MEDIA PSYCHOLOGY, 9, 473492 Copyright 2007, Lawrence Erlbaum Associates, Inc.

Dont Tease Me: Effects of Ending Type on Horror Film Enjoyment

Cynthia M. King and Nora Hourani
Center for Entertainment and Tourism Studies, California State Fullerton

This experimental study explores the influence of horror film ending type on audience enjoyment. Four horror films were manipulated to create versions with traditional endings (endings in which the evil/antagonist is destroyed) or teaser endings (endings in which the evil/antagonist revives). Analysis explored differences in the preferences for ending type based on watcher motivations. Results suggest that viewers, particularly viewers highly motivated by gore or thrill factors, rated traditional endings more favorably than teaser endings, although seemingly for different reasons.

INTRODUCTION
Wearing plain black clothes and a simplistic white mask, the unstoppable shape is the first and most chilling film mass murderer. He has no remorse, no weakness, and no motive. The film is masterfully made, but like all horror films, the devastating finale is followed by a camera zoom-in on the masked killer as he stalks the streets, ready to slaughter again. (Nashawaty, 1998).

Tales of horror have entertained adolescents and adults alike since the arrival of the Gothic novels in the eighteenth century. Antagonists such as Dracula and Frankenstein and later Michael Myers and Jason have horrified generations of fans. Originally, horror novels and films followed a traditional formula where good ultimately triumphs over evil. Following this formula, at the end of the story, the evil antagonist is unequivocally destroyed. Various film critics, however, argue that
Correspondence should be sent to Cynthia M. King, Center for Entertainment and Tourism Studies, California State University, Fullerton, P.O. Box 6850, Fullerton, CA 92834-6850. E-mail: cking@ fullerton.edu

474

KING AND HOURANI

many horror films no longer follow this tradition. Starting in the early 1970s with the success of Rosemarys Baby and The Exorcist, a new horror formula emerged featuring defenseless victims in terrifying no-win situations where the evil/antagonist triumphed over the forces of reason and rationalism (Rosenbaum, 1979). According to Gleiberman (1997), resurrection of the evil/antagonist at the end of the film paves the way for the sequels that follow.
In the 1980s teen horror movies were as unavoidable as garden weeds. One after another, they popped up in the multiplexes, perpetually recycling the masked killers, softcore-sexpot victims, and Who will be the next to die? formulas of Halloween, Friday the 13th and A Nightmare on Elm Street, just to name a few (p. 44).

Typically, these horror films contain teaser endings in which, although at first it may appear that the evil has been destroyed, a final scene reveals that the evil has survived, usually foreshadowing the anticipated sequel. Many popular horror film series such as Halloween, Friday the 13th, and A Nightmare on Elm Street contain teaser endings. In the past, these films have ranked high on movie lists; for example, I Know What You Did Last Summer (1997), grossed $9.7 million in its opening weekend (Gleiberman, 1997). However, more recently, many teaser films have not fared as well, such as Jason X (2001), the tenth film in the Halloween series which grossed only $6.5 million in its opening weekend followed by a steep 65% drop in sales to only $2.4 million on its second weekend. It is unclear, however, to what extent teasers may contribute to the relative success or failure of these films. Thus, this study seeks to explore what effect, if any, differences in ending type have on audience perceptions of the films. The Appeal of Horror Films Researchers have generated many theories and rationales to explain the appeal of horror. Several theoretical approaches suggest that horror may be inherently attractive or enjoyable for many viewers. These theories variously suggest that acts of mayhem and horror may create sensory delight, aesthetic enjoyment of destruction, novelty, fascination at the violation of social norms and/or thrills that appeal to sensation seekers (see Sparks & Sparks, 2000 for review). Other theorists suggest that vicariously sharing the horror experienced by film characters allows viewers to cathartically purge or validate their own fears (Carroll, 1990; Stein & Davis, 1982; Thomas, 1972). These theories might be interpreted to suggest that the more mayhem and horror a film contains, the more appealing it should be. Thus, if films with teaser endings contained more mayhem and horror, they might be predicted to be more appealing. Interestingly, however, while films with teaser endings might imply greater potential for future mayhem and violence, they do not necessarily contain any more actual mayhem and horror than traditional films.

DONT TEASE ME

475

Teasers depart from traditional films in that they violate conventional expectations of good conquering evil. According to McCauley (1998), many of the images in horror films violate social norms. He speculates that it is these violations that audiences find particularly fascinating. Similarly, Carroll (1990) contends that audiences are drawn to violence, mayhem, and horror because it is seen as novel or unusual. Therefore, audiences may become fascinated by films with teaser endings in which evil triumphs over good because they are seen as even more unusual, even more of a violation of social norms than traditional horror films. If, however, it is the novelty of teaser endings that audiences enjoy, then as these endings have become increasingly common, in fact, more the norm than the exception, we might anticipate that they would begin to lose their appeal. Indeed, some film critics have voiced their growing frustration with teaser endings. In his review of Jason X (2001), Mark Ramsey (2002) made his plea to the film studio directly: Hey New Line, somewhere along the way we stopped caring about what happens to Jason because we like beginnings, middles, and ends andHELLOJason, the unstoppable killing machine, will never end (paragraph 15). Several psychological and dramaturgical deconstructions of horrors appeal similarly suggest that audiences might generally prefer horror films with more traditional endings. Tamborini (1991) argued for the importance of the resolution of horror:
One of the most plausible explanations for the appeal of horror suggests that pleasure is derived from the successful resolution of the threat presented by the antagonist responsible for the creation of horror (p. 317).

As Tamborini (2003) notes, this storyline has been a successful formula for the horror film since the early 1930s (Sontag, 1966), and for horror in other media forms since the late eighteenth century (Hallie, 1969). This proposition is consistent with explanations of reactions to other forms of media entertainment. Traditional horror films usually contain the same elements found in most dramatic films. Following the exposition and climax, the protagonist succeeds in overcoming the threat of the evil/antagonist. In explaining the appeal of violent fare, Zillmann (1998) contends that a satisfying resolution is one that contains a positive outcome for the liked protagonist and a negative outcome for enemylike characters. According to Zillmanns disposition theory, we enjoy films in which good things happen to good people and bad things happen to bad people. Enjoyment in this situation is dependent on the condition of a negative outcome for the disliked antagonist. When the antagonist is destroyed in the end, the pleasure and satisfaction that are obtained become even greater (Tamborini & Stiff, 1987). Thus, this model suggests that the appeal of horror may be largely dependent upon the death and demise of the evil/antagonist and the victory and success of the films protagonist.

476

KING AND HOURANI

Excitation transfer theory as applied to media entertainment (Zillmann, 1980) supports similar predictions. According to this theory, arousal experienced from one event is transferred onto the next event encountered. Zillmann (1980) proposed that when film antagonists are endangered, distress and arousal increase in viewers, thus creating suspense. When the protagonist succeeds in overcoming the threat, the emotional distress dissipates, but the physiological arousal remains. Thus, the positive affect that is created by the triumph is intensified by the residual arousal from the distress. As a result, it is reasoned that the greater the danger the protagonist faces, the more intense the satisfaction will be with the successful resolution of the film. Applying these theories, the more havoc the evil/antagonist wreaks, the more aroused and negatively disposed toward the antagonist the audience becomes. In films with traditional endings, theory suggests that audience feelings and arousal are then transferred, intensifying audience enjoyment when the evil/antagonist is conquered. Films with teaser endings, however, evoke the same arousal and distress, but they do not provide a positive resolution, leaving individuals in a highly aroused, but presumably negative mood state. Consistent with this reasoning, Zillmann (1991) noted that [g]enerally speaking, it is the benefaction of good and liked protagonists and the just, punitive treatment of their transgressions and resented opponents, the antagonists, that evoke joyful reactions, citing as evidence Zillmann (1980); Zillmann and Bryant (1975); Zillmann and Cantor (1976, 1977). Hoffner and Cantor (1991) similarly found that under certain conditions, children found frightening film sequences more enjoyable when the sequence had resolved rather than unresolved endings. Although this may be true for many suspenseful genres, it is not clear whether it also holds true for horrora genre, which by definition, is designed to inspire more negative emotional experiences. Both dramatic theory and research would seem to suggest that audiences should find films with traditional endings more enjoyable than films with teaser endings. As intuitive as this may sound, given that teaser endings dominate contemporary horror, it seemed prudent to conduct a more empirical assessment of this claim. This logic forms the basis for the first hypothesis: H1: Audience enjoyment will be greater for horror films with traditional endings than it will be for horror films with teaser endings. Individual Differences in Horrors Appeal Although a preponderance of theory and research would suggest that audiences should generally prefer traditional endings, it is difficult to imagine that films with teaser endings would continue to enjoy such great box office success if they were so universally despised. Of course, not all viewers gain equal enjoyment from any given film. Researchers consider individual differences to be important determi-

DONT TEASE ME

477

nants of how we experience horror. A variety of personality characteristics sensation-seeking (Edwards, 1991; Tamborini & Stiff, 1987); empathy (Tamborini, Stiff, & Heidel, 1990; Zuckerman, 1996; Zuckerman & Litle, 1986), the Machiavellian trait of deceit (Tamborini, Stiff, & Zillmann, 1987), pyschoticism (Aluja-Fabregat, 2000; Weaver, 1991) and coping style (Sparks & Spirek, 1988) have been shown to influence cognitive and affective responses to horror and graphic violence. Noting the various traits that appeared to predict horrors appeal, Johnston (1995) speculated that not all audiences might enjoy horror for the same reasons. Based on data she collected from horror audiences, Johnston created a typology of watchers that reflected different motivations for viewing horror. Just as some people may generally enjoy horror films more than other people do, some viewers may enjoy teaser films while others do not. Different viewing motivations may result in different film preferences; thus ending type may have a greater influence on film enjoyment among viewers with certain motivations than it does for others. Research suggests that some viewers enjoy the destructiveness and excitement of horror, while others appreciate a just ending (Tamborini, Stiff, & Zillmann, 1987; Johnston, 1995). Logic would suggest that viewers who appreciate just endings should prefer traditional endings where the evil is vanquished, while those who enjoy the destructiveness of horror may not have as strong a preference for these endings. Indeed, these audiences may be more titillated by the anticipation of more mayhem foreshadowed by teaser endings. This difference embodies the distinction between thrill and gore-watcher types, two of the watcher types identified by Johnston (1995). Thrill-watchers are thought to enjoy being scared by horror via their empathy for film protagonists, while gore-watchers appear to enjoy the gore and intensity or destructiveness of horror. According to Johnston (1995), high levels of empathy and adventure-seeking characterize the thrill-watching personality. The thrill-watcher enjoys being startled and scared. Although Johnstons study found that thrill-watching was not significantly related to either killer or victim identification, thrill-watchers may empathize with the victim but do not perceive themselves at risk for victimization (Johnston, 1995). Tamborini (1996) similarly proposed that trait empathy might influence viewer experiences with horror. Research suggests that highly empathic individuals are more likely to experience suspense and distress while viewing suspenseful films than individuals with low empathy scores (de Weid, Hoffman, & Roskos-Ewoldsen, 1997). With their empathy and attraction to suspense, thrill-watchers motivations appear to be the most consistent with postulates of disposition and excitation transfer theories. This logic would suggest that these individuals feel greater concern for film protagonists, and thus, should exhibit greater appreciation for traditional endings, where the protagonists prevail. Gore-watchers are also characterized by high adventure seeking, but they depart from thrill-seekers in their low levels of empathy and fearfulness. This combi-

478

KING AND HOURANI

nation of personality traits make gore-watchers seek high arousal originating from graphic portrayals of blood, death, and even physical torture. Other researchers concur that the content and form of todays super-violent horror have great excitatory capacity that is enjoyable for some viewers (Mundorf & Mundorf, 2003; Tannenbaum, 1980; Tamborini, 2003; Tamborini, Miller, Stiff, & Heidel, 1988). This view is consistent with theories on arousal that assert that reinforcement, and in particular reward, can result in some circumstances from an increase in arousal (Berlyne, 1967, p. 30). As Mundorf and Mundorf (2003) suggest, horror may have great enjoyment for viewers who are simply seeking excitement for its own sake. Gore-watching motivations are thought to reflect a more cognitive curiosity about physical violence (the ways that people are killed), a vindictive interest in killing (victims get what they deserve), and an attraction to the grotesque (viewing blood and guts). Gore-watching is also related to high viewing frequency. This is consistent with findings that low-empathy people are most attracted to graphic horror (Tamborini, Stiff, & Heidel, 1990) and with findings that low fearfulness is associated with the appeal of graphic horror (Mundorf, Weaver, & Zillmann, 1989). Gore-watching was strongly associated with identification with the killer especially in males; males were less likely to identify with the victim. Identity with the killer and insensitivity toward the victim may make gore-watchers less concerned about whether or not the killer is ultimately destroyed. In fact, if their enjoyment is motivated by the gore and horror itself, as well as their identification with the killer, then gore-watchers might actually prefer that the killer survive, leaving open the possibility of more horror to come. Gore-watchers, therefore, might show less preference for traditional over teaser endings. Given these considerations, the following hypothesis is proposed: H2: The effects specified in H1 will be greater among thrill-watchers and diminished, if not reversed, among gore-watchers.

METHODOLOGY Overview Respondents were exposed to one of four different horror films that contained teaser endings. Half of the respondents viewed one of the films with its original teaser ending, while the other half viewed an edited version of one of the same films without the teaser ending. After viewing the film, respondents were asked to evaluate the films on a variety of scales. Results were analyzed in a 2 2 2 customized factorial design with ending type (traditional, teaser), gore-watcher (low, high), and thrill-watcher (low, high) functioning as independent-measure factors.

DONT TEASE ME

479

Respondents Respondents were 229 college students recruited from various departments in undergraduate classes at a large university. Free pizza and sodas were provided as compensation for all participants. Participation also served as one of several possible extra credit options for selected courses. A first round of data was collected from a sample of 104 students. Following preliminary analysis and review, the decision was made to collect a second round of data to increase the sample size and the power of the design. One hundred and twenty-five students participated in the second set of study sessions. Procedure The experiment was conducted in 24 sessions including three sessions for each of the eight films (the four original films with teaser endings and the four edited traditional versions of the same films). The sessions took place over two separate two-week periods in classrooms with video projection systems. Each session lasted approximately 2 hours. Respondents were informed that they would be participating in a horror film appreciation study and that they would watch and evaluate edited versions of horror films that may or may not contain graphic violence, adult language and/or some sexual situations, as might be seen in movies holding restricted ratings. They were given the opportunity to withdraw their participation at any time, without penalty, if they had objections to any part of the procedure. There were no refusals. Prior to viewing the film, respondents were asked to complete the first part of the questionnaire (which took approximately 15 min). Respondents then viewed one of the films. When the film ended, they completed the questionnaire. After finishing the questionnaire, respondents were dismissed and thanked for participating. The protocol was reviewed by the human subjects committee. Because students were informed that the films were edited (even though they were not told specifically what was edited), the study was judged not to include deception that would require formal debriefing. Thus, in order to prevent subject contamination, respondents were not debriefed immediately following the sessions regarding the study manipulations. They were, however, informed that they could pick up a report detailing the study and the results at the departmental office at the end of the semester or request that the information be mailed to them. The Film Stimulus Multiple films were used to increase generalizability and to avoid any potential confounding factors that might result from the idiosyncrasies of a single film. The

480

KING AND HOURANI

four films that were employed lasted approximately 90100 min. The selected films were chosen to fit the following criteria: more than 2 years old (respondents might be more familiar with the endings of recent films), less than 2 hours long (the allotted time for the study), and contain a teaser ending (that could be manipulated for the study). The four films that were selected were: Candyman, Needful Things, Leprechaun, and Pet Sematary. All of the four films contained what at first appears to be a final scene in which the evil is destroyed, however, a final segment followed this scene that reveals that the evil either survived or was somehow resurrected. Each film was manipulated to create traditional (or non-teaser ending) versions of the film by removing the teaser scene and rolling the credits, thus, giving the impression that the film ended with good conquering evil. Measures

Watcher Type
A set of 10 questions based on Johnstons (1995) watcher motivation scales was used to identify gore and thrill-watcher types. These scales included statements such as I enjoy scary movies because they contain a lot of blood and gore, (gore-watcher) and I watch horror films because I like to be scared, (thrill-watcher) that were assessed on five-point integer scales (2 to 2) ranging from strongly disagree to strongly agree.

Film Evaluations
Prior to evaluating the films, respondents were asked whether or not they had seen the film. Only five people indicated seeing the film, and these individuals were excluded from the analysis. Respondents were asked to evaluate the films on 16 adjective scales: disturbing, entertaining, suspenseful, enjoyable, predictable, interesting, upsetting, tragic, boring, depressing, violent, serious, distressing, scary, exciting, and funny. Evaluations for all the scales were made on 11-point integer scales ranging from not at all (0) to extremely (10). In order to gain a richer, more in-depth understanding of audience opinions of different ending types, respondents were given the opportunity to express their thoughts in answer to two open-ended questions about the endings of the films in the final part of the questionnaire. The first two questions focused on the respondents opinions about the film ending, including What did you like/not like about the ending? and How do you think the movie should have ended? These questions were included in anticipation that they might provide greater explanatory insight into closed-ended evaluations. The last question, What do you think the pur-

DONT TEASE ME

481

pose of this study is? was included as a manipulation check to insure that participants were not aware of the manipulation.

RESULTS Manipulation Checks At the end of the questionnaire, respondents were questioned about the purpose of the study to determine whether they were aware of the film manipulations. No responses suggested that respondents were aware of the film manipulations or the specific purpose of the study. As an additional manipulation check, analysis was run with the film itself (Candyman, Leprechaun, Pet Sematary, or Needful Things) included in the design as an independent measure. No significant interactions were found among the four films and experimental variables; thus, all four films were retained in the analysis. And as a final check, analysis was run comparing the data collected in the first and second wave of data collection to insure against threats of periodicity. No significant differences were found between the two data sets that might pose such threats. Data Reduction

Film Watcher Type

Measures of each watcher type were calculated based on the mean scores of the specified scale items. Viewers were separated into high and low groups for each watcher type by their mean split creating two dichotomous watcher type variables, gore (low, high) and thrill (low, high); thus, the watcher type variables were not mutually exclusive. The mean scores for each watcher type were as follows: gore (M = .26), and thrill (M = .37).

Film Evaluations
The film evaluations (16 adjective scales, averaged across the four horror films) were subjected to principle component analysis. Only scales that loaded > .66 for the factor in question and loaded < .50 for all other factors were retained. Four factors emerged for the evaluation of the films. Factor 1, labeled Entertaining, was characterized by high loadings on the scales entertaining (.88), enjoyable (.99), interesting (.86), serious (.76), exciting (.79). The combination of these scales yielded high internal consistency: Cronbachs = .92 and accounted for 37% of the variance. Upsetting (.88), tragic (.83), depressing (.766) and distressing (.78) formed Factor 2, labeled Distressing (= .86) which accounted for 25% of the vari-

482

KING AND HOURANI

ance. Scary (.79) and suspenseful (.729) formed Factor 3 ( = .88 ), labeled Scary, accounting for 8% of the variance, and the item Predictable (.87) loaded alone on Factor 4, accounting for 7% of the variance.

Film Ending Evaluations

Given time and resource constraints and original plans to use this data for more qualitative assessments, the open-ended data was only quantitatively analyzed for the 104 respondents who participated in the first round of data collection. Responses of the first two open-ended questions were content analyzed and placed in several clearly defined categories. The first question inquired whether respondents liked/disliked the ending of the film, thus the two main comment categories were: like and dislike. For those who viewed traditional endings, the subcategories under like included: evil is blown up in the fire, he gets what he deserves, interesting, legend is stopped/theme is discontinued, satisfying, and unpredictable. The subcategories under dislike were as follows: unrealistic, not enough suspense, legend is discontinued, and other. Responses for those who saw the teaser endings also were categorized under like and dislike. The subcategories under like included: evil is not blown up in the fire, [he] gets away, interesting, legend lives on/theme is continued, satisfying, predictable, and other. The subcategories under dislike included: evil is not destroyed, legend is continued, predictable, and other. The second open-ended question asked respondents how the film should have ended. Responses for those who viewed traditional endings were subcategorized into the following: evil should live, innocent character should live, more realistic, legend is continued along with the theme, and dont change. Among those who viewed teaser endings answers to the change question were subcategorized into the following: evil should be destroyed, kill off legend/ theme, and dont change. Intercoder Reliability To establish the reliability of coding for the open-ended questions, two coders analyzed the responses to the questions based on the established categories. The degree of agreement between the coders was 100% as calculated by Holstis formula. Data Analysis In addition to a priori power analysis, Cohen (1990); Cohen and Cohen; (1983) and Smith, Levine, Lachlan, and Fediuk (2002) recommend using no design or analysis more complex than needed in order to minimize error. Thus, a customized 2 2 2 factorial ANOVA design was employed that tested only the hypothesized main

DONT TEASE ME

483

effects of ending type (traditional, teaser), and two-way interaction effects for ending type with each watcher type, gore (low, high) and thrill (low, high). No three-way interactions were predicted, therefore, this interaction was excluded from the analysis. In addition, since main effects of watcher type were not the focus of this study, they were also excluded from the analysis. By limiting the contrasts in this design, we were able to minimize error while complying with human subjects review requirements to impose upon no more participants than necessary to meet minimum sample sizes. Power Analysis Following Lipseys (1990) recommendations, effect size and partial power analyses were calculated on the data collected from the first 104 respondents. Effect sizes for the specified comparisons ranged from .50 1.25, with most falling between .60 and .70. Based on this analysis, additional data was collected in order to increase the power of the design. Although access to additional subjects was limited, enough additional data was collected to result in an average cell size of 57, with a minimum cell size of 35 for the specified 2 2 comparisons. Based on Lipseys (1990) tables, power estimates ranged from .65 to .99, with most estimates falling in the .85.99 range. Observed power as calculated by SPSS is reported for each significant finding below. Film Evaluations Analysis of the Entertaining scale revealed interactions between ending type and each of the watcher types, gore (F(2, 198) = 9.87, p < .001; 2 = .10, observed power = .99) and thrill (F(2, 198) = 4.76, p < .01; 2 = .05, observed power = .79), as well as a main effect of ending type (F(1, 198) = 7.61, p < .002; 2 = .04, observed power = .89). Overall, films with traditional endings were judged more entertaining (M = 6.40) than those with teaser endings (M = 5.39); however this effect was qualified by interactions which revealed significant preferences for traditional endings only among high gore and thrill-watchers. In addition, traditional endings were rated as significantly more entertaining by high gore-watchers than low gore-watchers, and teaser endings were rated as significantly less entertaining by high thrill-watchers than low-thrill-watchers. Mean comparisons are reported in Table 1. Analysis of the Distress scale revealed interactions between ending type and each watcher type: gore (F(2, 198) = 6.87, p < .001; 2 = .10, observed power = .99) and thrill (F(2, 198) = 6.87, p < .001; 2 = .07, observed power = .92), although only simple effects of watcher type were found where traditional endings were less distressing for high gore-watchers (M = 2.99) than low gore-watchers (M = 4.46) and teaser endings were more distressing for high thrill-watchers (M = 4.75) than

484

KING AND HOURANI

low thrill-watchers (M =2.93). Both high gore and thrill-watchers found teaser endings more distressing than traditional endings, although these differences were not significant (see Table 1). Analysis of the Scary scale revealed interactions similar to those found for entertaining between ending type and each of the watcher types, gore (F(2, 198) = 11.18, p < .001; 2 = .10, observed power = .99) and thrill (F(2, 198) =12.00, p < .001; 2 = .11, observed power = 1.0), as well as a main effect of ending type (F(2, 198) = 31.18, p < .001; 2 = .14, observed power, 1.00). Overall, films with traditional endings were judged more scary (M = 7.47) than those with teaser endings (M = 5.52); however this effect was also qualified by interactions which revealed significantly higher scary ratings for traditional endings only among high gore and thrill-watchers. In addition, traditional endings were rated as significantly more scary by high gore-watchers than low gore-watchers, and teaser endings were rated as significantly less scary by high thrill-watchers than low thrill-watchers. Mean comparisons are reported in Table 1. Analysis of Predictability also revealed a similar pattern of interactions between ending type and each of the watcher types, gore (F(2, 198) = 5.99, p < .003; 2 = .08, observed power = .88) and thrill (F(2, 198) = 15.32, p < .001; 2 = .14, observed power = 1.0), as well as a main effect of ending type (F(2, 198) = 4.41, p < .037; 2 = .03, observed power, .60). Overall, films with teaser endings were judged more predictable (M = 7.60) than those with traditional ending (M = 6.99); however this effect was also qualified by interactions which revealed significantly higher predictable ratings for teaser endings only among high gore and low thrill-watchers. In addition, traditional endings were rated as significantly less predictable by high gore-watchers than low gore-watchers, and teaser endings were
TABLE 1 Mean Comparisons of Film Evaluation Measures by Ending and Watcher Type Film Evaluation Entertaining Scary Distressing Predictable Traditional Ending 7.67a 5.96b 6.29a 7.30b 6.99a Teaser Ending 7.54a 7.79a 8.77b 6.51a 7.60b

Watcher Traditional Teaser Traditional Teaser Traditional Teaser Type Ending Ending Ending Ending Ending Ending Gore Low High Thrill Low High Total Note. 5.70a 7.46b 6.30a 6.62a 6.40a 5.25a 5.90a 6.13a 4.72b 5.39b 6.75a 8.56b 7.36a 7.72a 7.47a 5.27a 6.46a 6.73a 4.41b 5.52b 4.46a 2.99b 3.77a,b 4.15a,b 3.88a 3.92a,b 3.71a,b 2.93a 4.75b 3.87a

Mean pairs not sharing a letter differ significantly by p < .05.

DONT TEASE ME TABLE 2 Summary Cross-tabulation of Open-Ended Opinions of Horror Film Ending Traditional ending (n) Liked Disliked Change Dont Change 85% (44) 15% (8) 31% (15) 69% (34)

485

Teaser ending (n) 31% (16) 69% (36) 78% (40) 22% (11)

rated as significantly less predictable by high thrill-watchers than low thrill-watchers. Mean comparisons are reported in Table 1.

Film Ending Evaluations

Consistent with the closed-ended measures, respondents who saw the films without the teaser endings expressed more favorable comments regarding the film ending. Chi Square analysis for the cross-tabulation between ending type (traditional, teaser) and comment category (like, dislike) was significant = 2 (1, N = 104) = 30.88, p < .001. More respondents who saw the traditional versions indicated they liked the ending, while more respondents who saw the teaser versions indicated that they didnt like the ending (See Table 2). Unpredictability was the most frequent reason offered for liking the traditional ending, while predictability was the most frequent reason offered for liking the teaser ending. Unrealistic was the most frequent reason offered for disliking the traditional ending, while predictability, again was the most common reason offered for disliking the teaser ending. Other response categories for both film versions, however, included comments with specific references to whether or not the evil was destroyed, and in combination, these answers figured most prominently (see Table 3 for a complete breakdown of responses). The cross-tabulation between ending type and desire to change the ending (change, dont change) was also significant 2 (1, N = 100) = 23.09, p < .001, with more respondents expressing desire to change the teaser ending than the traditional ending (see Table 2). In most cases, desire to change the ending specifically focused on the fate of the evil (see Table 3).

DISCUSSION As anticipated in Hypothesis 1, audience reactions to the horror films were significantly influenced by the nature of the film ending. Overall, viewers rated films with the traditional (non-teaser) endings as more enjoyable, scarier and less pre-

486
Traditional Ending (N = 52) # 9 16 16 11 5 22 79 11 20 20 14 6 29 100 17 31 31 21 10 42 85* 3 4 3 4 6 12 1 22 16 22 29 3 70 %# %N # Teaser Ending (N = 52) %# 9 12 9 12 19 36 3 100 23 31 42 4 100 %N 6 8 6 8 12 23 2 31* 31 42 56 6 69* 8 3 6 5 22 36 14 27 23 100 15 6 12 10 15* Liked Evil is not blown up in the fire He gets away Interesting Continuation of theme/legend Satisfying Predictable Other Total Disliked Evil is not destroyed Legend is continued Predictable Other Total

TABLE 3 Detailed Cross-tabulation of Open-Ended Opinions of Horror Film Ending

Liked Evil is blown up in the fire He gets what he deserves Interesting Legend/theme is discontinued Satisfying Unpredictable Total

Disliked Unrealistic Not enough suspense Legend/theme is discontinued Other Total

Traditional Ending (N = 49) # 3 5 4 4 16 Dont change 34 69* 11 19 31 25 25 100 6 10 8 8 31* Change Evil should be destroyed Discontinue legend/theme Change predictability Total %# %N #

Teaser Ending (N = 51) %# %N

Change Evil should live Innocent character should live Legend/theme is continued More realistic Total Dont change Total

24 22 9 55

43 41 16 100

47 42 18 78* 22*

# = Number of responses. Some participants offered multiple responses, so totals could equal more than the number of respondents. %# = Percentage of responses out of the total number of responses in the selected category (like, dislike, change, or dont change). %N = Percentage of responses out of the total number of respondents for film type (traditional or teaser ending). * = Since response codings were not mutually exclusive, the total percentage of respondents in each category does not equal the sum of the column percentages.

487

488

KING AND HOURANI

dictable than their teaser counterparts. Open-ended responses similarly suggested that more respondents enjoyed films in which the evil was destroyed. One respondent commented, Finally a film where the guy actually dies and gets what he deserved. [Mr. Gaunt] wreaked havoc to the extreme, its time that he died and people can get on with their lives (Female). Another stated, Im glad that the [Candyman] was destroyed after he ruined [Helens] life. He got what he deserved, I really liked that (Male). Respondents who viewed teaser endings were more likely to express dislike of the ending and want to change it than those who viewed traditional endings. These findings are consistent with disposition and suspense theories that postulate that viewers enjoy negative outcomes for film antagonists. The second hypothesis, however, was only partially supported. Given the preponderance of horror films with teaser endings, it was reasoned that at least some horror fans, possibly those who enjoyed the gore of horror, must prefer teasers. In this study, however, both high thrill and gore-watchers showed a stronger preference for traditional endings than teasers. And yet, while both gore and thrill viewers appeared to prefer traditional endings, additional patterns in the data reveal that they may appreciate them for somewhat different reasons. Responses from both closed and open-ended questions point to predictability as a determining factor for the appeal of horror for many viewers. The open-ended answers suggest that many viewers may dislike teaser endings because they have become the typical horror film ending and prefer traditional endings because they view them as an unexpected but refreshing change. An example of such a response is as follows, Very interesting to have [Candyman] to actually die in the fire. His legend died with him, and now the residents of [Cabrini Greene] can sleep without the nightmares. I still cant believe that he diedno blinking, no signs that he will come backvery interesting (Male). Indeed, this explanation is supported by the closed-ended analysis, where audiences generally rated the traditional films as significantly less predictable than the teasers. The consistent use of teaser endings in horror films may have made these endings so mundane that audiences expect them. In line with Carrolls (1990) claim that the appeal of horror lies in its novelty, teaser endings may have initially increased enjoyment because they were novel and unexpected, however, as teaser endings became more common and their novelty declined, so may have their appeal. High gore-watchers most strongly reflected this pattern, rating traditional endings as more entertaining and scary, but less predictable than teaser endings. High thrill-watchers similarly rated films with traditional endings more entertaining and scarier, but, in contrast to other viewers, they also rated traditional endings more predictable. Given that thrill-watchers tend to exhibit a strong preference for just endings, they may not only want the evil to be destroyed, but perhaps even expect the evil to be destroyed. Thus, in spite of the recent teaser trend, thrill-watchers may still find traditional endings more in keeping with their expectations, and therefore, rate them as both more enjoyable and more predictable.

DONT TEASE ME

489

Although many audiences may prefer traditional endings, open-ended comments suggest indicate there are some horror fans that enjoy teasers. As one respondent stated, Having [Gaunt] come out of the fire without a hair out of place, just makes it even more exciting. I would have liked it more if he turned into a devil right then and there, showing his true colors. Another respondent stated, I liked the fact that were left hanging with the idea that the [Leprechaun] will return with more adventures for the next time hell be back. Interestingly, predictability may also mediate these preferences. In open-ended responses, predictability was the most common answer offered for both liking and disliking teaser films. Some respondents similarly indicated that they enjoyed the teaser endings because they were realistic and disliked traditional endings because they were unrealistic, implying a belief and expectation that evil cannot be destroyed. Just as with traditional endings, some audiences may see the survival of the antagonist in teasers predictable, while others do not, and further, some of those people may appreciate predictability, while others do not. It would seem that audiences differ not only in terms of their expectations for horror films, but also as to whether they enjoy having those expectations met or violated. Notably, study manipulations had limited impacts on how distressing viewers found the films. Consistent with other film evaluations, there were some simple effects of watcher type on audience distress, but effects of ending type did not reach significance. This may be because the difference between traditional and teaser endings is only the anticipation of further mayhem. It is also possible that results were obscured because some audiences viewed distressing as a positive endorsement for a horror film, while others viewed it negatively. Another explanation may be that the self-reported measures of distress used in this study did not provide an accurate picture of audience distress reactions. Some respondents may have been hesitant to disclose their true reactions to the film. Sparks, Pellechia, and Irvines (1999) findings suggest combining physiological measures with self-reported data might help provide a more complete picture of audiences negative affective reactions to frightening films, particularly for those with repressive coping styles. Clearly, future research might continue to examine different viewers and viewing contexts to gain greater insights into horror appreciation. More specifically, efforts might focus on refining watcher typologies that better isolates viewer motivations related to specific horror preferences. Results suggest that researchers may want to pay particular attention to the influence of viewer perceptions and preferences for predictability of horror. Additional investigations might include physiological measures to enrich measures of affect such as distress. Such measures could also examine the role physiological arousal may play in mediating reactions to horror as predicted by excitation transfer and other theories. To better understand these mechanisms, researchers should consider applying a methodological process model to observe and measure reactions during the course of the viewing experience.

490

KING AND HOURANI

In summary, consistent with traditional theories of drama and suspense, this study found that most audiences appeared to have strong preferences for traditional endings. These preferences were particularly strong for those more inclined to watch horror, regardless of whether they were motivated by gore or thrills. Such preferences would seem to have significant implications for film makers in that they cut against the current trend of teaser endings. It may be that many audiences continue to watch horror with teaser endings simply because they have no choice! Some producers might maintain that teasers are necessary to promote blockbuster sequels; however, one could argue that the best promotion for a sequel is to make the initial film more enjoyable. Even absent teaser endings, the evil can always be revived in previews and opening segments of following films. Thus, filmmakers might want to consider reverting back to traditional endings in more films. Perhaps by not eliminating teaser endings completely, but instead by varying their formulas, filmmakers can provide something for everyone and still maximize the suspense that drives all drama.

REFERENCES
Aluja-Fabregat, A. (2000). Personality and curiosity about TV and films violence in adolescents. Personality and Individual Differences, 29, 379392. Berlyne, D. E. (1967). Arousal and reinforcement. In D. Levine (Ed.), Nebraska Symposium on Motivation (Vol. 15, pp. 1110). Lincoln; University of Nebraska Press. Carroll, N. (1990). The philosophy of horror, or paradoxes of the heart. New York: Routledge. Cohen, J. (1990). Things I have learned (so far). American Psychologist, 45, 13041312. Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the behavioral sciences (2nd Ed.). Hillsdale, NJ: Lawrence Erlbaum Associates. De Weid, M., Hoffman, K., & Roskos-Ewoldson, D. R. (1997). Forewarning of graphic portrayal of violence and the experience of successful drama. Cognition and Emotion, 11, 481494. Edwards, E. (1991). The ecstasy of horrible expectations: Morbid curiosity, sensation seeking, and interest in horror movies. In B. Austin (Ed.), Current research in film: Audience, economics, and law (Vol. 5 pp. 1938). Norwood, NJ: Ablex. Gleiberman, O. (1997, October). Killing time. Entertainment Weekly, 402, 44. Hallie, P. P. (1969). The paradox of cruelty. Middletown, CT: Wesleyan University press. Hoffner, C., & Cantor, J. (1991). Factors affecting childrens enjoyment of a frightening film sequence. Communication Monographs, 58(1), 4163. Johnston, D. (1995). Adolescents motivations for viewing graphic horror. Human Communication Research, 21(4), 522552. Lipsey, M. W. (1990). Design sensitivity: Statistical power for experimental research. Newbury Park: Sage. McCauley, C. (1998). When screen violence is not attractive. In J. H. Goldstein (Ed.), Why we watch: The attractions of violent entertainment (pp. 141162). New York: Oxford University Press. Mundorf, N., & Mundorf, J. (2003). Gender Socialization of Horror. In J. Bryant, D. Roskos-Ewoldsen & J. Candor (Eds.). Communication and emotion: Essays in Honor of Dolf Zillmann (pp. 155178).Mahwah, NJ: Lawrence Erlbaum Associates.

DONT TEASE ME

491

Mundorf, N, Weaver, J., & Zillmann, D. (1989). Effects of gender roles and self perceptions on affective reactions to horror films. Sex Roles, 20, 655673. Nashawaty, C. (1998, August). Final cut twenty years after Halloween carved her name in lights, Jamie Lee Curtis takes on Michael Myers one last time (we think) in Halloween: H20. Entertainment Weekly, 445, 2734. Ramsey, M. (2002, April 25). Jason X - Friday the Umpteenth. [On-line]. Available: http:// www.moviejuice.com/2002/jasonx.htm. Rosenbaum, R. (1979, September). Gooseflesh. Harpers, 259:1552, 8692. Smith, R. A, Levine, T. R., Lachlan, K. A., & Fediuk, T. A. (2002). The high cost of complexity in experimental design and data analysis: Type I and Type II error rates in multiway ANOVA. Human Communication Research, 28(4), 515531. Sontag, S. (1966). Against interpretation. New York: Farrar, Straus & Giroux. Sparks, G., Pellechia, M., & Irvine, C. (1999). The repressive coping style and fright reactions to mass media. Communication Research, 26(2), 176192. Sparks, G. G, & Sparks C. W. (2000). Mayhem and horror. In D. Zillmann & P. Vorderer (Eds.), Media Entertainment: The Psychology of Its Appeal (pp. 7391). Mahwah, NJ: Lawrence Erlbaum Associates. Sparks, G., & Spirek, M. (1998). Individual differences in coping with stressful mass media: An activation-arousal view. Human Communication Research, 15, 195216. Stein, M., & Davis, J. (1982). Therapies for adolescents: current treatments for problem behaviors. NY: Jossey Bass. Tamborini, R. (2003). Enjoyment and social functions of horror. In J. Bryant, D. Roskos-Ewoldsen, & J. Cantor. Communication and emotion: Essays in honor of Dolf Zillmann. Hillsdale, NJ: Lawrence Erlbaum Associates. Tamborini, R. (1996). A model of empathy and emotional reactions to horror. In J. Weaver & R. Tamborini (Eds.), Horror films: Current research on audience preferences and reactions (pp. 103124). Mahwah, NJ: Lawrence Erlbaum Associates. Tamborini, R. (1991). Responding to horror: Determinants of exposure and appeal. In J. Bryant & D. Zillmann (Eds.). Responding to the screen: Reception and reaction processes (pp. 305327). Hillsdale, NJ: Lawrence Erlbaum Associates. Tamborini, R., Miller, K., Stiff, J., & Heidel, C. (1988). Predictors of emotional reactions to audio and visual elements in graphic horror: A time series analysis. Paper presented at the annual conference of the Speech Communication Association, New Orleans, LA. Tamborini, R., & Stiff, J. (1987). Predictors of horror film attendance and appeal: An analysis of the audience for frightening films. Communication Research, 14, 15436. Tamborini, R., Stiff, J., & Heidel, C. (1990). Reacting to graphic horror: A model of empathy and emotional behavior. Communication Research, 17, 616640. Tamborini, R., Stiff, J., & Zillmann, D. (1987). Preference for graphic horror featuring male versus female victimization: Individual differences associated with personality characteristics and past film viewing experiences. Human Communication Research, 13, 529552. Tannenbaum, P. H. (1980). Entertainment as vicarious emotional experience. In P. H. Tannenbaum (Ed.). The entertainment functions of television (pp. 107131). Hillsdale, NJ: Lawrence Erlbaum Associates. Thomas, J. (1972). Gobble, gobbleone of us! In R. Huss & T. J. Ross (Eds.), Focus on the horror film (pp. 135138). Englewood Cliffs, NJ: Prentice-Hall. Weaver, J. B. (1991). Effects of opposite-gender companions affect to horror on distress, delight, and attraction. Journal of Personality and Social Psychology, 51, 586594. Zillmann, D. (1998). The psychology of the appeal of portrayals of violence. In J. H. Goldstein (Ed.). Why we watch: The attractions of violent entertainment (pp. 179211). New York: Oxford University Press.

492

KING AND HOURANI

Zillmann, D. (1991). Empathy: Affect from bearing witness to the emotion of others. In J. Bryant & D. Zillmann (Eds.), Sports, games, and play: Social and psychological viewpoints (2nd Ed., pp. 241278). Hillsdale, NJ: Lawrence Erlbaum Associates. Zillmann, D. (1980). The entertainment functions of television. Hillsdale, NJ: Lawrence Erlbaum Associates. Zillmann, D., & Bryant, J. (1975). Viewers moral sanction of retribution in the appreciation of dramatic presentations. Journal of Experimental Social Psychology, 11, 572582. Zillmann, D., & Cantor, J. (1976). A disposition theory of humor and mirth. In T. Chapman & H. Foot (Eds.), Humor and laughter: Theory, research, and application (pp. 93115). London: Wiley. Zillmann, D., & Cantor J. (1977). Affective responses to the emotions of a protagonist. Journal of Research in Personality, 8, 335349. Zuckerman, M. (1996). Sensation seeking and the taste for vicarious horror. In J. B. Weaver & R. Tamborini (Eds.), Horror films: Current research on audience preferences and reactions (pp. 147160). Mahwah, NJ:Lawrence Erlbaum Associates. Zuckerman, M., & Litle, P. (1986). Personality and curiosity about morbid sexual events. Personality and Individual Differences, 2, 4956.