollment as the baseline group (continued)
Immediate Enrollment Delayed Enrollment
Independent Variable B Stand
Error
Exp(B) B Stand
Error
Exp(B)
High School Support
     Yes 0.47 0.10 1.59*** -0.02 0.11 0.98
     No (ref.)
School Type
Private School 1.56 0.35 4.76*** 1.07 0.37 2.91
Public (ref.)
Free and Reduced Lunch
1
st
 Quartile (Low %) -0.61 0.33 0.54 -0.99 0.34 0.37
2
nd
  Quartile -0.29 0.34 0.75 -0.71 0.34 0.49
3
rd
 Quartile -0.60 0.36 0.55 -1.14 0.36 0.32
4
th
 Quartile (ref.)
Mother?s PSE Expectations
Missing -1.35 0.18 0.26*** -0.34 0.19 0.71
None -1.88 0.22 0.15*** -1.18 0.24 0.31***
Two-Year -1.12 0.17 0.33*** -0.47 0.18 0.63
Bachelor?s -0.61 0.16 0.54*** -0.12 0.17 0.89
Grad School (ref.)
Par Involve w/ School -0.00 0.05 1.00 0.02 0.05 1.02
Peer Encouragement 0.04 0.05 1.04 -0.12 0.05 0.89
Educ Materials in Home
     Yes -0.40 0.25 0.67 -0.75 0.25 0.47
No (ref.)
Took Art, Music, or Dance
Almost daily -0.10 0.25 0.91 -0.39 0.29 0.68
1 to 2x Week 0.18 0.24 1.20 0.39 0.25 1.48
1x Wk -0.34 0.26 0.70 -0.18 0.28 0.84
     Never (ref.)
Number of Cases in 
Analyses
6,825
change -2 log likelihood 8,122
Pseudo R
2 
(Cox & Snell) 0.30
Percent Classified Correctly 75.4
Notes: Enrollment status is relative to not enrolling in college.
Data weighted by normalized F4F2 panel weight.
***p<.001
Source: Analyses of NELS data
161
REFERENCES 
1992 Unemployment Rates. (2004). Retrieved December 26, 2004, from www.bls.gov
Adelman, C. (2002). The relationship between urbanicity and educational outcomes. In 
W. G. Tierney & L. S. Hagedorn (Eds.), Increasing access to college: Extending 
possibilities for all students (pp. 35-63). Albany, NY: State University of New 
York Press.
Alwin, D. F., Hauser, R. M., & Sewell, W. H. (1973). Colleges and Earnings (Working 
Paper 73-23).Unpublished manuscript, Madison, WI.
Aslanian, C. B. (2001). Adult students today. New York: The College Board.
Becker, G. (1993). Human capital: A theoretical and empirical analysis with special 
reference to education (3rd ed.). Chicago: The University of Chicago Press.
Bourdieu, P. (1986). The forms of capital (R. Nice, Trans.). In J. F. Richardson (Ed.), 
Handbook of theory and research for the sociology of education (pp. 241-258). 
New York: Greenword Press.
Bourdieu, P., & Passeron, J. C. (1977). Reproduction: In education, society, and culture.
London: Sage Publications.
Bowen, H. R. (1997). Investment in learning: The individual and social value of 
American higher education. Baltimore: The Johns Hopkins University Press.
Burt, R. S. (2001). Structural holes versus network closure as social capital. In N. Lin, R. 
S. Burt & K. Cook (Eds.), Social capital: Theory and research (pp. 31-56). 
Hawthorne, NY: Aldine de Gruyter.
Cabrera, A. F., & La Nasa, S. M. (2001). On the path to college: Three critical tasks 
facing America's disadvantaged. Research in Higher Education, 42, 119-149.
162
Choy, S. P. (2002). Condition of education 2000: Nontraditional undergraduates. 
Washington, DC: NCES.
Choy, S. P., & Bobbit, L. (2000). Low-income students: Who they are and how they pay 
for their education. Washington, DC: National Center for Education Statistics.
Cohen, J., & Cohen, P. (1983). Missing data. In J. Cohen & P. Cohen (Eds.), Applied 
multivariate regression/Correlation analyses for the behavioral sciences (2nd ed., 
pp. 275-300). Hillsdale, NJ: Lawrence Erlbaum.
Coleman, J. S. (1988). Social capital in the creation of human capital. American Journal 
of Sociology, 94 (Supplement), 95-120.
College Opportunity and Career Help Program. (2005). Retrieved March 21, 2005, from 
www.ksg.harvard.edu/coach/index.htm
Cook, B., & King, J. E. (2004). Low-Income adults in profile: Improving lives through 
higher education. Washington, DC: American Council on Education Center for 
Policy Analysis.
Croninger, R. G., & Lee, V. E. (2001). Social capital and dropping out of high school: 
Benefits to at-risk students of teachers' support and guidance. Teachers College 
Record, 103, 548-581.
Curtin, T. R., Ingels, S. J., Wu, S., & Heuer, R. (2002). User's manual: NELS:88 Base-
year to fourth follow-up data file user's manual. Washington, DC: National 
Center for Education Statistics.
DiMaggio, P., & Mohr, J. (1985). Cultural capital, educational attainment, and marital 
selection. American Journal of Sociology, 90, 1231-1261.
163
Douglass, G. K. (1997). Economic returns on the investment in higher education. In H. R. 
Bowen (Ed.), Investment in Learning: The individual and social value of 
American higher education (pp. 359-387). Baltimore: The Johns Hopkins 
University Press.
Downey, D. B., & Powell, B. (1993). Do children in single-parent households fare better 
living with same-sex parents? Journal of Marriage and the Family, 55(1), 55-71.
Fitzgerald, B. K., & Delaney, J. A. (2002). Educational opportunity in America. In D. E. 
Heller (Ed.), Condition of access: Higher education for lower income students
(pp. 3-24). Westport, CT: American Council on Education/Praeger.
Gandara, P. (2002). Meeting common goals: Linking K-12 and college interventions. In 
W. G. Tierney & L. S. Hagedorn (Eds.), Increasing access to college: Extending 
possibilities for all students (pp. 81-103). Albany, NY: State University of New 
York Press.
Gandara, P., & Bial, D. (2001). Paving the way to postsecondary education: K-12 
intervention programs for underrepresented youth. Washington, DC: National 
Postsecondary Education Cooperative.
Hamrick, F., A., & Stage, F. K. (1998). High minority enrollment, high school-lunch 
rates: Predisposition to college. Review of Higher Education, 21, 343-357.
Hearn, J. C. (1991). Academic and nonacademic influences on the college destinations of 
1980 high school graduates. Sociology of Education, 64, 158-171.
Hearn, J. C. (1992). Emerging variations in postsecondary attendance patterns: An 
investigation of part-time, delayed, and nondegree enrollment. Research in 
Higher Education, 33, 657-687.
164
Heller, D. E. (1997). Student price response in higher education: An update to Leslie and 
Brinkman. Journal of Higher Education, 68, 624-651.
Horn, L., & Carroll, C. D. (1996). Nontraditional undergraduates: Trends in enrollment 
from 1986 to 1992 and persistence and attainent among 1989-1990 beginning 
postsecondary students. Washington, DC: National Center for Education 
Statistics.
Horvat, E. M. (2001). Understanding equity and access in higher education: The potential 
contribution of Pierre Bourdieu. In J. C. Smart (Ed.), Higher education: 
Handbook of theory and research (pp. 195-238). New York: Agathon Press.
Horvat, E. M. (2003, Spring). The Interactive Effects of Race and Class in Educational 
Research: Theoretical Insights from the Work of Pierre Bourdieu. Retrieved Fall, 
2004, from 
http://www.urbanedjournal.org/archive/Issue3/articles/article0009.html.
Hossler, D., Braxton, J., & Coopersmith, G. (1989). Understanding student college 
choice. In J. C. Smart (Ed.), Higher education: Handbook of theory and research
(Vol. 5, pp. 231-288). New York: Agathon Press.
Hossler, D., & Gallagher, K. S. (1987). Studying student college choice: A three phase 
model and the implications for policymakers. College and University, 62, 207-
221.
Hossler, D., Schmidt, & Vesper. (1999). Going to college: How social, economic, and 
educational factors influence the decisions students make. Baltimore: Johns 
Hopkins University Press.
165
Hurtado, S., Inkelas, K. K., Briggs, C., & Rhee, B. S. (1997). Differences in college 
access and choice among racial/ethnic groups: Identifying continuing barriers. 
Research in Higher Education, 38, 43-75.
IHEP. (2005). The investment payoff: A 50-state analysis of the public and private 
benefits of higher education. Washington, DC: The Institute for Higher Education 
Policy.
Kane, T. J. (1994). Race, college attendance and college completion. Washington, DC; 
Cambridge, MA: Brookings Institution; Kennedy School of Government Harvard 
University.
Kane, T. J. (1999). The price of admission: Rethinking how Americans pay for college. 
Washington, DC/New York: Brookings Institution Press/Russell Sage 
Foundation.
Lamont, M., & Lareau, A. (1988). Cultural capital: Allusions, gaps and glissandos in 
recent theoretical develpments. Sociological Theory, 6, 153-168.
Lareau, A. (1987). Social class differences in family-school relationships: The 
importance of cultural capital. Sociology of Education, 60, 73-85.
Lareau, A., & Horvat, E. M. (1999). Moments of social inclusion and exclusion: Race, 
class and cultural capital in family-school relationships. Sociology of Education, 
72(1), 37.
Lin, N. (2001a). Building a network theory of social capital. In N. Lin, K. Cook & R. S. 
Burt (Eds.), Social capital: Theory and research (pp. 3-30). Hawthorne, NY: 
Aldine de Gruyter.
166
Lin, N. (2001b). Social capital: A theory of social structure and action. Cambridge: 
Cambridge University Press.
Manski, C. F., & Wise, D. A. (1983). College choice in America. Cambridge, MA: 
Harvard University Press.
Marx, F. (2002). Women in community development (WICD), higher education for low-
income women: A real route out of poverty. Retrieved April 30, 2002, from 
www.wcwonline.org
Matus-Grossman, L., & Gooden, S. (2001). Opening doors to earning credentials: 
Impressions from community college access and retention from low-wage 
workers. Oakland, CA: Manpower Demonstration Research Corporation.
McDonough, P. M. (1997). Choosing colleges: How social class and schools structure 
opportunity. Albany, NY: State University of New York Press.
McNeal, R. B. (1999). Parental involvement as social capital: Differential effectiveness 
on science achievement, truancy, and dropping out. Social Forces, 78, 117-144.
Morrow, V. (1999). Conceptualizing social capital in relation to the well-being of 
children and young people: a critical review. Sociological Review, 47, 745-765.
NCES. (1993). Digest of Education Statistics. Washington, DC: National Center for 
Education Statistics.
Paulsen, M. B. (2001). The economics of human capital and investment in higher 
education. In M. B. Paulsen & J. C. Smart (Eds.), The finance of higher 
education: Theory, research, policy, and practice (pp. 55-96). New York: 
Agathon Press.
167
Paulsen, M. B., & St. John, E. P. (2002). Social class and college costs: Examining the 
financial nexus between college choice and persistence. Journal of Higher 
Education, 73, 189-236.
Pedhazur, E. J., & Schmelkin, L. (1991). Measurement, design, and analysis: an 
integrated approach. Hillsdale, NJ: Lawrence Erlbaum Associates.
Perna, L. W. (2000). Differences in the decision to attend college among African 
American, Hispanics, and Whites. Journal of Higher Education, 71, 117-140.
Perna, L. W. (2001). The relationship between family responsibilities and employment 
status among college and university faculty. Journal of Higher Education, 72, 
584-611.
Perna, L. W. (2004). The key to college access: Rigorous academic preparation. In W. G. 
Tierney, Z. B. Corwin & J. E. Colyar (Eds.), Preparing for college: Nine elements 
of effective outreach. Albany, NY: SUNY Press.
Perna, L. W., & Titus, M. A. (2004). Understanding differences in the choice of college 
attended: The role of state public policies. Review of Higher Education, 27, 501-
525.
Perna, L. W., & Titus, M. A. (in press). The relationship between parental involvement as 
social capital and college enrollment: An examination of racial/ethnic group 
differences. Journal of Higher Education.
Plank, S. B., & Jordan, W. J. (2001). Effects of information, guidance, and actions on 
postsecondary destinations: A study of talent loss. American Educational 
Research Journal, 38, 947-979.
168
Portes, A. (1998). Social capital: Its origins and applications in modern sociology. Annual 
Review of Sociology, 24, 1-24.
Schwartz, J. B. (1985). Student financial aid and the college enrollment decision: The 
effects of public and private grants and interest subsidies. Economics of Education 
Review, 4, 129-144.
A shared agenda: A leadership challenge to improve college access and success. (2004). 
Boston, MA: Pathways to College Network.
SPSS. (1999). SPSS: Advanced models 9.0. In. Chicago: SPSS.
St. John, E. P. (2003). Refinancing the College Dream. Baltimore: Johns Hopkins 
University Press.
Stanton-Salazar, R. D. (1997). A social capital framework for understanding the 
socialization of racial minority children and youths. Harvard Eductional Review, 
67, 1-40.
Swail, W. S., & Perna, L. W. (2002). Pre-college outreach programs: A national 
perspective. In W. G. Tierney & L. S. Hagedorn (Eds.), Increasing access to 
college: Extending possibilities for all students (pp. 15-34). Albany, NY: State 
University of New York Press.
Teachman, J. D., & Polonko, K. A. (1988). Marriage, parenthood, and the college 
enrollment of men and women. Social Forces, 512-523.
Thomas, S. L., & Heck, R. H. (2001). Analysis of large-scale secondary data in higher 
education research: Potential perils associated with complex sampling designs. 
Research in Higher Education, 42, 517.
169
Twenty-first century scholars 1990-2001 progress report. (2002). Retrieved March 28, 
2005, from www.in.gov/ssaci/news/21progressreport2002.pdf
Venezia, A. (2003). Connecting the systems: What can postsecondary education do to 
work with K-12 to help students better prepare for college? Peer Review, 5, 27-
30.