I know from personal experience that it's true! How perceived realism in video games and television series may affect real-world beliefs

Transcription

1 I know from personal experience that it's true! How perceived realism in video games and television series may affect real-world beliefs Anne-Wil Coumou Universiteit van Utrecht (bachelor thesis) Keywords: Absorption, Real-world beliefs, Television Series, Video Games, Identification, Realism, Reality, Believability. Supervisors: rs: F. Hakemulder, M.M. Kuijpers.

2 Anne-Wil Coumou 2 Abstract From the idea that a narrative can be absorbing, bring about a transporting experience, make the spectator identify with characters and in turn experience the narrative as realistic comes our main hypothesis that perceived realism can affect real-world beliefs. We subjected our 70 participants to a questionnaire exploring this and two supporting hypotheses, making use of the individual's memories of a favourite series or video game. We then explored our results using multiple regression and came to mixed yet interesting and revealing conclusions, both for our hypotheses and as well for the mediumspecifc aspects of the process we propose.

3 How perceived realism in video games and television series may affect real-world beliefs 3 Introduction Sons of Anarchy is an American television show produced by FX Networks, named after the motorcycle club it revolves around. The MC struggles with their gun-running business, the law and internal conflict. I enjoyed watching the show from episode one, but was somewhat startled when in my hometown, in rural Holland, an MC appeared a year ago. I had heard of the Hell's Angels, naturally, but I did not realize Motorcycle Clubs where a widespread phenomenon to the point where I would find them in my direct environment. I was fascinated, and became convinced the Demons operated in the same way the Sons did. They wore 'cuts', drove around upon the same loud motorbikes; they set up shop in a bar and caused police trouble. They had Prospects and Men of Mayhem; they held to the same hierarchical structure as the Sons. And most importantly, they were all bearded. As I am not a member nor very close to this MC in any other way, it is a fact that I do not know how the Demons operate. However, I took the knowledge I had gained watching a fictional motorcycle club in a television series, and I projected it upon an actual one. That seems to be quite an irrational thing to do, however, it may prove to be something of a common tendency. The idea that stories and their media may be so persuasive as to affect ideas or actual behavior has received considerable academic attention. There is, for instance, a broad spectrum of effect-research which advocates that vivid media may be able to affect behavior (mainly negatively) such as research revolving around videogames and their tendency to evoke aggressive behavior (Dietz, 1998; Dill & Thill, 2007; McDonald & Kim, 2001; Miller & Summers, 2007). More directly, there is a body of research interested in the way a story in any medium may be persuasive (Brock, Strange & Green (eds.), 2002; Bussele & Bilandzic, 2009; Duchan, Bruder & Hewitt (eds.), 1995) and may consequently affect certain 'real-world beliefs' (Green & Brock, 2000; Busselle & Bilandzic, 2005). But still, there are some questions that remain unanswered: do all media evoke this behavior? What are the preconditions? What role does realism play in all this? The current investigation attempts to begin to clarify such issues. First, we will elaborate the research that has inspired us and clarify some terms we will use frequently. Real-world belief structures Real-world beliefs are suppositions a person has about real life and the manner in which the world 'works'. For instance, it is my belief that one cannot just leave their bike unattended and unlocked in the middle of the city, for it will inevitably get stolen. Whether I have or haven't had my bike stolen in the city before does not have to matter (for instance, I haven't), because I 'just know' this is how the world works. It is our understanding these beliefs may be affected by a compelling narrative. Many a time has it been attempted to show a link between a compelling narrative and a changing of real-world beliefs to be story-consistent (Green, 2008; Green, 2004; Green & Brock, 2002; Shapiro et.al. 2010; Strange en Leung, 1999; Wheeler, Green & Brock, 1997) Despite compelling theories that knowledge gained from a narrative would be stored separately from real-world knowledge in the brain (Potts, St. John & Kirson, 1989) these investigations succeeded in showing a link between narrative fiction and real-world belief change. It is supposed this link will be found again in this investigation. We then propose that there must be some element in the narrative that allows for the spectator to draw a link between the story and the actual world (Green, 2008). We would like to propose that what is required for belief change to occur is an element of realism. Defining such an element, however, is tricky business. What is realism? Realism In 2007, Pauliot and Owen wrote compellingly about the shortcomings of previous work conducted on the topic of realism and its relation to attitude or behavioral change. They concluded that whereas some research was able to conclude that perceived realism had some effect, other studies found no effect. In part, this seemed to be due to methodological errors, mainly related to the lack of a manipulation check and the subsequent alternative explanations for findings. This puts previous evidence given to suggest that there is a

4 Anne-Wil Coumou 4 relation between perceived realism and real-world belief change in a negative light. However, there does exist valid evidence to support the idea that realism plays an important part in the narrative experience. In 2004, Wood and colleagues conducted an open-ended survey among self-proclaimed videogamers, finding that 'realism in terms of sounds and vision' played a pivotal role in the gamers' appreciation of the videogame. The mental models perspective also follows that when attention is shifted towards inexplicable inconsistencies in the story, or 'unrealism', the pleasant absorbing experience will be abruptly broken. (Busselle and Bilandzic, 2008: 256.) This study, then, was designed in part to investigate the role realism plays in the narrative experience and the subsequent alteration of real-world beliefs. An exploratory study that seeked to find out how participants deduced realism is a study conducted by Alice Hall in Hall's theory is a thorough investigation of what realism means to viewers, conducting group-interviews with 47 participants asking them to elaborate on televised realism. Hall discerned six dimensions of realism: plausibility (events or behaviors that have potential to occur in the real world), typicality (the type or range of people whom the media portrayal resembled), factuality (for instance, Schindler's List was mentioned as a factual historic occurrence captured), involvement (the extent to which one can relate to the media characters, whether it 'feels' real), narrative consistency (the internal coherence) and finally, perceptional persuasiveness (impressive, convincing visuals). The term realism may confuse and become automatically linked with reality or the actual. It is important to note here that from these accounts follows that realism does not necessarily have something to do with reality, but more so with probability, credibility or believability. Naturally, not all of the six dimensions will apply to every mediaproduct. A fantasy-film will not have high plausibility or typicality and no factuality, but may have high involvement, narrative consistency and perceptional persuasiveness to compensate. It seems that very many mediaproducts may be explained using this comprehensive model, even such shows that leave us wondering sometimes about why 'we' seem to find them credible enough, like overproduced soaps (high involvement, high narrative consistency) or extreme realitytelevision shows (high plausibility, factuality and perhaps involvement). This account of realism-judgments is comprehensive and seems methodologically sound. That is not to discard other, meaningful, efforts that have been made to define the nature of perceived realism or the realism-judgment. Hall notes the similarities between her work and previous, and we may add to that here. For instance, the notion of narrative consistency is similar to Shapiro et.al.'s conceptualization of relative realism (2002) and Busselle and Bilandzic's notion of narrative realism (2005) Plausibility in turn is similar to Busselle and Bilandzic's concept of external realism (2005). We could list very many other meaningful contributions here and note similarities (Potter, 1988; Busselle & Greenberg, 2000; Green & Brock, 2002; Shapiro et.al. 2010; Green et.al. 2006) but it seems sufficient to remark that many contributions have been made and are recognized. However, none seem to reach the level of completion Hall's framework does. Hall's account is logically sound, seems capable of capturing a broad spectrum of mediaproducts and, most importantly, follows directly from the mouths of participants. Mental Models perspective It has been proposed that viewers construct mental models of a narrative when consuming it, constructing these models actively as the story progresses (Busselle and Bilandzic, 2009). These models represent settings, characters and situations and are created by combining information from the text with knowledge the reader or viewer already possesses about life in general as well as about the specific topics and genre of the narrative. Preexisting knowledge originated in real world experience (e.g. schemas and stereotypes) and from other fictional and nonfictional mediated experiences. (p.332). What this perspective may add to Hall's account is an explanation of the process in which the viewer is working with and working out the realism of the story world. The viewer is taking cues from the story, mixing them with his knowledge of the world but also of genre-conventions, and creating a universe in which the story is set. The only reason the viewer would be reminded of the level of realism of the story is when something goes wrong. As

5 How perceived realism in video games and television series may affect real-world beliefs 5 Busselle and Bilandzic point out, it should only be when there is a matter of an inconsistency, or when the narrative becomes either too complex or too predictable (boring) for the viewer that awareness of the constructed nature of the story should begin to dawn on the viewer. The viewer may then lose engagement with the story, because it has become difficult to smoothly make sense of the narrative. It has been noted that little research focuses on on-line realism judgments (Busselle & Bilandzic, 2005), being judgments regarding realism made while consuming the mediaproduct instead of thereafter (offline judgments) and this study will not include these on-line judgments again. It seems that when the absorbing experience is successful, and mental models are being constructed smoothly, no mention of realism should come to the viewer's mind while watching, reading or playing. Absorption and Transportation We define absorption as a sensation of being cut off from the world, forgetting time and the self temporarily. Transportation is a way in which the state of absorption may manifest itself. Indebted to Melanie C. Green for this understanding, we define transportation as an altered state of awareness in which individuals may, due to a compelling story, feel that they are completely immersed in the world of the narrative. (Green, 2004). This world of the narrative is that universe which we actively create using mental models. This idea of a 'shift' in perceptual presence is in line with what is called Deictic Shift Theory. According to this theory, audience members switch to the time and location of the narrative, and to the subjective world of the characters. This is necessary because some information makes sense only from the 'deictic center' of the story, in other words: from where, when and how the story is set (Busselle & Bilandzic, 2009). Flow A concept closely related to absorption is the concept of flow as discerned by Mihaly Csikszentmihaly. He defines flow as state of intense concentration, a focus on the task at hand which takes up all attention and none is left to think of anything else. Selfconsciousness disappears and the sense of time becomes distorted. This experience should feel good, rewarding, and people should want to return to it. (Csikszentmihalyi, 1990) Indeed, there are several similarities between the narrative absorption experience and the concept of flow. As Sherry (2004) points out, flow occurs when there is a balance between an individual's skill and the difficulty of the task at hand. If the balance does not exist, flow will not occur. (p.336). As previously mentioned, it is of pivotal importance when constructing a mental model of the story, that the spectator isn't challenged too much or too little by the narrative. Csikszentmihalyi's focus however, seems to be more on the physical state of being, on ways of achieving flow which can be quite diverse. Our focus here is upon using a narrative in particular to become absorbed and transported. Narrative engagement could be one way to achieve a sense of flow, but the concept is broader as described by Csikszentmihaly. An absorbing experience with a narrative is inherently different from flow in the sense that when engaged in an athletic activity, for instance, one is intensely focused and may lose awareness but does not feel transported to any story-world. Indeed, as Bussele and Bilandzic point out, the transportation aspect of the narrative experience, the 'deictic shift' to another time and place, is unique to the narrative flow experience (Buselle & Bilandzic, 2009). Realism and absorption As mentioned, we hypothesize that realism and absorption have a reciprocal relationship. Mental models are constructed along the paths the narrative takes, and these may differ on the dimensions of realism incorporated. The story must conform to some level of realism or consistency, otherwise the viewer will become consistently aware of the fabricated nature of the narrative, and feel unable to get into the flow of the story, to become absorbed by it. Here we find the interaction between absorption and realism. Only when participants becomes absorbed, can they experience a sense of realism and only when they

6 Anne-Wil Coumou 6 experience a sense of realism, can they become absorbed. First of all, this hypothesis makes logical sense. When viewers are not absorbed, they will experience the narrative differently than when they are absorbed. Their attention is not fully devoted, they may miss segments because they become distracted, and so on. This lack of absorption and transportation prevents the viewer from making sense of the narrative from its deictic center. The viewers remains on the outside looking in, remaining aware of the fact that they are watching a television show or playing a videogame. Naturally then, in such a state the viewer will not feel that the story has high realism in any dimension. The reason the absorbed state does not come about can be any number of things. It could be an external factor like stress or fatigue. Or it could be an internal factor, like a lack of realism. When viewers engage with a narrative, they will try to construct mental models in a smooth manner, progressing as the story moves along. When there is a lack of realism, this process will no longer run smoothly. For instance, when the perceptional persuasiveness of a film is low, it may bring the viewer back to reality because the viewer can see how the film was made. Second, research has found a positive correlation between transportation and perceived realism. (Green, 2004.) and there have been indications that perceived unrealism will limit transportation (Busselle & Bilandzic, 2008). Furthermore, there have been findings to suggest a positive correlation between absorption and story-consistent beliefs (Busselle & Bilandzic, 2009). Identification We hypothesize that there is a correlation between identification and absorption as well. Defining identification, we follow Zillmann's account (Zillmann, 1994) and hold that identification with characters works in roughly the same way real-life interactions with loved ones do. We find the idea of identification as empathy, feeling exactly what the character is feeling at the time of the feeling not to be very credible (See: A. Coumou, in press, for an elaboration). We propose that identification as loving sympathy for a character is a strong mechanism of narrative engagement, but it will only come about when the viewer is absorbed. Again, when viewers remain on the outside looking in, they will not experience the events in the story from its deictic center, and they will not connect with the story or the characters in a profound way. Considering that identification may only occur when the viewer is in the absorbed state, we thus hypothesize that identification may be an indicator of absorption. Mediumspecificity We have elaborated upon the absorbing experience, and what chain of events may be set in motion by this experience. It is our understanding that although this process may occur for any medium, there are mediumspecific elements to each experience. We use Donald Norman's interpretation of the term affordances to highlight this (Norman, 1988). According to Norman an affordance is something a medium allows the viewer or user to do. An affordance of a tablet is that it allows the user to touch the screen, move around on the screen; the device will respond to this. This means some media have little affordances, like traditional television (although you may zap), whereas videogames have quite some affordances (the gamer is constantly using his 'remote', if you will). We conclude from our previous work (A. Coumou, in press) that these affordances determine not whether a viewer or gamer has an absorbing experience, but they determine what kind of absorbing experience. In the current investigation, we have compared the absorbing experiences television series and video games instigate. We have chosen these two specific media because of their individual differences. Previous research (A. Coumou, in press) has shown that television watching is often perceived as a very passive and relaxing activity which does not require much, if anything at all, from its user. Playing a videogame however, requires much more of a lean-forward and active attitude. We wonder in what way these two diverse media may bring about the same absorbing, real-world belief affecting experience.

7 How perceived realism in video games and television series may affect real-world beliefs 7 Summing up our theory, we hypothesize (1) that only when viewers are experiencing absorption, can they perceive realism in a fictional narrative. Being absorbed, the viewer may experience transportation, a deictic shift, and feel present inside the story world. This world the viewer will actively construct using mental models, made up of real-world knowledge, story knowledge and knowledge of genreconventions. If this constructing goes well, the viewer may grow attached to the story world and the characters therein, this process we call identification. Hence we hypothesize (2) that only when viewers are absorbed, can they experience the process of identification, and that identification can be an indicator of absorption. The viewer will make no on-line judgment about realism if the narrative unraveles in a challenging yet comprehendible manner, leaving no more than a few loose ties. The viewers may then feel they have had a pleasurable, consuming or meaningful experience. The knowledge and experience from the story-world may come to feel real and the viewer may tie it up with the real-world knowledge, thus leading us to our main hypothesis (3) that real-world beliefs may only be affected when the narrative is perceived as realistic. Broken down into a simple model, our theory looks as follows. Fig. 1: Model Mediumspecifc factors Absorption/flow Identification Mental Models Transportation/deictic shift Perceived realism RW-beliefs

8 Anne-Wil Coumou 8 Method Participants Participants were 72 students between the ages of 14 and 17 (M = 15.10, SD =.561), in groups of Two questionnaires were not filled out with care and removed from our sample, leaving N = 70. Of the participants, 33 were male and 39 were female. Our sample was taken from a school in rural Holland, in a southern province called Zeeland. In Holland, we distinguish three educational phases. There is primary school from the age of four until twelve, secondary school until sixteen or eighteen depending on the learning capacity of the student (a higher education means a longer education) and then there is tertiary education like craftsman school or university, again depending on the skill of the student. Our 72 students were taken from secondary school, where they were enrolled in 'levels' one and two (basis and kader) of a total of six levels, six (VWO) being for the most talented students. They participated from the kindness of their hearts. Their education being quite practically oriented, the students were part of four classes each oriented towards a career type. Group one (N= 19, 18 females and 1 male) was oriented towards healthcare and appearance. Group two (N= 11, 5 females and 6 males) was oriented towards metal- and vehicles engineering. Group three (N = 18, all males) was oriented towards trades and administration, and finally group four (N= 21, 17 female and 4 male) was a mix of orientations towards both healthcare and appearance, and theater. Materials Our participants filled out a questionnaire containing 30 7-point Likert scales, divided into two sections. Since we wanted to compare mediumspecific aspects of the absorbing experience and so forth, we proposed virtually the same fifteen scales once for each medium. The first section of scales were devoted to television series and the second section to video games. These scales will be elaborated upon in more detail later. We asked our participants to remember a series or a videogame they thoroughly enjoyed. This investigation has chosen not to expose the subjects to specific mediafragments for several reasons. First of all, this way we run no risk of showing a series that does not appeal to the viewer. In turn, we run no risk of boring the participant, since we were warned about the students' limited attention span. Second, we run no risk of showing a series to someone that naturally doesn't like series. Because the questionnaire was divided into two sections concerning two different media, those students that did not watch series much or did not enjoy videogames did not fill our that part of the questionnaire. Only those students who enjoy one, the other or both media thus told us about their experience with that medium. Those who have a natural predisposition not to like (and with that perhaps not be so susceptible to absorption, for instance) a series or videogame, were this way excluded. Third, we also ran no risk of showing a series that whereas it may be liked by the participant, it may not have a profound effect on the viewer because of the classroom setting. Usually the classroom is used for other purposes of course, so it may be hard to understand the purpose of this undertaking and attain an appropriate attitude. Asking for personal experiences will assure the most profound experience to be reported on (as it is likely to immediately come to mind when asked about profound experiences related to mediacontent), fitting the individual. Apparatus As said, our questionnaire consisted of thirty scales divided over two media. There were four categories of scales: identification scales, absorption scales, believability scales and real-world belief scales. They were randomized for each medium to prevent order-effects. The first two scales were taken from previous research (M.M. Kuijpers, in press). However, we did make some alterations based on our own preliminary research, which revealed differences in the identification process between series and games. We discovered that whereas we identify identification as a process similar to feeling for loved ones in good or bad places, identification may be of a different nature

9 How perceived realism in video games and television series may affect real-world beliefs 9 for videogames. Because some games allow the gamer to customize the icon or avatar to the extent that gamers may indeed feel they are the character, we believe identification as empathy is a possibility for videogames. Therefore, we included two extra identification items for videogames, namely the propositions (1) In the game, I am able to customize the character to my individual wishes: hairand skin color, sex, and so on. and(2) When I am playing the videogame, I often feel I am the character myself. These items were incorporated alongside the other three identification items that were used for both series and videogames. This lead to series having three identification items and games having a total of five. The division of the items then came down to fourteen for series and sixteen for video games. After filling out each part of the survey, the students were asked to fill out the name of the series or the game they had kept in mind while filling out the survey. The believability scales and the real-world belief scales were designed for the purpose of this investigation. We based our design of these scales on a translation of our theoretical excepts into comprehendible, clear propositions that we hoped the students would instantly understand. We designed three believability items and three real-world belief items. We will elaborate on these scales here, because they seem to be quite similar and this may require some theoretical elaboration. The following table shows each proposition clearly. Fig. 2: Scales Believability I find the story of this series/videogame to be credible. There are many contradictions in this story. When I find contradictions in this story, I don't really mind. Real-world beliefs I feel this series/videogame expands my understanding of the world. (RW2) I think situations in the story happen in reality as well. (RW1) I feel that by watching the series/playing the game I am learning more about the world. (RW2) As mentioned in the paragraph concerning 'realism', we do not necessarily link realism with reality. This is why the scale pertaining to realism is called 'believability.' It is our understanding from a theoretical point of view that what causes a positive realism-judgment to occur is the idea that the story can happen, may happen, could happen: that the story is credible, believable. This has been elaborated upon thoroughly in our theoretical section. Hence our propositions pertaining to realism do not ask for reality as much as for credibility of the story. But then why does the second real-world beliefs-proposition ask for reality? Because the vital difference between believability and realworld beliefs is that real-world beliefs are about what people think happens in the world; what can actually happen within the laws of nature, social relations, economy, and so on. Whereas realism as it is understood here is related to what could happen within the story, within the laws of the story and whether the story is consistent in this way. What we aim to find out, then, is if what happens in the story has an effect on what people think can happen in the real world. Doing some math, the reader may have noticed that two items have not been accounted for as if yet. It is our understanding that participants will have the most profound absorbing experiences if they enjoy the narrative alone. We suspect that narrative engagement will be limited in the social context, because recipients may be distracted by how they appear towards the other social actors. We then presume that the optimal absorbing experience will occur when one is alone, free of any judgments or social rules. To test this understanding, we asked each participant to fill out on a seven-point scale their opinion of the proposition: I usually watch this series alone and I usually play this game alone.

10 Anne-Wil Coumou 10 Procedure The students filled our questionnaire in their familiar classrooms. The attention span of these students is not very great, hence we wanted to not excite them too much with the presence of an unfamiliar researcher in their classroom. We kept the setting as normal as possible, the students filling out the questionnaire handed out by a familiar teacher, right after they had made a test for school. Before students started to fill out the scales, they were asked for their age and sex. We chose to ask for this information at the beginning of the survey to calm the minds of the participants somewhat. These students are not familiar with research projects, hence we felt it would be best to start off with a comprehendible, non-intimidating question. There was a risk that the students would not have enough concentration left after the test to fill out the questionnaire with care, but it turned out that the quiet and focused atmosphere encouraged the students to dedicate some more focus to the questionnaire. This was enforced by the fact that every student had to wait in silence until the end of the class before anyone could leave.

11 How perceived realism in video games and television series may affect real-world beliefs 11 Results Before we will have a look at how we went about finding clues to our hypotheses, we must look at our apparatus closely and investigate its statistical validity. For this reason we will first investigate the validity of our scales and the normality and homogeneity of our data. After this, we will look at our hypotheses and from these explain the path along certain tests we have taken. Naturally, at the very beginning of the elaboration of our results, we start with our most basic descriptive statistics. Because of the synoptic qualities of a table we feel it can be more clear than a textual elaboration of our data, hence we have chosen this method to elaborate our basic descriptive statistics. Descriptive Statistics Fig. 3: Descriptive statistics for television series N Valid Missing Absorption Identification Believability Real-world beliefs Real-world believability Mean Median Mode Std. Dev Min Max Fig. 4: Descriptive statistics for video games N Valid Missing Absorption Identification Believability Real-world beliefs Real-world believability Mean Median Mode Std. Dev Min Max Scale validation Without means to test our scale before conducting the experiment, we could only be sure of the validity of our scales after the experiment. Our absorption scale for television series, comprised of four items, was consistent (Cronbach's α =.730). Deleting one item would make the scale more consistent (α =.745), but we found this difference not noteworthy enough to do so. We expected the absorption scale to be reliable, since this scale had already been validated in previous research (M.M. Kuijpers, in press).

12 Anne-Wil Coumou 12 For video games however, the absorption scale appeared less consistent (Cronbach's α =.658) which would increase (α =.694) if one item was deleted. This was the same item that would increase the alpha for television series. We decided to remove this item from the video game absorption scale: 'The story is so captivating that I can isolate myself from what is going on around me.' Our identification scale had also been validated in the past already (M.M. Kuijpers, in press), and was validated here again (Cronbach's α =.789) for television series. As already described, we added two items to our identification scale for video games because we proposed identification may be more complex for videogames. We included two items related to identification as empathy. The totality of this scale proved relatively consistent (Cronbach's α =.674.) This could be increased by removing one item (α =.770) and we found this increase worth removing the item from our scale: In the game, I am able to customize the character to my individual wishes: hair- and skin color, sex, and so on. Our real-world belief scale and believability scale were designed for the purpose of this study. The real-world belief scale for television series could almost pass for consistency (Cronbach's α =.645). This number would not increase if an item were removed. There are however, significant (p < 0.05) correlations between the three items on this scale, as shown in figure one. The questions accompanying the codes (RW1, RW2 and RW3) are explicated in figure 5. Moreover, when validating this scale the item statistics showed a total correlation well above the required level (r > 0.3). Fig. 5: Correlations RW3 RW1 RW2 RW3 Pearson Correlation P N RW1 Pearson Correlation P N RW2 Pearson Correlation ** 1 P N For videogames, the real-world belief scale approached consistency (Cronbach's α =.645) which would increase if one item (RW1B, see RW1 but for video games) were removed (α =.702) Our believability scale for television series was less successful (Cronbach's α =.440) and the same went for this scale for videogames (Cronbach's α =.535) The believability scales were thus somewhat disappointing. It is however possible that the distinction between real-world beliefs and believability we have explicated earlier, did not translate well into the questions we proposed to our participants. We will explore this in more detail in the discussion. We do not, however, want to discard the answers given by our

13 How perceived realism in video games and television series may affect real-world beliefs 13 participants so we propose to combine the items from both scales. Putting the two scales together as one scale does make a consistent scale (Cronbach's α =.669) for television series and a moderately consistent scale for videogames (Cronbach's α =.664). We will be discussing here the results from both the individual scales and the new, combined scale. In the discussion section we will go into how these results must be weighed. Test of normality Using the Shapiro-Wilk test from the explore-option in SPSS, we tested for normal distribution among the four different groups the questionnaires were taken in. We used the Shapiro-Wilk test instead of the Kolmogorov-Smirnoff test because of our limited sample size. The Shapiro-Wilk is recommended for group sizes of under fifty participants, and our groups certainly were. We found non-normal distribution for identification in group one for television (p =.014). For videogames the distribution approached non-normal (p =.065). For real-world beliefs for videogames, group one approached non-normal distribution as well (p =.051). We conclude that identification for both media and real-world beliefs for videogames are non-normally distributed within group one (N=19). Over the entire sample however (N = 70), identification and real-world beliefs for videogames were normally distributed (p >0.05). For this reason, we will take no specific measures when it comes to these scales. Identification for television series proved to be non-normally distributed for both group one and the entire sample. (p = 0.001). Looking then at gender differences, it seems identification is normally distributed among males from the entire sample (p =.373), but not among females (p =.021). For this reason, we used gender as a covariate in the following tests to make sure the non-normal distribution of data did not affect our results. Group differences Because the surveys were taken in four different classrooms, among students going in different career-directions, it is important to make sure there are no significant differences between our groups. If there are significant differences between our groups, this would mean we cannot submit all of our participants together to certain tests, but that certain groups (those that are not homogen) should be tested separately. We tested the homogeneity of the different groups and their responses to our scales and there were no statistically significant differences to be found (Levene's p >.05), except when it came to identification for television series (F (1, 59) = 8.461, p =.005, p2 =.125). The post-hoc Games-Howell test revealed, however, that there were no significant differences in answers between the groups for identification (p > 0.05). All by myself We previously hypothesized that absorption would be limited if the participant is not watching or playing by himself. Socialization may get in the way of the absorbing narrative experience. To test this hypothesis, we included two items in our questionairre asking the participants whether they mostly watched or played alone or not. From these results, an independent variable t-test revealed a statistically significant higher absorption rate when watching alone (M = 5.24, SD = 1.03), than when watching not-alone (M = 4.36, SD = 1.05), t (57) = 3.07, df = 55, p < For videogames, the same test revealed that playing alone matters less when becoming absorbed, showing no significantly higher absorption rate when watching alone (M = 4.47, SD = 1.30) as opposed to watching not-alone (M = 4.67, SD = 1.40), t (29) =.003, df = 27, p =.998.

14 Anne-Wil Coumou 14 Our hypotheses and results We found only positive correlations between our scales. We will highlight several of these that were statistically significant in light of our hypotheses. We will start with our hypothesis that identification can be an indicator of absorption. From this it would follow that there is a positive correlation between absorption and identification. We indeed found a considerable positive Pearson correlation between absorption and identification (r (28) =.423, p < 0.05) for video games and again for television series (r (61) =.516, p <0.05). A multiple regression analysis then showed that absorption is a fine predictor of identification (26,6% of the variance in identification can be accounted for by absorption, β =.516) for television series. For video games, absorption proved a moderate predictor of identification (17,9% of variance in identification can be accounted for by absorption, β =.423). The second hypotheses that will be discussed is the idea that only when absorption is experienced, realism can be perceived. From this hypothesis it would follow that there is a positive correlation between absorption and believability. For video games we found a considerable Pearson correlation between absorption and believability (r (36) =.400, p < 0.05) and an almost identical one for television series (r (62) =.401, p < 0.05). Using a multiple regression analysis we tested absorption as a predictor of believability in television series, but found only a moderate connection (16,6% of the variance in believability can be accounted for by absorption, β =.401). We also explored identification as a predictor of believability and found an almost identical connection (15,9% of the variance in believability can be accounted for by identification, β =.399 ). The new scale, combining believability and real-world beliefs revealed a considerable correlation to absorption(r (61) =.499, p < 0.05). The regression analysis showed that absorption and identification are fine predictors of this new scale (33,3% of the variability in the real-world believability scale can be accounted for by absorption [β =.328] and identification [β =.336]). Unconventional as this approach may be, we checked to see if this number could be improved upon using a nonlinear (or curvilinear) regression for absorption as a predictor of real-world believability. We tested a quadratic effect using a hierarchical multiple regression approach, which measures to see if the results would change significantly if a bend in the line (hence: non-linear) would be allowed, but this revealed no significant change (R² Change =.005, p =.540). We also tested absorption as a predictor of believability for video games and again found a moderate connection (16% of the variance in believability can be accounted for by absorption, β =.400). We explored identification as a predictor of believability and found barely any relation (6,2% of variance in believability can be accounted for by identification, β =.248). The new, 'real-world believability'-scale revealed a considerable correlation between absorption and 'real-world believability' (r (35) =.476, p < 0.05). The regression analysis in turn showed that identification and absorption were fine predictors of real-world believability (24,9% of variance in the real-world believability scale can be accounted for by the total of absorption [β =.437] and identification [β =.121]). And again we applied the unconventional method of seeking out a non-linear regression, using a nonlinear (or curvilinear) regression

15 How perceived realism in video games and television series may affect real-world beliefs 15 for absorption as a predictor of real-world believability. We tested a quadratic effect using a hierarchical multiple regression approach, again revealing no significant change (R² Change =.000, p =.978). Our third and final hypothesis is that real-world beliefs may only be affected when the involving narrative is perceived as realistic. This hypotheses proposes positive correlations between believability (and in turn absorption and identification, as elaborated before) and real-world beliefs. We found a moderate positive correlation between believability and real-world beliefs (r (61)=.367, p <0.05). For video games however, we found only a non-significant correlation between believability and real-world beliefs (p> 0.05). A step-wise regression analysis for television series including absorption, identification and believability as predictors of real-world beliefs showed a moderate connection (26,8 % of the variance in real-world beliefs can be accounted for by the total of absorption [β =.311] and identification [β =.285]). Believability however, was excluded from this analysis because it did not reveal a significance in relation to real-world beliefs whereas absorption and identification did (p < 0.05). For videogames, we conducted a regression analysis with absorption, identification and believability as predictors of real-world beliefs and this showed a very moderate connection (21,9% of variability in real-world beliefs can be accounted for by absorption [β =.189], identification [β =.237] and believability [β =.118]).

16 Anne-Wil Coumou 16 Discussion Before we may interpret our results, we must discuss the shortcomings of our apparatus. Two of the scales we used here were validated in previous research and proved sound in this investigation again. We removed one item from our absorption scale (A1B: 'The story is so captivating that I can isolate myself from what is going on around me.'), because this would improve the consistency of the scale to a level we were more comfortable with. However, this removal did not bring any theoretical inconsistency to light and because of this will be discussed in no more depth here. What we will put some focus on, is the removal of an item from our identification scale (I2). Our identification scale is a mix of the scale devised by M. Kuijpers (in press), and two items we added to this scale for videogames because we hypothesized that identification as empathy is possible for videogames. Because our identification scale asked for identification as sympathy like the kind one could feel for a friend, we felt the need to add two items to this scale so it could include both definitions of identification. The two items we added were the following: (I1) When I am playing the videogame, I often feel I am the character myself. and (I2) In the game, I am able to customize the character to my individual wishes: hair- and skin color, sex, and so on. We finally removed item I2, because it proved inconsistent with the rest of the items on the scale. It has occurred to us that perhaps these two items, for they ask for another type of identification than the other three items on this scale do, should form a separate scale. Conducting such an analysis reveals a negative Cronbach's alpha (α = -.100). This indicates that there could be a negative correlation between these two scales. We found a very small negative correlation at a nonsignificant level (r (41) = -.049, p =.762). Interpreting this may lead to speculation, so let us conclude that this is not the direction we should be looking in to try and find some conclusive results. Inspecting the descriptive statistics for these scales, we find quite a grave difference between the mean and the mode for I2 (M = 4.34, MO = 7). The same goes for I1 (M = 3.46, MO = 1). Looking at the mode for a minute, it now seems that there was a high level of adjustability of character, we see the mode for I2 being at the top of the scale at seven. However, we also find that there was a very low level of feeling to be the character, since the mode for I1 is all the way at the bottom of the scale at one. In this way, we can after all deduce from these items that our hypothesis that identification as empathy may be possible for videogames is not confirmed here. Games that have characters with high adjustability do not necessarily make players feel more like they themselves are the character. Unlike our absorption and identification scale which we took from previous research, the believability and real-world belief scales were devised for the current investigation. Only after conducting this study were we able to validate these scales, which proved complicated. The consistency of these scales was somewhat low, and we propose that this is due to the lack of understanding of these scales from our participants. This is something we did nor could anticipate. Perhaps our ability to adequately deal with the level of education of our participants was not great enough. The level of education was quite low, and it is therefore possible that we were not prepared to formulate questions that would fit our participants' comprehensive abilities. Our believability scale asks for contradictions in the narrative. It may be that this question came across to our participants as a question pertaining to understandability, they may have interpreted it as some control-question, instead of a question pertaining to how noticeable narrative inconsistencies were in the story. If it was misunderstood in this way, then it may have been expected that participants would score quite high on these items. After all, it is a common tendency to think we are smarter, more literate or more capable of understanding than we may actually be. Looking at the modes for these items, we see that for television series the modes are indeed quite high (for B2 = 6, B3 =7) but for videogames, these modes are terribly low (for both questions, 1). It thus proves difficult to adequately interpret what may or may not have gone wrong processing these scales by our participants. We therefore decided to hold onto these scales individually and also to mold the believability and the real-world belief scales into one new scale: real-world believability. This is because we propose the distinction between the two scales may not have

17 How perceived realism in video games and television series may affect real-world beliefs 17 been as evident to our participants as it has been to us. The main difference between the two scales is that believability pertains to the consistency or credibility of the story and the real-world belief scale asks for the applicability of the story-world knowledge to the actual world. The believability scale thus questions the realism of the story and the real-world belief scale questions the connection to reality of the story. We suspect our believability and real-world belief scale may not have been note worthily different to our participants. Holding onto the individual scales and also creating the new, combined scale allows us to compare results and draw adequate conclusions. Summing up our theory briefly once more before interpreting our results in light of it, we previously hypothesized that (1) only when absorption is experienced, realism can be perceived. We also hypothesized that (2) identification could be an indicator of absorption because identification would only find the space to come about during a profound and absorbing narrative experience. And finally we hypothesized that (3) our real-world beliefs may only be affected when the involving narrative is perceived as realistic. This discussion will discuss each of these hypotheses separately before moving on to making more general claims about the current investigation. Hypothesis one: Only when absorption is experienced, can realism be perceived The first question we will need to ask in order to come closer to an answer to this hypothesis is: was absorption experienced? Yes, more convincingly for television series than for video games, but for both nonetheless as the descriptive statistics clearly show, absorption is part of the narrative experience for our participants. The second and equally obvious question we will need to ask is: was realism experienced? This is a more difficult question to answer, because of the position our believability scale is in. Looking at the believability scales descriptives, we can see that for television series the scores were quite high and for videogames they were only slightly less so. However, we have already discussed the difficulties interpreting this scale correctly presents, so we compare these results to our combined real-world believability scale. These scales score slightly lower than the believability scales. Should we conclude that little realism was experienced? There is one thing that speaks to the contrary of this, namely one item from the believability scale (B1/B1B) which goes: 'I find the story of this series/this video game credible.' The other two items on the believability scale have been elaborated before as potentially unclear, but this item seems not to leave much for the imagination to play with: is it credible, or not so much? One could argue that the nature of 'credibility' could be unclear, and we have certainly elaborated before that realism or credibility can have many dimensions, but it seems probable here that the participant took to his own interpretation of the word 'credible' and scored his favorite series or videogame accordingly. Whatever this interpretation of realism or credibility may be is interesting but not of pivotal importance, because whatever realism means to her or him, the media product either is or is not realistic in that way. And in this way, we do measure credibility. Looking at the descriptives for this item, then, we see that for television series, believability scores quite high (M = 5.47, SD = 1.29, Mode = 6). Video games however, score note worthily lower on believability and results are less in alignment as well, judging from the differences between mean and mode and looking at the standard deviation which is considerably higher than for television series (M = 3.7, SD = 1.99, Mode = 1). It is here we find clear differences between the two media, and it is from these results that I would propose to conclude that realism was experienced for television series, but not so much for video games at all. We in turn see that absorption was experienced for television series, but less so for video games. Finally, the third question we will need to ask is: is there a correlation between these two entities? Certainly, as discussed in the results-section, we found a decent positive significant Pearson correlation between absorption and