Abstract
This study examined young viewers’ evaluations of the social and physical aggression that is endemic in tween sitcoms. Preadolescents (N = 176) were randomly assigned to watch a tween sitcom that featured a protagonist or antagonist committing social or physical aggression (two exemplars at each level). As suggested by recent work on disposition theory and moral development, participants’ moral evaluations of the depicted aggression were related to their liking of the perpetrator, not their perceptions of the severity of the aggression. Furthermore, there was support for the causal chain proposed by Raney’s formulation of disposition theory, whereby protagonist liking predicted moral indifference, which further predicted self-reported likelihood of imitating the aggression. This path was stronger at higher ratings of funniness. When aggressors were perceived as likable and their actions as funny, viewers reported more indifference and greater likelihood of imitation.
In 1989, a new sitcom, Saved by the Bell, aired on NBC. The program featured six teen characters who humorously navigated the social and academic challenges of high school. The program represented NBC’s gamble that it would be lucrative to pull out of the Saturday morning cartoon lineup for children and focus on a newly conceptualized demographic of preadolescents (or “tweens”). By 1993, three of NBC’s live-action teen programs were among the top-five favorites with youth viewing audiences (Donlon, 1993), and other networks quickly created their own versions. Disney found success with programs like Lizzie McGuire, Hannah Montana, and Wizards of Waverly Place; Nickelodeon created the immensely popular Victorious and iCarly (Busis, 2012).
Even as the number of shows catering to this demographic has increased over the last 20 years, the archetypes and situations featured have remained unchanged. In every episode, physically attractive, likable, and funny protagonists encounter prototypical teen challenges and deal with an array of secondary, sometimes villainous, characters. Consider, for example, a few episodes of iCarly: in one, a villainous perpetrator ties up her victims and throws plates and glasses at their heads; in another, a lovable but roguish perpetrator goes online to humiliate a friend by revealing that he has never kissed anyone before. Both acts were intentional rather than accidental and neither was particularly justified. In yet another episode, the aggression was provoked—the same lovable but roguish perpetrator wrestles a bully to the ground to save her friend from physical aggression. Thus, the nature of these portrayals is complicated because the aggression is sometimes justified and sometimes not; sometimes committed by likable protagonists and sometimes by unlikable villains.
The immense popularity of these programs poses a puzzle and raises a question about effects. How do young viewers evaluate these hostile interactions, and what might be the consequences for their own actions? Prior experimental, cross-sectional, and longitudinal research has linked exposure to tween sitcoms to young viewers’ physical and social aggression (Coyne & Archer, 2005; Coyne, Archer, & Eslea, 2004; Linder & Gentile, 2009; Martins & Wilson, 2012b). Thus far, this research has focused on effects on attribution biases as an explanatory mechanism—Those with more exposure to such content have more hostile attributions about ambiguous situations (Martins, 2013) and are, in turn, more likely to act aggressively (Gentile, Coyne, & Walsh, 2011).
The current study seeks to build on this literature by focusing on young viewers’ interpretations of the content (rather than on exposure per se), particularly their moral evaluations of the characters’ actions, and their perceptions of the depicted aggression as funny. In doing so, we seek to reunite this area of research with Bandura’s (2002) focus on moral responses to media depictions of aggression and to consider the implications of affective dispositions toward characters. Perhaps, in addition to shaping social attributions, the partisan nature of tween sitcom narratives (likable protagonists and repugnant antagonists hurt each other for comedic effect) contributes to biased moral evaluations of characters’ aggression and thereby to increased likelihood of imitation.
We begin by describing youthful moral reasoning about physical and social aggression and consider prior evidence that young viewers’ moral judgments can be swayed by media depictions. We then use the framework of disposition theory to consider the potential roles of severity of depicted aggression and liking of the perpetrator in shaping such moral evaluations. Finally, we consider how perceived funniness may contribute to moral responses.
Moral Reasoning in Youth
Moral reasoning involves thinking through and articulating explanations for choices about how one ought to behave (Eisenberg, 1986; Richardson, 2014). Kohlberg’s (1969, 1976) work in this area demonstrated that an individual’s ability to evaluate and respond to moral dilemmas becomes more sophisticated with age. At the preconventional level, children (approximately ages 2-9) tend to judge actions by the consequences they bring to the self or by simple principles of exchange (e.g., it is wrong to do it because I will get punished, or it is okay to do it because she did it to me). At the conventional level (typical of adolescents), individuals tend to judge the morality of actions in terms of societal views and expectations (e.g., I should not do it because peers would think I am bad and that would make me feel bad about myself, or I should not do it because it is against the law). At the postconventional/principled level (most likely to occur in adulthood), there is an increasing focus on broad principles of social rights and a recognition that an individual’s moral judgments based on these principles may take precedence over society’s views (e.g., I need to protest unjust laws, or I need to fight to bring down an unfair regime).
More recent work in this area (e.g., Haidt, 2001) suggests that moral judgments are often guided by quick, emotion-based intuitions, rather than a process of extended reasoning and reflection. Haidt’s (2001) social intuitionist model posits that these quick moral judgments are influenced by (a) an individual’s perceptions of what his or her friends think and do (i.e., relatedness motives) and (b) an individual’s desire to be consistent with his or her present worldview (i.e., coherence motives). Once these issues are considered, individuals quickly try to rationalize and reason their way into the already-constructed moral judgment. Given this model’s assumptions, Haidt (2001) also contended that peer networks have the strongest influence on morality during late childhood and adolescence.
Reasoning About Physical and Social Aggression
Research on preadolescents’ and adolescents’ moral reasoning suggests that they tend to apply similar standards and types of reasoning for both physical and social aggression, both of which can be used to cause harm to peer relationships. As defined by Galen and Underwood (1997), social aggression includes a wide array of behaviors intended to hurt the victim’s social standing rather than inflict physical pain (Galen & Underwood, 1997). Such behaviors may be direct or covert, manipulative or straightforwardly hostile, verbal or nonverbal. Thus, the construct covers the range of behaviors depicted on tween sitcoms better than the two related, but narrower, constructs of indirect aggression (which focuses on covert acts; Lagerspetz, Bjorkvist, & Peltonen, 1988) and relational aggression (which focuses on social manipulation; Crick et al., 1999).
Murray-Close, Crick, and Galotti (2006) asked fourth and fifth graders to evaluate scenarios involving physical or relational (e.g., exclusion) aggression. Although children judged physically aggressive behaviors as more wrong and harmful than relationally aggressive behaviors, they identified both forms as hurtful and as moral issues (rather than issues of social convention, prudence, or personal choice). Goldstein and Tisak (2010) found that 11- to 15-year-olds (regardless of age) rated gossiping and physical aggression (i.e., hitting or shoving) as equally wrong, and they raised similar moral considerations about the victim’s well-being and rights in both instances (e.g., for gossiping: “They have a right to know what you are saying about them,” p. 475). Social exclusion was also judged as morally problematic; however, it was considered less wrong than gossiping or physical aggression, particularly if it was in retaliation for similar treatment. Similarly, an extensive line of research by Killen and colleagues (see Killen, 2007; Rutland, Killen, & Abrams, 2010) has documented adolescents’ use of principled judgments about exclusion (e.g., it is wrong to exclude on the basis of race or gender), combined with considerations of peer relationships (e.g., it might be okay to exclude someone if they did not really fit in with the group).
Taken together, this body of research suggests that across the preteen and adolescent years, individuals tend to judge both physical and social forms of intentional aggression as moral violations and that they rely on a blend of judgments about social relationships, exchange/retaliation motives, and principles about rights. There is some prior indication that these judgments can be altered by exposure to media depictions of aggression.
Effects of Media Depictions on Youthful Moral Reasoning
Krcmar and Valkenburg (1999) found that 6- to 12-year-olds with higher levels of exposure to fantasy violence gave higher ratings of the acceptability of using physical violence in response to provocation (e.g., attacks on a family member) and that they gave less advanced moral justifications for their judgments. Krcmar and Curtis (2003) reported that 5- to 14-year-olds who saw a fantasy program with a violent resolution, relative to a control group, were significantly more likely to choose an aggressive solution to a vignette. In doing so, they were more likely to offer hedonistic, self-focused reasoning to support their choice rather than considering the perspective of others (see also Krcmar & Vieira, 2005).
Also relevantly, Mares and Braun (2013) found that experimental exposure to high-conflict scenes in tween sitcoms altered fifth graders’ moral judgments about the acceptability of social exclusion (e.g., jocks not allowing a skater to join their team), by heightening perceptions that individuals from different social groups might not function well together. Indeed, this latter finding is consistent with Haidt’s (2001) contention about the importance of relatedness motives for youthful moral reasoning.
Taken together, these studies suggest that young viewers’ moral judgments about depictions of physical and social aggression are swayed by the nature of the content, and that these moral judgments can, in turn, have implications for what is considered appropriate behavior. Other research on parasocial relationships with media characters (Rubin & McHugh, 1987; Tian & Hoffner, 2010) suggests that favorite characters in tween sitcoms may perhaps serve as surrogate peer networks. Perhaps then, young viewers’ moral reasoning about aggression committed by beloved characters would be biased by relatedness motives, rather than being rational and partisan, and this biased reasoning would have implications for their own future actions. Thus far, no studies have systematically considered the role of tween viewers’ affective dispositions toward characters in shaping their moral judgments. However, work on disposition theory can speak to these issues.
Disposition Theory and Moral Judgments
Disposition-based theories of media enjoyment posit that enjoyment of media fare is predicted by viewers’ affective disposition toward media characters and the outcomes experienced by those characters (for review, see Raney, 2003). Specifically, disposition theories propose that enjoyment increases when good things happen to highly liked characters, bad things happen to disliked characters, or the combination of the two. Conversely, viewer enjoyment decreases when liked characters experience negative events or disliked characters experience good fortune (Zillmann, 1994). Empirical support for disposition theory of drama (Zillmann & Cantor, 1976, 1977), fright-evoking content (Hoffner & Cantor, 1991), and reality programming (Oliver, 1996) abounds. Indeed, the research demonstrates that viewers’ feelings, or dispositions, toward the characters are essential to viewer enjoyment.
How do we choose which characters to love and which ones to hate? Zillmann (2000) argued that when evaluating dramatic entertainment content, viewers are “untiring moral monitors” who continually make decisions about how morally correct or incorrect a character’s actions are (Zillmann, 2000). We come to like characters whose actions are deemed morally right and dislike characters whose actions are deemed morally wrong. Dramatic narrative structures often involve good overcoming evil (or lesser forms thereof), and we side with the forces of good. This moral monitoring is flexible; it can change as the characters’ motivations and behaviors in the program are revealed. Thus, some of this early work aligns with a Kohlbergian perspective, such that viewers’ moral judgments are the product of conscious moral cognition and reasoning.
However, Raney (2004) has questioned whether viewers always engage in conscious moral contemplation of characters’ behaviors and motivations. In his expansion of disposition theory, Raney (2004) reasoned that viewers are not devoid of information and anticipations when they sit down to watch a television program; rather, their preexisting story schemas “provide viewers with cognitive pegs on which to hang their initial interpretations and expectations of characters” (p. 354). In this account, the automatic use of story schemata sometimes precedes our moral judgments, and viewers of dramas can usually identify the “good guys” versus the “bad guys” (hence who to like) even before the story really unfurls. As such, this perspective parallels Haidt’s (2001, 2007) work and suggests that moral intuitions precede moral evaluations.
Raney’s (2004) extension of disposition theory further proposes that viewers’ desire to enjoy the program may sometimes cause them to engage in biased evaluations and attributions about the behaviors of liked versus disliked characters. In the pursuit of pleasure, they may excuse the bad behaviors of liked characters as temporary, nondiagnostic aberrations that do not reflect the characters’ more enduring positive traits, while judging the negative actions of disliked characters more harshly. Such favorable interpretations of liked characters’ negative behaviors may, in turn, have further consequences. Raney (2004) concluded his article by suggesting that the tendency to “stretch our moral code” in the media world may spill over into the real world (p. 364), resulting in greater tolerance of real life violence or even greater willingness to commit acts of aggression oneself.
The Current Study
To summarize, research has demonstrated that exposure to tween sitcoms increases youthful aggression but has not focused on moral reasoning as an avenue for such effects. However, there is evidence to suggest that media depictions sway young viewers’ moral judgments about aggression, and recent developmental theorizing together with recent formulations of disposition theory suggest that these judgments may be driven by viewers’ feelings about the characters. In contrast, earlier formulations of disposition theory propose that characters are liked or disliked based on the morality of their actions.
In the current study, we sought to unite these disparate bodies of literature and to examine the predictions afforded by different versions of disposition theory. How do young viewers reason about the aggressive actions of tween sitcom characters? Are there instances when their evaluations of depicted aggression are such that they might be encouraged to consider acting similarly? If tween viewers’ evaluations of aggressive actions are the product of conscious moral cognition and reasoning, their assessment of the level of aggression should predict their moral evaluation of the actions regardless of their feelings for the perpetrator (Kohlberg, 1969; Zillmann, 2000). However, if tweens’ evaluations are based on relatedness motives and predispositions toward characters (Haidt, 2001; Raney, 2004), then viewers may be more likely to excuse the aggressive actions of liked characters and to regard only those of less likable characters as wrong.
Obviously, both conscious judgments and dispositions may sway moral reasoning. We were interested in the relative strength of these two types of considerations on moral indifference, our main outcome. Based on the work of Bandura (2002) and of Krakowiak and Tsay-Vogel (e.g., Krakowiak & Tsay-Vogel, 2013; Tsay & Krakowiak, 2011), we conceptualized this outcome as involving tolerant evaluations of intentional harm/injustice combined with apathy about the victim’s suffering. We see it as related to, but not synonymous with, the much more complex concept of moral disengagement. The latter, as laid out by Bandura (2002), involves a variety of possible processes by which individuals engage in cognitive reassessment of inhumane conduct, disavow the actor’s agency, minimize harm suffered, and dehumanize or blame the victim.
Thus, we were focused on examining moral indifference as a response to depictions of characters intentionally inflicting social and physical pain on one another. We conceive of this project as situated within the particular developmental period (tween, preadolescent) that forms the target audience for these programs, rather than focusing on age comparisons. Based on the theorizing of Haidt (2001) and recent formulations of disposition theory (Raney, 2004), we predicted the following hypotheses:
Alternatively, what if viewers do not like the perpetrator (e.g., if the aggression is committed by a villainous antagonist), but instead like the victim? Zillmann and Cantor (1977) found that second and third graders felt good when a morally bad protagonist suffered, but they responded empathically to both positive and negative outcomes for morally good or ambiguous protagonists. In subsequent writing, Zillmann (1994) argued that character liking “virtually controls empathy” (p. 44).
If liking the victim of aggression leads to empathic pity for the victim’s sufferings rather than empathic pleasure at the perpetrator’s actions, then moral indifference should be averted. Accordingly, we predicted the following hypothesis:
The Potentially Complicating Role of Perceived Funniness
One possible limitation of the above reasoning, and of most prior research on tween sitcoms, is that they ignore the fact that these programs are intended to be comedic. For example, in one clip that we used, a perpetrator is shown dunking her friend’s face in a bowl of soup; such a portrayal is meant to induce mirth, not outrage. Zillmann and Cantor (1972) and Raney (2003) argued that audiences treat moral considerations in comedic situations quite differently than dramatic ones. Following Freud (1989), humor scholars (Meyer, 2000; Zillmann & Stocking, 1976) contend that the structure of intended humorous communication—the linguistic and situational elements known as “joke work”—provides the audience with a measure of moral amnesty to find pleasure in the disparagement of another. That is, humans recognize situations as funny by various common elements they contain, permitting them to treat the actions and motivations therein as nonserious from a moral or social perspective. One such element is the laugh track, which is inserted liberally throughout situation comedies such as tween sitcoms. Thus, joke work may serve as an externally offered moral disengagement cue in media content (Bandura, 2002; Hartmann, Krakowiak, & Tsay-Vogel, 2014).
There is some indication that funniness serves to alter adolescents’ moral responses to real-world instances of aggression. In two experiments, Jones, Newman, and Bautista (2005) found that sixth and eighth graders used contextual cues such as level of friendship between perpetrator and victim to decide whether examples of insults (about weight, appearance, and academic performance) were funny and “joking” rather than “mean.” They argued that participants used intended funniness to distinguish teasing (seen as relatively acceptable) from harassment and bullying.
However, evidence for the effects of perceived funniness on evaluations of media depictions of aggression is mixed. On the one hand, Snow (1974) found that only 16% of 9- to 12-year-olds thought that a clip from a violent, funny cartoon (The Road Runner Show) contained violence, compared with 70% of those who watched a Western and 100% of those who watched news about the Vietnam War. In contrast, Haynes (1978) found that fifth and sixth graders perceived a funny cartoon to be more violent than a nonviolent cartoon, and King (2000) found that funny wisecracks by the hero of an action film increased (rather than decreased) women’s distress while viewing. Finally, Mares, Braun, and Hernandez (2012) examined fifth graders’ responses to physical and verbal conflict in tween sitcoms. They found that participants rated low- and high-conflict episodes as equally amusing; conflict neither detracted from (nor enhanced) ratings of how funny the content was. Thus, there is currently little consensus on the role of perceived funniness in responses to media depictions of aggression, and there is no systematic examination of the role of perceived funniness for moral evaluations of depicted relational aggression as well as physical aggression.
To the extent that viewers like the perpetrator of the depicted aggressive acts and can find the acts funny rather than excessive or upsetting, they should demonstrate higher levels of moral indifference. Accordingly, we predicted the following hypothesis:
Extending the Causal Chain
What about the final link in the causal chain? Bandura (2002) argued that moral disengagement in response to violent acts increases the likelihood of imitation by decreasing internal inhibitions. In our study, if liking the perpetrator increases moral indifference, that in turn could increase the likelihood of engaging in such acts. Based on the reasoning behind H3, we would further predict that this indirect relationship would be moderated by perceived funniness.
Method
Design
This was initially conceived as an experiment with a 2 (protagonist vs. antagonist) × 2 (social vs. physical aggression) × 2 exemplar (iCarly vs. Victorious) design. Participants were randomly assigned to one of eight conditions with gender balanced in each one. As discussed below, unanticipated exemplar interactions emerged, shifting the focus of hypotheses and analyses to the effects of subjective perceptions, controlling for experimental condition.
Stimuli
We chose two situation comedies in which the protagonist or antagonist was shown engaging in either social or physical aggression. iCarly (2007-2012) and Victorious (2010-2013) are popular teen sitcoms that continue to air on Nickelodeon. Both shows are popular with tween audiences (Gorman, 2010a, 2010b). Twelve clips from 11 episodes that featured either a protagonist or antagonist perpetrating either social or physical aggression were selected for pretesting. The episodes were edited from 24 minutes to an average of 12 minutes by removing the subplots and shortening the opening and closing credits.
Pretesting of stimuli
The 12 clips were pretested with 141 undergraduate students (53% female) in introductory communication classes who were offered extra credit to watch one randomly assigned clip. We used undergraduates because their availability allowed for repeated pretesting as the stimuli were being evaluated and selected. Although we expected that there might be main effects of age for various ratings, we also expected that undergraduates would show the same pattern as tweens in responding to the various clips (i.e., no age by condition interactions), rendering them suitable as a sample for evaluating which clips to retain and which to reject.
Selection of clips for experimental conditions
The clips for the study were selected to achieve statistically significant differences in ratings of the perpetrator’s social and physical aggression (for social vs. physical conditions), and in ratings of how likable the perpetrator was (for protagonist vs. antagonist conditions). Clips had to be equivalent across all four conditions in how funny they were. Based on these criteria, the following clips were selected.
Social aggression by protagonist
In the iCarly episode (“iKiss”; Season 2, Episode 10), Sam overhears Freddie confess to Carly that he has never been kissed, and Sam reveals the secret on Carly’s web show. Sam’s actions result in Freddie being ridiculed at school. In the Victorious episode (“Freak the Freak Out”; Season 1, Episode 13), Cat and Jade enter a karaoke contest at a local club but lose to the club owner’s daughters. When Cat and Jade learn that the contest was fixed, they seek help from Tori to win the next contest. The girls embarrass their rivals by tricking them into another contest with Tori performing lead vocals. Tori’s performance turns the karaoke audience against the daughters, who get booed off stage.
Social aggression by antagonist
In the iCarly episode (“iReunite With Missy”; Season 2, Episode 20), Carly’s old friend Missy comes back to town. Jealous of Sam’s relationship with Carly, Missy attempts to poison their relationship by excluding her from activities with Carly. In the Victorious episode (“Pilot”; Season 1, Episode 1), Tori is a new student at her performing arts high school. Jade’s boyfriend befriends Tori. A jealous Jade retaliates by humiliating Tori in their improv class.
Physical aggression by protagonist
In the iCarly episode (“iMake Sam Girlier”; Season 2, Episode 16), Sam tries to be more feminine than usual to attract a boy, but when the school bully starts to pick on Carly and physically hurt her, Sam tackles the bully to the floor and beats her up. In the Victorious episode (“Jade Dumps Beck”; Season 1, Episode 5), Robbie is reviewing Trina’s one-woman show for the school news blog. The show is terrible, and Robbie is honest about it in his review. Trina physically harms Robbie by shoving his face into a bowl of soup. She throws his desk across the room and physically threatens him throughout the show to ensure he writes a good review.
Physical aggression by antagonist
In the iCarly episode (“iRescue Carly”; Season 7, Episode 5), Carly goes to a “smash party” hosted by Dana, one of Sam’s old juvie friends. Carly and her two friends are tied up and have plates and glasses thrown at them until Sam comes and rescues them. In the Victorious episode (“Crazy Ponnie”; Season 3, Episode 13), Cat accidentally waxes off Jade’s eyebrows. Jade physically attacks Cat, and Cat falls asleep in study hall and shaves off all her hair.
Participants
A total of 176 fourth to sixth graders (52% female) participated during the late spring or summer of 2013 (M age = 10.82, SD = 1.03; 30 fourth graders, 75 fifth graders, 71 sixth graders). This age range falls within the parameters of the target audience for tween programming (Levin, 2007). All participants were fluent in English. They were recruited from schools, summer camps, and community programs from two Midwestern cities. Only children whose parents gave consent participated.
Procedure
Children were randomly assigned to watch a clip on a laptop computer. This clip either contained physical aggression or social aggression that was committed either by a protagonist or antagonist. Then, a trained interviewer worked one on one with each child by reading the questions aloud and entering the tweens’ responses into the computer. The total time for each session lasted about 45 minutes. Afterward, the interviewer thanked and debriefed the child. Depending on the setting, the child received a gift card or the school received a small donation that was noncontingent upon the number of children participating.
Measures
The order of questions in the interviews was constant across viewing conditions. The interviewer asked about the participant’s age, perceptions of the perpetrator and victim, perceptions of the aggressive act including how justified and funny they thought the aggression was, likelihood of imitation, overall television viewing, and exposure to specific tween sitcoms. The measures are described below in the order of the hypotheses. Descriptive statistics are provided in Table 1.
Descriptive Statistics for Key Variables for the Entire Sample and Within Protagonist/Antagonist Conditions.
Indicates a significant difference by factor (e.g., protagonist vs. antagonist, or social vs. physical aggression), at p < .05.
Physical and social aggression ratings
Two questions asked how aggressive the perpetrator was: “Sometimes people are violent, like punching, slapping, and hitting. How violent is [the perpetrator] in this clip?” and “Sometimes people are socially aggressive, like trying to hurt people’s feelings by teasing or hurting people’s friendships with others. How socially aggressive is [the perpetrator] in this clip?” Response options ranged from (0) not at all to (4) very violent/socially aggressive.
Character liking
Perpetrator liking scores were generated by averaging responses to three questions that asked how much the participant liked the perpetrator, how funny the perpetrator was, and how nice the perpetrator was (α = .80). Response options ranged from (0) not at all to (4) very much. Victim liking scores were created from three parallel questions (α = .90).
Moral indifference
Moral indifference was measured with questions taken from previous research (Bandura, 2002; Krakowiak & Tsay-Vogel, 2013; Tsay & Krakowiak, 2011). We showed a picture of the most relevant socially or physically aggressive scene and asked, “How much did [victim] deserve it? How okay was it for [perpetrator] to do that? Did you feel bad for [the victim] in some parts?” Response options for these questions ranged from (0) not at all to (4) very much, with reverse coding for the question about feeling bad for the victim. Scores for the three questions were averaged to create an index of moral indifference (α = .85).
Likelihood of imitation
We asked the children, “How likely is it that you would do something like that?” Response options for this question ranged from (0) not at all to (4) very much.
Ratings of funniness of the aggression
The interviewer showed the child a picture of the most relevant aggressive scene and asked, “How funny was that part?” Response options for this question ranged from (0) not at all to (4) very much.
Television viewing
Television viewing was assessed with Harrison’s (2000) measure. Children were asked how many hours of television they watched yesterday (a) “in the morning before school”; (b) “during school, in class”; (c) “after school but before dinner”; and (d) “after dinner.” For Saturday and Sunday, they were asked, “how many hours of television do you watch” (a) “in the morning, before lunch”; (b) “in the afternoon, before dinner”; and (c) “after dinner.” Responses to the weekday questions were multiplied by 5 and added to the Saturday and Sunday responses, and averaged to create an index of daily television exposure.
Prior exposure to tween programs
Children indicated how often they watched iCarly and Victorious on a scale from (0) never, (1) hardly ever, (2) sometimes, to (3) often. Children also indicated how often they watched each of the following tween sitcoms: Wizards of Waverly Place, A.N.T. Farm, Jessie, Ned’s Declassified, Austin & Ally, Shake It Up, Good Luck Charlie, So Random!, and Drake & Josh. These programs were chosen because they were popular with the tween viewing audience and likely to feature both physically and socially aggressive interactions by protagonists and antagonists (Mares et al., 2012; Martins & Wilson, 2012a). Responses options for these and the experimental programs were summed to create a tween viewing score for each participant (α = .91).
Results
Comparing Undergraduate and Tween Evaluations
To evaluate our assumption that there would be no age by condition interactions (which would have invalidated our choice of undergraduates for pretesting the clips), we conducted a 2 (protagonist vs. antagonist perpetrator) × 2 (social vs. physical aggression) × 2 (iCarly vs. Victorious exemplar) × 2 (tweens vs. undergraduate) MANOVA. The four dependent variables that we measured with both groups were liking of the perpetrator, ratings of perpetrator’s level of social and physical aggression, and funniness of the aggression. As expected, this analysis revealed a significant multivariate main effect of age, Wilks’s Λ = .80, F(4, 314) = 18.07, p < .01,
Analytic Strategy for Tween Experiment
We began by confirming random assignment to condition and found no significant differences in gender, age, overall TV viewing, or prior exposure to the programs. Then, to assess the influence of character type and aggression type on children’s responses to aggression, we began by conducting a 2 (protagonist vs. antagonist) × 2 (social aggression vs. physical aggression) × 2 (iCarly vs. Victorious exemplar) MANOVA. Children’s moral evaluation of the aggressive scene served as the dependent variable. Unfortunately, for the majority of outcomes, there were significant interactions between exemplar condition and one or both of the other experimental factors (aggression type and likability). Such interactions suggested that experimental effects of character type and aggression conditions were at least partly attributable to specific characteristics of individual clips rather than generalizing across exemplars. As such, they posed serious challenges to interpretation of the experimental effects.
Given these issues, we shifted toward a regression approach to examine the influence of participants’ subjective ratings while controlling for the three experimental factors, all the two-way interactions, and the three-way interaction. This procedure allowed us to tease apart the variance explained by experimental conditions (including potential confounds of character condition and aggression type with exemplar) from the variance explained by children’s subjective perceptions of how physically and socially aggressive the perpetrator was, and the extent to which they liked the perpetrator or victim. For example, we could examine whether, controlling for experimental assignment to condition (and all other main effects and interactions), greater liking of the perpetrator was associated with a more lenient evaluation of the aggression.
Based on these considerations, we ran a series of hierarchical regression analyses to predict children’s moral indifference. In Step 1, we entered participant sex, age, exposure to tween television series, and overall television exposure to control for these individual differences. In Step 2, we entered the experimental main effects and interaction terms. In Step 3, we entered our focal predictors or the tweens’ subjective ratings.
We tested three models to predict children’s moral indifference. In Model 1, we examined perceptions of intensity of perpetrator aggression as well as ratings of perpetrator likability (H1). In Model 2, we examined perceptions of perpetration aggression and ratings of victim likability (H2). In Model 3, we examined perceptions of funniness and perpetrator likability (H3).
As shown in Table 3, we began by running all three models within the entire sample. By controlling for experimental factors and the two- and three-way interactions, we were able to examine the roles of aggression level, character liking, and funniness, over and above effects of perpetrator type, aggression type, and plot details idiosyncratic to particular exemplars. Next, we ran the three models within the subgroups of those who saw a protagonist or an antagonist commit the aggression to examine whether the processes of giving moral judgments differed for these two character types. Perhaps severity of aggression would be more relevant to moral evaluations of antagonists’ actions, and liking would be more relevant to moral evaluations for protagonists. Finally, we ran the three models yet again, within the subgroups of those who saw physical or social aggression, to examine whether the predictors of moral indifference were the same for both forms of hostility.
Predicting Evaluations by Gender, Age, Prior Viewing, and Experimental Conditions
Table 2 shows the coefficients and R2 change for the first two steps, which were identical across all three models. As a set, none of the individual differences entered on Step 1, including age, significantly predicted children’s moral indifference. Step 2 contained the main effects and interactions of the experimental conditions. As shown in Table 2, there were numerous significant interactions between the experimental conditions and with exemplar, making it inappropriate to interpret the main effects or two-way interaction between liking and aggression type. The coefficients were large because the main effects and interaction terms were entered on the same step, but the goal of this step was to statistically control for these terms rather than to estimate their effects.
Predicting Moral Indifference by Sex, Age, Prior Viewing, and Experimental Conditions.
Note. Cell entries (apart from R2) are standardized coefficients. For Step 2, the coefficients are large because of the associations between the conditions. Coefficients and their associated significance are reported from the blocks in which that set of variables was first entered.
p < .05. **p < .01. ***p < .001.
H1: Perceived Social and Physical Aggression, Perpetrator Liking, and Moral Indifference
H1a posited that moral indifference would be negatively predicted by perceptions of the perpetrator’s levels of social and physical aggression. The results of these tests are summarized in Model 1, Table 3. In the overall sample, neither perceived physical aggression nor social aggression significantly predicted moral indifference. Within the antagonist experimental condition, however, perceived social aggression was negatively related to moral indifference. Perceived physical and social aggression did not significantly predict moral indifference in any of the other experimental conditions. Thus, H1a was not supported.
Prediction of Moral Indifference by Character Aggression, Liking, and Perceived Funniness.
Note. Steps 1 and 2 (same for all models) in Table 2. Cell entries (apart from R2) are standardized coefficients. Coefficients and their associated significance are reported from the steps in which that set of variables was first entered. The coefficients for the first two steps are shown in Table 2. SA = ratings of how social aggressive perpetrator was; PA = ratings of how physically aggressive perpetrator was.
p < .05. **p < .01. ***p < .001.
H1b posited that tweens’ moral indifference would be predicted by liking of the perpetrator. As can be seen from Model 1 in Table 3, more liking of the perpetrator was associated with more indifference. This significant relationship was observed both in the whole sample analyses and within experimental comparisons, except for when the perpetrator was the antagonist, lending partial support to H1b.
H2: Victim Liking, Perceived Level of Social and Physical Aggression, and Moral Indifference
H2 predicted that moral indifference would be negatively predicted by liking of the victim. As can be seen in Model 2, Table 3, there was no support for H2 either within the complete sample or within each experimental condition.
H3: Perceived Funniness, Character Liking, and Moral Indifference
H3 predicted an interaction between liking of the perpetrator and ratings of perceived funniness, such that moral indifference would be highest when liking is high and perceived funniness is high. The hypothesized interaction between liking the perpetrator and perceived funniness was significant in the whole sample analysis and in three of the four experimental conditions. These interactions were probed using Hayes’s (2013) PROCESS macro to estimate the simple slopes of the relationship between perpetrator liking and moral indifference at three levels of perceived funniness (−1 SD below, the mean, and +1 SD above), including the same set of control variables as well as the other interaction terms in the analysis. The simple slope analysis conducted for the sample overall revealed that when perceived funniness was high, there was a significant positive relationship between character liking and moral indifference (+1 SD: b = 1.27, p < .05). There were no significant relationships at low or average levels of perceived funniness. The same pattern emerged within the protagonist experimental condition. At high levels of perceived funniness, there was a positive relationship between character liking and moral indifference (+1 SD: b = 0.39, p < .05). This relationship was not significant at low or average levels of perceived funniness. Despite the significant interaction terms in the hierarchical regression analysis within the social and physical aggression conditions, the simple slope analysis revealed no significant differences at three levels of perceived funniness (−1 SD, the mean, and +1 SD). Thus, H3 was partially supported.
H4: A Causal Chain From Liking to Imitation, via Concern Moderated by Perceived Funniness
H4a and H4b predicted a causal chain moderated by perceived funniness such that at high levels of perceived funniness, liking of the perpetrator would promote moral indifference; this in turn would predict likelihood of imitation. We tested a moderated mediation model using the PROCESS macro (Model 7) by Hayes (2013). We entered perpetrator liking as a predictor, moral indifference as our mediator, perceived funniness as the moderator, and likelihood of imitation as the outcome variable. In addition, we included the same set of control variables as well as the other interaction terms in the analysis. The results indicated the hypothesized moderated mediation. The mediator variable model revealed a significant character liking by perceived funniness interaction predicting moral indifference (b = 2.96, p < .05), and the dependent variable model revealed a significant main effect of moral indifference on likelihood of imitation (b = −2.74, p < .05). The significance of the indirect effect of moral indifference was verified across each level of perceived funniness with bootstrapped standard errors and 95% confidence intervals (CIs). As predicted, the indirect effect of moral indifference was significant among tweens at average (b = 0.03, SE = .02) and high levels (b = 0.08, SE = .04) of perceived funniness (because the limits of the 95% CI did not include zero), but not among tweens at low levels of perceived funniness (see Figure 1). Thus, H4 was supported, consistent with the argument that liking the aggressive perpetrator stretched tweens’ moral codes; in turn, this resulted in greater likelihood of imitating the aggression (H4a), especially when tweens perceived the aggressive scene to be funny (H4b).

Moderated mediation analysis for character liking, moral indifference, and imitation, at three levels (low, average, and high) of perceived funniness. The numbers in the parentheses are standardized coefficients.
Discussion
The goal of this study was to examine how tween audiences interpret the conflicts and hostilities that are the hallmarks of tween sitcoms. We sought to build on prior research by examining how moral evaluations of the characters’ actions are swayed by features of the content, including severity of aggression, likableness of the characters, and funniness. Up until now, research on tween sitcoms has focused primarily on measuring exposure rather than viewers’ interpretations, and on the multistep implications of exposure for attribution biases and aggression. We sought to focus more closely on immediate evaluations and responses as an additional possible route toward negative effects.
We turned to the various formulations of disposition theory for predictions about young viewers’ moral judgments. If preadolescents were continuously and consciously monitoring characters’ aggressive behavior as Zillmann (2000) proposed, their moral evaluations of the aggressive scenes (and, in turn, their perceptions of possibly imitating such actions) should be predicted primarily by their perceptions of the level of aggressiveness. From this perspective, we anticipated (H1a) that adolescents’ perceptions of the perpetrator’s level of social and physical aggression would negatively predict moral indifference. Our results show that, with a few exceptions, children’s subjective ratings of the extent of social and physical violence were largely unrelated to moral judgments about the acts.
In contrast, these moral evaluations were significantly predicted by ratings of how much the viewer liked the aggressive perpetrator (H1b), both across the whole sample where we controlled for all experimental effects and within each type of aggression (relational or physical). This pattern, whereby character liking (rather than severity of behavior) predicted moral evaluations, is consistent with Haidt’s (2001) assertion that moral judgments are guided by quick, emotion-based intuitions (including relatedness motives), rather than being rational and nonpartisan. As such, it also bolsters Raney’s (2004) claim that viewers are cognitive misers who rely on simple genre cues about who is nice and who is to be considered nasty, and that these affective dispositions toward characters alter evaluations of their actions.
The one condition where character liking did not predict moral indifference (and where perceived level of social aggression was a significant predictor) was when the perpetrator was an antagonist. In that condition, character liking was low and showed relatively little variance (see Table 1), leaving little chance of covariation between level of liking and moral indifference. As a whole, the results point to the power of siding with a character in shaping tween viewers’ responses to the characters’ aggressive acts.
Do these responses simply reflect a preconventional rather than conventional level of moral development (Kohlberg, 1969) focused on good or bad consequences for the self? Were these viewers simply too immature to care about aggressive acts toward media characters? If that were the case, there should be little variability in moral indifference, but that was not what was observed. Rather, the range of scores indicated that some children saw the aggressive actions as not okay, as hurtful, and as unjustified by preceding events, and this variability corresponded with liking of the character. To illustrate, consider the following responses made by preadolescents who liked the perpetrators. When asked whether they thought it was “okay” for the perpetrator to use aggression against the victim, tweens who liked the perpetrators typically said, “Yes because [the victim] really deserved it” or that the victim was “being a jerk.” On the other hand, preadolescents who did not like the antagonist answered this question by saying things like “Two wrongs don’t make a right; the [perpetrator] did not have to be mean” and “It’s not okay because violence is never the answer.”
An additional consistent finding that emerged from these data concerned the null effects for victim liking (H2). This is not to say that dispositions toward the victim are unimportant or do not matter; however, feelings about the perpetrator seemed to be more salient. It could be that tween sitcoms minimize perspective-taking and foster simplistic depictions of conflict. Future research should examine whether these findings can be replicated with dramas, a genre that is known for depicting conflicting sides of conflict (e.g., both perpetrator and victim perspectives) within the narrative.
We predicted an interaction between liking of the perpetrator and perceived funniness such that moral indifference would be highest when liking is high and perceived funniness is high (H3). This hypothesis was supported for the overall sample as well as within the protagonist experimental condition. As outlined in the introduction, Raney argued that viewers sometimes form alliances with characters based on heuristics and story schemata, rather than the character’s moral goodness. If a liked protagonist then behaves in negative ways, viewer enjoyment may be jeopardized because the viewer (if rational) should start assigning blame to that liked character. In the context of a sitcom, it may feel problematic to keep laughing when a favorite character behaves like a jerk. One method of preserving enjoyment may be to enlist various forms of moral indifference; if no harm was done, or the victim deserved it, then the viewer is free to laugh. Indeed, it appears that the “joke work” provided in tween sitcoms gives the audiences a measure of moral amnesty to find pleasure in the ridicule of another, particularly when the disparagement was at the hands of a liked protagonist. The results of this study replicate and extend prior work in this area by considering the moral evaluations of youths (rather than adults) and by examining evaluations of social as well as physical aggression.
Finally, we found support for the causal chain suggested in Raney’s (2004) expansion of disposition theory (H4). Consistent with Bandura’s (2002) reasoning that moral disengagement in response to violent acts increases the likelihood of imitation, we found that protagonist liking predicted moral indifference when the aggression was perceived as funny, which in turn predicted self-reported likelihood of imitating the aggression. Future research should continue to examine the proposition that moral indifference plays a role in the enjoyment and imitation of immoral behaviors, particularly among young viewers.
These are novel findings for this area. The results suggest an additional route to antisocial responses (moral indifference as well as previously established effects on attribution biases) and help establish boundary conditions for this additional route. At least with regard to moral indifference, it appears that overall amount of exposure to aggression in tween sitcoms may be less important than the degree to which the aggression is carried out by likable characters and is perceived as funny. To the extent that it is depicted as enacted by unlikable antagonists and/or is not perceived as funny, it does not appear to promote moral indifference or perceived likelihood of imitation.
In addition, these findings contribute to work on disposition theory, by examining the predictions in the context of a tween audience; by adding evidence in support of more recent formulations of the theory; by providing an initial, exploratory test of the proposed causal chain; and by considering the moderating role of funniness. Consistent with recent versions of disposition theory (Raney, 2004) and moral reasoning (Haidt, 2001), these data suggest that affective dispositions toward characters, rather than perceived aggression, were the primary determinants of moral evaluations. Young viewers were indifferent to the aggression of liked characters while condemning the aggression of disliked characters. These responses, in turn, predicted their self-reported likelihood of committing such acts of aggression themselves—an outcome that is predicted by disposition theory but has not often been examined directly. Moreover, the perceived funniness of the aggressive act moderated this causal chain, strengthening the indirect effect of liking on willingness to imitate.
As with any study, there are limitations to acknowledge. One major issue was that we measured likelihood of imitation (and perceived funniness) with one-item measures. An obvious flaw in the use of single-item measures is that they cannot yield estimates of internal consistency (Wanous, Reichers, & Hudy, 1997); therefore, it is not possible to estimate error. However, studies on other constructs such as self-esteem, for example, have compared single-item with multiple-item measures and have found satisfactory correlations between the different measures (e.g., Gardner, Cummings, Dunham, & Pierce, 1998; Wanous & Reichers, 1996). Likewise, single-item measures are beneficial for researchers interested in an overall impression of the construct of interest (for some recent examples, see Balmas, 2012; Weenig, Wilke, & ter Mors, 2011). Practically speaking, the single-item measures were a simple way to assess some of these core measures without overtaxing the children in our sample. Nonetheless, we present the findings of effects on self-reported likelihood of imitation as exploratory and in need of replication and extension with stronger measures. Ideally, such further work would include both hypothetical statements of imitation and measures of behavior. In reviewing the literature on cartoon violence, Kirsh (2006) noted that researchers failed to find significant effects of funny cartoon violence on elementary school children’s aggressive behaviors, relative to control groups (Hapkiewicz & Roden, 1971; Hapkiewicz & Stone, 1974) but did find effects on boys’ responses to hypothetical questions about their future actions (Nathanson & Cantor, 2000). Thus, further work is needed to determine the links between reported possibilities and actual behaviors.
In addition, the significant interactions between exemplar condition and one or both of the other experimental factors (likability and aggression type) posed significant challenges to the interpretation of experimental effects. Although we are confident that we were able to examine differences in moral reasoning using a regression approach, an important next step would be to replicate the findings in an experiment that can test the effects of both liking and aggression across exemplars. A second limitation is that we examined preadolescents’ moral evaluations of characters only in the context of tween sitcoms. We picked this particular genre of programming because of its popularity with this particular audience. Nonetheless, future work should examine whether the findings reported here can be replicated in studies examining other television genres.
Despite these limitations, this study builds on previous research by highlighting the importance of affective dispositions in preadolescents’ moral judgments of liked and disliked characters. Although this study produced little support for the idea that tweens are engaging in careful moral monitoring of television characters, future study is needed to understand how dispositional valence is initially established and what other factors may play a role in its formation. Given that these sitcoms routinely feature liked protagonists who perpetrate social and physical aggression (Martins & Wilson, 2012a), such work would be fruitful in helping researchers understand routes by which violent depictions may encourage young people’s moral indifference, thereby increasing the probability of aggressive responses.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Funding for this project came, in part, from the Hamel Family Research Fund administered by the Department of Communication Arts at the University of Wisconsin-Madison.
