ABSTRACT Title of Dissertation: PERSUASIVE EFFECTS OF NARRATIVES IN IMMERSIVE MEDIATED ENVIRONMENTS Zexin Ma, Doctor of Philosophy, 2018 Dissertation directed by: Professor Xiaoli Nan, Department of Communication This dissertation proposed a theoretical framework to model the persuasive effects of narratives in immersive mediated environments (IMEs) and the underlying psychological mechanisms. Drawing upon previous research on narrative persuasion and immersive media, this dissertation hypothesized that narratives presented in immersive (vs. non-immersive) mediated environments would lead to higher levels of spatial presence, social presence, transportation, and identification. Spatial and social presence were hypothesized to mediate the effects of media format (i.e., IMEs vs. non-IMEs) on transportation and identification, respectively. Furthermore, media format was predicted to have an indirect effect on counterarguing through spatial presence, social presence, transportation, and/or identification. These psychological mechanisms were also predicted to mediate the effects of media format on story- consistent attitudes and behavioral intentions/willingness. In addition, trait empathy was proposed as a moderator to influence the relationship between media format and viewers’ spatial presence, social presence, transportation, and identification. A controlled experiment was conducted to test the above hypotheses in two video contexts. Results provided relatively consistent evidence across the two contexts. In the driving under the influence (DUI) context, narratives presented in immersive (vs. non-immersive) mediated environments led to a higher level of spatial presence, which in turn promoted greater transportation. Media format had an approaching significant indirect effect on counterarguing through spatial presence and transportation. The model as a whole explained a large amount of variance in participants’ attitudes toward DUI and behavioral intentions to engage in DUI. Similarly, in the malaria context, narratives presented in immersive (vs. non- immersive) mediated environments led to higher levels of spatial presence and social presence. Spatial and social presence were found to mediate the effect of media format on transportation and identification, respectively. The indirect effect of media format on counterarguing through spatial presence and transportation was significant. The proposed model accounted for a relatively large amount of variance in attitudes and behavioral willingness to help people who need protection from malaria, as well. Theoretical contributions, practical implications, ethical issues, limitations, and directions for future research are discussed. PERSUASIVE EFFECTS OF NARRATIVES IN IMMERSIVE MEDIATED ENVIRONMENTS by Zexin Ma 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 2018 Advisory Committee: Professor Xiaoli Nan, Chair Professor Robert Feldman Professor Dale J. Hample Professor Anita Atwell Seate Professor Leah Waks © Copyright by Zexin Ma 2018 Dedication This dissertation is dedicated to my late grandma. Without the love and inspirations from you, none of this would have been possible. ii Acknowledgements I would like to take a moment to express my thanks to people who provided guidance and support to me, not only on this project but also throughout my graduate training. First, I want to express the deepest appreciation to my advisor, Dr. Xiaoli Nan, for her guidance and mentorship. You not only set a high bar for me, but also helped me achieve it. Thank you for nurturing my enthusiasm for communication and for providing me with tremendous academic support. I must thank my dissertation committee. Special thanks to Dr. Dale J. Hample, who patiently answered every single question that I raised and constantly reminded me what a cool project I have been working on. Thanks to Dr. Anita Atwell Seate for her insightful and encouraging comments that made this dissertation a better one. To Dr. Leah Waks, whose humor and warm words supported me throughout my dissertation process and beyond. And to Dr. Robert H. Feldman for providing expertise in the realm of health communication. I am also hugely appreciative to Dr. Gregory R. Hancock, who helped guide my statistical analysis and made statistics learning so much fun. Similarly, my gratitude goes to my fellow graduate students and colleagues. Special mention to those who helped advertise my study and recruit participants. I also want to thank the Center for Health and Risk Communication at the University of Maryland, which provided the equipment for me to conduct experiments. Mom and dad, where I would be without your love and support. Who believes that a 27-year-old still shares every laughter and tear with her mom despite 12-hour iii time difference. You are a combination of a great mother and a best friend. Dad, you not only provide solid support that allows me to fly high and free, but also build a home where I can always rest and recharge. Finally, thank you to my best friend, Zhiren. Thank you for taking me to eat spicy Chinese food (my absolute cure to release stress), helping proofread my paper late at night, and being a good listener and the biggest fan. I am so grateful that our paths cross. iv Table of Contents Dedication ................................................................................................................. ii Acknowledgements .................................................................................................. iii Table of Contents ...................................................................................................... v List of Tables .......................................................................................................... vii List of Figures ........................................................................................................ viii Chapter 1: Introduction .............................................................................................. 1 Chapter 2: Conceptual Background ........................................................................... 5 2.1 Narrative Persuasion ........................................................................................ 5 2.1.1 Defining Narrative Persuasion................................................................... 5 2.1.2 Explaining Narrative Effects ..................................................................... 6 2.2 Immersive Mediated Environments ................................................................ 12 2.2.1 Defining Immersive Mediated Environments .......................................... 12 2.2.2 Presence .................................................................................................. 15 2.3 Modeling Narrative Effects in Immersive Mediated Environments ................ 19 2.3.1 The Persuasive Narrative Theory in Immersive Mediated Environments (PENTIMEs) ................................................................................................... 19 2.3.2 Trait Empathy as a Moderator ................................................................. 24 Chapter 3: Pilot Studies ........................................................................................... 27 3.1 Pilot Study 1 .................................................................................................. 27 3.1.1 Study Objectives ..................................................................................... 27 3.1.2 Participants and Apparatus ...................................................................... 28 3.1.3 Design and Procedure ............................................................................. 28 3.1.4 The 360° Video ....................................................................................... 29 3.1.5 Results .................................................................................................... 30 3.1.6 Conclusions ............................................................................................ 31 3.2 Pilot Study 2 .................................................................................................. 31 3.2.1 Participants and Apparatus ...................................................................... 32 3.2.2 Design and Procedure ............................................................................. 32 3.2.3 The 360° Video ....................................................................................... 32 3.2.4 Key Dependent Variables ........................................................................ 33 3.2.5 Results .................................................................................................... 35 3.2.6 Conclusions ............................................................................................ 37 Chapter 4: Method of Main Study ........................................................................... 38 4.1 Participants and Apparatus ............................................................................. 38 4.2 Design and Procedure .................................................................................... 38 4.3 360° Videos ................................................................................................... 39 4.3.1 The 360° Video on DUI .......................................................................... 39 4.3.2 The 360° Video on Malaria ..................................................................... 40 4.4 Measures ....................................................................................................... 41 4.4.1 Phase I Measures .................................................................................... 41 4.4.2 Phase II Measures ................................................................................... 43 Chapter 5: Results of Main Study ........................................................................... 48 v 5.1 Sample Characteristics ................................................................................... 48 5.2 Checking Order Effect ................................................................................... 49 5.2.1 DUI......................................................................................................... 49 5.2.2 Malaria ................................................................................................... 50 5.3 Test of Multivariate Normality....................................................................... 50 5.4 Test of Discriminant Validity ......................................................................... 53 5.4.1 DUI......................................................................................................... 53 5.4.2 Malaria ................................................................................................... 54 5.5 Testing Overall Persuasion ............................................................................ 54 5.5.1 DUI......................................................................................................... 54 5.6.1 Malaria ................................................................................................... 55 5.6 Main Model Testing ...................................................................................... 56 5.6.1 DUI......................................................................................................... 58 5.6.2 Malaria ................................................................................................... 68 5.7 Interaction Effects Testing ............................................................................. 79 5.7.1 DUI......................................................................................................... 80 5.7.2 Malaria ................................................................................................... 88 Chapter 6: General Discussion ................................................................................ 96 6.1 Summary of Experimental Findings ............................................................... 96 6.1.1 The Main Model ..................................................................................... 96 6.1.2 The Main Model with Trait Empathy as a Moderator ............................ 101 6.2 Limitations and Future Directions ................................................................ 103 6.3 Practical Implications .................................................................................. 106 6.4 Ethical Issues ............................................................................................... 107 6.5 Conclusions ................................................................................................. 108 Appendix A: Video Transcripts ............................................................................. 111 The DUI video................................................................................................... 111 The malaria video .............................................................................................. 112 Appendix B: Measures .......................................................................................... 114 References ............................................................................................................. 119 vi List of Tables Table 5.1. Univariate and Multivariate Skew and Kurtosis Test Statistics 52 Table 5.2. Error Variance for Each Variable 57 Table 5.3. Correlation Matrix for Key Variables in the Model (DUI Context) 59 Table 5.4. Standardized and Unstandardized Parameter Estimates for the Main Model (DUI Context) 60 Table 5.5. Correlation Matrix for Key Variables in the Model (Malaria Context) 70 Table 5.6. Standardized and Unstandardized Parameter Estimates for the Main Model (Malaria Context) 71 Table 5.7. Standardized and Unstandardized Parameter Estimates for the Interaction Model (DUI Context) 82 Table 5.8. Standardized and Unstandardized Parameter Estimates for the Interaction Model (Malaria Context) 89 vii List of Figures Figure 2.1. The theorized main model. 24 Figure 2.2. The theorized interaction model. 26 Figure 5.1. The main model in the DUI context. 68 Figure 5.2. The main model in the malaria context. 79 Figure 5.3. The interaction model in the DUI context. 87 Figure 5.4. The interaction model in the malaria context. 94 Figure 5.5. Interactions of media format and trait empathy on social presence. 95 viii Chapter 1: Introduction Please close your eyes. Imagine that you are standing in a street of Syria. A bomb exploded a moment ago. You look around. You see the scared faces of little kids. You hear the sounds of people’s screaming. You find yourself in a helpless situation. You are frightened, worried, and anxious. All of it looks so real to you. This is neither a fiction nor a nightmare. This is a new way of storytelling in immersive mediated environments (IMEs), environments that allow individuals to perceive themselves to be completely enveloped with the aid of immersive technologies. Viewing a 360° video on head mounted displays (HMDs), for example, brings users this new mediated experience. Many mainstream media companies, such as the New York Times and the Guardian, have adopted this immersive storytelling technique. Social media websites, such YouTube and Facebook, also support uploading and playback of 360° spherical videos. IMEs are transforming storytelling. Narrative, or storytelling, is a basic mode of human interaction. Since the ancient times, our ancestors have used stories to pass on wisdoms from generations to generations. In our daily life, we tell stories to communicate our thoughts and feelings with one another. In an era of mass media, journalists use stories to report what is going on in the world; advertisers use stories to convince consumers to buy their products; screen writers use stories to entertain the public. Stories are largely used to inform and to persuade. The latter function, termed as narrative persuasion (Hinyard & Kreuter, 2007), has received a substantial body of research in communication. Research suggests that compared to non-narrative messages (i.e., didactic arguments or statistical information), narrative messages are more effective in 1 preventing personal drug use (Banerjee & Greene, 2012), promoting HPV vaccination (Hopfer, 2012; Nan, Dahlstrom, Richards, & Rangarajan, 2014), and increasing favorable attitudes toward mental illnesses (Chang, 2008; Ma & Nan, in press). Two major mechanisms, narrative involvement and character involvement, have been proposed to explain narrative’s persuasive influence (Green & Brock, 2000; Moyer- Gusé, 2008; Slater & Rouner, 2002). Narrative involvement emphasizes the idea of being immersed in the story world, whereas character involvement focuses on the audience members’ responses toward the characters. Narratives exert persuasive impact through audience involvement with the narrative as a whole and with characters, which can reduce persuasion resistance (Moyer-Gusé, 2008; Slater & Rouner, 2002). However, the majority of earlier studies inquired narrative persuasion in the traditional mediated environments, such as examining the effects of stories published in newspaper or portrayed in entertainment television shows. The persuasive effects of narratives in IMEs have rarely been examined. As immersive media have the potential to significantly amplify the persuasive effects of narratives through its immersive nature, it is important to study narratives in IMEs. The primary goal of this dissertation is to investigate whether and how narratives presented in IMEs affect audience members’ attitudes and behavioral intentions. Furthermore, as most research studied narrative effects by comparing narrative versus non-narrative messages, it is hard to keep the messages identical across different conditions. This study is the first study to examine narrative persuasion by keeping the message content equivalent. 2 Immersive media research proposes that immersive virtual technologies have the potential to persuade and influence through intensifying the users’ psychological experience in the IMEs (Grigorovici, 2003; Kim & Biocca, 1997). For example, Kim and Biocca (1997) argued that “when the users feel present in the virtual environment, they are also likely to feel persuaded.” Their study also demonstrated that presence had a positive effect on memory and purchasing intention (Kim & Biocca, 1997). Grigorovici (2003) further stated that being in the mediated environments triggered users to respond to the environment as real and unmediated, thus considering the experience in the mediated environments as direct or first-hand experience. By integrating literature in narrative persuasion and immersive media, this dissertation proposes a theoretical framework to model the persuasive processes and effects of narratives in IMEs - the persuasive narrative theory in immersive mediated environments (PENTIMEs). PENTIMEs suggests that narratives presented in immersive (vs. non-immersive) mediated environments are likely to enhance users’ perception of presence, which will in turn strengthen their involvement with the story and characters. As story and character involvement are found to have a negative relation to persuasion resistance, involvement will reduce counterarguing, a common type of persuasion resistance, which will lead to greater story-consistent attitudes and behavioral intentions/willingness. An experiment with two conditions was conducted to assess PENTIMEs in two video contexts: preventing driving under the influence (DUI) and promoting efforts to fight against malaria. DUI is an important public health problem in the 3 United States. In 2015, 10,265 deaths resulted from alcohol-impaired crashes (National Highway Traffic Safety Administration, NHTSA, 2016). This amounts to approximately 28 deaths involved in alcohol-impaired crashes every day in the United States. The damages and deaths resulted from DUI contribute to a cost of $52 billion annually. Continuous campaigns and efforts are needed to prevent DUI. Malaria is one of the world’s oldest and deadliest diseases. According to World Health Organization (2017), an estimated 216 million cases of malaria occurred globally in 2016, with an estimated 445,000 deaths. The majority of the malaria cases (90%) and deaths (91%) were from the African region. These two topics address two important public health issues. Understanding narrative effects in IMEs has practical implications for designing effective campaign messages to reduce people’s intentions to engage in DUI and to promote individuals’ willingness to help fight against malaria. The chapters that follow provide a detailed description of the theoretical arguments and empirical evidence. Specifically, Chapter 2 starts with a review of theories in narrative persuasion and immersive media. Based on previous literature, it advances a preliminary theoretical framework, PENTIMEs, to explain the effects of narratives in IMEs. Chapter 3 reports two pilot studies that were conducted preceding the main experiment. Next, the methods and results of the main study are presented in Chapter 4 and 5, respectively. In Chapter 6, a summary of findings, limitations to the study, practical implications, related ethical issues, and theoretical contributions are discussed. 4 Chapter 2: Conceptual Background 2.1 Narrative Persuasion 2.1.1 Defining Narrative Persuasion Narrative has been defined as “a representation of connected events and characters that has an identifiable structure, is bounded in space and time, and contains implicit or explicit messages about the topic being addressed” (Kreuter et al., 2007, p. 222). The formats of narratives include testimonials, news exemplars, entertainment-education (E-E) programs such as TV soap operas, etc. The use of narratives to influence individuals’ beliefs, attitudes, and behaviors is defined as narrative persuasion (de Graaf, Hoeken, Sanders, & Beentjes, 2012). Narrative persuasion differs from traditional persuasion, which often takes the form of didactic arguments or statistical information (Hinyard & Kreuter, 2007; Nan, Futerfas, & Ma, 2017). A great deal of research has attested to the power of narratives versus non-narratives to change beliefs, attitudes, and behaviors. For example, de Wit, Das, and Vet (2008) demonstrated that homosexuals had a higher risk susceptibility of infection with hepatitis B virus and intention to obtain vaccination when they read a personal account by a peer-group member than statistical prevalence information. In another study, Chang (2008) found that narrative (vs. argument) advertising could induce more sympathy toward people with depression, generate greater willingness to seek help if they are depressed, and improve efficacy in identifying friends or family members who may suffer from depression. Moreover, studies have also shown that fictional narrative films are more persuasive than 5 nonfiction non-narrative films. Participants who watched a narrative film featuring a woman with cervical cancer had a significant increase in their knowledge about cervical cancer, and were more likely to have had or scheduled a Pap test than those who watched a non-narrative film about cervical cancer (Murphy et al., 2015). These studies revealed the advantages of narrative persuasion over traditional persuasion. Due to its ability to engage less involved audience and make complex information comprehensible, narrative also received success in real-world practice as a promising tool to promote public health. For example, entertainment TV dramas have been found to increase public health literacy (e.g., Valente et al., 2007) and intentions to adapt healthy behaviors (Collins, Elliott, Berry, Kanouse, & Hunter, 2003). More recently, Centers for Disease Control and Prevention (CDC) applied narrative strategy into their national campaigns (e.g., Tips from Former Smokers & Inside Knowledge). These campaigns profile real stories of real people, demonstrating the daily life of patients who live with severe health consequences resulted from unhealthy behaviors. Research found that the Tips from Former Smokers campaign motivated approximately 1.64 million smokers to quit smoking and more than 100,000 smokers to refrain from smoking. In addition, over 6 million nonsmokers talked about the dangers of smoking with their friends and family after being exposed to this campaign (McAfee, Davis, Alexander Jr, Pechacek, & Bunnell, 2013). 2.1.2 Explaining Narrative Effects The superiority of narrative effects has been explained by several theoretical frameworks, including the transportation-imagery model (Green & Brock, 2000; 2002), the extended elaboration likelihood model (E-ELM, Slater & Rouner, 2002), 6 and the entertainment overcoming resistance model (EORM, Moyer-Gusé, 2008; Moyer-Gusé & Nabi, 2010). These frameworks suggest that two distinct but interrelated mechanisms, narrative involvement and character involvement, explain narrative influence. Narrative involvement describes the experience in which individuals are completely focused on the events portrayed in the story and temporarily lose awareness of the real world around them. It is often labelled as transportation, engagement, or absorption interchangeably in the literature (Green & Brock, 2000; Slater & Rouner, 2002). Green and Brock (2000) conceptualized transportation as “a distinct mental process, an integrative melding of attention, imagery, and feelings” (p. 701). It is a multi-dimensional construct that consists of attention focusing, imagery evoking, and emotional arousing. When readers are transported, they attend to the story, have a vivid image of the story in their mind, and experience emotions portrayed by the story. Imagery, however, is limited to text- or audio-based narratives. The consequence of transportation is losing awareness of the real world around them. It seems that audience members become part of the story they are reading or viewing. The extent to which an individual is transported to the narrative world is predicted by media modality (Walter, Murphy, Frank, & Baezconde-Garbanati, 2017), prior knowledge (Green, 2004), and personality traits such as transportability (Bilandzic & Busselle, 2008) and trait empathy (Hall & Bracken, 2011). A substantial body of studies suggested that individuals who were more transported into the narrative message had greater belief and attitude change (Kim, 7 Bigman, Leader, Lerman, & Cappella, 2012; Murphy, Frank, Chatterjee, & Baezconde-Garbanati, 2013; Murphy, Frank, Moran, & Patnoe-Woodley, 2011). Kim et al. (2012) found that smokers who read a news article with an exemplar on successful smoking cessation experienced greater narrative engagement, which in turn, increased smoking cessation intentions than those who read a news article without an exemplar. Additionally, Murphy et al. (2011) showed that transportation was the best individual predictor of change in story relevant knowledge, attitudes, and behaviors after exposure to a lymphoma drama storyline. An important mechanism through which transportation exerts influence on persuasive effects is reducing persuasion resistance, such as counterarguing. Counterarguing, studied most in the persuasion literature, refers to “the generation of thoughts that dispute or are inconsistent with the persuasive argument” (Slater & Rouner, 2002, p. 180). All three models, including the transportation-imagery model, E-ELM, and EORM, propose that transportation and counterarguing are fundamentally incompatible with each other. Being transported to the narrative world reduces the audience’s motivation and ability to generate counter arguments. It is because narrative transportation is supposed to be an enjoyable experience, which allows audience members to resonate with events described in the story. Moreover, when transported, individuals devote his or her cognitive and affective efforts to the events occurred in the story. If message recipients are consciously generating disagreements with the message, they are not transported by definition. It is because transportation emphasizes the loss of awareness of the real world. Empirical research also provided support that counterarguing mediated the relationship between 8 transportation and persuasiveness (e.g., McQueen, Kreuter, Kalesan, & Alcaraz, 2011). Character involvement is concerned with how audience members interact with characters. The most studied type of character involvement is identification. The literature has identified two types of identification, wishful identification and identification. Wishful identification differs from identification, with the former focusing on the psychological experience after the media exposure and the latter describing the experience during the exposure to the narrative. Wishful identification occurs when a viewer wants to “become like a media character” (Hoffner & Buchanan, 2005, p. 327). Wishful identification has often been studied to investigate the long-term consequence of the media exposure. For example, Boon and Lomore’s (2001) survey indicated that young adults made changes in their appearances, values, and other characters to emulate admired celebrity idols. Identification refers to “an imaginative process through which an audience member assumes the identity, goals, and perspective of a character” (Cohen, 2001, p. 261). It can be broadly viewed as sharing the feelings and thoughts of the character. When identifying with a media character, the viewer momentarily forgets about his or her own identity and becomes that character. Through identification, the viewer feels what the character feels and understands the events just like how the character understands them. This study focuses on responses that audience members have during a media presentation. It is because the goal of the current study is to investigate whether the level of immersion of media technology impacts viewers’ responses when they are 9 actually exposed in the mediated environments. This study excludes any psychological experience that might occur after the media exposure. Wishful identification, which describes the long-term effect of the media exposure, is not relevant in this research. Thus, identification is conceptualized in this study as emotional and cognitive process by which an audience member imagines him or herself as a particular story character. The antecedents of identification include technical production features, audience characteristics, and character attributes (Cohen, 2001). It is important to note that in Cohen’s (2001) seminal work on identification, he operationalized identification as four dimensions: affective involvement (sharing the feelings with the character), cognitive involvement (sharing the perspective of the character), motivational involvement (internalizing the goal of the character), and absorption (losing self-awareness during narrative exposure). However, a study published later by Tal-Or and Cohen (2010) argued that identification should be operationalized by measuring emotional and cognitive perspective taking, which was closest to its theoretical definition. The authors reasoned it was because Cohen (2001) devised the items of identification without comparing them to the scale of transportation, leading to the overlaps between the two scales. Another important issue that needs to be addressed here is that transportation and identification are conceptually different from each other. Transportation refers to the degree of the absorption with the narrative as a whole, whereas identification describes the involvement with the specific story character. Operationally, they tap on different aspects. Transportation focuses on the experience of being emotionally 10 aroused by the story, focusing attention to the story, and losing awareness of the surroundings. Identification emphasizes developing an emotional connection and taking the cognitive perspective of the character. Nonetheless, these two processes often concur (Moyer-Gusé, 2008; Moyer-Gusé & Nabi, 2010)1. Similar with transportation, identification has also been found to have a positive effect on persuasive outcomes (de Graaf et al., 2012; Hoeken & Fikkers, 2014; Igartua, 2010). For instance, Igartua (2010) showed that identification with a fictional film character predicted the change in relevant attitudes and beliefs. de Graaf et al. (2012) further tested the causal effect of identification on attitudes by manipulating the character perspective. The authors found that identification mediated the effect of perspective on readers’ attitudes. Furthermore, identification was also proposed and empirically tested to have a negative relation to counterarguing (Moyer-Gusé, 2008; Moyer-Gusé & Nabi, 2010). As indicated earlier, the majority of narrative communication studies were conducted in the traditional mediated environments, such as asking participants to read a short story in paper or watching a video on computers. Stories presented in IMEs were rarely examined. To the best of my knowledge, only two studies that exist in the literature investigated immersive storytelling (Fonseca & Kraus, 2016; Sundar, Kang, & Oprean, 2017). These studies provided preliminary empirical evidence that narratives presented in immersive (vs. non-immersive) mediated environments had greater effects on viewers’ story-related perceptions and cognitions (please see next 1 The correlations of transportation and identification were significant in both contexts, .33 in the DUI context and .32 in the malaria context. 11 section for a detailed description of these two studies). However, they did not investigate the relations between user experience triggered by technologies and audience’s involvement with the narrative and characters, which are two major components in 360-degree storytelling (Elmezeny, Edenhofer, & Wimmer, 2018). Thus, in order to advance our theoretical understanding of narrative persuasion in IMEs, it is necessary to integrate literature in narrative persuasion and immersive media. The following section presents a discussion on IMEs and the important psychological constructs. 2.2 Immersive Mediated Environments 2.2.1 Defining Immersive Mediated Environments IMEs have often been used to describe mediated environments generated by virtual reality (VR) in previous studies (Blascovich et al., 2002; Cummings & Bailenson, 2016; Fox, Arena, & Bailenson, 2009). VR refers to a technological system, which “includes a computer capable of real-time animation, controlled by a set of wired gloves and a position tracker, and using a head-mounted stereoscopic display for visual output” (Steuer, 1992, p. 74). The goal of VR experience is to replace the cues of the physical environment in which one is actually present with the environment created by virtual technology (Fox et al., 2009). VR has been around for a few decades, but it did not gain popularity until recently when the hardware became more affordable and available to consumers. It is expected that the global VR hardware revenue will grow at a compound annual growth rate of 54.84% from 2018 12 to 2023 (Research and Markets, 2017). The opportunities that VR brings are enormous. In the meantime, the rise of VR has brought 360° videos to the market. 360° videos are the video recordings where a view in every direction is recorded at the same time. They are typically shot using either a collection of multiple cameras or a camera that contains multiple camera lenses. Unlike traditional videos that only allow viewers to see what the directors choose; 360° videos offer an immersive view that viewers can choose where to look at by themselves. Viewers can watch 360° videos via computers, mobile devices, or head-mounted displays (HMDs). When viewed on a computer, viewers can pan around the video by clicking and dragging the mouse. On a mobile device, viewers can drag their fingers across the screen or move it around in different directions. When viewed on HMDs (e.g., Samsung Gear VR), viewers can move their heads to look around. In particular, the experience of viewing 360° videos on HMDs offers users the opportunity to fully immerse themselves in the videos through shutting themselves out from their own physical reality. This makes 360° videos viewing similar to VR experience. An important clarification that needs to be addressed is that 360° videos are not VR because they are not computer- generated simulations. However, when 360° videos are displayed with HMDs, the mediated environments it creates share similarities with virtual environments. In this dissertation, I argue that IMEs should not exclusively refer to virtual experience. I define IMEs as mediated environments that allow individuals to perceive themselves to be completely enveloped with the aid of immersive 13 technologies. Immersive technologies are inclusive, extensive, surrounding, and vivid. In defining these important constructs, Slater and Wilbur (1997, p. 3) state: Inclusive (I) indicates the extent to which physical reality is shut out. Extensive (E) indicates the range of sensory modalities accommodated. Surrounding (S) indicates the extent to which this virtual reality is panoramic rather than limited to a narrow field. Vivid (V) indicates the resolution, fidelity, and variety of energy simulated within a particular modality (for example, the visual and colour resolution). Vividness is concerned with the richness, information content, resolution and quality of the displays. (p. 3) Furthermore, immersive technologies are also able to match the display of information with participant’s proprioceptive feedback about body movements (Slater & Wilbur, 1997). For example, visual displays are changed accordingly with a turn of the head. Steuer (1992) also noted that immersive technologies were interactive in a sense that users could influence the content or form of the mediated environments. Although these characteristics are originally used to describe VR, they can also be applied to 360° videos when viewed on HMDs. Viewing 360° videos on HMDs creates an IME that is inclusive, extensive, surrounding, and vivid. Specifically, viewing 360° videos on HMDs only allows viewers to see the content displayed on HMDs, which shut them out from their immediate environments. In other words, users’ physical reality (such as light and sound) are blocked off as they enter the immersive environments. This illustrates its inclusive feature. In addition, most 360° videos include both audio and visuals, which can stimulate users’ auditory and visual senses. Other senses, such as the touch sense, 14 can also be added through haptic feedback to make the experience more extensive. Moreover, 360° videos provide a panoramic view that allows viewers to explore the surroundings in 360 degrees. These videos are made with cameras with higher visual and color resolution, which meets the vivid criterion. The displayed images will be changed accordingly with the head movements, as well. Hence, 360° videos displayed on HMDs generate an inclusive, extensive, surrounding, and vivid mediated environment. In contrast to the viewing experience on HMDs, viewing 360° videos on flat computer screens or mobile devices creates a mediated environment that is extensive, surrounding, and vivid; however, it does not satisfy the inclusiveness criterion. When sitting in front of a computer or holding a mobile device, the physical reality is not completely shut off. Users can still be aware of the physical surroundings around them: they can see the desk over here, the chairs over there, etc. In addition, they have to see everything through the computer screens, as they are watching the video within the boundaries of a frame. Whereas when watching 360° videos on HMDs, the frame disappears, enabling viewers to enter the mediated world. Therefore, using computers or mobile devices to watch 360° videos does not create an immersive mediated environment. 2.2.2 Presence A key psychological experience that is generated by IMEs is the feeling of presence (Biocca, 1997; Heeter, 1992; Lee, 2004; Lombard & Ditton, 1997; Slater & Wilbur, 1997; Steuer, 1992). There are three types of presence revealed in the 15 literature: spatial presence (telepresence), social presence (co-presence), and self- presence (Aymerich-Franch, Karutz, & Bailenson, 2012; Biocca, 1997; Lee, 2004). Spatial presence refers to the sense of being located in the mediated environments. For example, Steuer (1992) defined spatial presence as “the extent to which one feels present in the mediated environment, rather than in the immediate physical environment” (p. 76). Witmer and Singer (1998) had a similar definition, “the subjective experience of being in one place or environment, even when one is physically situated in another” (p. 225). Immersive media technologies trigger the sense of spatial presence among users. Viewing a 360° video on HMDs, for example, cuts off the real environment from users and puts them in the environments created by media. Thus, users form the conviction that he or she is physically located in the mediated environments. Spatial presence occurs as long as when users feel they are situated within the spatial environments generated by the media. It does not require a story plot or a media character to produce the sense of spatial presence. That is to say, spatial presence is conceptually different from transportation. In particular, transportation concerns viewers’ attention focusing, imagery evoking, and emotional arousing, which are the responses to narratives. In contrast, spatial presence is not necessarily generated by the specific media message, but by the media technological features. Previous research has found that special presence is an important mechanism that drives social influence and persuasion in IMEs (Fonseca & Kraus, 2016; Grigorovici, 2003; Li, Daugherty, & Biocca, 2002; Sundar et al., 2017). Li et al. (2002) conducted two experiments to explore whether three-dimensional (3-D) 16 advertising is more effective than two-dimensional (2-D) advertising through spatial presence. The authors stated that 3-D advertising allowed consumers to fully explore the products on the Web as if they were in the retail store, hence creating a seemingly virtual experience. Their findings supported the hypotheses that 3-D (vs. 2-D) advertising resulted in a greater sense of presence and ultimately greater product knowledge, more favorable brand attitude, and increased purchase intention. In a more recent study, Fonseca and Kraus (2016) compared the effects of 360° videos presented in head-mounted versus hand-held displays on viewers’ attitude and behavior change. The authors found that the video displayed on HMDs led to greater presence than the one displayed on tablet screens. In addition, participants who viewed the 360° video with HMDs reported more message-consistent attitudes and performed greater message-consistent behavior than those with tablets. Similar findings were obtained by Sundar and colleagues (2017). According to the meta-analysis conducted by Cummings and Bailenson (2016), spatial presence is affected by immersive features, such as image and sound quality, stereoscopic vision, tracking level, etc. Additionally, because spatial presence is inherently a user’s experience, it has been found to be influenced by individual characteristics, including trait empathy and immersive tendencies (Wallach, Safir, & Samana, 2010). Similar to spatial presence, social presence is a psychological experience triggered by technological attributes. It is defined as “the sense of being with another”, or more specifically, “the sense of another through a medium” (Biocca, Harms, & Burgoon, 2003, p. 456). In IMEs, users experience social presence through 17 their interactions with another media character or virtual avatar/agent or the mere perception that he or she exists in the same place with the character or avatar/agent Unlike spatial presence, social presence occurs when at least another media character or virtual avatar/agent exists in the mediated environments. Social presence also facilitates attitudinal and behavioral change (Fortin & Dholakia, 2005; Skalski & Tamborini, 2007). Fortin and Dholakia (2005) found that interactive web-based advertisement had a moderate effect on social presence, which in turn affected advertising effectiveness. In another study, Skalski and Tamborini (2007) examined whether interactive and attractive virtual agents affected social presence and persuasion. Their results demonstrated that interactive (vs. non- interactive) virtual agents increased the sense of social presence, which in turn influenced attitude and behavioral intentions. Likewise, social presence is also affected by immersive qualities, such as audio quality (Skalski & Whitbred, 2010), and individual differences, such as transportability (Lee & Shin, 2014). Self-presence is a unique psychological experience created by VR. It is defined as “the psychological state in which virtual self is experienced as the actual self” (Aymerich-Franch et al., 2004, p. 1). This dimension of presence emphasizes how users feel connected with his or her virtual body, identity, and emotions (Biocca, 1997; Ratan & Hassler, 2010) in the virtual environments. Because the current study focuses on the IMEs generated by 360° videos, which do not provide a virtual avatar for the users; self-presence is not discussed further here. However, future research that investigates the effects of narratives in immersive virtual environments should take self-presence into consideration. 18 In short, spatial and social presence are two core psychological mechanisms generated by immersive media technologies. They are affected by technological affordances (i.e., technological attributes of media) as well as user characteristics. Presence also mediates the relationship between mediated environments and persuasive outcomes. 2.3 Modeling Narrative Effects in Immersive Mediated Environments 2.3.1 The Persuasive Narrative Theory in Immersive Mediated Environments (PENTIMEs) By integrating the theoretical frameworks of narrative persuasion and immersive media, the current study proposes the persuasive narrative theory in immersive mediated environments (PENTIMEs) (see Figure 2.1). The major assumption of PENTIMEs is that the psychological experience (e.g., spatial and social presence) triggered by technological affordances intensifies the viewers’ involvement with the story (e.g., transportation and identification). The enhanced story involvement reduces persuasion resistance, which in turn promotes greater story-consistent attitudes and behavioral intentions. This assumption is based on the Modality-Interactivity-Agency-Navigability (MAIN) model (Sundar, 2008). In the MAIN model, Sundar (2008) proposed that technological affordances served as cues to trigger cognitive heuristics, which would then affect users’ perceptual experience with the media. He defined a heuristic as a judgment rule that could influence users’ perceptions toward the media content. For example, media modality (i.e., the channel through which the information is 19 presented) is perhaps one of the most fundamental technological affordances. 360° videos delivered via HMDs differ in modality with those presented in flat computer screens. In fact, the former is richer in modality than the latter. As demonstrated in the previous section, the experience of viewing 360° videos on HMDs affords users the opportunity to fully immerse themselves in the mediated environments through shutting them out from their own physical reality. In other words, watching 360° videos on HMDs replaces the cues of the physical environment in which one is actually present. In IMEs, all reminders of the physical environments, such as light and sound, are blocked off. Therefore, the complete immersion afforded by HMDs serves as a cue to trigger users’ sense of being there in the mediated environments, which in turn affects their responses to the media content, such as the emotional and cognitive involvement with the message. In PENTIMEs, I propose that media format (i.e., IMEs vs. non-IMEs) will have a direct positive effect on spatial presence and social presence based on previous research (Fonseca & Kraus, 2016; Sundar et al., 2017). In addition, narratives presented in IMEs (vs. non-IMEs) will lead to a higher level of transportation and identification due to its greater immersive nature. H1: Exposure to a narrative presented in immersive (vs. non-immersive) mediated environments will lead to a higher level of a) spatial presence and b) social presence. H2: Exposure to a narrative presented in immersive (vs. non-immersive) mediated environments will lead to a higher level of a) transportation and b) identification. 20 Furthermore, spatial presence mediates the relation between media format and transportation. In a similar fashion, social presence mediates the association between media format and identification. It is noteworthy that spatial presence and social presence intensifies transportation and identification, respectively. It is because spatial presence refers to the mere perception of being situated in the mediated environments, whereas social presence describes the feeling of being together with a media character in the mediated environments. In other words, spatial presence concerns the perceptions toward the mediated environments in general, while social presence deals with the cognitive experience with another virtual avatar or agent in the mediated environments. As distinguished early, transportation and identification are two related but distinct constructs. Transportation refers to the degree of the absorption with the narrative as a whole, whereas identification describes the involvement with the specific story character. Both spatial presence and transportation are not necessarily impacted by the media character. However, social presence and identification deal specifically with the media character. In order to delineate the specific impact of different types of presence on different kinds of involvement with the narrative, this model proposes that spatial and social presence distinctively influences transportation and identification. H3: Media format will have an indirect effect on transportation through spatial presence. H4: Media format will have an indirect effect on identification through social presence. 21 Moreover, as counterarguing has been found to have a negative relation to transportation and identification (McQueen et al., 2011; Moyer-Gusé & Nabi, 2010), media format is predicted to have an indirect effect on counterarguing through spatial presence, social presence, transportation, and/or identification. As seen from Figure 2.1, there are four specific indirect effects from media format on counterarguing: one through transportation, one through identification, one through spatial presence and transportation, and one through social presence and identification. H5: Media format will have an indirect effect on counterarguing through a) spatial presence, b) social presence, c) transportation, and/or d) identification. As recent studies found that 360° video messages presented with HMDs (vs. flat computer screens) led to greater message-consistent attitudes and behaviors (Fonseca & Kraus, 2016; Sundar et al., 2017), PENTIMEs also proposes that narratives presented in IMEs (vs. non-IMEs) will lead to greater persuasiveness. In addition, media format will have an indirect effect on story relevant attitudes/behavioral intentions through proposed psychological mechanisms. There are 10 specific indirect effects from media format on persuasive outcomes: one through spatial presence, one through social presence, one through transportation, one through identification, one through spatial presence and transportation, one through social presence and identification, one through transportation and counterarguing, one through identification and counterarguing, one through spatial presence, transportation, and counterarguing, and one through social presence, transportation, and counterarguing. Finally, it is not clear whether media format will have a direct effect on story-consistent a) attitudes and b) behavioral intention/willingness after 22 controlling for spatial presence, social presence, transportation, identification, and counterarguing. The following hypotheses and research questions are proposed. H6: Exposure to a narrative presented in immersive (vs. non-immersive) mediated environments will lead to more story-consistent a) attitudes and b) behavioral intention/willingness. H7: Media format will have an indirect effect on story-consistent attitudes through a) spatial presence, b) social presence, c) transportation, d) identification, and/or e) counterarguing. H8: Media format will have an indirect effect on story-consistent behavioral intentions/willingness through a) spatial presence, b) social presence, c) transportation, d) identification, and/or e) counterarguing. RQ1: Will media format have a direct effect on story-consistent a) attitudes and b) behavioral intention/willingness after controlling for spatial presence, social presence, transportation, identification, and/or counterarguing? 23 Figure 2.1. The theoretical main model. Spatial presence and social presence are correlated. Transportation and identification are correlated. Attitudes and intentions/willingness are correlated. 2.3.2 Trait Empathy as a Moderator As discussed earlier, psychological experience triggered by media affordances or generated by narrative experience are affected by individual differences. It is because these psychological mechanisms are inherently users’ experience and can be influenced by user characteristics. For presence, the literature has identified several personality traits that influences presence. These include trait empathy (Nicovich, Boller, & Cornwell, 2005; Sas & O’Hare, 2003), immersive tendencies (Wallach et al., 2010), trait absorption (Wirth, Hofer, & Schramm, 2012), etc. In the narrative persuasion literature, previous scholars found that need for affect (Appel & Richter 2010), need 24 for cognition (Zwarun & Hall, 2012), transportability (Bilandzic & Busselle, 2008), and trait empathy (Hall & Bracken. 2011) impacted the extent to which individuals were transported to the narrative world or identified with the narrative character. A common personality trait identified by the literature in narrative persuasion and immersive media is trait empathy. Hence, this study also tests the moderating role of the trait empathy. Trait empathy is a personality trait that describes individuals’ tendencies to experience empathy. Davis (1980) claimed that trait empathy was a multidimensional construct, including perspective-taking, fantasy empathy, empathic concern, and personal distress. Perspective-taking refers to individuals’ tendencies to adapt the point of view of other people. Fantasy empathy reflects a tendency to be involved in the fictional world of books, movies, and plays. Empathic concern taps on the tendencies of individuals to feel sympathy, compassion, and concern for unfortunate others. Personal distress describes individuals’ feelings of anxiety, unease, and discomfort in intense interpersonal contexts. The scales of perspective-taking and fantasy empathy measure individuals’ empathetic cognitive responses. Empathic concern and personal distress describe individuals’ tendencies in affective responses. As earlier studies found a positive relation between fantasy (a sub-dimension of the trait empathy) and presence (Nicovich et al., 2005; Sas & O’Hare, 2003; Wallach et al., 2010). I propose trait empathy will moderate the impact of media format on spatial and social presence. Spatial and social presence will be more strongly influenced by media format among those high (vs. low) in trait empathy. Likewise, transportation and identification will be more strongly influenced by media 25 format among those high (vs. low) in trait empathy. Please see Figure 2.2 for the interaction model. H9: Trait empathy will moderate the impact of media format on spatial presence, social presence, transportation, and identification such that a) spatial presence, b) social presence, c) transportation, and d) identification will be more strongly influenced by media format among individuals high (vs. low) in trait empathy. Figure 2.2. The theoretical interaction model. Spatial presence and social presence are correlated. Transportation and identification are correlated. Attitudes and intentions/willingness are correlated. 26 Chapter 3: Pilot Studies Two pilot studies were conducted preceding the main experimental study. This chapter describes the two pilot studies. The pilot studies and the main study were approved by the Institutional Review Board at the University of Maryland. 3.1 Pilot Study 1 3.1.1 Study Objectives The first objective of the current pilot study was to investigate if the picture and sound quality differed when the same 360° video were viewed on two devices: HMDs and flat computer screens. The major concern was that the laptops (DELL Latitude E6440) provided in our lab were outdated. In particular, the pixel density, which describes the sharpness and clarity of the displays, is relatively low (i.e., 112 pixels per inch, ppi) compared to newer models of laptops. As a comparison, the pixel density of Dell Latitude 12 7000 is 352.47ppi (Laptop Pixel Density (PPI) List, n.a.)2. The low pixel density of the laptops provided in our lab might lead to poorer picture quality for the non-IMEs condition than the IMEs condition, which could potentially pose internal validity threat to the study. Therefore, a pilot study should be conducted to compare the picture and sound quality across two conditions. The second objective was to identify any potential problems that might occur during the lab session. In particular, the current study involves viewing a 360° video, a new video format, 2 The pixel density of Samsung Galaxy S7 smartphone is 577ppi. It is important to note that smartphones are built with higher pixel density than laptops in general. The reason is that laptops are designed to use in 2 to 4 feet away from the human eyes but the designed distance for smartphones is 1 to 2 feet. 27 which some participants might not have experienced before. The pilot study, therefore, aimed to improve the efficiency of the main study data collection through identifying and addressing the logistical issues. 3.1.2 Participants and Apparatus 36 undergraduates participated in the study in exchange for class extra credits. The mean age of the participants was 19.61 years old. About half of the participants were White (47.2%), followed with 19.4% Asian, 16.7% Black or African American, 11.1% Hispanic, and 5.6% other. There were slightly more female (52.8%) than male (47.2%) participants. 3.1.3 Design and Procedure Pilot study 1 consisted of two phases. Phase I was an online questionnaire. After digitally signing the consent form, participants were asked to complete a questionnaire that included measures of demographics (age, sex and race), trait empathy, and baseline attitudes toward DUI. Then they were invited to participate in phase II, a laboratory-based study. The average gap between participating in phase I & II was 7 days, ranging from 0 day to 15 days. Participants were randomly assigned to watch the 360° video either in the IMEs condition or the non-IMEs condition. Participants in the IMEs condition watched the video on Samsung Gear VR 2016 edition with Samsung Galaxy S7 smartphone. They were instructed to move their heads to look around when wearing the headset. Before they watched the main video, participants in the IMEs condition also watched a pilot video to adjust and fit the headset. Introducing the Daily 360 was used as the pilot video, which was a one-and- 28 half-minute long video produced by New York Times’ journalists. The video portrays the events happening around the world in 360 degrees. For participants in the non- IMEs condition, they watched the video on DELL Latitude E6440 laptop. They were instructed to pan around the video by moving around on the touchpad. Both groups of participants used Samsung Galaxy S7 earphones when viewing the stimulus. After watching the video, participants were asked to complete questions measuring their attitudes toward DUI and psychological experience3. They were also asked to rate the picture and sound quality of the media experience. At the end of the study, participants were debriefed and thanked. 3.1.4 The 360° Video A 360° video named “Decisions: a 360° virtual reality drunk driving experience” was used as the experimental stimulus. This video is part of Johnnie Walker’s Join the Pact campaign established in 2016, aiming to prevent driving under the influence of alcohol (DUI). The video features stories of different groups of people in three vehicles. In the first vehicle, there was a young woman, Samantha, driving after she had several drinks celebrating success at her job. In another vehicle, there was a couple going out for their first date night after their daughter was born. In the last vehicle, there was a group of college-aged friends enjoying a night out. Due to the influence of alcohol, Samantha’s car collided with two other vehicles. The car crash led to five innocent people to be severely injured or dead. This video is 4 minutes and 41 seconds long. See Appendix A for the video transcript. 3 Measures were not presented in this section because they were not focal point in the first pilot study. 29 3.1.5 Results The first goal of the current pilot study was to test whether the picture and sound quality differed across two experimental condition. Participants were asked to indicate how they thought about the picture and sound quality, respectively, during the media experience on a seven-point scale (1= very poor, 7 = very good). The results showed there were no significant differences between the groups in terms of picture quality, F(1, 34) = .068, p = .80, partial h2 = .002, nor sound quality, F(1, 34) = .004, p = .95, partial h2 = .000. The non-significant results might be due to the small sample size. Nonetheless, the descriptive statistics indicated that participants in the IMEs condition, M = 4.32, SD = 1.67, rated the picture quality slightly better than those in the non-IMEs condition, M = 4.18, SD = 1.51. Sound quality was rated almost the same by two groups of participants: IMEs condition, M = 5.79, SD = 1.13, and non-IMEs condition, M = 5.76, SD = 1.30. The descriptive statistics suggested that difference in terms of picture quality displayed with our laptops vs. smartphones was worrisome. Therefore, I should consider using a flat computer screen with higher pixel density. The second goal was to identify potential issues that might occur during the lab-session data collection. First, the use of touchpad to pan around the 360° video emerged as a significant issue. In the IMEs condition, participants watched the video by moving their heads in different directions. This level of interactivity was different from viewing the video by moving their figures on the touchpad. Therefore, a more comparable control group would be using a mobile device, such as an iPad, which affords individuals to look around by moving their hands that hold the mobile 30 devices. In the next pilot study and main study, I decided to use iPads for the non- IMEs condition. iPads have a higher pixel density (264ppi) and allow users to move their hands. This level of interactivity (i.e., moving hands) is comparable with that in the IMEs condition (i.e., moving heads). Another issue was concerned with participants who were not able to focus the video in the IMEs condition. In this case, I asked them to take off the headsets and quit the study. It gave me a hint that in my future data collection, I should warn the participants in the IMEs condition beforehand that they might not be able to adjust the focus successfully. If that happened, they should let me know and pause the study. These participants shall be excluded from the data analysis. 3.1.6 Conclusions In summary, two lessons were learned from pilot study 1. First, iPads should replace laptops in the non-IMEs condition. It is because iPads have a higher pixel density and afford users to watch the 360° video by moving their hands. Second, I need to ask participants to take off their headsets if they are not able to adjust the focus of the video. 3.2 Pilot Study 2 Because I decided to replace laptops with iPads in the non-IMEs condition, I conducted a second pilot study to identify potential problems involved with in-lab data collection. Moreover, I wanted to examine the item reliability of the main outcome measures. Finally, I wanted to provide a pilot test of the key hypotheses. In particular, I wanted to pretest whether participants in the IMEs condition had a higher 31 level of spatial presence, social presence, transportation, and identification than those in the non-IMEs condition. 3.2.1 Participants and Apparatus Pilot study 2 was conducted during the summer of 2017. Those who completed pilot study 1 were not allowed to sign up for pilot study 2. A total of 16 undergraduate students participated in both phases of the second pilot study. Participants were in the average age of 22.44 years old, with majority female students (68.8%). Among them, 62.5% were White, 25.0% Black, 6.3% Hispanic, and 6.3% Asian. 3.2.2 Design and Procedure The design and procedure of the second pilot study was similar to the first pilot study. The only difference was that in the non-IMEs condition, participants watched the video on Apple iPad Air 2, which has a pixel density of 264 ppi. Participants were instructed to move the iPads with their hands to view the video from different angles. The average gap between participating in phase I & II was 7 days, ranging from 1 day to 14 days. 3.2.3 The 360° Video The 360° video played in the first pilot study was used again as the experimental stimulus. 32 3.2.4 Key Dependent Variables Spatial presence. Four items were adapted from the spatial presence scale of Temple Presence Inventory (TPI, Lombard et al., 2009). Sample items included “How much did it seem as if the objects and people you saw/heard had come to the place you were?”, and “How much did it seem as if you could reach out and touch the objects or people you saw/heard?” Responses to the four items were averaged into a composite score (M = 3.97, SD = 1.31, Cronbach’s α = .81). Social presence. To assess social presence, four items were adapted from the social presence – actor with medium scale of Temple Presence Inventory (TPI, Lombard et al., 2009). Example items included “How often did you have the sensation that people you saw/heard could also see/hear you?”, and “To what extent did you feel you could interact with the person or people you saw/heard?” An overall score of social presence was created by averaging the four items (M = 3.14, SD = 1.35, Cronbach’s α = .88). Transportation. To assess the extent to which participants were transported into the narrative world, five items culled from previous research (e.g., Green & Brock, 2000; Ma & Nan, in press) were used. These items captured its major dimensions of transportation, including emotional involvement with the story, cognitive attention to the story, and lack of awareness of the physical environments. Examples of the items were “The story affected me emotionally,” and “I was mentally involved in the story while watching it.” An average score was obtained (M = 5.49, SD = 1.15, Cronbach’s α = .78). 33 Identification. Identification was measured with four items adapted from Cohen (2001) which tapped on viewers’ affective and cognitive involvement with the character. Sample items included “When I was watching the story, I could feel the emotions the character(s) expressed;” and “I was able to understand the events in this video in a manner similar to that in which the character(s) understood them.” (M = 5.30, SD = 0.92, Cronbach’s α = .85). Counterarguing. Counterarguing was measured with three different questions: “While watching this video, I found myself actively disagreeing with it;” “I found myself actively disagreeing with the video,” and “I was looking for flaws in the way information was presented in the video.” The items were used in previous studies (e.g., Nabi, Moyer-Gusé, and Byrne, 2007) and averaged to create an overall score of counterarguing (M = 2.23, SD = 1.13, Cronbach’s α = .78). Posttest attitudes toward DUI. Posttest attitudes toward DUI were measured with four semantic differential items. The items followed the stem, “Now, please think about how you feel about driving under the influence of alcohol at this moment. Would you say driving under the influence of alcohol is…” and included “bad-good, foolish-wise, negative-positive, unfavorable-favorable.” Responses were averaged across the four items to create an overall measure of posttest attitudes (M = 1.24, SD = 0.66, Cronbach’s α = .98). Posttest intentions to engage in DUI. Three questions were adapted from previous research to assess participants’ posttest intentions to engage in DUI (Burrows & Blanton, 2016). The three questions were “How likely is it that, in the next 30 days, you would drive a vehicle when you feel a ‘slight buzz’ from drinking 34 alcohol?”; “How likely is it that, in the next 30 days, you would help to drive a friend home when you’re slightly feeling some initial effects from having a few alcoholic drinks?”; and “How likely is it that, in the next 30 days, you would be a passenger in a car driven by someone who had been drinking?” Responses were on 10-point scales (1 = likely; 10 = extremely likely) and averaged to form an overall index (M = 2.31, SD = 2.05, Cronbach’s α = .95). 3.2.5 Results The major issue was concerned with blurry images displayed on iPads. The video was played on YouTube application, which required Internet. However, the Wi- Fi in the research lab was not stable, which caused the blurring image. Therefore, in the main study data collection, I needed to download the video and play it on a local VR player4. In addition, one participant also reported that he was not able to focus the video and was excluded from the study. In addition, the main outcome measures used in the study were reliable. As seen from the measurement section, the measures for all major outcome variables had a Cronbach’s α that was greater than .75, which indicated good reliability. Furthermore, a multivariate analysis of covariance (MANCOVA) was conducted to provide a pilot test of the major hypotheses. The experiment condition was entered as the predictor, spatial presence, social presence, transportation, and identification as outcome variables, and gender as the covariate. Gender was included as the covariate because literature identified that gender affected users’ formation of 4 The video played in the IMEs condition was downloaded and stored in the Samsung VR app. 35 spatial and social presence (Felnhofer et al., 2014; Lombard & Ditton, 1997; Nicovich et al., 2005). Possibly due to the small sample size, media format did not have a significant effect on spatial presence, F(1, 13) = .66, p = .43, partial h2 = .009, nor social presence, F(1, 13) = .28, p = .61, partial h2 = .000, nor transportation, F(1, 13) = .19, p = .67, partial h2 = .033, nor identification, F(1, 13) = .03, p = .86, partial h2 = .001. However, the descriptive statistics showed that the mean of spatial presence was higher in the IMEs condition (M = 4.26, SE = .47) than the mean in the non-IMEs condition (M = 3.74, SE = .42). Similar pattern was shown for social presence such that the mean of social presence was higher in the IMEs condition (M = 3.33, SE = .47) than the mean in the non-IMEs condition (M = 3.00, SE = .41). Moreover, participants in the IMEs condition (M = 5.35, SE = .38) experienced a higher level of identification than those in the non-IMEs condition (M = 5.26, SE = .33). However, participants in the IMEs condition (M = 5.34, SE = .45) experienced a lower level of transportation than those in the non-IMEs condition (M = 5.60, SE = .39). Overall, the descriptive statistics from the pilot study showed that IMEs led to a higher mean of spatial presence, social presence, and identification than non-IMEs. These findings were consistent with the hypothesis. Transportation, however, was found to have a lower mean in the IMEs (vs. non-IMEs). I suspect that transportation might be impacted by cybersickness. Cybersickness is a well-documented side effect associated with VR devices, such as HMDs (Howarth & Costello, 1997; Mon- Williams, Warm, & Rushton, 1993). It includes symptoms like general and vision discomfort. In the current study, participants in the IMEs condition might experience 36 cybersickness, which could potentially decrease their level of transportation. Therefore, in the main study, cybersickness should be measured and controlled in the statistical analysis. 3.2.6 Conclusions The first lesson learnt from pilot study 2 was that the video stimuli should be downloaded and played on a local VR player in iPads. Moreover, pilot study 2 showed that the main outcome measures used in the study were reliable. Pilot study 2 also provided some preliminary evidence that supported major hypothesis. Finally, cybersickness should be measured and controlled in the statistical analysis. 37 Chapter 4: Method of Main Study 4.1 Participants and Apparatus Participants were undergraduate students recruited from communication classes at a large Mid-Atlantic university in exchange for extra credit. Participants who completed pilot study 1 or pilot study 2 were not qualified to sign up for the main study. For phase I online questionnaire, a total of 394 students participated in the study. For phase II laboratory session, 216 out of the 394 participants signed up and completed the study (retention rate is 54.82%). Therefore, the sample size for the main study is 216. Participants averaged the 19.42 years of age, ranging from 18 years to 36 years old. The sample was 53.2% White, 22.2% Asian, 12.5% Black or African American, 8.8% Hispanics, and 3.2% reporting other racial backgrounds. Slightly more women (54.6%) than men (45.4%) participated in the research5. 4.2 Design and Procedure Similar with pilot study 1 and 2, the main study involved two phases. Phase I was an online questionnaire. After digitally signing the consent form, participants were asked to complete a questionnaire that included measures of demographics (age, sex and race), trait empathy, and social desirability. Their baseline attitudes toward DUI and general attitudes toward helping others in need were also included. It took 5 For those who dropped out of the study (N = 178), their mean age was 19.58, ranging from 17 years to 37 years old. There were 56.2% White, 20.2% Asian,14.0% Black or African American, 5.1% Hispanics, and 4.5% reporting other racial backgrounds. More females (57.3%) dropped out of study than males (42.7%). 38 about 15 minutes to complete phase I. During phase II, participants were invited to a communication research lab. The average time lag between phase I and II was 7 days, ranging from 1 day to 70 days. Participants were randomly assigned to one of two conditions: watching the 360° narrative videos in IMEs (on Samsung Gear VR) or in non-IMEs (on iPad). All participants in both conditions watched two videos: the video on DUI and the video on malaria. The reason to use two videos was to test the hypotheses and research questions in different contexts. The videos were randomly played for each participant. After watching each video, participants were asked to complete a set of the questions on iPads based on the previous video viewing experience. Measures of spatial presence, social presence, transportation, identification, counterarguing, and cybersickness were identical for both video experience. Measures of persuasive outcomes were unique for each video context. For example, if a participant watched the DUI video first, then he/she was asked to complete a set of questions based on the DUI video viewing experience. Then he/she watched the malaria video and completed a second set of questions based on the malaria video experience. At the end of the study, participants were debriefed and thanked. Phase II took about an hour to an hour and a half to finish. 4.3 360° Videos 4.3.1 The 360° Video on DUI A 360° video named “Decisions: a 360° virtual reality drunk driving experience” was used as the experimental stimulus. This video is part of Johnnie Walker’s Join the Pact campaign established in 2016, aiming to prevent driving 39 under the influence of alcohol (DUI). The video features stories of different groups of people in three vehicles. In the first vehicle, there was a young woman, Samantha, driving after she had several drinks celebrating success at her job. In another vehicle, there was a couple going out for their first date night after their daughter was born. In the last vehicle, there was a group of college-aged friends enjoying a night out. Due to the influence of alcohol, Samantha’s car collided with two other vehicles. The car crash led to five innocent people to be severely injured or dead. This video is 4 minutes and 41 seconds long. See Appendix A for the video transcript. 4.3.2 The 360° Video on Malaria United Nations Virtual Reality released a 360° video named “Under the Net” in March 2017. This video is part of the Nothing but Nets campaign whose goal is to raise awareness, funds, and voices to fight malaria. The video features a story of a 11- year-old girl, Amisa, a refugee who lives in the Nyarugusu Refugee Camp in Tanzania with her mother and siblings. From a first-person perspective, Amisa tells the story of her life living in the refugee camp. Due to the lack of protection, Amisa and her siblings are infected with malaria. Life gets better when her family moves into a new tent, receives treatment for malaria, and gets bed nets to protect them from malaria-carrying mosquitoes. The original video is 8 minutes and 28 seconds long. After editing, the video used in the experiment lasts 7 minutes and 3 seconds. See Appendix A for the video transcript. 40 4.4 Measures Unless otherwise noted, all questions were measured on a 1-7 scale (1 = strongly disagree/not at all, 7 = strongly agree/very much). See Appendix B for the full instrument. 4.4.1 Phase I Measures Baseline attitudes toward DUI. Baseline attitudes toward DUI were measured with four semantic differential items. Sample items included “In your view, driving under the influence of alcohol is bad/good;” and “In your view, driving under the influence of alcohol is foolish/wise.” The items were scored on a seven-point scale from 1 = “bad/foolish/negative/unfavorable” to 7 = “good/wise/positive/favorable”. Ratings were averaged across the four items to create an overall measure of baseline attitudes toward DUI (M = 1.08, SD = 0.27, Cronbach’s α = .73). This variable served as a covariate in the main statistical analysis in the DUI context, as it allowed me to test the effect of media format on posttest attitudes toward DUI after controlling the baseline attitudes. Attitudes toward helping people. Attitudes toward helping people were gauged with four items. Examples of the items included “People should be willing to help others who are less fortunate,” and “Helping troubled people with their problems is very important to me.” These items were used in previous studies that promote humanitarian aid (e.g., Webb, Green, & Brashear, 2000). A composite score was created by averaging the four items (M = 5.80, SD = 0.84, Cronbach’s α = .89). This variable served as a covariate in the main statistical analysis in the malaria context, as it allowed me to test the effect of media format on posttest attitudes toward helping 41 people who need protection after controlling the attitudes toward helping people in general. Social desirability. To assess social desirability, nine items were adapted from Crowne and Marlowe (1960). Example items included “I’m always willing to admit it when I make a mistake,” and “I have never deliberately said something that hurt someone’s feelings.” The nine items were averaged to create an overall score of social desirability (M = 4.50, SD = 0.81, Cronbach’s α = .72). Social desirability was used as a covariate in the statistical analysis in both contexts to control for response bias when self-reporting attitudes and behavioral intentions. Trait empathy. Davis’ (1980) trait empathy scale was used to assess participants’ individual differences in empathy. Because previous research found fantasy empathy was related to presence and transportation (Hall & Bracken. 2011; Nicovich et al., 2005; Sas & O’Hare, 2003; Wallach et al., 2010), only the fantasy scale was used in the current study. Examples of the fantasy scale included “I daydream and fantasize, with some regularity, about things that might happen to me;” and “When I am reading an interesting story or novel, I imagine how I would feel if the events in the story were happening to me.” Participants’ responses were averaged into an overall index of trait empathy (M = 5.11, SD = 0.90, Cronbach’s α = .76). Demographic and social information. Participants’ demographic information (e.g., gender, ethnicity, and age) was included in the phase I questionnaire. Participants’ Greek Life affiliation was also collected by indicating if they are affiliated with a fraternity/sorority. Greek Life affiliation was controlled in the statistical analysis in the DUI context, as previous research found that 42 fraternity/sorority membership was associated with driving after drinking (LaBrie, Kenney, Mirza, & Lac, 2011). 4.4.2 Phase II Measures Spatial presence. Four items were adapted from the spatial presence scale of Temple Presence Inventory (TPI, Lombard et al., 2009). Sample items included “How much did it seem as if the objects and people you saw/heard had come to the place you were?”; and “How much did it seem as if you could reach out and touch the objects or people you saw/heard?” Participants responded to special presence measure twice, one for each video. Responses to the four items were averaged into a composite score for each video (for DUI video: M = 3.60, SD = 1.52, Cronbach’s α = .86; for malaria video: M = 3.72, SD = 1.52, Cronbach’s α = .85). Social presence. To assess social presence, four items were adapted from the social presence – actor with medium scale of Temple Presence Inventory (TPI, Lombard et al., 2009). Example items included “How often did you have the sensation that people you saw/heard could also see/hear you?”; and “To what extent did you feel you could interact with the person or people you saw/heard?” An overall score of social presence for each video was created by averaging the four items (for DUI video: M = 2.68, SD = 1.40, Cronbach’s α = .87; for malaria video: M = 3.60, SD = 1.56, Cronbach’s α = .83). Transportation. To assess the extent to which participants were transported into the narrative world, five items culled from previous research (e.g., Green & Brock, 2000; Ma & Nan, in press) were used. These items captured its major dimensions of transportation, including emotional involvement with the story, 43 cognitive attention to the story, and lack of awareness of the physical environments. Examples of the items were “The story affected me emotionally,” and “I was mentally involved in the story while watching it.” An average score was obtained for each video (for DUI video: M = 5.64, SD = 0.97, Cronbach’s α = .73; for malaria video: M = 5.50, SD = 1.02, Cronbach’s α = .75). Identification. Identification was measured with four items adapted from Cohen (2001) which tapped on viewers’ affective and cognitive involvement with the character. Sample items for DUI video include “When I was watching the story, I could feel the emotions the character(s) expressed;” and “I was able to understand the events in this video in a manner similar to that in which the character(s) understood them.” Because there were several characters involved in the DUI video, participants were asked to rate the items first and then write down the name(s) or a brief description of character(s) they had in their mind when they rated the above statements (M = 5.21, SD = 0.99, Cronbach’s α = .74). The majority of the participants (83.94%) had the main character in mind when they answered the questions. Items for the malaria video were identical with those for the DUI video, except that “character(s)” was replaced with “Amisa”, the name of the main character in the malaria video. A composite score of identification with Amisa was created by averaging the four items (M = 4.91, SD = 1.16, Cronbach’s α = .77). Counterarguing. Counterarguing was measured with three different questions: “While watching this video, I found myself actively disagreeing with it;” “I found myself actively disagreeing with the video,” and “I was looking for flaws in the way information was presented in the video.” The items were used in previous studies 44 (e.g., Nabi et al., 2007) and averaged to create an overall score of counterarguing (for DUI video: M = 2.32, SD = 1.20, Cronbach’s α = .74; for malaria video: M = 1.90, SD = 0.96, Cronbach’s α = .74). Cybersickness. Participants were asked to indicate to what extent they felt uncomfortable, dizzy, fatigue, and lightheaded, and experienced vision discomfort and eyestrain during the media experience. These items were adapted from the Virtual Reality Symptom Questionnaire (Ames, Wolffsohn, & Mcbrien, 2005). An overall index of cybersickness was created by averaging the six items (for DUI video: M = 2.92, SD = 1.62, Cronbach’s α = .92; for malaria video: M = 2.77, SD = 1.64, Cronbach’s α = .93). Posttest attitudes toward DUI. Posttest attitudes toward DUI were again measured with four semantic differential items. The items followed the stem, “Now, please think about how you feel about driving under the influence of alcohol at this moment. Would you say driving under the influence of alcohol is…” and included “bad-good, foolish-wise, negative-positive, unfavorable-favorable.” Responses were averaged across the four items to create an overall measure of posttest attitudes (M = 1.08, SD = 0.24, Cronbach’s α = .88). Posttest intentions to engage in DUI. Three questions were adapted from previous research to assess participants’ posttest intentions to engage in DUI (Burrows & Blanton, 2016). The three questions were “How likely is it that, in the next 30 days, you would drive a vehicle when you feel a ‘slight buzz’ from drinking alcohol?”; “How likely is it that, in the next 30 days, you would help to drive a friend home when you’re slightly feeling some initial effects from having a few alcoholic 45 drinks?”; and “How likely is it that, in the next 30 days, you would be a passenger in a car driven by someone who had been drinking?” Responses were on 17-point scales (1 = completely unlikely; 17 = completely likely) and averaged to form an overall index (M = 2.52, SD = 2.57, Cronbach’s α = .78). In order to detect greater variability among the extreme values, the anchors of the response metric were expanded (Pelham & Blanton, 2012). Posttest attitudes toward helping people who need protection from malaria. Attitudes toward helping people were measured by asking participants whether they thought that helping people like Amisa to get bed nets as a protection from malaria was: bad-good, foolish-wise, negative-positive, unfavorable-favorable. Participants rated the items on 7-point semantic differential scales. An index of posttest attitudes was constructed by calculating the means of the four items (M = 6.69, SD = 0.70, Cronbach’s α = .92). Posttest willingness to help people who need protection from malaria. Four items were constructed to evaluate participants’ posttest willingness to engage in actions to help people who need protection from malaria. Similar items were used in published articles that measure willingness to help in the humanitarian aid context (e.g., Peng, Lee, & Heeter, 2010). Participants were asked to indicate how willing they were to a) follow Nothing But Nets campaign on social media to learn more about the fight to defeat malaria; b) share this video with their family/friends to disseminate the message about the situation of the people who need protection from malaria; c) donate money to send nets and help save lives; and d) join the Nothing But Nets Champions Council and become a leader to defeat malaria. Responses were on 46 17-point scales (1 = completely unwilling; 17 = completely willing) and averaged to create a composite score of behavioral willingness to engage in actions to help people who need protection from malaria (M = 10.25, SD = 3.42, Cronbach’s α = .85). In order to detect greater variability among the extreme values, the anchors of the response metric were expanded (Pelham & Blanton, 2012). 47 Chapter 5: Results of Main Study 5.1 Sample Characteristics Several statistical tests were conducted to further compare sample characteristics (i.e., demographic information and baseline attitudes) between participants who only completed phase I and those who completed both phases. First, an independent-samples t-test was conducted to investigate whether there were age differences between these two groups. Results showed that there was not a significant difference in age for participants who only completed phase I (M = 19.58, SD = 2.38) and those who completed both phases (M = 19.42, SD = 2.41), t(392) = .69, p = .50. A logistic regression was carried out to compare if the two groups of participants differed in gender. Findings showed that there was not a significant gender difference between participants who only completed phase I and those who completed both phases, Wald test = .28, p = .60. Furthermore, results from a multinomial regression indicated that two groups of samples did not differ in racial or ethnical background, 𝜒# (0,4) = 2.89, p = .58. Moreover, two independent-samples t-tests were conducted to examine group differences in terms of baseline attitudes toward two topical issues. For baseline attitudes toward DUI, there was a significant difference for participants who only completed phase I (M = 1.20, SD = .73) and those who completed both phases (M = 1.08, SD = .27), t(236) = 2.40, p = .02. Specifically, participants who dropped out of the study after phase I had more approving attitudes toward DUI than those who completed both phases of the study. In addition, for baseline attitudes toward helping 48 people, there was not a significant difference for participants who only completed phase I (M = 5.64, SD = 1.14) and those who completed both phases (M = 5.80, SD = .84), t(317) = -1.59, p = .11. 5.2 Checking Order Effect A multivariate analysis of variance (MANOVA) was employed to determine whether video order had an effect on endogenous variables for each video context. Across two MANOVAs, the independent variable was order (DUI video played first vs. Malaria video played first). 5.2.1 DUI In the first MANOVA, the dependent variables included spatial presence, social presence, transportation, identification, counterarguing, cybersickness, posttest attitudes and posttest intentions after watching the DUI video. At the multivariate level, the result suggested that order did not have a significant effect on endogenous variables in the proposed model for DUI context, F(8, 207) = .742, p = .655, partial h2 = .028. At the univariate level, order did not have a significant effect on all dependent variables: spatial presence, F(1, 214) = .586, p = .445, partial h2 = .002; social presence, F(1, 214) = .307, p = .580, partial h2 = .001; transportation, F(1, 214) = 1.515, p = .220, partial h2 = .007; identification, F(1, 214) = .210, p = .647, partial h2 = .001; counterarguing, F(1, 214) = .869, p = .352, partial h2 = .004; cybersickness, F(1, 214) = .112, p = .738, partial h2 = .001; posttest attitudes toward drunk driving, F(1, 214) = .033, p = .855, partial h2 = .000; and posttest intentions to engage in DUI, F(1, 214) = .543, p = .462, partial h2 = .003. 49 5.2.2 Malaria In the second MANOVA, the dependent variables included spatial presence, social presence, transportation, identification, counterarguing, cybersickness, posttest attitudes, and posttest willingness measured after watching the malaria video. At the multivariate level, the result suggested that order did not have a significant effect on endogenous variables in the proposed model for malaria context, F(8, 207) = 1.399, p = .199, partial h2 = .051. At the univariate level, order did not have a significant effect on spatial presence, F(1, 214) = .730, p = .394, partial h2 = .003; social presence, F(1, 214) = .070, p = .792, partial h2 = .000; transportation, F(1, 214) = .649, p = .421, partial h2 = .003; identification, F(1, 214) = .499, p = .481, partial h2 = .002; cybersickness, F(1, 214) = .231, p = .632, partial h2 = .001; posttest attitudes, F(1, 214) = 1.261, p = .263, partial h2 = .006; and posttest willingness, F(1, 214) = 2.831, p = .094, partial h2 = .013. Order had a significant effect on counterarguing, F(1, 214) = 4.366, p = .038, partial h2 = .020. Because the effect size of the order on counterarguing was relatively small, partial h2 = .020, the order effect was not considered in the rest of the analyses. 5.3 Test of Multivariate Normality Multivariate normal distribution is a major assumption for using Maximum Likelihood (ML), the default estimation method of Mplus, to build models. If the normality assumption is violated, the 𝜒# statistic, data-model fit indices based on 𝜒# statistic, and standard errors of the parameter estimates are biased (Finney & DiStefano, 2006). One strategy to address the violation of the normality assumption is 50 to use MLM as the estimator. MLM refers to the maximum likelihood parameter estimates with standard errors and a mean-adjusted chi-square test statistic that are robust to non-normality (Finney & DiStefano, 2006). I used the modified SPSS macro (Cain, Zhang, & Yuan, 2017) that was initially developed by DeCarlo (1997) to obtain the univariate and multivariate skew and kurtosis test statistics (see Table 5.1). Research suggests that problems could occur when the absolute value of univariate skewness is greater than 2, the absolute value of univariate kurtosis is greater than 7 (Curran, West, & Finch, 1996), and when Mardia’s normalized multivariate kurtosis is greater than 3 (Bentler, 2004). As you can see from Table 1, although the univariate skew and kurtosis test statistics did not meet the above cutoff values for the majority of the variables, some variables reached the cutoff value. What’s more, the coefficient of Mardia’s kurtosis was 11.84 and 15.66 for two video contexts, respectively. Both values were greater than 3, suggesting that the assumption of multivariate normality is violated. Therefore, MLM was used as the estimator for the measurement models and path analyses presented below. 51 Table 5.1 Univariate and Multivariate Skew and Kurtosis Test Statistics Variable Univariate Univariate Mardia’s kurtosis skewness kurtosis DUI context a Spatial presence 0.10 -0.80 Social presence 0.84 -0.03 Transportation -0.64 0.12 Identification -0.65 0.90 11.84 Counterarguing 0.98 0.66 Posttest attitudes 3.46 11.83 Posttest intentions 2.02 3.20 Malaria context b Spatial presence 0.07 -0.88 Social presence 0.10 -0.81 Transportation -0.68 0.05 Identification -0.61 0.39 15.66 Counterarguing 0.98 0.21 Posttest attitudes -3.20 12.90 Posttest intentions -0.50 0.42 a The standard error of skewness was .17. The standard error of kurtosis was .33. b The standard error of skewness was .17. The standard error of kurtosis was .33. 52 5.4 Test of Discriminant Validity In testing for evidence of discriminant validity among spatial presence, social presence, transportation, and identification, confirmatory factor analysis was conducted in Mplus 7.11 (Muthén & Muthén, 2013). In particular, I compared a model in which these four factors correlate freely with one in which they are perfectly correlated for each video context. The greater the difference between the value of 𝜒# and CFI, the stronger the support for proof of discriminant validity (Byrne, 2012). Criteria for model fit indices were based on Hu and Bentler’s (1998) study: the value of mean-square error of approximation (RMSEA) equals to or less than .06, the value of standardized root-mean-square residual (SRMR) equals to or less than .08, and the value of comparative fit index (CFI) equals to or greater than .95. It is noteworthy that the 𝜒# value for MLM cannot be used for 𝜒# difference testing in the regular way. Following the formulas provided by Satorra and Bentler (2010), I computed the 𝜒# difference test by hand. 5.4.1 DUI The measurement model in which spatial presence, social presence, transportation, and identification correlated freely (Model 1) showed acceptable fit: RMSEA = .045; SRMR = .063; CFI = .973; and 𝜒# (99) = 141.622. However, the measurement model in which the four factors perfectly correlated (Model 2) did not result in a good fit: RMSEA = .119; SRMR = .109; CFI = .793; and 𝜒# (105) = 428.655. Suggested by modification indices, both Model 1 and 2 included 14 pairs of residual correlations among observed items. A comparison between these two models yielded a statistically significant 𝜒# difference, 𝜒# (6) = 26698.5, p < .001. In 53 addition, the difference in practical fit was large, ∆CFI = 0.180. Therefore, the results provided strong evidence for the discriminant validity in the DUI video context. 5.4.2 Malaria The measurement model in which spatial presence, social presence, transportation, and identification correlated freely (Model 3) showed acceptable fit: RMSEA = .052; SRMR = .076; CFI = .960; and 𝜒# (103) = 163.828. However, the measurement model in which the four factors perfectly correlated (Model 4) did not result in a good fit: RMSEA = .135; SRMR = .117; CFI = .719; and 𝜒# (109) = 540.165. Suggested by modification indices, both Model 3 & 4 included 10 pairs of residual correlations among observed items. The value of 𝜒# difference was statistically significant, 𝜒# (6) = 25456.79, p < .001. In addition, the difference in practical fit was large, ∆CFI = 0.241. Therefore, the results provided strong evidence for the discriminant validity in the malaria video context. 5.5 Testing Overall Persuasion Two multivariate analysis of covariance (MANCOVAs) were conducted to test whether narratives presented in immersive (vs. non-immersive) mediated environments led to greater persuasiveness overall. 5.5.1 DUI In the first MANCOVA, experimental condition was entered as the independent variable, and attitudes toward DUI and intentions to engage in DUI were entered as the outcome variables. Participants’ gender, affiliation with Greek life, social desirability, baseline attitudes toward DUI, and cybersickness measured after 54 watching the DUI video were included as covariates. The results showed that in the DUI context, at the multivariate level, experimental conidiation had a significant effect on attitudes and intentions, F(2, 208) = 4.267, p = .015, partial h2 = .039. At the univariate level, experimental condition had a significant effect on attitudes toward DUI, F(1, 209) = 4.328, p = .039, partial h2 = .020, and on intentions to engage in DUI, F(1, 209) = 5.803, p = .017, partial h2 = .027. Viewers in the immersive condition, M = 1.04, SE = .02, had less favorable attitudes toward DUI than those in the non-immersive condition, M = 1.11, SE = .02. In addition, viewers in the immersive condition, M = 2.10, SE = .24, had fewer intentions to engage in DUI than those in the non-immersive condition, M = 2.95, SE = .24. 5.6.1 Malaria In the second MANCOVA, experimental condition was entered as the independent variable, and attitudes and willingness toward helping people who need protection from malaria were entered as the outcome variables. Participants’ gender, social desirability, baseline attitudes toward helping people, and cybersickness measured after watching the malaria video were included as covariates. The results indicated that in the malaria context, at the multivariate level, experimental conidiation did not have a significant effect on attitudes and willingness, F(2, 209) = .822, p = .441, partial h2 = .008. At the univariate level, experimental condition did not have a significant effect on attitudes toward helping people who need protection, F(1, 210) = 1.611, p = .206, partial h2 = .008, nor on willingness to help people who need protection, F(1, 210) = .023, p = .880, partial h2 = .000. 55 5.6 Main Model Testing Path analysis was used to test hypothesized main model in Mplus 7.11 (Muthén & Muthén, 2013). Error variance was specified for each variable included in the model (Schumacker & Lomax, 2010). The error variance of a single variable was obtained by multiplying the sample variance of an observed variable by (1 – reliability). Please see Table 5.2 for error variance for each variable. Criteria for model fit indices were based on Hu and Bentler’s (1998) study: the value of mean-square error of approximation (RMSEA) equals to or less than .06, the value of standardized root-mean-square residual (SRMR) equals to or less than .08, and the value of comparative fit index (CFI) equals to or greater than .95. 56 Table 5.2 Error Variance for Each Variable Reliability Sample Error variance variance DUI context Baseline attitudes toward .73 0.07 0.0189 DUI Social desirability .72 0.66 0.1848 Cybersickness .92 2.63 0.2104 Spatial presence .86 2.30 0.322 Social presence .87 1.97 0.2561 Transportation .73 0.93 0.2511 Identification .74 0.99 0.2574 Counterarguing .74 1.44 0.3744 Posttest attitudes .88 0.06 0.0072 Posttest intentions .78 6.6 1.452 Malaria context Attitudes toward helping .89 0.71 0.0781 people Social desirability .72 0.66 0.1848 Cybersickness .93 2.70 0.189 Spatial presence .85 2.30 0.345 57 Reliability Sample Error variance variance Social presence .83 2.43 0.4131 Transportation .75 1.03 0.2575 Identification .77 1.35 0.3105 Counterarguing .74 0.91 0.2366 Posttest attitudes .92 0.49 0.0392 Posttest intentions .85 11.73 1.7595 Note. The error variance of a single variable was obtained by multiplying the sample variance of an observed variable by (1 – reliability). 5.6.1 DUI Model 5 aimed to test the hypothesized main model in the DUI context. Participants’ gender, affiliation with Greek life, social desirability, baseline attitudes toward DUI, and cybersickness measured after watching the DUI video were included as covariates for all paths in the model. Correlations among key variables in the model are presented in Table 5.3. As demonstrated in Figure 5.1, the model resulted in a satisfactory model fit: RMSEA = .000, 90% CI [.000, .090], SRMR = .010, CFI = 1.000, and 𝜒# (5) = 4.498, p = .48. Results for standardized and unstandardized parameters are presented in Table 5.4. 58 Table 5.3 Correlation Matrix for Key Variables in the Model (DUI Context) Spatial Social Transportation Identification Counterarguing Posttest Posttest presence presence attitudes intentions Spatial presence 1 Social presence .75** 1 Transportation .34** .20** 1 Identification .27** .30** .33** 1 Counterarguing -.08 .00 -.28** -.07 1 Posttest attitudes -.09 .04 -.16* -.07 .30** 1 Posttest intentions -.07 .09 -.02 .12 .12 .34** 1 Note. * p < .05; ** p < .01. Tests are two-tailed. 59 Table 5.4 Standardized and Unstandardized Parameter Estimates for the Main Model (DUI Context) Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Hypothesized parameters Media format ® spatial presence .18 (.08)* 0.51 (0.22)* Media format ® social presence .07 (.08) 0.18 (0.20) Media format ® transportation .08 (.07) 0.13 (0.12) Media format ® identification -.10 (.08) -0.16 (0.14) Media format ® attitudes toward DUI -.10 (.06) -0.05 (0.03) Media format ® intentions to engage -.10 (.09) -0.47 (0.42) in DUI Spatial presence ® transportation .38 (.07)*** 0.22 (0.05)*** Spatial presence ® attitudes toward -.13 (.18) -0.02 (0.03) DUI Spatial presence ® intentions to -.62 (.24)** -1.01 (0.41)* engage in DUI Social presence ® identification .40 (.07)*** 0.26 (0.05)*** Social presence ® attitudes toward .09 (.17) 0.02 (0.03) DUI 60 Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Social presence ® intentions to .54 (.24)* 0.93 (0.44)* engage in DUI Transportation ® counterarguing -.46 (.12)*** -0.57 (0.16)*** Transportation ® attitudes toward .06 (.09) 0.02 (0.03) DUI Transportation ® intentions to engage .03 (.14) 0.08 (0.37) in DUI Identification ® counterarguing .13 (.15) 0.16 (0.19) Identification ® attitudes toward DUI -.03 (.08) -0.01 (0.02) Identification ® intentions to engage .21 (.11) 0.56 (0.30) in DUI Counterarguing ® attitudes toward .29 (.09)** 0.06 (0.02)** DUI Counterarguing ® intentions to .11 (.11) 0.23 (0.25) engage in DUI Control variable parameters Gender ® spatial presence .19 (.08)* 0.52 (0.21)* Affiliation with Greek life ® spatial -.11 (.07) -0.44 (0.28) presence Social desirability ® spatial presence .08 (.09) 0.16 (0.18) 61 Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Baseline attitudes toward DUI ® -.03 (.06) -0.15 (0.37) spatial presence Cybersickness ® spatial presence .02 (.09) 0.02 (0.08) Gender ® social presence .07 (.08) 0.17 (0.21) Affiliation with Greek life ® social -.13 (.08) -0.51 (0.30) presence Social desirability ® social presence .14 (.09) 0.26 (0.18) Baseline attitudes toward DUI ® .10 (.08) 0.59 (0.43) social presence Cybersickness ® social presence .11 (.09) 0.09 (0.08) Gender ® transportation .15 (.07)* 0.25 (0.12)* Affiliation with Greek life ® -.16 (.08)* -0.38 (0.18)* transportation Social desirability ® transportation -.01 (.09) -0.02 (0.11) Baseline attitudes toward DUI ® -.12 (.08) -0.43 (0.31) transportation Cybersickness ® transportation -.33 (.08)*** -0.17 (0.04)*** Gender ® identification -.09 (.08) -0.15 (0.14) Affiliation with Greek life ® -.09 (.08) -0.22 (0.20) identification 62 Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Social desirability ® identification .09 (.10) 0.11 (0.12) Baseline attitudes toward DUI ® -.13 (.09) -0.47 (0.31) identification Cybersickness ® identification .01 (.09) 0.003 (0.05) Gender ® counterarguing .18 (.09)* 0.36 (0.19) Affiliation with Greek life ® .05 (.09) 0.15 (0.26) counterarguing Social desirability ® counterarguing .05 (.09) 0.07 (0.14) Baseline attitudes toward DUI ® .09 (.10) 0.41 (0.46) counterarguing Cybersickness ® counterarguing .04 (.10) 0.03 (0.06) Gender ® attitudes toward DUI -.01 (.08) -0.003 (0.03) Affiliation with Greek life ® attitudes .03 (.08) 0.02 (0.05) toward DUI Social desirability ® attitudes toward -.06 (.08) -0.02 (0.02) DUI Baseline attitudes toward DUI ® .65 (.13)*** 0.63 (0.17)*** attitudes toward DUI Cybersickness ® attitudes toward .04 (.09) 0.01 (0.01) DUI 63 Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Gender ® intentions to engage in DUI .12 (.08) 0.55 (0.38) Affiliation with Greek life ® -.10 (.10) -0.64 (0.64) intentions to engage in DUI Social desirability ® intentions to -.32 (.09)*** -1.04 (0.33)** engage in DUI Baseline attitudes toward DUI ® .33 (.11)** 3.23 (1.20)** intentions to engage in DUI Cybersickness ® intentions to engage -.02 (.11) -0.03 (0.16) in DUI Note. * p < .05; ** p < .01; *** p < .001. H1 predicted that exposure to a narrative presented in immersive (vs. non- immersive) mediated environments would lead to a higher level of a) spatial presence and b) social presence. Consistent with H1(a), the story presented in IMEs led to greater spatial presence than that presented in non-IMEs, b = .18, p = .02.6 However, social presence did not differ between narratives viewed in IMEs and non-IMEs, b = .07, p = .36. Therefore, H1(a) was supported but H1(b) was not supported in the DUI context. H2 predicted that exposure to a narrative presented in immersive (vs. non- immersive) mediated environments would lead to a higher level of a) transportation and b) identification. The data indicated that media format did not have a direct 6 All parameters reported in the text are standardized parameters under the STDYX output. Unstandardized parameters are reported in the Table 5.4 & 5.6. 64 significant effect on transportation, b = .08, p = .28, nor identification, b = -.10, p = .25. Both H2(a) and (b) were not supported. H3 predicted that media format would have an indirect effect on transportation through spatial presence. Consistent with H3, media format had a significant indirect effect on transportation through spatial presence, b = .07, p = .04, 95% CI [.003, .134]7. Specifically, narratives presented in IMEs, compared to those presented in non-IMEs, increased spatial presence, b = .18, p = .02, which then led to a higher level of transportation, b = .38, p < .001. The results suggested that although media format did not directly influence transportation, it indirectly affected transportation through spatial presence. The total effect of media format on transportation was approaching significance, b = .15, p = .054, 95% CI [-.003, .299]. H4 predicted that media format would have an indirect effect on identification through social presence. Inconsistent with H4, media format did not have a significant indirect effect on identification through social presence, b = .03, p = .37, 95% CI [- .033, .089]. Nonetheless, social presence increased identification, b = .40, p < .001. The total effect of media format on identification was not significant, b = -.07, p = .45, 95% CI [-.240, .105]. H5 predicted that media format would have an indirect effect on counterarguing through a) spatial presence, b) social presence, c) transportation, and/or d) identification. As we can see from the Figure 5.1, there were four specific indirect effects from media format to counterarguing: one through transportation, one 7 The frequentist settings are SYMMETRIC. BOOTSTRAP is not available for MLM. 65 through spatial presence and transportation, one through identification, and one through social presence and identification. The total indirect effect approached significance, b = -.08, p = .05, 95% CI [-.154, .000]. Among the four specific indirect effects, only the one through spatial presence and transportation approached significance, b = -.03, p = .055, 95% CI [-.064, .001]. In particular, participants in the IMEs condition reported a greater level of spatial presence, b = .18, p = .02, which led to higher level of transportation, b = .38, p < .001. Transportation decreased counterarguing, b = -.46, p < .001. Because media format was not hypothesized to have a direct effect on counterarguing, the total effect of media format on counterarguing equaled to the total indirect effect, b = -.08, p = .05, 95% CI [-.154, .000]. H6 predicted that exposure to a narrative presented in immersive (vs. non- immersive) mediated environments would lead to more story-consistent a) attitudes and b) behavioral intentions/willingness. H7 predicted that media format would have an indirect effect on story-consistent attitudes through a) spatial presence, b) social presence, c) transportation, d) identification, and/or e) counterarguing. H8 predicted that media format would have an indirect effect on story-consistent behavioral intentions through a) spatial presence, b) social presence, c) transportation, d) identification, and/or e) counterarguing. RQ1 asked whether media format will have a direct effect on story-consistent a) attitudes and b) behavioral intentions/willingness after controlling for spatial presence, social presence, transportation, identification, and counterarguing. 66 In terms of the attitudes toward DUI, results showed that the total effect of media format on attitudes was significant, b = -.13, p = .02, 95% CI [-.241, -.020], supporting H6(a). Specifically, participants in the IMEs condition demonstrated less favorable attitudes toward DUI than those in the non-IMEs condition. However, the total indirect effect of media format on attitudes toward DUI through the proposed mediators was not significant, b = -.03, p = .31, 95% CI [-.083, .026]. There was no significant specific indirect effect, either. H7 was not supported. Moreover, media format did not have a significant direct effect on attitudes toward DUI after controlling for the meditators, b = -.10, p = .10. For behavioral intentions to engage in DUI, results indicated that the total effect of media format on behavioral intentions was significant, b = -.20, p = .02, 95% CI [-.359, -.035], supporting H6(b). Specifically, participants in the IMEs condition demonstrated fewer behavioral intentions to engage in DUI than those in the non- IMEs condition. In addition, the total indirect effect of media format on intentions was significant, b = -.09, p = .04, 95% CI [-.185, -.003]. Among the 10 specific indirect effects, only the one through spatial presence was marginally significant, b = -.11, p = .06, 95% CI [-.232, .005]. IMEs increased spatial presence, b = .18, p = .02, which then reduced behavioral intentions to drive under the influence of alcohol, b = - .62, p = .01. Finally, media format did not have a significant direct effect on intentions to engage in DUI after controlling for the meditators, b = -.10, p = .26. As shown in Figure 5.1, the R-square for spatial presence was .099, social presence .083, transportation .289, identification .197, counterarguing .203, posttest attitudes toward DUI .568, and posttest intentions to drive under the influence of 67 alcohol .426. The results indicated that the model explained a relatively large amount of variance in persuasive outcomes (Cohen, 1988). Figure 5.1. The main model in the DUI context. Significant paths are indicated by solid lines. Nonsignificant paths are indicated by dashed lines. Standardized parameters are reported, with standard errors in the parentheses. Spatial presence and social presence are correlated. Transportation and identification are correlated. Attitudes and behavioral intentions are correlated. Participants’ gender, affiliation with Greek life, social desirability, baseline attitudes toward DUI, and cybersickness are controlled in the model. RMSEA = .000, 90% CI [.000, .090], SRMR = .010, CFI = 1.000, and 𝜒# (5) = 4.498, p = .48. 5.6.2 Malaria A similar path analysis was built to test the hypothesized main model in the malaria context (Model 6). Participants’ gender, social desirability, attitudes toward 68 helping people, and cybersickness measured after watching the malaria video were included as control variables for all paths in the model. Correlations among key variables in the model are presented in Table 5.5. As demonstrated in Figure 5.2, the model for the malaria context resulted in a satisfactory model fit: RMSEA = .000, 90% CI [.000, .080], SRMR = .008, CFI = 1.000, and 𝜒# (5) = 3.628, p = .60. Results for standardized and unstandardized parameters are presented in Table 5.6. 69 Table 5.5 Correlation Matrix for Key Variables in the Model (Malaria Context) Spatial Social Transportation Identification Counterarguing Posttest Posttest presence presence attitudes intentions Spatial presence 1 Social presence .76** 1 Transportation .37** .28** 1 Identification .34** .32** .32** 1 Counterarguing -.17* .09 -.35** -.07 1 Posttest attitudes -.01 -.04 .16* -.08 -.45** 1 Posttest intentions .16* .17* .39** .31** -.26** .09 1 Note. * p < .05; ** p < .01. Tests are two-tailed. 70 Table 5.6 Standardized and Unstandardized Parameter Estimates for the Main Model (Malaria Context) Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Hypothesized parameters Media format ® spatial presence .26 (.07)*** 0.71 (0.21)*** Media format ® social presence .32 (.08)*** 0.91 (0.22)*** Media format ® transportation .04 (.08) 0.07 (0.15) Media format ® identification -.02 (.09) -0.04 (0.17) Media format ® attitudes toward .07 (.07) 0.09 (0.10) helping people Media format ® willingness to help -.12 (.08) -0.78 (0.50) people Spatial presence ® transportation .43 (.07)*** 0.27 (0.05)*** Spatial presence ® attitudes toward -.11 (.28) -0.05 (0.13) helping people Spatial presence ® willingness to -.74 (.39) -1.67 (0.87) help people Social presence ® identification .46 (.08)*** 0.33 (0.06)*** Social presence ® attitudes toward .03 (.29) 0.01 (0.14) helping people 71 Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Social presence ® willingness to .68 (.40) 1.53 (0.89) help people Transportation ® counterarguing -.46 (.13)*** -0.43 (0.12)*** Transportation ® attitudes toward -.03 (.13) -0.02 (0.10) helping people Transportation ® willingness to help .30 (.11)** 1.08 (0.41)** people Identification ® counterarguing .08 (.13) 0.07 (0.11) Identification ® attitudes toward -.07 (.10) -0.05 (0.07) helping people Identification ® willingness to help .23 (.11)* 0.72 (0.34)* people Counterarguing ® attitudes toward -.51 (.11)*** -0.42 (0.11)*** helping people Counterarguing ® willingness to -.18 (.11) -0.68 (0.41) help people Control variable parameters Gender ® spatial presence .16 (.08)* 0.46 (0.22)* Social desirability ® spatial presence .02 (.10) 0.03 (0.20) 72 Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Baseline attitudes toward helping .06 (.08) 0.10 (0.14) people ® spatial presence Cybersickness ® spatial presence -.06 (.08) -0.05 (0.07) Gender ® social presence .09 (.08) 0.25 (0.22) Social desirability ® social presence .03 (.09) 0.06 (0.19) Baseline attitudes toward helping -.05 (.08) -0.08 (0.15) people ® social presence Cybersickness ® social presence -.09 (.08) -0.08 (0.07) Gender ® transportation .04 (.07) 0.06 (0.13) Social desirability ® transportation -.06 (.10) -0.08 (0.12) Baseline attitudes toward helping .26 (.09)** 0.29 (0.09)** people ® transportation Cybersickness ® transportation -.19 (.09)* -0.11 (0.05)* Gender ® identification -.07 (.07) -0.15 (0.15) Social desirability ® identification .32 (.09)*** 0.48 (0.13)*** Baseline attitudes toward helping .12 (.08) 0.15 (0.10) people ® identification Cybersickness ® identification -.16 (.09) -0.10 (0.06) Gender ® counterarguing .04 (.09) 0.07 (0.15) Social desirability ® counterarguing .14 (.11) 0.17 (0.13) 73 Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Baseline attitudes toward helping -.12 (.10) -0.12 (0.10) people ® counterarguing Cybersickness ® counterarguing .09 (.09) 0.05 (0.05) Gender ® attitudes toward helping -.10 (.07) -0.13 (0.10) people Social desirability ® attitudes toward -.19 (.09)* -0.19 (0.10) helping people Baseline attitudes toward helping .24 (.07)** 0.20 (0.07)** people ® attitudes toward helping people Cybersickness ® attitudes toward .03 (.08) 0.01 (0.03) helping people Gender ® willingness to help people .24 (.07)** 1.54 (0.46)** Social desirability ® willingness to .003 (.09) 0.01 (0.42) help people Baseline attitudes toward helping .29 (.08)*** 1.16 (0.35)** people ® willingness to help people Cybersickness ® willingness to help .11 (.07) 0.22 (0.15) people Note. * p < .05; ** p < .01; *** p < .001. 74 H1 predicted that exposure to a narrative presented in immersive (vs. non- immersive) mediated environments would lead to a higher level of a) spatial presence and b) social presence. In support of H1(a), the story presented in IMEs led to greater spatial presence than that presented in non-IMEs, b = .26, p < .001. This result is in line with findings from DUI context. In addition, the data also supported H1(b) that the story presented in IMEs led to greater social presence than that presented in non- IMES, b = .32, p < .001. H2 predicted that exposure to a narrative presented in immersive (vs. non- immersive) mediated environments would lead to a higher level of a) transportation and b) identification. Similar with findings from the DUI context, the data for malaria context showed that media format did not have a direct significant effect on transportation, b = .04, p = .63, nor on identification, b = -.02, p = .80. The findings did not support H2 (a) and (b). H3 predicted that media format would have an indirect effect on transportation through spatial presence. Consistent with H3, media format had a significant indirect effect on transportation through spatial presence, b = .11, p = .003, 95% CI [.037, .181]. Specifically, narratives presented in IMEs, compared to those presented in non-IMEs, increased spatial presence, b = .26, p < .001, which then led to a higher level of transportation, b = .43, p < .001. The total effect of media format on transportation was not significant, b = .15, p = .08, 95% CI [-.017, .315]. Results from both video contexts showed that the effect of media format on transportation was mediated by spatial presence. 75 H4 predicted that media format would have an indirect effect on identification through social presence. In contrast with findings from the DUI context, the data for the malaria video supported the hypothesis that media format had a significant indirect effect on identification through social presence, b = .14, p = .001, 95% CI [.060, .232]. Specifically, narratives presented in IMEs, compared to those presented in non-IMEs, increased social presence, b = .32, p < .001, which then led to a higher level of identification, b = .46, p < .001. The total effect of media format on identification was not significant, b = .12, p = .15, 95% CI [-.043, .292]. H5 predicted that media format would have an indirect effect on counterarguing through a) spatial presence, b) social presence, c) transportation, and/or d) identification. As we can see from the Figure 5.2, there were four specific indirect effects from media format to counterarguing: one through transportation, one through spatial presence and transportation, one through identification, and one through social presence and identification. The total indirect effect was not significant, b = -.08, p = .15, 95% CI [-.138, .021]. However, among the four specific indirect effects, the one through spatial presence and transportation was significant, b = -.05, p = .01, 95% CI [-.091, -.010]. In particular, narratives viewed in IMEs resulted in a greater sense of spatial presence, b = .26, p < .001, which led to a higher level of transportation, b = .43, p < .001. Transportation reduced counterarguing, b = - .46, p < .001. In the DUI context, the specific indirect effect through spatial presence and transportation was also approaching significance. The total effect equaled to the total indirect effect, b = -.08, p = .15, 95% CI [-.138, .021]. 76 H6 predicted that exposure to a narrative presented in immersive (vs. non- immersive) mediated environments would lead to more story-consistent a) attitudes and b) behavioral intentions/willingness. H7 predicted that media format would have an indirect effect on story-consistent attitudes through a) spatial presence, b) social presence, c) transportation, d) identification, and/or e) counterarguing. H8 predicted that media format would have an indirect effect on story-consistent behavioral intentions through a) spatial presence, b) social presence, c) transportation, d) identification, and/or e) counterarguing. RQ1 asked whether media format will have a direct effect on story-consistent a) attitudes and b) behavioral intentions/willingness after controlling for spatial presence, social presence, transportation, identification, and counterarguing. In terms of attitudes toward helping people with malaria, results showed that the total effect of media format on attitudes was not significant, b = .06, p = .31, 95% CI [-.058, .184]. H6 (a) was not supported in the malaria context. Moreover, the total indirect effect of media format on attitudes toward helping people through the proposed mediators was not significant, b = -.002, p = .96, 95% CI [-.075, .071]. This result was consistent with the result in the DUI context, which also found a non- significant total indirect effect. However, the specific indirect effect through spatial presence, transportation, and counterarguing was significant, b = .03, p = .046, 95% CI [.000, .051]. Participants in the IMEs condition experienced a greater level of spatial presence, b = 0.26, p < .001, which in turn enhanced feelings of transportation, b = .43, p < .001. Transportation then reduced counterarguing, b = -.46, p < .001, which promoted more favorable attitudes toward helping people who need protection 77 from malaria, b = -.51, p < .001. After controlling for the mediators, media format did not have a significant direct effect on attitudes toward helping people who need protection from malaria, b = .07, p = .38. For willingness to help people who need protection from malaria, results indicated that the total effect of media format on behavioral willingness was not significant, b = -.01, p = .90, 95% CI [-.147, .129]. H6 (b) was not supported in the malaria context, either. The total indirect effect was marginally significant, b = .11, p = .064, 95% CI [-.006, .234]. Among the 10 specific indirect effects, the one through spatial presence and transportation was marginally significant, b = -.03, p = .064, 95% CI [-.002, .067]. Moreover, the one through social presence and identification was also marginally significant, b = -.03, p = .067, 95% CI [-.002, .070]. The results showed inconsistencies with those in the DUI context. After controlling for the mediators, media format did not have a significant direct effect on willingness to help people who need protection from malaria, b = -.12, p = .12. As illustrated in Figure 5.2, the model explained 9.5% of variance in spatial presence, 10.7% of variance in social presence, 30.1% of variance in transportation, 34.8% of variance in identification, 26.4% of variance in counterarguing, 38.9% of variance in posttest attitudes toward helping people, and 50.7% of variance in posttest willingness to help people who need protection from malaria. 78 Figure 5.2. The main model in the malaria context. Significant paths are indicated by solid lines. Nonsignificant paths are indicated by dashed lines. Standardized parameters are reported, with standard errors in the parentheses. Spatial presence and social presence are correlated. Transportation and identification are correlated. Attitudes and behavioral willingness are correlated. Participants’ gender, social desirability, attitudes toward helping people, and cybersickness are controlled in the model. RMSEA = .000, 90% CI [.000, .080], SRMR = .008, CFI = 1.000, and 𝜒# (5) = 3.628, p = .60. 5.7 Interaction Effects Testing H9 predicted that trait empathy would moderate the impact of media format on spatial presence, social presence, transportation, and identification such that a) spatial presence, b) social presence, c) transportation, and d) identification would be more strongly influenced by media format among individuals high (vs. low) in trait 79 empathy. In order to test the moderating effect of trait empathy, an interaction term was created by multiplying experiment manipulation and trait empathy. Both variables were centered before creating the interaction term8. Path analysis was again used to test hypothesized interaction model by adding trait empathy and interaction term on four endogenous variables: spatial presence, social presence, transportation, and identification. All variables and paths in the main model were kept. Error variance was specified for each variable included in the model (Schumacker & Lomax, 2010). The error variance of a single variable was obtained by multiplying the variance of an observed variable by (1 – reliability). For the interaction term, I followed formula provided by (Aiken & West, 1991) to calculate the error variance, which equaled to .0984. 5.7.1 DUI As demonstrated in Figure 5.3, the model (Model 7) with interaction term resulted in a satisfactory model fit: RMSEA = .000, 90% CI [.000, .064], SRMR = .014, CFI = 1.000, and 𝜒# (11) = 9.491, p = .58. Adding the trait empathy and interaction term did not change the overall pattern of the relations compared to the main model (Model 5). Because the current analysis focused on the moderation effect of trait empathy on spatial presence, social presence, transportation, and identification, only the results on these four endogenous variables were reported here. Please see Table 5.7 for standardized and unstandardized parameters for the whole model. 8 Experiment manipulation was coded as 0 (non-IMEs) and 1 (IMEs). It was treated as a continuous variable because 1 indicated a greater level of immersion than 0. 80 The results showed that the interaction term did not have any significant effect on the four variables, b = .06, p = .56 for spatial presence, b = .15, p = .18 for social presence, b = -.10, p = .27 for transportation, and b = -.16, p = .10 for identification. H9 was not supported in the DUI context. Trait empathy only influenced identification, b = .21, p = .01. Those with greater trait empathy were more likely to identify with the characters in the video. Media format, again, had a significant positive effect on spatial presence, b = .18, p = .02, which was consistent with the findings from the main model. 81 Table 5.7 Standardized and Unstandardized Parameter Estimates for the Interaction Model (DUI Context) Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Hypothesized parameters Media format ® spatial presence .18 (.08)* 0.51 (0.22)* Media format ® social presence .07 (.08) 0.17 (0.20) Media format ® transportation .08 (.07) 0.13 (0.12) Media format ® identification -.09 (.08) -0.15 (0.14) Media format ® attitudes toward DUI -.10 (.06) -0.05 (0.03) Media format ® intentions to engage -.10 (.09) -0.46 (0.42) in DUI Trait empathy ® spatial presence .07 (.09) 0.13 (0.17) Trait empathy ® social presence .01 (.10) 0.02 (0.17) Trait empathy ® transportation .04 (.08) 0.05 (0.08) Trait empathy ® identification .21 (.09)* 0.23 (0.09)** Media format * Trait empathy ® .06 (.11) 0.28 (0.48) spatial presence Media format * Trait empathy ® .15 (.11) 0.62 (0.48) social presence 82 Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Media format * Trait empathy ® -.10 (.10) -0.27 (0.24) transportation Media format * Trait empathy ® -.16 (.10) -0.42 (0.26) identification Spatial presence ® transportation .39 (.07)*** 0.23 (0.05)*** Spatial presence ® attitudes toward -.15 (.18) -0.02 (0.03) DUI Spatial presence ® intentions to -.62 (.23)** -1.00 (0.40)* engage in DUI Social presence ® identification .41 (.07)*** 0.27 (0.05)*** Social presence ® attitudes toward .11 (.16) 0.02 (0.03) DUI Social presence ® intentions to .53 (.23)* 0.92 (0.43)* engage in DUI Transportation ® counterarguing -.45 (.12)*** -0.56 (0.16)*** Transportation ® attitudes toward .06 (.09) 0.02 (0.03) DUI Transportation ® intentions to engage .03 (.14) 0.07 (0.38) in DUI Identification ® counterarguing .12 (.15) 0.15 (0.19) 83 Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Identification ® attitudes toward DUI -.04 (.08) -0.01 (0.02) Identification ® intentions to engage .23 (.12) 0.60 (0.31) in DUI Counterarguing ® attitudes toward .28 (.09)** 0.06 (0.02)** DUI Counterarguing ® intentions to .11 (.12) 0.24 (0.25) engage in DUI Control variable parameters Gender ® spatial presence .18 (.08)* 0.52 (0.21)* Affiliation with Greek life ® spatial -.11 (.07) -0.44 (0.28) presence Social desirability ® spatial presence .08 (.09) 0.16 (0.19) Baseline attitudes toward DUI ® -.02 (.06) -0.11 (0.36) spatial presence Cybersickness ® spatial presence .02 (.09) 0.01 (0.08) Gender ® social presence .08 (.08) 0.21 (0.22) Affiliation with Greek life ® social -.15 (.08) -0.55 (0.29) presence Social desirability ® social presence .13 (.10) 0.24 (0.19) 84 Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Baseline attitudes toward DUI ® .11 (.08) 0.60 (0.41) social presence Cybersickness ® social presence .10 (.09) 0.09 (0.08) Gender ® transportation .13 (.07) 0.22 (0.12) Affiliation with Greek life ® -.14 (.08) -0.35 (0.18) transportation Social desirability ® transportation .002 (.10) 0.002 (0.12) Baseline attitudes toward DUI ® -.12 (.09) -0.41 (0.32) transportation Cybersickness ® transportation -.33 (.08)*** -0.17 (0.04)*** Gender ® identification -.14 (.08) -0.23 (0.14) Affiliation with Greek life ® -.06 (.07) -0.16 (0.18) identification Social desirability ® identification .13 (.09) 0.16 (0.12) Baseline attitudes toward DUI ® -.11 (.09) -0.42 (0.33) identification Cybersickness ® identification .004 (.08) 0.002 (0.05) Gender ® counterarguing .17 (.09) 0.36 (0.19) Affiliation with Greek life ® .05 (.09) 0.15 (0.26) counterarguing 85 Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Social desirability ® counterarguing .06 (.09) 0.08 (0.14) Baseline attitudes toward DUI ® .10 (.10) 0.43 (0.46) counterarguing Cybersickness ® counterarguing .04 (.10) 0.03 (0.06) Gender ® attitudes toward DUI -.004 (.08) -0.002 (0.03) Affiliation with Greek life ® attitudes .03 (.08) 0.02 (0.05) toward DUI Social desirability ® attitudes toward -.06 (.07) -0.02 (0.02) DUI Baseline attitudes toward DUI ® .64 (.13)*** 0.63 (0.17)*** attitudes toward DUI Cybersickness ® attitudes toward .04 (.09) 0.01 (0.01) DUI Gender ® intentions to engage in DUI .12 (.08) 0.56 (0.38) Affiliation with Greek life ® -.10 (.10) -0.63 (0.65) intentions to engage in DUI Social desirability ® intentions to -.32 (.09)*** -1.05 (0.33)** engage in DUI 86 Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Baseline attitudes toward DUI ® .33 (.11)** 3.24 (1.19)** intentions to engage in DUI Cybersickness ® intentions to engage -.02 (.11) -0.03 (0.18) in DUI Note. * p < .05; ** p < .01; *** p < .001. Figure 5.3. The interaction model in the DUI context. Significant paths are indicated by solid lines. Nonsignificant paths are indicated by dashed lines. Standardized parameters are reported, with standard errors in the parentheses. Spatial presence and social presence are correlated. Transportation and identification are correlated. Attitudes and behavioral intentions are correlated. Participants’ gender, affiliation with Greek life, social desirability, baseline attitudes toward DUI, and cybersickness 87 are controlled in the model. RMSEA = .000, 90% CI [.000, .064], SRMR = .014, CFI = 1.000, and 𝜒# (11) = 9.491, p = .58. 5.7.2 Malaria Model 8 was built to test the moderating effect of trait empathy on spatial presence, social presence, transportation, and identification in the malaria video context. As seen from Figure 5.4, the model had a good model fit: RMSEA = .039, 90% CI [.000, .086], SRMR = .018, CFI = .992, and 𝜒# (11) = 14.630, p = .20. Compared with the main model (Model 6), findings for the relations in Model 8 had a consistent pattern. Please see Table 5.8 for standardized and unstandardized parameters for the whole model. The results demonstrated that the interaction variable had a significant effect on social presence, b = -.21, p = .03, but not on spatial presence, b = -.02, p = .84, nor on transportation, b = -.05, p = .65, nor on identification, b = .09, p = .43. Trait empathy had a significant effect on spatial presence, b = .24, p = .005, and an approaching significant effect on social presence, b = .15, p = .075. Individuals with greater tendencies to empathize experienced a higher level of spatial presence. Media format had a consistent positive effect on spatial presence, b = .24, p = .001, and social presence, b = .32, p < .001. 88 Table 5.8 Standardized and Unstandardized Parameter Estimates for the Interaction Model (Malaria Context) Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Hypothesized parameters Media format ® spatial presence .24 (.07)** 0.67 (0.21)** Media format ® social presence .32 (.08)*** 0.91 (0.22)*** Media format ® transportation .04 (.08) 0.08 (0.15) Media format ® identification -.03 (.09) -0.07 (0.17) Media format ® attitudes toward .06 (.07) 0.08 (0.10) helping people Media format ® willingness to help -.11 (.07) -0.69 (0.47) people Trait empathy ® spatial presence .24 (.09)** 0.43 (0.16)** Trait empathy ® social presence .15 (.08) 0.26 (0.15) Trait empathy ® transportation .01 (.09) 0.01 (0.10) Trait empathy ® identification .12 (.10) 0.16 (0.13) Media format * Trait empathy ® -.02 (.10) -0.09 (0.42) spatial presence Media format * Trait empathy ® -.21 (.09) -0.91 (0.41)* social presence 89 Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Media format * Trait empathy ® -.05 (.10) 0.01 (0.10) transportation Media format * Trait empathy ® .09 (.11) 0.28 (0.36) identification Spatial presence ® transportation .42 (.08)*** 0.27 (0.05)*** Spatial presence ® attitudes toward -.12 (.21) -0.06 (0.10) helping people Spatial presence ® willingness to -.54 (.26)* -1.22 (0.57)* help people Social presence ® identification .45 (.08)*** 0.32 (0.06)*** Social presence ® attitudes toward .04 (.22) 0.02 (0.10) helping people Social presence ® willingness to .48 (.26) 1.07 (0.57) help people Transportation ® counterarguing -.46 (.13)*** -0.43 (0.12)*** Transportation ® attitudes toward -.03 (.12) -0.02 (0.09) helping people Transportation ® willingness to help .27 (.11)* 0.97 (0.40)* people Identification ® counterarguing .07 (.13) 0.06 (0.11) 90 Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Identification ® attitudes toward -.06 (.10) -0.04 (0.07) helping people Identification ® willingness to help .27 (.11)* 0.83 (0.34)* people Counterarguing ® attitudes toward -.51 (.11)*** -0.41 (0.11)*** helping people Counterarguing ® willingness to -.18 (.11) -0.67 (0.41) help people Control variable parameters Gender ® spatial presence .14 (.08) 0.38 (0.22) Social desirability ® spatial presence .07 (.10) 0.13 (0.21) Baseline attitudes toward helping -.02 (.08) -0.04 (0.15) people ® spatial presence Cybersickness ® spatial presence -.03 (.08) -0.03 (0.07) Gender ® social presence .05 (.08) 0.14 (0.22) Social desirability ® social presence .08 (.10) 0.17 (0.20) Baseline attitudes toward helping -.09 (.09) -0.15 (0.16) people ® social presence Cybersickness ® social presence -.08 (.09) -0.08 (0.08) Gender ® transportation .03 (.07) 0.05 (0.13) 91 Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Social desirability ® transportation -.05 (.10) -0.07 (0.13) Baseline attitudes toward helping .26 (.09)** 0.28 (0.10)** people ® transportation Cybersickness ® transportation -.20 (.09)* -0.11 (0.05)* Gender ® identification -.07 (.07) -0.15 (0.15) Social desirability ® identification .34 (.09)*** 0.50 (0.14)*** Baseline attitudes toward helping .07 (.09) 0.09 (0.11) people ® identification Cybersickness ® identification -.15 (.09) -0.10 (0.06) Gender ® counterarguing .04 (.09) 0.07 (0.15) Social desirability ® counterarguing .16 (.11) 0.19 (0.13) Baseline attitudes toward helping -.12 (.10) -0.13 (0.10) people ® counterarguing Cybersickness ® counterarguing .09 (.09) 0.05 (0.05) Gender ® attitudes toward helping -.10 (.07) -0.13 (0.10) people Social desirability ® attitudes toward -.22 (.09)* -0.21 (0.10)* helping people 92 Path Standardized path Unstandardized path coefficients (SE) coefficients (SE) Baseline attitudes toward helping .25 (.07)*** 0.21 (0.07)** people ® attitudes toward helping people Cybersickness ® attitudes toward .03 (.08) 0.01 (0.03) helping people Gender ® willingness to help people .23 (.07)** 1.47 (0.42)*** Social desirability ® willingness to -.02 (.09) -0.08 (0.40) help people Baseline attitudes toward helping .28 (.07)*** 1.11 (0.32)*** people ® willingness to help people Cybersickness ® willingness to help .10 (.07) 0.20 (0.14) people Note. * p < .05; ** p < .01; *** p < .001. 93 Figure 5.4. The interaction model in the malaria context. Significant paths are indicated by solid lines. Nonsignificant paths are indicated by dashed lines. Standardized parameters are reported, with standard errors in the parentheses. Spatial presence and social presence are correlated. Transportation and identification are correlated. Attitudes and behavioral willingness are correlated. Participants’ gender, social desirability, attitudes toward helping people, and cybersickness are controlled in the model. RMSEA = .039, 90% CI [.000, .086], SRMR = .018, CFI = .992, and 𝜒# (11) = 14.630, p = .20. SPSS Process (Hayes, 2012) was used to probe the interaction. Media format was entered as the independent variable, trait empathy as the moderator, and social presence as the dependent variable. Gender, attitudes toward helping people, social desirability, cybersickness, and spatial presence were kept as covariates. The relation between media format and social presence was estimated at three values of trait 94 empathy, one standard deviation below the mean (i.e., predominantly low trait empathy), the mean (i.e., moderate trait empathy), and one standard deviation above the mean (i.e., predominantly high trait empathy) (Aiken & West, 1991). Results showed that at one standard deviation below the mean of trait empathy, participants in the immersive condition had a higher level of social presence than those in the control condition, β = .77, p < .05. At the value of the mean, participants in the immersive condition had a higher level of social presence than those in the control condition, β = .38, p = .01. However, at one standard deviation above the mean, participants in both conditions did not differ in social presence, β = -.02, p = .93. The findings demonstrated that social presence was more strongly influenced by media format among individuals low (vs. high) in trait empathy, which was opposite to H8 (b). The interaction pattern is plotted in Figure 5.5. 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 IMEs Non-IMEs Low trait empathy High trait empathy Figure 5.5. Interactions of media format and trait empathy on social presence. 95 Social presence Chapter 6: General Discussion This chapter has a few objectives. First, the chapter provides a summary of the findings across two video contexts. Second, the chapter addresses the limitations and discusses the future research directions. Third, despite the limitations, it discusses the practical implications of this dissertation. Finally, related ethical issues are discussed and conclusions made from this research are presented. 6.1 Summary of Experimental Findings 6.1.1 The Main Model Drawing upon literature in narrative persuasion and immersive media, this dissertation proposed a new model, the persuasive narrative theory in immersive mediated environments (PENTIMEs). PENTIMEs aimed to model narrative effects and its underlying psychological mechanisms in IMEs. PENTIMEs predicted that narratives presented in immersive (vs. non-immersive) mediated environments would lead to a higher level of spatial presence and social presence. In addition, narratives presented in immersive (vs. non-immersive) mediated environments would lead to a higher level of transportation and identification. Spatial presence would mediate the relation between media format and transportation. Similarly, social presence would mediate the relation between media format and identification. Furthermore, media format would have an indirect effect on counterarguing through spatial presence, social presence, transportation, and/or identification. The model also predicted that media format would have an indirect effect on persuasive outcomes through the proposed mediators. 96 A laboratory experiment was conducted to test PENTIMEs across two video contexts. The results from both video contexts showed a relatively consistent pattern, supporting the majority of the hypotheses proposed in the model. First of all, individuals who viewed the 360° video stories in IMEs had a greater sense of presence than those in the non-IMEs in general, which is consistent with previous research (Fonseca & Kraus, 2016; Sundar et al., 2017). In the DUI context, narratives presented in IMEs (vs. non-IMEs) led to a higher level of spatial presence but not social presence. In the malaria context, narratives presented in IMEs (vs. non-IMEs) led to a higher level of spatial presence and social presence. Because social presence emphasizes the sense of being in the same space with media characters, the insignificant link between media format and social presence in the DUI context could be explained by the multiple characters and its production techniques. Unlike the malaria video, the DUI video involves multiple characters in three moving vehicles. The four-and-a-half-minute video cuts among the three vehicles about eight times, forcing the viewers to feel constantly changing their seats in different vehicles in a very short time. The “constant seats changing” might reduce the viewers’ feeling of being together with the characters in the same space. Moreover, viewers were supposed to be seated in the moving vehicles. Without sitting in a vibrating chair that moves around in sync with the vehicles, viewers could not feel a greater sense of social presence when they were watching the video on HMDs versus iPads. In contrast, for the malaria video, when viewers were situated in a relatively non-changing non-moving space, they felt a greater sense of social presence in IMEs versus non-IMEs. In addition, the proposed model explained about 97 10% of variance in presence, indicating that the immersive storytelling had a relatively small effect on presence (Cohen, 1988) Second, narratives presented in immersive (vs. non-immersive) mediated environments did not directly lead to a higher level of transportation and identification. However, IMEs (vs. non-IMEs) increased transportation through enhanced spatial presence in both video contexts. Moreover, IMEs (vs. non-IMEs) magnified identification through improved social presence in the malaria context, but not in the DUI context. Nonetheless, social presence led to a greater level of identification in the DUI context. These results were in line with the MAIN model (Sundar, 2008) that proposed technological affordances served as cues to trigger cognitive heuristics, which would in turn affect users’ perceptual experience with the media content. Immersive technologies took audience members to the mediated world from their office, home, or wherever they physically were, and made them to feel they were in the exact place in which the story occurred. Compared to the story presented in non-IMEs, the one presented in IMEs led to a higher level of spatial presence, which in turn increased transportation. In other words, when viewers felt they were located in the mediated environments, they were more likely to attend to the events happening in the mediated environments and resonated their emotions with the events. Furthermore, the feeling of being in the same space with the story characters increased the likelihood of the audience taking emotional and cognitive perspective with the story characters (i.e., identification). The proposed model explained 28.9% of variance in transportation and 19.7% of variance in identification in the DUI context, and 30.1% of variance in transportation and 34.8% of variance in 98 identification in the malaria context. Overall, the model had a medium-sized effect on transportation and identification (Cohen, 1988). Because one of the major advantages of narrative persuasion is its ability to overcome persuasion resistance (Green & Brock, 2000; Moyer-Gusé, 2008; Slater & Rouner, 2002), PENTIMEs included counterarguing, a common form of persuasion resistance studied in the narrative persuasion literature, as an important psychological mechanism to explain narrative effects in IMEs. PENTIMEs proposed that narratives presented in immersive (vs. non-immersive) mediated environments would reduce counterarguing through spatial presence, social presence, transportation, and /or identification. There were four specific indirect effects from media format to counterarguing: one through transportation, one through spatial presence and transportation, one through identification, and one through social presence and identification. The total indirect effects of media format on counterarguing was significant in the DUI context, but not in the malaria context. However, the results revealed a similar pattern for the specific indirect effect. The indirect effect of media format on counterarguing through spatial presence and transportation approached significance in the DUI context and was significant in the malaria context. These findings suggested narratives presented in the immersive (vs. non-immersive) mediated environments indirectly reduced counterarguing through increasing spatial presence and transportation. The model explained 20.3% and 26.4% of variance in counterarguing in the DUI and malaria context, respectively. Finally, PENTIMEs predicted that narratives presented in the immersive (vs. non-immersive) mediated environments would lead to greater persuasiveness, and 99 this effect would be mediated through spatial presence, social presence, transportation, identification, and/or counterarguing. In the DUI context, participants in the immersive (vs. non-immersive) mediated environments reported less favorable attitudes toward DUI and fewer intentions to drive under the influence of alcohol. The effect of media format on attitudes toward DUI was not mediated by the variables proposed in the model. Hence, it is not clear on why messages presented in IMEs led to less favorable attitudes toward DUI than those in non-IMEs. However, the results showed a significant total indirect effect of media format on intentions to engage in DUI. Among the 10 specific indirect effects, the one through spatial presence was marginally significant. Specifically, IMEs increased spatial presence, which then reduced behavioral intentions to drive under the influence of alcohol. The model explained 56.8% of variance in attitudes toward DUI and 42.6% of variance in intentions to engage in DUI. In the malaria context, participants did not differ in attitudes and willingness to help people who need protection from malaria across two experimental conditions. Nonetheless, the results indicated media format had a significant indirect effect on attitudes toward helping people through spatial presence, transportation, and counterarguing. Furthermore, media format had an approaching significant indirect effect on willingness to help people through the proposed mediators in the model. It is noteworthy that although the current study did not find a statistically significant impact of media format on attitudes and willingness in the malaria context, the model explained a moderate amount of variance in attitudes toward helping people (38.9%) and a large amount of variance in willingness to help people (50.7%). The non- 100 significant link could be a result of small sample size or a lack of control of individual differences such as familiarity with the malaria issue. Compared to the drunk driving issue that participants might have experienced themselves or heard of others experienced, the topic on malaria might not be much relevant to the current sample. It is speculated that the familiarity of the topical issue could impact how participants respond in different mediated environments. Future research could inquire the role of the topical familiarity in immersive storytelling. In short, the dissertation demonstrated that narratives presented in immersive mediated environments were effective in influencing viewers’ attitudes and behavioral intentions/willingness. These results were consistent with previous research that showed immersive virtual reality was a promising persuasive tool (Ahn, 2015; Ahn et al., 2016; Rosenberg, Baughman, & Bailenson, 2013). In addition, the persuasive effects of storytelling in IMEs could be explained by several psychological mechanisms, including spatial presence, social presence, transportation, identification, and counterarguing. 6.1.2 The Main Model with Trait Empathy as a Moderator Because trait empathy is an important personality trait that influences psychological experience in IMEs (Nicovich et al., 2005; Sas & O’Hare, 2003; Wallach et al., 2010), the current study also proposed that trait empathy would moderate the impact of media format on spatial presence, social presence, transportation, and identification. In the DUI context, the results demonstrated that trait empathy did not affect the influence of media format on spatial presence, social presence, transportation, nor 101 identification. In the malaria context, the results showed that trait empathy moderated the impact of media format on social presence, but not on three other variables. Specifically, narratives presented in the immersive (vs. non-immersive) mediated environments led to a higher level of social presence among individuals with low and moderate trait empathy. However, among those who are high in trait empathy, social presence did not differ between two conditions. In other words, social presence was more strongly influenced by media format among individuals with low (vs. high) trait empathy, which was opposite to what was predicted. Although previous studies found that the higher trait empathy an individual had, the greater spatial presence (e.g., Wallach et al., 2010) and narrative transportation one experienced in the same mediated environments (e.g., Hall & Bracken, 2011), the current study indicated trait empathy, name fantasy, did not affect users’ level of spatial presence, transportation, and identification when being exposed to different mediated environments (i.e., IMEs vs. non-IMEs). Furthermore, in the malaria context, narratives presented in immersive (vs. non-immersive) mediated environments led to greater social presence among participants with low (vs. high) trait empathy. This result was opposite to the hypothesis. It was speculated that low- trait-empathy users needed more external cues to experience a higher level of social presence. Watching the 360° video story on HMDs (vs. iPads) provided low-trait- empathy viewers with greater immersion. However, for viewers with high trait empathy, they might be able to experience a relatively high level of social presence without too many environmental cues. Therefore, social presence generated in two mediated environments did not differ among individuals with high trait empathy. 102 6.2 Limitations and Future Directions There are a few limitations to this dissertation that merit discussion. The major limitations arise from the complex experimental designs that contain combinations of between-subject and within-subject variables. Due to the limited sample size, the current study conducted two separate path analyses for each video context instead of one structural equation model that could control error variances among the repeated measured items. Because the major constructs such as spatial presence and social presence were measured twice, one for each context, this could threaten internal validity. However, if there were a larger sample size, the measurement errors among the repeated measured items could be correlated in one model, thus controlling the error variances. In addition, such model would have allowed statistical comparison between the results from the two contexts. The current separate path analysis could only provide a qualitative description of the general pattern. Another limitation suffered from the experimental design was the effect of video order on counterarguing. Although the effect size of the video order was small, the results should be interpreted with caution. Second, the head or hand movements were not tracked due to the technological difficulties. Although all participants were instructed to move their heads or hands in different directions before they watched the video, some participants were more likely to move than others. The different degree of head or hand movements could possibly influence individual’s psychological experience in the mediated environments, and should be controlled in the statistical analyses. Future 103 research should use devices that can measure the users’ head or hand movements to obtain more precise results. Third, there are a few drawbacks in terms of the measurement. Self-report questionnaire was employed in this dissertation to measure attitudes/behavioral intentions. There could be a ceiling effect for participants to report their attitudes and/or behavioral intentions toward drunk driving or helping others. Although participants’ social desirability was controlled in the analyses, more covert measures of attitudes/behavioral intentions could be used in the future. For example, the implicit association test can be employed to assess attitudes toward drunk driving and intentions to drive under the influence. Moreover, the current study did not measure participants’ actual behaviors. For instance, in the context of providing humanitarian aid, participants could be given an opportunity to donate money, which can serve as a behavioral outcome. More research is needed to test the effects of narratives presented in the IMEs on actual behaviors. This dissertation also implemented the questionnaires immediately after the message exposure. Therefore, the long-term effects of narrative exposure were not determined. Furthermore, the proposed model was tested with 360° videos that shot real people. VR, which is made of computer-generated simulations, also produces IMEs. Future studies may investigate narrative effects in immersive virtual environments. For example, VR can provide users with an opportunity to play the role of a story character. It would be interesting to examine how this role-play experience affects narrative effects and psychological experience. In addition, researchers should continue to identify factors that potentially affect users’ psychological experience 104 when exposed to the narratives in IMEs. Trait empathy, namely fantasy, was not found to be an important moderator. Other personality traits, such as immersive tendencies and transportability, might be considered in the future inquiries. Last but not least, there are a few issues with regard to the sample. Because the current sample consisted of young adults attending a large research university in North America, the extent to which the findings are generalizable to other populations is not clear. As previous research indicated that age affected spatial and social presence (Siriaraya & Siang Ang, 2012), future research could replicate this study in an older population. Moreover, culture might also affect viewers’ experience in IMEs. As found in the current study, users had a greater sense of being in the same place with Amisa (the main character in the malaria video) in immersive (vs. non- immersive) mediated environments. In other words, immersive media enables viewers to feel they have a relatively short distance with the narrative character or even sit or stand right next to him/her. However, the preferred interpersonal distances differ across the world (Sorokowska et al., 2017); researchers could investigate how users in areas other than North America respond to narratives presented in IMEs. Moreover, the independent-t test showed that participants who dropped out of the study after completing phase I questionnaire had more approving attitudes toward DUI than those who completed both phases. Therefore, it is not clear if immersive stories have a similar effect on individuals who are more favorable toward DUI. Researchers could further investigate this area by targeting at individuals who have relatively more positive attitudes toward DUI. 105 6.3 Practical Implications Narrative persuasion and immersive media were independently proposed as promising persuasive strategies. Findings from the current study indicated that a combination of these two strategies achieved greater persuasive effects, which hold important practical implications. The major implication is that using immersive media to tell stories is a powerful persuasive tool. As technology advancement makes immersive media content relatively inexpensive to create, 360° videos stories can be easily implemented in the public campaigns. For example, United Nations SDG Action Campaign utilizes immersive media to tell stories of vulnerable groups, which aims to spread awareness and influence decision makers. As narratives have been increasingly used to promote public health and safety, immersive storytelling is of great potential to change individual’s health-related beliefs, attitudes, and behaviors. This creates a tremendous opportunity for public health educators and practitioners to incorporate immersive storytelling in their interventions. Moreover, the current study demonstrated that 360° video stories were more effective when viewed on HMDs than on mobile devices. Practitioners might want to distribute inexpensive VR headsets, such as Google Cardboards (about $3 each), to achieve greater persuasiveness. For instance, the New York Times sent out Google Cardboards to its readers to experience immersive storytelling with their “NYT VR” application. In addition, the high mobility of the HMDs affords the stories to be played at any location of choice, ranging from streets to classrooms. If customers were offered to watch an immersive story of a drunk driving case before they left 106 bars, it would largely decrease the likelihood of them driving under the influence. Evidence showed that a fundraising conference helped raise 3.8 billion dollars, over 70 percent more than what was projected, after showing donors an immersive story of a Syrian refugee (Watercutter, 2016). Findings from this project also offer a few suggestions to design effective immersive stories. To create a story that enhances both spatial and social presence, video producers might not change the story settings too often. The DUI video used in the study cuts among three vehicles about eight times within four and half minutes. This forced viewers to constantly “change” their seats in different vehicles, thus reducing their feeling of being together with the characters in the same space. In contrast, the malaria video does not cut through different settings frequently, which induces a significantly higher level of spatial presence when viewed in immersive (vs. non-immersive) mediated environments. In addition, the vehicles are moving in the DUI video. Providing viewers with vibrating chairs that move around in sync with the vehicles could offer them a more immersive experience. 6.4 Ethical Issues The current research found that immersive media indirectly reduced users’ counterarguing, the generation of thoughts that are inconsistent with persuasive attempt, through spatial presence and transportation. Reduced counterarguing then increased message persuasiveness. For a prosocial message, using immersive media to reduce users’ counterarguing to achieve greater persuasiveness is desirable. However, it is concerned that immersive media might be used as a method of 107 brainwashing, which is to introduce new and unwanted thoughts to individuals’ mind after purposely reduce their critical thinking ability. An ethical issue that needs to be discussed here is that whether researchers are responsible for the potential negative effects of their work. My answer to this question is yes. When the foreseeable harms outweigh the benefits of a research project, the research should not be initiated, such as human cloning. In this research, the effect of immersive media on users’ generation of counter arguments is indirect and small. Therefore, it is hard to draw the conclusion that immersive media can be used as a brainwashing tool. If the future research found that immersive media was indeed an effective brainwashing tool, regulations and codes of ethics should be enforced to address this technology-related ethical dilemma (Herschel & Andrews,1997). 6.5 Conclusions This dissertation project proposed and empirically tested the persuasive narrative theory in immersive mediated environments (PENTIMEs). The major thesis was that the cognitive heuristics generated by IMEs affected the story recipients’ involvement with the narrative and the characters, which in turn had an impact on persuasion resistance and ultimately persuasive effects. In addition, this project also tested the moderating role of trait empathy. Results from two video contexts provided relatively consistent evidence that supported PENTIMEs. Trait empathy was not found to be an important moderator. This dissertation made several important theoretical contributions to the communication research. First and foremost, this project proposed a theoretical model 108 through integrating literature in narrative persuasion and immersive media. Although previous research suggested that narrative communication was a promising persuasive tool, almost all the studies were conducted in traditional/non-immersive mediated environments. To my best knowledge, the current study was among the first to investigate the effects of persuasive narratives in IMEs, which extended the scholarship in narrative persuasion. Second, as most research studied narrative effects by comparing narrative versus non-narrative messages, it was hard to keep the messages identical across different conditions. This study was the first study to examine narrative persuasion by using the same narrative message across experimental conditions. Furthermore, IMEs have the ability to amplify key psychological experiences such as spatial and social presence, which then affects story involvement and ultimately persuasive outcomes. This dissertation was the first to empirically test this central thesis, hence advancing our understanding of the psychological mechanisms underlying narrative persuasion in IMEs. Finally, while immersive media have been proposed as the ultimate empathy machine (Milk, 2015), there was a lack of understanding on how individuals’ empathic tendency influenced their psychological responses in IMEs. The dissertation was the first to test the moderating role of trait empathy on psychological experience including presence and story involvement. In conclusion, this project demonstrated the promise of using immersive media to provide vivid and engaging narrative experience that addresses important issues. A theoretical model was proposed and tested in two contexts. Findings from the study provided several important theoretical contributions to the communication 109 theory and practical implications for addressing issues related to safety and charity behavior. 110 Appendix A: Video Transcripts The DUI video [This story is set in three vehicles. There is one girl, Samantha, in the first vehicle. There is a couple in the second vehicle. And there are two young men and one young woman in the third vehicle.] Vehicle 1: Samantha: Hello. Man (from phone): Hi Samantha. Samantha: Yes, who’s this? Man (from phone): Darren Echoes from Specter Fashion. Saw your portfolio and loved it. I’m going to grab a drink with the design team. Are you free to stop by the bar? Vehicle 2: (The couple is sitting in the car.) Vehicle 3: Boy 1: You know think about letting my beard grow, I’m talking like two years out. Boy 2: I think Stacey would love that. Girl: What do you think? Boy 2: Dude, that’s what I do. Girl: Can you even grow facial hair? Vehicle 1: Women (from phone): Wait so you’ve already met the design team at the bar? What happened? Samantha: They have bottle service like I didn’t know that was possible before 7. Woman (from phone): They don’t waste time. Samantha: I tried to keep up with them, but they’re pros. I just damn hope I didn’t look like too much of a wuss. Woman (from phone): Are you okay to drive? Vehicle 3: Boy 1: Stacey sent me a snap and she’s already at the party. Why aren’t we at the party? Girl: Taking the long way so your beard has a few more minutes. Vehicle 2: Wife: I love you as a dad. You are supposed to say I love you as a mom. Husband: I do. I do. Vehicle 3: Boy 1: You think we’re being cool if I ask Stacey for directions? 111 Girl: What do you think? (Samantha speeds up and passes the couple’s car.) Vehicle 2: Wife: Someone is in a hurry. Vehicle 3: Girl: Look on the bright side. Your five o’clock shadow just hit. Ahhh…. (Car crashes) Voice over (police): Three vehicles. Multiple casualties. Voice message (from the couple’s babysitter): Hey guys, will you be home soon? I thought you said 9pm. It’s almost ten and Lily just woke up crying. I think she just wants her mom. The malaria video [Music] Amisa: My name is Amisa. I’m 11 years old. I live in Nyarugusu with my mom, five sisters, and a little brother. We are originally from the Congo. There is a lot of violence back home. [Music] When the boys came with guns in the night and took my father, the rest of my family came here to be safe. I miss my daddy every day and hope he is okay. My mom says it is a lot safer here, but there is something else that scares me, malaria. I caught malaria two months ago. I was really sick and scared. My little sisters are ill now. They have high fevers and chills. Chasiline cries all the time. There are so many sick people in the hospital, and we have to wait our turn to sign in. The nurse takes my sister’s blood. [Music] The doctor says they have malaria. I’m worried, but the doctor gives them medicine and says they should get well soon. But a lot of kids here are fighting malaria, and some people don’t make it. Living in the big tent with so many people around, it is easy for mosquito to spread malaria. But today is a special day for me and my family. Today is our time to move into our very new tent in a different part of the country. [Music] We have waited for a long time for this. I hope my best friend Foraha and our other neighbors can get their own tent soon. So many people are here and everything they own. The compca helps us build the new tent. [Music] It has been a very long time since we had our own space. I love our new home. It is so clean and quiet. Once we register in the neighborhood, we need to sign up for the school. I get to meet my new headmaster, I will start 4th grade soon. I cannot wait. I love learning. I want to be a nurse when I grow up to help deliver babies and keep people safe. But I have to stay healthy if I want to stay in school. That is why I’m so thankful to get a new tent. The health workers say the net will prevent malaria. I wish everyone here could have one. [Music] It is very important to 112 hang up the mosquito net properly and always sleep under the net. When the night falls, the mosquitos come out. Darkness is a most dangerous time of our day. I’m so happy now. My sisters are already feeling better. We have a new home and a new mosquito net to keep us safe from malaria. My family and I have travelled really far to get here. It has not been easy. But now we have hope. Finally, we can sleep safe. We can dream good dreams, and pray for our daddy to come and find us soon. [Music] 113 Appendix B: Measures *Unless otherwise noted, use 1-7 semantic or Likert scales. PHASE I -- BASELINE MEASURES -- Demographics How old are you? ________ Which of the following best describes your racial/ethnic background? • White • Black or African-American • Spanish, Hispanic, or Latino • Asian • Other Are you: • Male • Female Greek affiliation Are you affiliated with a fraternity/sorority? • Yes • No Baseline attitudes toward helping others • People should be willing to help others who are less fortunate. • Helping troubled people with their problems is very important to me. • People should be more charitable toward others in society. • People in need should receive support from others. Baseline attitudes toward drunk-driving In your view, driving under the influence of alcohol would be • bad/good • foolish/wise • negative/positive 114 • unfavorable/favorable Trait empathy Fantasy Items [taps respondents’ tendencies to transpose themselves imaginatively into the feelings and actions of fictitious characters in books. movies, and plays]. • I daydream and fantasize, with some regularity, about things that might happen to me. • I really get involved with the feelings of the characters in a novel. • I am usually objective when I watch a movie or play, and I don’t often get completely caught up in it. (R) • Becoming extremely involved in a good book or movie is somewhat rare for me. (R) • After seeing a play or movie, I have felt as though I were one of the characters. • When I watch a good movie, I can very easily put myself in the place of a leading character. • When I am reading an interesting story or novel, I imagine how I would feel if the events in the story were happening to me. Social desirability • I’m always willing to admit it when I make a mistake. • I never resent being asked to return a favor. • I have never been irked when people expressed ideas very different from my own. • I have never deliberately said something that hurt someone’s feelings. • I never hesitate to go out of my way to help someone in trouble. • I have never intensely disliked anyone. • When I don’t know something I don’t at all mind admitting it. • I am always courteous, even to people who are disagreeable. • I would never think of letting someone else be punished for my wrong doings. -- RANDOMIZATION – Participants were randomly assigned to one of the two groups: phase 2 (cat) or phase 2 (dog) study. They were asked to sign up for the according phase 2 study based on the animal picture they get here (i.e., either a cat or a dog). Phase 2 (cat) study asked participants to view the 360-degree on the iPads and phase 2 (dog) study asked them to view the same video on Samsung Gear VR. SONA ID 115 What is your anonymous 6-digit SONA ID code (example: ‘123493’)? Please do NOT type into your user name. Only type into your 6-digit SONA ID code. PHASE II EXPOSURE TO STIMULUS POST-EXPOSURE MEASURES – PROCESSES AND PERSONALITIES Spatial presence • To what extent did you experience a sense of “being there” inside the environment you saw/heard? • How much did it seem as if the objects and people you saw/heard had come to the place you were? (Not at all/Very much) • How much did it seem as if you could reach out and touch the objects or people you saw/heard? (Not at all/Very much) • How often when an object seemed to be headed toward you did you want to move to get out of its way? (Never/Always) Social presence • How often did you have the sensation that people you saw/heard could also see/hear you? • To what extent did you feel you could interact with the person or people you saw/heard? • How much did it seem as if you and the people you saw/heard were together in the same place? • How often did it feel as if someone you saw/heard in the environment was talking directly to you? Transportation • The story affected me emotionally. • I was mentally involved in the story while watching it. • While I was watching the story, activity going on in the room around me was on my mind. (R) • While I was watching the story, I forgot my daily affairs. • During watching the story, I did not think for a while about the things that had been on my mind lately. Identification • When I was watching the story, I could feel the emotions the character(s) expressed. 116 • I was able to understand the events in this video in a manner similar to that in which the character(s) understood them. • When I was watching the story, I felt I could really get inside the character(s)’ head. • At key moments in this story, I felt I knew exactly what the character(s) were going through. Counterarguing • While watching this video, I found myself actively disagreeing with it. • I was looking for flaws in the way information was presented in this video. • I found myself actively disagreeing with the video. Cybersickness • To what extent did you feel uncomfortable during the media experience? • To what extent did you feel dizzy during the media experience? • To what extent did you feel fatigue during the media experience? • To what extent did you feel lightheaded during the media experience? • To what extent did you experience vision discomfort during the media experience? • To what extent did you experience eyestrain during the media experience? POST-EXPOSURE MEASURES – KEY OUTCOMES Posttest attitudes toward DUI Now, please think about how you feel about driving under the influence of alcohol at this moment. Would you say driving under the influence of alcohol is: • bad/good • foolish/wise • negative/positive • unfavorable/favorable Behavioral intentions to engage in DUI [1=Completely unlikely; 3=Extremely unlikely; 5=Quite unlikely; 7=Slightly unlikely; 9=Neutral; 11=Slightly likely; 13=Quite likely; 15=Extremely likely; 17=Completely likely] • How likely is it that, in the next 30 days, you would drive a vehicle when you feel a “slight buzz” from drinking alcohol? • How likely is it that, in the next 30 days, you would help to drive a friend home when you’re slightly feeling some initial effects from having a few alcoholic drinks? • How likely is it that, in the next 30 days, you would be a passenger in a car driven by someone who had been drinking? 117 Posttest attitudes toward helping people who need protection from malaria Think about how you feel about the issue presented in the video at this moment. Would you say that helping people like Amisa to get bed nets as a protection from malaria is: • bad/good • foolish/wise • negative/positive • unfavorable/favorable Behavioral willingness to help people who need protection from malaria How willing are you to engage in the following actions? [1=Completely unwilling; 3=Extremely unwilling; 5=Quite unwilling; 7=Slightly unwilling; 9=Natural; 11=Slightly willing; 13=Quite willing; 15=Extremely willing; 17=Completely willing] • Follow Nothing But Nets campaign on social media to learn more about the fight to defeat malaria. • Share this video with your family/friends to disseminate the message about the situation of the people who need protection from malaria. • Donate money to send nets and help save lives. • Join the Nothing But Nets Champions Council and become a leader to defeat malaria. SONA ID What is your anonymous 6-digit SONA ID code (example: ‘123493’)? Please do NOT type into your user name. Only type into your 6-digit SONA ID code. 118 References Ahn, S. J. G. (2015). Incorporating immersive virtual environments in health promotion campaigns: A construal level theory approach. Health Communication, 30, 545-556. doi:10.1080/10410236.2013.869650 Ahn, S. J. G., Bostick, J., Ogle, E., Nowak, K. L., McGillicuddy, K. T., & Bailenson, J. N. (2016). Experiencing nature: Embodying animals in immersive virtual environments increases inclusion of nature in self and involvement with nature. Journal of Computer-Mediated Communication, 21, 399-419. doi:10.1111/jcc4.12173 Ames, S. L., Wolffsohn, J. S., & Mcbrien, N. A. (2005). The development of a symptom questionnaire for assessing virtual reality viewing using a head- mounted display. Optometry and Vision Science, 82, 168-176. doi:10.1097/01.OPX.0000156307.95086.6 Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Thousand Oaks: California: Sage. Appel, M., & Richter, T. (2010). Transportation and need for affect in narrative persuasion: A mediated moderation model. Media Psychology, 13, 101-135. doi:10.1080/15213261003799847 Aymerich-Franch, L., Karutz, C., & Bailenson, J. N. (2012). Effects of facial and voice similarity on presence in a public speaking virtual environment. Proceedings of the International Society for Presence Research Annual Conference, USA, 24-26. 119 Banerjee, S. C., & Greene, K. (2012). Role of transportation in the persuasion process: Cognitive and affective responses to antidrug narratives. Journal of Health Communication, 17, 564-581. doi:10.1080/10810730.2011.635779 Bentler, P. M. (2004). EQS 6 Structural equations program manual. Encinco, CA: Multivariate Software, Inc. Bilandzic, H., & Busselle, R. W. (2008). Transportation and transportability in the cultivation of genre-consistent attitudes and estimates. Journal of Communication, 58, 508-529. doi:10.1111/j.1460-2466.2008.00397.x Biocca, F. (1997). The cyborg’s dilemma: Progressive embodiment in virtual environments. Journal of Computer-Mediated Communication, 3. doi:10.1111/j.1083-6101.1997.tb00070.x Biocca, F., Harms, C., & Burgoon, J. K. (2003). Toward a more robust theory and measure of social presence: Review and suggested criteria. Presence: Teleoperators & Virtual Environments, 12, 456–480. doi:10.1162/105474603322761270 Blascovich, J., Loomis, J., Beall, A. C., Swinth, K. R., Hoyt, C. L., & Bailenson, J. N. (2002). Immersive virtual environment technology as a methodological tool for social psychology. Psychological Inquiry, 13, 103–124. doi:10.1207/S15327965PLI1302_01 Boon, S. D., & Lomore, C. D. (2001). Admirer-celebrity relationships among young adults: Explaining perceptions of celebrity influence on identity. Human Communication Research, 27, 432-465. doi:10.1111/j.1468- 2958.2001.tb00788.x 120 Burrows, C. N., & Blanton, H. (2016). Real-world persuasion from virtual-world campaigns How transportation into virtual worlds moderates in-game influence. Communication Research, 43, 542-570. doi:10.1177/0093650215619215 Byrne, B. M. (2013). Structural equation modeling with Mplus: Basic concepts, applications, and programming. New York, NY: Routledge. Cain, M. K., Zhang, Z., & Yuan, K. H. (2017). Univariate and multivariate skewness and kurtosis for measuring nonnormality: Prevalence, influence and estimation. Behavior Research Methods, 49, 1716-1735. doi:10.3758/s13428- 016-0814-1 Chang, C. (2008). Increasing mental health literacy via narrative advertising. Journal of Health Communication, 13, 37–55. doi:10.1080/10810730701807027 Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillside, NJ: Erlbaum Cohen, J. (2001). Defining identification: A theoretical look at the identification of audiences with media characters. Mass Communication & Society, 4, 245– 264. doi:10.1207/S15327825MCS0403_01 Collins, R. L., Elliott, M. N., Berry, S. H., Kanouse, D. E., Kunkel, D., Hunter, S. B., & Miu, A. (2004). Watching sex on television predicts adolescent initiation of sexual behavior. Pediatrics, 114, e280-e289. doi:10.1542/peds.2003-1065-L Crowne, D. P., & Marlowe, D. (1960). A new scale of social desirability independent of psychopathology. Journal of Consulting Psychology, 24, 349-354. doi:10.1037/h0047358 121 Cummings, J. J., & Bailenson, J. N. (2016). How immersive is enough? A meta- analysis of the effect of immersive technology on user presence. Media Psychology, 19, 272–309. doi:10.1080/15213269.2015.1015740 Curran, P. J., West, S. G., & Finch, J. F. (1996). The robustness of test statistics to nonnormality and specification error in confirmatory factor analysis. Psychological Methods, 1, 16-29. doi:10.1037/1082-989X.1.1.16 Davis, M. H. (1983). Measuring individual differences in empathy: Evidence for a multidimensional approach. Journal of Personality and Social Psychology, 44, 113-126. doi:10.1037/0022-3514.44.1.113 DeCarlo, L. T. (1997). On the meaning and use of kurtosis. Psychological Methods, 2, 292-307. doi:10.1037/1082-989X.2.3.292 de Graaf, A., Hoeken, H., Sanders, J., & Beentjes, J. W. J. (2012). Identification as a mechanism of narrative persuasion. Communication Research, 39, 802–823. doi:10.1177/0093650211408594 de Wit, J. B., Das, E., & Vet, R. (2008). What works best: objective statistics or a personal testimonial? An assessment of the persuasive effects of different types of message evidence on risk perception. Health Psychology, 27, 110- 115. doi:10.1037/0278-6133.27.1.110 Elmezeny, A., Edenhofer, N., & Wimmer, J. (2018). Immersive Storytelling in 360- Degree Videos: An Analysis of Interplay Between Narrative and Technical Immersion. Journal for Virtual Worlds Research, 11, 1-13. doi:10.4101/jvwr.v11i1.7298 122 Felnhofer, A., Kothgassner, O. D., Hauk, N., Beutl, L., Hlavacs, H., & Kryspin- Exner, I. (2014). Physical and social presence in collaborative virtual environments: Exploring age and gender differences with respect to empathy. Computers in Human Behavior, 31, 272–279. doi:10.1016/j.chb.2013.10.045 Finney, S. J., & DiStefano, C. (2006). Non-normal and categorical data in structural equation modeling. In G.R. Hancock & R.O. Mueller (Eds.), Structural equation modeling: A Second Course (pp. 269-314). Charlotte, NC: Information Age Publishing, Inc. Fonseca, D., & Kraus, M. (2016, October). A comparison of head-mounted and hand- held displays for 360° videos with focus on attitude and behavior change, Proceedings of the 20th International Academic Mindtrek Conference, Finland, 287-296. doi:10.1145/2994310.2994334 Fortin, D. R., & Dholakia, R. R. (2005). Interactivity and vividness effects on social presence and involvement with a web-based advertisement. Journal of Business Research, 58, 387-396. doi:10.1016/S0148-2963(03)00106-1 Fox, J., Arena, D., & Bailenson, J. N. (2009). Virtual reality: A survival guide for the social scientist. Journal of Media Psychology: Theories, Methods, and Applications, 21, 95–113. doi:10.1027/1864-1105.21.3.95 Green, M. C. (2004). Transportation into narrative worlds: The role of prior knowledge and perceived realism. Discourse Processes, 38, 247-266. doi:10.1207/s15326950dp3802_5 123 Green, M. C., & Brock, T. C. (2000). The role of transportation in the persuasiveness of public narratives. Journal of Personality and Social Psychology, 79, 701- 721. doi:10.1037/0022-3514.79.5.701 Green, M. C., & Brock, T. C. (2002). In the mind’s eye: Transportation-imagery model of narrative persuasion. In M. C. Green, J. J. Strange, & T. C. Brock (Eds.), Narrative impact: Social and cognitive foundations (pp. 315-341). Mahwah, NJ: Lawrence Erlbaum. Grigorovici, D. (2003). Persuasive effects of presence in immersive virtual environments. In G. Riva, F. Davide, & W.A. IJsselsteijn (Eds). Being there: Concepts, effects and measurement of user presence in synthetic environments (pp. 192-205). Amsterdam, The Netherlands: Ios Press. Hall, A. E., & Bracken, C. C. (2011). “I really liked that movie”: Testing the relationship between trait empathy, transportation, perceived realism, and movie enjoyment. Journal of Media Psychology: Theories, Methods, And Applications, 23, 90-99. doi:10.1027/1864-1105/a000036 Hayes, A. F. (2012). PROCESS: A versatile computational tool for observed variable mediation, moderation, and conditional process modeling [White paper]. Retrieved from http://www.afhayes.com/ public/process2012.pdf Heeter, C. (1992). Being there: The subjective experience of presence. Presence: Teleoperators and Virtual Environments, 1, 262–271. doi:10.1162/pres.1992.1.2.262 124 Herschel, R. T., & Andrews, P. H. (1997). Ethical implications of technological advances on business communication. The Journal of Business Communication, 34 160-170. doi:10.1177/002194369703400203 Hinyard, L. J., & Kreuter, M. W. (2007). Using narrative communication as a tool for health behavior change: A conceptual, theoretical, and empirical overview. Health Education & Behavior, 34, 777–792. doi:10.1177/1090198106291963 Hoeken, H., & Fikkers, K. M. (2014). Issue-relevant thinking and identification as mechanisms of narrative persuasion. Poetics, 44, 84-99. doi:10.1016/j.poetic.2014.05.001 Hoffner, C., & Buchanan, M. (2005). Young adults’ wishful identification with television characters: The role of perceived similarity and character attributes. Media Psychology, 7, 325-351. doi:10.1207/S1532785XMEP0704_2 Hopfer, S. (2012). Effects of a narrative HPV vaccination intervention aimed at reaching college women: A randomized controlled trial. Prevention Science, 13, 173-182. doi:10.1007/s11121-011-0254-1 Howarth, P. A., & Costello, P. J. (1997). The occurrence of virtual simulation sickness symptoms when an HMD was used as a personal viewing system. Displays, 18, 107-116. doi:10.1016/S0141-9382(97)00011-5 Hu, L. T., & Bentler, P. M. (1998). Fit indices in covariance structure modeling: Sensitivity to underparameterized model misspecification. Psychological Methods, 3, 424-453. doi:10.1037/1082-989X.3.4.424 125 Igartua, J. J. (2010). Identification with characters and narrative persuasion through fictional feature films. The European Journal of Communication Research, 35, 347–373. doi:10.1515/comm.2010.019. Kim, H. S., Bigman, C. A., Leader, A. E., Lerman, C., & Cappella, J. N. (2012). Narrative health communication and behavior change: The influence of exemplars in the news on intention to quit smoking. Journal of Communication, 62, 473–492. doi:10.1111/j.1460-2466.2012.01644.x Kim, T., & Biocca, F. (1997). Telepresence via television: Two dimensions of telepresence may have different connections to memory and persuasion. Journal of Computer-Mediated Communication, 3, doi:10.1111/j.1083- 6101.1997.tb00073.x Kreuter, M. W., Green, M. C., Cappella, J. N., Slater, M. D., Wise, M. E., Storey, D., ... & Woolley, S. (2007). Narrative communication in cancer prevention and control: A framework to guide research and application. Annals of Behavioral Medicine, 33, 221-235. doi:10.1007/BF02879904 Pixel Density Display Listing (n.a.). Laptop Pixel Density (PPI) List. Retrieved from http://pixensity.com/list/laptop/ LaBrie, J. W., Kenney, S. R., Mirza, T., & Lac, A. (2011). Identifying factors that increase the likelihood of driving after drinking among college students. Accident Analysis & Prevention, 43, 1371-1377. doi:10.1016/j.aap.2011.02.011 Lee, E. J., & Shin, S. Y. (2014). When the medium is the message: How transportability moderates the effects of politicians’ Twitter 126 communication. Communication Research, 41, 1088-1110. doi:10.1177/0093650212466407 Lee, K. M. (2004). Presence, explicated. Communication Theory, 14, 27–50. doi:10.1111/j.1468-2885.2004.tb00302.x Li, H., Daugherty, T., & Biocca, F. (2002). Impact of 3-D advertising on product knowledge, brand attitude, and purchase intention: The mediating role of presence. Journal of Advertising, 31, 43-57. doi:10.1080/00913367.2002.10673675 Lombard, M., & Ditton, T. (1997). At the heart of it all: The concept of presence. Journal of Computer-Mediated Communication, 3. doi:10.1111/j.1083- 6101.1997.tb00072.x Lombard, M., Ditton, T. B., & Weinstein, L. (2009, November). Measuring presence: The temple presence inventory, Proceedings of the 12th Annual International Workshop on Presence, USA, 1-15. Ma, Z., & Nan, X. (in press). Role of narratives in promoting mental illnesses acceptance. Atlantic Journal of Communication. McAfee, T., Davis, K. C., Alexander Jr, R. L., Pechacek, T. F., & Bunnell, R. (2013). Effect of the first federally funded US antismoking national media campaign. The Lancet, 382, 2003-2011. doi:10.1016/S0140-6736(13)61686-4 McQueen, A., Kreuter, M. W., Kalesan, B., & Alcaraz, K. I. (2011). Understanding narrative effects: The impact of breast cancer survivor stories on message processing, attitudes, and beliefs among African American women. Health Psychology, 30, 674-682. doi:10.1037/a0025395 127 Milk, C. (2015). How virtual reality can create the ultimate empathy machine. TED. Retrieved from https://www.ted.com/talks/chris_milk_how_virtual_reality_can_create_the_ul timate_empathy_machine?language=en Mon-Williams, M., Warm, J. P., & Rushton, S. (1993). Binocular vision in a virtual world: visual deficits following the wearing of a head-mounted display. Ophthalmic and Physiological Optics, 13, 387-391. doi:10.1111/j.1475-1313.1993.tb00496.x Moyer-Gusé, E. (2008). Toward a theory of entertainment persuasion: Explaining the persuasive effects of entertainment-education messages. Communication Theory, 18, 407–425. doi:10.1111/j.1468-2885.2008.00328.x Moyer-Gusé, E., & Nabi, R. L. (2010). Explaining the effects of narrative in an entertainment television program: Overcoming resistance to persuasion. Human Communication Research, 36, 26–52. doi:10.1111/j.1468- 2958.2009.01367.x Murphy, S. T., Frank, L. B., Chatterjee, J. S., & Baezconde-Garbanati, L. (2013). Narrative versus nonnarrative: The role of identification, transportation, and emotion in reducing health disparities. Journal of Communication, 63, 116- 137. doi:10.1111/jcom.12007 Murphy, S. T., Frank, L. B., Chatterjee, J. S., Moran, M. B., Zhao, N., Amezola de Herrera, P., & Baezconde-Garbanati, L. A. (2015). Comparing the relative efficacy of narrative vs nonnarrative health messages in reducing health 128 disparities using a randomized trial. American Journal of Public Health, 105, 2117-2123. doi:10.2105/AJPH.2014.302332 Murphy, S. T., Frank, L. B., Moran, M. B., & Patnoe-Woodley, P. (2011). Involved, transported, or emotional? Exploring the determinants of change in knowledge, attitudes, and behavior in entertainment-education. Journal of Communication, 61, 407-431. doi:10.1111/j.1460-2466.2011.01554.x Muthén, L. K., & Muthén, B. O. (2013). Mplus 7.11. Los Angeles, CA: Muthén & Muthén. Nabi, R. L., Moyer-Gusé, E., & Byrne, S. (2007). All joking aside: A serious investigation into the persuasive effect of funny social issue messages. Communication Monographs, 74, 29-54. doi:10.1080/03637750701196896 National Highway Traffic Safety Administration (2016). Drunk driving. Retrieved from https://www.nhtsa.gov/risky-driving/drunk-driving Nan, X., Dahlstrom, M. F., Richards, A., & Rangarajan, S. (2015). Influence of evidence type and narrative type on HPV risk perception and intention to obtain the HPV vaccine. Health communication, 30, 301-308. doi:10.1080/10410236.2014.888629 Nan, X., Futerfas, M., & Ma, Z. (2017). Role of narrative perspective and modality in the persuasiveness of public service advertisements promoting HPV vaccination. Health communication, 32, 320-328. doi:10.1080/10410236.2016.1138379 Nicovich, S. G., Boller, G. W., & Cornwell, T. B. (2005). Experienced presence within computer-mediated communications: Initial explorations on the effects 129 of gender with respect to empathy and immersion. Journal of Computer- Mediated Communication, 10. doi:10.1111/j.1083-6101.2005.tb00243.x Pelham, B. W., & Blanton, H. (2012). Conducting research in psychology: Measuring the weight of smoke (4th ed.). Pacific Grove, CA: International Thompson Publishing. Peng, W., Lee, M., & Heeter, C. (2010). The effects of a serious game on role-taking and willingness to help. Journal of Communication, 60, 723-742. doi:10.1111/j.1460-2466.2010.01511.x Ratan, R., & Hasler, B. S. (2010). Exploring self-presence in collaborative virtual teams. PsychNology Journal, 8, 11-31. Research and Markets (2017). Virtual reality (VR) marketplace: VR technologies, companies, solutions, devices, components, applications and services 2018 – 2023. Retrieved from https://www.researchandmarkets.com/research/whsf88/virtual_reality Rosenberg, R. S., Baughman, S. L., & Bailenson, J. N. (2013). Virtual superheroes: Using superpowers in virtual reality to encourage prosocial behavior. PLoS ONE, 8, e55003, doi:10.1371/journal.pone.0055003 Sas, C., & O’Hare, G. M. (2003). Presence equation: An investigation into cognitive factors underlying presence. Presence: Teleoperators and Virtual Environments, 12, 523-537. doi:10.1162/105474603322761315 Schumacker, R. E., & Lomax, R. G. (2010). A beginner’s guide to structural equation modeling: 3rd edition. New York, NY: Taylor & Francis Group. 130 Siriaraya, P., & Siang Ang, C. (2012). Age differences in the perception of social presence in the use of 3D virtual world for social interaction. Interacting with Computers, 244, 280-291. doi:10.1016/j.intcom.2012.03.003 Skalski, P., & Tamborini, R. (2007). The role of social presence in interactive agent- based persuasion. Media psychology, 10, 385-413. doi:10.1080/15213260701533102 Skalski, P., & Whitbred, R. (2010). Image versus sound: A comparison of formal feature effects on presence and video game enjoyment. PsychNology Journal, 8, 67-84. Slater, M. D., & Rouner, D. (2002). Entertainment—education and elaboration likelihood: Understanding the processing of narrative persuasion. Communication Theory, 12, 173-191. doi:10.1111/j.1468- 2885.2002.tb00265.x Slater, M., & Wilbur, S. (1997). A framework for immersive virtual environments (FIVE): Speculations on the role of presence in virtual environments. Presence: Teleoperators and Virtual Environments, 6, 603-616. doi:10.1162/pres.1997.6.6.603 Sorokowska, A., Sorokowski, P., Hilpert, P., Cantarero, K., Frackowiak, T., Ahmadi, K., ... & Blumen, S. (2017). Preferred interpersonal distances: A global comparison. Journal of Cross-Cultural Psychology, 48, 577-592. doi:10.1177/0022022117698039 131 Steuer, J. (1992). Defining virtual reality: Dimensions determining telepresence. Journal of Communication, 42, 73–93. doi:10.1111/j.1460- 2466.1992.tb00812.x Sundar S. S. (2008). The MAIN model: A heuristic approach to understanding technology effects on credibility. In M.J. Metzger and A.J. Flanagin (Eds.), Digital media, youth, and credibility (pp. 72–100). Cambridge, MA: The MIT Press. Sundar, S. S., Kang, J., & Oprean, D. (2017). Being there in the midst of the story: How immersive journalism affects our perceptions and cognitions. Cyberpsychology, Behavior, and Social Networking, 20, 672-682. doi:10.1089/cyber.2017.0271 Tal-Or, N., & Cohen, J. (2010). Understanding audience involvement: Conceptualizing and manipulating identification and transportation. Poetics, 38, 402-418. doi:10.1016/j.poetic.2010.05.004 Valente, T. W., Murphy, S., Huang, G., Gusek, J., Greene, J., & Beck, V. (2007). Evaluating a minor storyline on ER about teen obesity, hypertension, and 5 a day. Journal of Health Communication, 12, 551-566. doi:10.1080/10810730701508385 Wallach, H. S., Safir, M. P., & Samana, R. (2010). Personality variables and presence. Virtual Reality, 14, 3–13. doi:10.1007/s10055-009-0124-3 Walter, N., Murphy, S. T., Frank, L. B., & Baezconde-Garbanati, L. (2017). Each medium tells a different story: The effect of message channel on narrative 132 persuasion. Communication Research Reports, 34, 161-170. doi:10.1080/08824096.2017.1286471 Watercutter, A. (2016). VR films work great for charity. What about changing minds? Retrieved from https://www.wired.com/2016/03/virtual-reality-social-change- fundraising/ Webb, D. J., Green, C. L., & Brashear, T. G. (2000). Development and validation of scales to measure attitudes influencing monetary donations to charitable organizations. Journal of the Academy of Marketing Science, 28, 299-309. doi:10.1177/0092070300282010 Witmer, B. G., & Singer, M. J. (1998). Measuring presence in virtual environments: A presence questionnaire. Presence: Teleoperators and Virtual Environments, 7, 225–240. doi:10.1162/105474698565686 Wirth, W., Hofer, M., & Schramm, H. (2012). The role of emotional involvement and trait absorption in the formation of spatial presence. Media Psychology, 15, 19-43. doi:10.1080/15213269.2011.648536 World Health Organization (2017). World malaria report 2017. Retrieved from http://www.who.int/malaria/publications/world-malaria-report-2017/en/ Zwarun, L., & Hall, A. (2012). Narrative persuasion, transportation, and the role of need for cognition in online viewing of fantastical films. Media Psychology, 15, 327-355. doi:10.1080/15213269.2012.70059 133