SUPER GOOD OR SUPER BAD?
MEGAN JANSEN2
Super good or Super bad?
An experiment exploring the effects of the series Supergirl on prosocial behaviour and aggressive attitudes of late adolescents
Megan Jansen 11406577
Master’s Thesis
Graduate School of Communication
Master’s programme Communication Science: Entertainment Communication
S.R. Sumter, date of completion: 02-02-2018
3 Abstract
Up to now, the influence of superhero entertainment media, and female superheroes in particular, has been understudied. Existing literature has resulted in contradicting findings regarding the effect of superhero content. On the one hand studies find that superhero content invokes more prosocial behaviour, other studies find that superhero content leads to more aggression. Therefore, the current study looks at positive and negative effects of superhero content on adolescent viewers. An experiment was conducted among 156 Dutch 16-19-year-olds. The experiment had three conditions (violent Supergirl, non-violent Supergirl and the control condition), which generated mixed results. No effect of the violent Supergirl video on aggressive attitudes was found. The non-violent video resulted in lower levels of altruistic intentions and when moderated by gender, only defending behaviour was affected. For defending behaviour girls scored high overall except in the non-violent condition and boys scored higher in the non-violent condition compared to the control condition. Character identification did not have a moderating effect. The findings suggest that individual differences play a role in the effect of media content.
4
Super good or Super bad?
An experiment exploring the effects of the series Supergirl on prosocial behaviour and aggressive attitudes of late adolescents
Over the past few years Marvel has released a large number of superhero movies and even more are coming. In 2018 alone ten more superhero movies will be released
(boxofficemojo.com, 2017). On TV, series like Arrow and The Flash have become very popular with a wide variety of new shows and seasons coming out almost every month
(Bacon, 8 august 2017). In short, superhero content has exploded and an increasing number of people are viewing superhero based stories.
The popularity of superhero content has been ascribed to different explanations. One of these explanations for the popularity of superhero content is that in uncertain and
dangerous times superheroes offer support (Free, 2016; Hagley & Harrison, 2014). The terrorist attacks of 9/11 have started an increasing awareness of terrorism and in more recent years the attacks of ISIS have caused a lot of fear. It has been argued that in such uncertain times people have a need to feel safer, superhero content could provide this feeling of safety. Superheroes use violence to defy villains, which is often accepted by viewers and not seen as treating. Research on why this superhero violence is accepted by viewers revealed that the reasoning of consumers is often more complex than the awareness of the violence being fake. In short, violence by superheroes and some other fictional characters is not just accepted because ‘it is just a movie’ or ‘it is just a game’ (Hartmann, Toz, & Brandon, 2010; Hartmann & Vorderer, 2010). More in depth research found that the reasons why superheroes use
violence plays a role in the acceptability of the viewers (Krakowiak & Tsay-Vogel, 2013). Violence is deemed okay if the superhero uses it with a good moral reasoning, for instance to prevent greater damage done by the villain. Thus, the viewers are okay with violence from superheroes if the moral reasoning behind it is sound and for a good cause (Krakowiak &
5
Tsay-Vogel, 2013; Hartmann, Toz, & Brandon, 2010). Superhero stories present viewers with realistic storylines, a clear beatable villain and a superhero that uses acceptable violence to defy the villain. The superhero story presents the viewers with relatable content, i.e. evil that needs defeating, in a controlled setting, the superhero movie. For the above mentioned
reasons, superhero content is very enjoyable to watch in the current unstable time (Free, 2016; Hagley & Harrison, 2014).
The majority of superhero movies star a male protagonist. However, with the recent release of Wonder Woman, Marvel offers today’s mainstream audience a female superhero. Wonder Woman was certainly not the first female superhero, female superheroes have been around for quite some time. The Cat being one of the first female superheroes Marvel introduced. Up to now, female superheroes were often cast as sexy sidekicks to the male superhero of the stories, they were dressed in very provocative outfits and a lot of focus was placed on specific parts of their bodies like their breasts and buts (Peppard, 2017).
In recent years, however, female characters have been portrayed in a more positive way. Katniss Everdeen in The Hunger Games (2012) is a good example of a strong female protagonist. Katniss makes her own choices based on the morals she believes in. This positive portrayal of female characters has also spilled over in the superhero world; in Wonder Woman Diana is a superhero in her own right rather than a sidekick to a male superhero. The release of Wonder Woman seems the beginning of a lot more female superhero content, like Dark
Phoenix (2018), Wonder Woman 2 (2019), Gotham City Sirens (2020) and of course the
female superheroes that are part of superhero teams like Black Widow and Scarlet Witch in the new (still untitled) instalment of The Avenger saga (2019), and Harley Quin in Suicide
Squad 2 (TBA).
Although superhero content has become increasingly popular, little is known about how superhero content influences its viewers and even less is known about female
6
superheroes. Recently, there has been a call for more research on exposure to superhero content, as it is argued that it is possible that viewing “the helping and triumphing superheroes inspires adolescents to help others” (de Leeuw & Buijzen, 2016, p. 42). This assumption is based on studies which indicate that the prosocial behaviour displayed by superheroes can cause viewers to be more altruistic (Rosenberg, Baughman, & Bailenson, 2013). However, research on media violence would suggest that the aggressive behaviour superheroes display could potentially result in aggression rather than altruism among viewers (Huesmann, Eron, Klein, Brice, & Fischer, 1983). Up to now, it is unclear whether there are positive or negative effects of viewing superhero content. Most current studies seem to find conflicting results and have failed to address the upcoming popularity of female superheroes or if the gender of the superhero, or the viewers, plays a role in the effect of superhero media consumption. For this reason the main aim of the study is to answer the following question:
Does exposure to female superhero content have positive or negative effects on late adolescents?
Contemporary theorizing on media effects has pointed out that to better understand how people are influenced by media, different processes like character identification need to be considered. Thus, the second aim is to view whether character identification moderates the effect of superhero content. Lastly, since the gender of the superhero is a key point in this research the third aim is to explore effects of the respondents’ gender on the main effect. These two possible moderating effects will be explored separately, for both the aggressive (negative) and the prosocial (positive) outcomes. The study is conducted among late adolescents as they enjoy and are able to understand more morally complex characters (Valkenburg & Piotrowski, 2017), and during this age adolescents start to learn and develop their own values in life. In addition, adolescents of 12-18 years old are the primary target audience for superhero content (Macdonald, 2012).
7 Superhero content: a negative or positive effect?
Although research on the effects of superhero media is limited, at least seven studies have investigated the relationship between superhero content and a variety of positive and negative outcomes. The results of these studies will be discussed below.
One set of studies has investigated the positive effects of exposure to superhero media, these effects included increasing prosocial behaviour (Rosenberg, Baughman, & Bailenson, 2013; Peña & Chen, 2017), reducing aggressive attitudes (Huesmann, Eron, Klein, Brice, & Fischer, 1983), and defending behaviour against bullying (Coyne et al., 2017). With regard to prosocial behaviour, two studies showed that when participants were primed with superhero stimuli this resulted in participants showing more helping behaviour and more prosocial behaviour. One study did an experiment, in this experiment college aged participants were asked to play a video game in which they had the power to fly. In the game they were tasked with helping to find a child that needed medical attention (Rosenberg, Baughman, &
Bailenson, 2013). The participants that played the video game in this superhero setting showed more helping behaviour towards the researcher, who would drop pens after the
participants played the game, compared to the group that had a visual tour though the game in a touring bus (Rosenberg, Baughman, & Bailenson, 2013). In addition, Peña and Chen (2017) did a similar study. In this study 412 undergraduate students (81% female) participated in an experiment where they first were presented with five attributes of either superheroes,
supervillains or a dorm room for the control group (Peña & Chen, 2017). After being primed to one of the three conditions the participants were asked to play a game. During the game the participants were also asked to stay in a specific physical position that was expected to
influence the experiment. After this game the researcher dropped pens. The participants who had been exposed to the superhero condition showed more helping behaviour towards the researcher dropping the pens (Peña & Chen, 2017).
8
In regards to reducing aggressive attitudes, Huesmann et al. (1983) conducted an experiment among 169 first and third grade children. This experiment had the intention to reduce the effects of TV violence. The children were exposed to different hero stories, which were divided by boy and girl shows (Huesmann et al., 1983). In addition to exposure to a hero story the children participated in training sessions that were meant to reduce behavioural copying and make the children aware of special effects that could influence the storyline. After the experiment it was found that the experimental group was described to be less aggressive by peers. Violence on TV also affected the children less (Huesmann et al., 1983).
With regards to defending behaviour, a longitudinal study among 240 preschool children (49% male) and their parents, looked at the effect superheroes had on defending behaviour of children when it came to bullying (Coyne et al., 2017). The parents had to fill out a survey about the use of media of their children and the outcomes. In contrast to the studies above, this study demonstrated that superheroes had no effect on prosocial and defending behaviour (Coyne et al., 2017). The study did however find that the exposure to superhero content did relate to more relational and physical aggression (Coyne et al., 2017). In contrast to the earlier mentioned studies, this study shows absence of positive effects and that negative effects of superhero content are also possible, these negative effects will be discussed next.
Thus, although positive effects were found in some studies, other studies argue that superhero content can cause negative effects too. A study that explored the relationship of superhero powers and the effects of these powers provided some interesting insights. In the study 572 respondents were presented with the hypothetical situation of having superpowers (Das-Friebel, et al., 2017). The respondents could choose between six kinds of superpowers, three positive and three negative. The respondents had to indicate what they would do with the powers, use them for either altruistic benefits or for selfish reasons. Results suggested that
9
unlike superheroes themselves the respondents (94%) would use these powers for personal gain, men were more likely to want both positive and negative powers and women were more likely than men to want to use powers for personal gain (Das-Friebel, et al., 2017). Unlike the previous discussed studies were the gain of a superpower brought our more prosocial
behaviour, this study proves that superpowers would not always bring out the best in people. Other studies have researched the effect of violence present in entertainment media, for instance video games, to see if the mere presence of violence was enough to cause
aggressive behaviour. Konijn, Nije Bijvank and Bushman (2007) did a study on the effects of violent video games on 112 Dutch adolescent boys. This research revealed that identification with a violent game character made the respondents more aggressive (Konijn, Nije Bijvank & Bushman, 2007). Heightened aggressive behaviour after exposure to violence media content is often caused by the process of aggression imitation. The weapon effect is a phenomenon that also invokes more violent behaviours. A meta-analysis combined several studies that investigated the effects weapons have on aggressiveness (Benjamin, Kepes, & Bushman, 2017). The analysis stated that weapons are associated with aggression and that the use of weapons in media primes negative thoughts which causes the viewers to be more aggressive (Benjamin, Kepes, & Bushman, 2017). It does not matter if a ‘good’ character of ‘bad’ character uses the weapons, the presence of weapons alone causes this negative priming effect. Superhero content regularly uses weapons, and thus it is very plausible to assume that superhero content involving weapons will make viewers more aggressive.
The contradictory findings of the studies above might be related to the amount of media violence that was part of the stimuli. In the positive effects studies there are conditions of superheroes without violence, in these studies the participants ‘are a superhero’ and then do good, this gives the positive effects. In the negative effects studies the participants are
10
the violence is present. No study has yet looked at whether the presence or absence of
violence in the content has an effect, this will be specifically tested in this study. Considering the contradicting results of the studies about the positive effects of superhero content on the one hand and the negative effects of superhero content on the other hand the next hypotheses were formulated (a conceptual model of can be found in the Appendix):
H1a: Supergirl content with non-violent superhero behaviour results in higher levels of altruistic intentions than the violent and control group.
H1b: Supergirl content with violent superhero behaviour will cause more aggressiveness than the non-violent and control group.
The effect of Supergirl on both the violent and non-violent outcomes, is most likely not going to be a direct effect. Across the width variety of concepts that can influence whether there is a (positive or negative) effect of Supergirl on adolescents, this study will focus on character identification and gender. Both gender and character identification are hypothesized to be moderators.
Character identification
Character identification has been identified as one of the processes that could possibly explain why some viewers are more heavily influenced by media content than others (Konijn, Nije Bijvank, & Bushman, 2007). This section will therefore focus on the concept of character identification, in its different forms, and specify which possible effects character identification can have on viewers.
Character identification is one of the different forms in which viewers can engage with media characters, others include wishful identification, parasocial relationships and imitation (Cohen, 2014). Character identification, however similar to these other concepts does have some specific features. First of all, character identification is a one-sided connection to the media character, a viewer experiences the identification logically a reaction from the media
11
character is not possible (Igartua, 2010). Secondly, and this makes character identification different from a parasocial relationship, is the level of immersion in the story. Forming a parasocial relationship with a character, means that the viewers feels like he or she has a bond with the character, like a friendship, this is also a one-sided experience (Hoffner, 1996). However, the experience of identification is not forming a one-sided bond, the differences is that identification makes the viewers feel as if they are the character. This immersion to the point where viewers feel like they themselves experience the story is often mentioned when describing character identification (Igartua, 2010). The central point of identification with a character is that, for a limited time, viewers forget that they are in the position of audience, and that the viewers feel as if they themselves are experiencing the story (Cohen, 2014; Igartua, 2010). The feeling of ‘being’ the character makes character identification different from wishful identification, which means that the viewers want to be like the characters because they feel some level of admiration for the characters (Greenwood, 2007). In short character identification in the most basic form is a one-sided experience in which the viewers are so immersed into the story that they feel like they are the character.
Viewers can identify with characters on different levels, physically, mentally and emotionally. Physical identification happens when a physical trait is shared with the character, like being of the same gender or having the same kind of distinct clothing styles (e.g., gothic). In this study physical identification, girls identifying more with Supergirl because of shared gender, is hypothesized to lead to more identification. So when character identification is mentioned, this implies the physical identification of shared gender. Character identification can be either similarity identification, were the viewers identifies with the characters based on shared traits, and wishful identification, were the viewer desires to be like the character (Konijn, Nije Bijvank, & Bushman, 2007). So, for example a girl can identify with Barbie because they are both female, this would be (physical) similarity identification. But she could
12
also want to be like Barbie because she is a very good dancer, this is wishful identification. Like mentioned before in this study physical similarity identification is looked at.
Now that the basis of identification is defined some studies that researched the effects of character identification on viewers will be explored. A number of studies researched the effect of identification with a character on aggressive attitudes. For instance it was found by Huesmann, Moise-Titus, Podolski and Eron (2003) that when viewers identify more with an aggressive character that this heightened their aggressive attitudes. Another study researched wishful identification and found that boys who played a game involving violence displayed more aggressiveness when they wanted to be like the characters in this violent game (Konijn, Nije Bijvank, & Bushman, 2007). A similar effect was found in a study on idealization by Eyal and Rubin (2003). They found that participants that wanted to be like aggressive characters displayed more aggressiveness compared to those who had lower levels of idealisation for the aggressive characters (Eyal & Rubin, 2003). Lastly, a study done using female characters only found that wishful identification also had a strong effect on women (Greenwood, 2007). In this study female participants only were exposed to female characters and it was again found that wishful identification with female action heroes leads to more aggressive attitudes, surprisingly this study found no effect of similarity identification. Most research focused on whether identification with a character could cause more aggression and aggressive attitudes. It is therefore interesting to explore if identifying with a superhero can also have more positive effects, like inducing more prosocial behaviour. Therefore the following hypothesis was constructed:
H2a: The positive outcome effects of Supergirl will be stronger for respondents that have high levels of identification with Supergirl.
H2b: The positive outcome effects of Supergirl will be stronger for female respondents (due to physical character similarity).
13
H3a: The negative outcome effects of Supergirl will be weaker for respondents that have high levels of identification with Supergirl.
H3b: The negative outcome effects of Supergirl will be weaker for female respondents (due to physical character similarity).
The current study will aim to provide new insights on the possible effects of female superhero content on late adolescent boys and girls, a topic that has been understudied. The study will look at aggression, defending behaviour, pro-social behaviour and altruistic
intentions and explore the role of character identification and gender. This will be done using an experimental design.
Method Sample
The sample was collected through convenience sampling. Participants were not offered any incentives and could drop out at any time during the survey. If participants chose to drop out after participating they could contact me by email within 48 hours to retract their data. The confidentially of the data was safeguarded at all times.
Participants were recruited from two schools. Het Vlietland College from Leiden participated with six classes (i.e. four 5th grade classes, two 6th grade classes) and ROC
Tilburg with one class (MBO-2 second grade class). The total number of responses was 282,
after filtering out the non-relevant responses, missings, and respondents who did not meet the age specifications, 156 respondents remained. Of these respondents 68 (43.6%) were male and 88 (56.4%) female. The mean age of the respondents was M =16.90 years, SD = 0.85. HAVO and VWO were the two biggest groups with 30.1% (n = 47) HAVO and 56.4% (n = 88) VWO, and a small group of (n = 21) MBO respondents. Most respondents were in either year 5 (n = 93) or year 6 (n = 40). The level of English generally was average or above average, namely M = 3.56, SD = 0.95. Their level of English was assessed by asking the
14
participants to give a score to their own level of English. The level of English participants could report ranged from poor (1) to good (5), with the middle being average (3).
Design and procedure
This study used a three part online experiment to explore the effects of viewing female superhero content. At the start of the experiment the participants were presented with a short text explaining the rules and terms of 1) the research 2) the rules for dropping out and 3) the terms of confidentially. After agreeing to participate the respondents could either drop out or start the survey. This was the first part of the experiment.
In part two, participants were randomly assigned to one of three Supergirl viewing conditions, namely (1) violent Supergirl, in this condition Supergirl used violence to defy the villain, (2) Non-violent Supergirl, in this condition Supergirl used a non-violent way to defy villain, the villain was talked down, and (3) Control: in this condition the participants were first asked questions regarding their defending behaviour, prosocial behaviour, altruistic intentions and aggressive attitudes. Following these questions the respondents were then shown the non-violent Supergirl video. A short summary of each episode can be found in the Appendix. All edited episodes lasted for 17 minutes.
In part three, after viewing of the episode, participants were asked questions on their character identification, defending behaviour, prosocial behaviour, altruistic intentions, aggressive attitudes and some storyline compression questions.
Measures
Storyline Comprehension check. After viewing the edited episodes, in all conditions, the respondents were asked two questions about the storyline. This was done to see if the respondent watched the video and understood the story. This was the manipulation check. These two questions were ‘In the video you just watched you follow Kara who is Supergirl in secret, which accessory does Kara use to hide the fact that she is Supergirl?’
15
Respondents could choose between three answer options: ‘a hat’, ‘a pair of glasses’ and ‘she doesn’t use any’. The second question was ‘Supergirl works with the secret organisation the DEO together with… ?’ the respondents could choose from ‘Her adoptive sister’, ‘Her best friend’ and ‘A colleague’. Both of these questions were easy to answer after viewing the video which made them good storyline comprehension checks. Almost all respondents answered these questions correctly.
Character identification. Character identification was measured with an adjusted version of the EDI scale by Igartua (2010). This scale consisted of 14 items. An example of the items include: ‘Thought I was like Supergirl or very similar to her’. Each item was assessed on a 5-point Likert scale ranging from completely disagree (1) to completely agree (5). The reliability of the scale was good, Cronbach’s α = 0.93. To test the difference between high and low levels of character identification this variable was dichotomized using the median split method into Low Character Identification and High Character Identification. The median for character identification was 2.57. All responses of 2.57 and below were coded as low (0) levels character identification and all responses higher than 2.57 were coded as high (1) levels of identification.
Defending behaviour. Defending behaviour was measured with one subscale ‘defending behaviour’ on the larger scale by Pozzoli and Thornberg (2016). The items were rephrased to reflect the intention to help rather than past behaviour. For example ‘I help or comfort classmates who are excluded from the group and isolated’ was changed to ‘I would help or comfort classmates who are excluded from the group and isolated’. Each item was assessed on a 5-point Likert scale ranging from completely disagree (1) to completely agree (5). The reliability of the scale was good, Cronbach’s α = 0.83. To create an overall score for defending behaviour the items were averaged, low scores indicate lower levels of defending behaviour.
16
Prosocial behaviour. Prosocial behaviour was measured with the subscale ‘Prosocial behaviour’ on the larger ‘Strengths and Difficulties Questionnaire’ by Goodman (1997). This subscale included five items, an example is ‘I am considerate of other people's feelings’. Each item was assessed on a 5-point Likert scale ranging from completely disagree (1) to
completely agree (5). The reliability of the scale was acceptable, Cronbach’s α = 0.78. To
create an overall score for prosocial behaviour the items were averaged, low scores indicate lower levels of prosocial behaviour.
Altruistic intentions. Altruistic intentions was measured with an adjusted version of the ‘Altruistic Personality and Self-report Altruism scale’ (Rushton, Chrisjohn, & Fekken, 1981). Items were rephrased to reflect the intention to act altruistically rather than past behaviour. For example, ‘I have offered to help a handicapped or elderly stranger across a street’ was changed to ‘I would offer to help a handicapped or elderly stranger across street’. Each item was assessed on a 5-point Likert scale ranging from completely disagree (1) to
completely agree (5). The reliability of the scale was good, Cronbach’s α = 0.85. To create an
overall score for altruistic intentions the items were averaged, low scores indicate lower levels of altruistic intentions.
Aggressive attitudes. Aggressive attitudes was measured by two subscales of the original scale ‘State Hostility Scale’ by Anderson, Deuser and DeNeve (1995). This scale included 34 items and was subdivided into 4 subscales by Anderson and Carnagey (2009). The two subscales that were used are Aggression: feeling mean (in short: Feeling Mean) and Aggression: feeling aggravated (in short: Feeling Aggravated). Two example items are respectively ‘I feel mean’ and ‘I feel like swearing.’ Each item was assessed on a 5-point Likert scale ranging from completely disagree (1) to completely agree (5). The reliability of both subscales was good, i.e. Cronbach’s α = 0.91 for Feeling Mean and Cronbach’s α = 0.89 for the subscale Feeling Aggravated. To create an overall score for Feeling Mean the items
17
were averaged, this was also done for Feeling Aggravated. Low scores indicate lower levels of Feeling Mean and Feeling Aggravated.
Results Main effect of condition on pro- and anti-social outcomes
Prosocial outcomes. To test whether the non-violent Supergirl video had an effect on
prosocial outcomes (H1a) an one-way MANOVA test was conducted with condition as fixed factor and defending behaviour, prosocial behaviour and altruistic intentions as dependent variables. For the means and standard deviations see Table 1. The multivariate main effect of condition was significant, Wilks’ Λ = 0.91, F (6, 302) = 2.48, p = .023, partial η2 =0.05. This
means that there is an effect of condition on the positive outcome variables. Table 1
Mean and Standard Deviations for Positive Outcome Variables
Condition Defending behaviour Prosocial behaviour Altruistic intentions
M (SD) M (SD) M (SD)
Control 3.97 (0.80) 4.16 (0.61) 3.70 (0.63)a
Violent 3.89 (0.69) 4.07 (0.58) 3.42 (0.58)b
Non-violent 4.02 (0.63) 3.94 (0.66) 3.60 (0.41)
Condition had a significant effect on altruistic intentions F (2, 153) = 3.52, p = .032, partial η2 = 0.04, but not on prosocial behaviour F (2, 153) = 1.62, p = .202, partial η2 = 0.02, nor on defending behaviour F (2, 153) = 0.40, p = .673, partial η2 = 0.01. A post hoc
comparison was conducted to explore the effect of condition on altruistic intentions. There was only a significant difference between respondents in the control and the violent condition. Respondents in the control condition reported higher levels of altruistic intentions compared to the violent condition, p = .026.
Anti-social outcomes. To test whether the violent Supergirl video had an effect on
18
factor and Feeling Mean and Feeling Aggravated as dependent variables. For the means and standard deviations see Table 2. The multivariate main effect of condition was
non-significant, Wilks’ Λ = 0.94, F (4, 302) = 2.23, p = .065, partial η2 = 0.03. This means that there was no effect of condition on the negative outcome variables.
Table 2
Mean and Standard deviations for negative outcome variables
Aggression
Condition Feeling Mean Feeling Aggravated
M (SD) M (SD)
Control 1.73 (0.71) 1.95 (0.90)
Violent 1.98 (0.77) 2.28 (0.92)
Non-violent 1.98 (0.69) 2.06 (0.97)
Do character identification and gender moderate the relationship between Supergirl and prosocial outcomes?
To test whether character identification moderated the relationship between the non-violent Supergirl video and the positive outcomes (H2a) a two-way MANOVA was
conducted with condition and character identification (low/high) as fixed factors and
defending behaviour, prosocial behaviour and altruistic intentions as dependent variables. For the relevant means and standard deviations see Table 3. The multivariate interaction effect was non-significant, Wilks’ Λ = 0.95, F (6, 226) = .99, p = 0.43, partial η2 =0.03. This means that there is no moderating effect of character identification on the relationship of the non-violent Supergirl video and the positive outcome variables.
19 Table 3
Average Scores on Positive outcomes by condition and character identification
Defending behaviour Prosocial behaviour Altruistic intentions Condition Character Identification M (SD) M (SD) M (SD) Control Low 3.63 (0.94) 4.05 (0.61) 3.59 (0.74) High 4.21 (0.69) 4.16 (0.49) 3.86 (0.41) Violent Low 3.80 (0.78) 4.01 (0.64) 3.26 (0.61) High 4.00 (0.59) 4.15 (0.52) 3.58 (0.51) Non-violent Low 4.06 (0.73) 3.98 (0.73) 3.63 (0.35) High 4.03 (0.50) 3.90 (0.63) 3.58 (0.46)
To test whether gender moderated the relationship between the non-violent Supergirl video and the positive outcomes (H2b) a two-way MANOVA was conducted with condition and gender as fixed factors and defending behaviour, prosocial behaviour, and altruistic intentions as dependent variables. For the relevant means and standard deviations see Table 4. The multivariate interaction effect was significant, Wilks’ Λ = 0.90, F (6, 296) = 2.80, p = .011, partial η2 =0.05.
Table 4
Average Scores on Positive outcomes by condition and gender
Defending behaviour Prosocial behaviour Altruistic intentions
Condition Gender M (SD) M (SD) M (SD) Control Girls 4.20 (0.56) 4.27 (0.51) 3.88 (0.41) Boys 3.46 (1.00)a 3.91 (0.73) 3.34 (0.83) Violent Girls 3.95 (0.58) 4.10 (0.41) 3.65 (0.44) Boys 3.83 (0.79) 4.04 (0.71) 3.20 (0.61) Non-violent Girls 4.01 (0.53) 4.17 (0.50) 3.71 (0.37) Boys 4.03 (0.71)b 3.73 (0.73) 3.49 (0.42)
20
Gender moderated the relationship between condition and defending behaviour, F (2, 150) = 4.45, p = .013, partial η2 = 0.06. However, the interaction effect between gender and condition was not significant for prosocial behaviour F (2, 150) = 1.29, p = .279, partial η2 = 0.02 and altruistic intentions F (2, 150) = 1.20, p = .305, partial η2 = 0.02.
Figure 1. Interaction effect of condition on defending behaviour by gender
Figure 1 shows the means for defending behaviour by condition and gender, girls seem to report a higher level of defending behaviour in all conditions except in the non-violent condition. To statistically compare the effect of condition on defending behaviour by gender an one-way ANOVA test was conducted for boys and girls separately. The ANOVA test was non-significant for girls F (2, 1.185) = 1.90, p = .156, but marginally significant for boys F (2, 3.626) = 2.57, p = .084. Post hoc tests showed that the effect for boys was only marginally significant when comparing the control and non-violent condition, p = .082. Thus, when viewing the non-violent condition boys display higher levels of defending behaviour compared to the control group.
Do character identification and gender moderate the relationship between Supergirl and anti-social outcomes?
To test whether character identification moderated the relationship between the violent
3 3,25 3,5 3,75 4 4,25 4,5 4,75 5
Control Violent Non-Violent
De fe nding be ha viour Condition Girls Boys
21
Supergirl video and the negative outcomes (H3a) a two-way MANOVA test was conducted with condition and character identification (low/high) as fixed factors and Feeling Mean and Feeling Aggravated as dependent variables. For the relevant means and standard deviations see Table 5. The multivariate interaction was non-significant, Wilks’ Λ = 0.99, F (2, 114) = 0.67, p = .658, partial η2 =0.01. This means that there is no moderating effect of character identification on the relationship of the violent Supergirl video and the negative outcome variables.
Table 5
Average Scores on Negative outcomes by condition and character identification
Aggression
Feeling Mean Feeling Aggravated Condition Character Identification M (SD) M (SD)
Control Low 1.79 (0.87) 1.75 (1.08) High 1.49 (0.49) 1.73 (0.76) Violent Low 1.94 (0.72) 1.96 (0.96) High 2.02 (0.85) 2.03 (0.88) Non-violent Low 2.00 (0.77) 2.26 (0.95) High 1.93 (0.63) 2.22 (0.98)
To test whether gender moderated the relationship between the violent Supergirl video and the negative outcomes (H3b) a two-way MANOVA test was conducted with condition and gender as fixed factors and Feeling Mean and Feeling Aggravated as dependent variables. For the relevant means and standard deviations see Table 6. The multivariate interaction effect was non-significant, Wilks’ Λ = 0.95, F (4, 296) = 2.09, p = .082, partial η2 =0.03. This means that there is no moderating effect of gender on the relationship of the violent Supergirl video and the negative outcome variables.
22 Table 6
Average Scores on Negative outcomes by condition and gender
Aggression
Feeling Mean Feeling Aggravated
Condition Gender M (SD) M (SD) Control Girls 1.59 (0.86) 1.81 (0.84) Boys 2.03 (0.89) 2.22 (0.99) Violent Girls 2.09 (0.69) 2.25 (0.91) Boys 1.88 (0.84) 1.79 (0.88) Non-Violent Girls 1.99 (0.75) 2.50 (1.01) Boys 1.96 (0.65) 2.07 (0.92) Discussion
This study aimed to explore the effects of female superhero content on both male and female adolescent viewers. In particular, the study tried to establish whether the presence or absence of violence in the television series Supergirl would have any effect on the prosocial behaviour or aggressive attitudes of the viewers. In short, exposure to female superhero content resulted in both a small positive and negative effect, but only for the non-violent condition. This means that media containing violence will not always cause aggressive attitudes even though this is generally believed to be a fact and that non-violent content can induce prosocial behaviour. The findings will be discussed in more detail by addressing the hypotheses that guided the study.
Prosocial outcomes. It was expected that in line with earlier research (e.g., Peña &
Chen, 2017; Rosenberg, Baughman, & Bailenson, 2013) watching a non-violent Supergirl episode would lead to more prosocial outcomes. In contrast to what was hypothesized, respondents in the non-violent condition reported lower levels of altruistic intentions compared to the violent and control condition. In addition, no effects were found between
23
condition and the other two prosocial outcomes, i.e. defending behaviour and prosocial behaviour.
The effect of condition on altruistic intentions contradicts existing research on the effects of superhero content and prosocial outcomes (e.g. Rosenberg, Baughman, &
Bailenson, 2013). The difference between literature and the findings of this study can possibly be explained by the type of media that was studied. Most earlier studies that found a positive relationship between superhero content and prosocial outcomes included interactive media content. In these studies participants were actively involved with the content instead of passively viewing a video. For example, participants imagined they had superpowers (Peña & Chen, 2017) or played a video game where they had a superhero ability (Rosenberg,
Baughman, & Bailenson, 2013). Comparing the results of these studies with the finding of the present study shows that maybe not the superhero content itself had an effect on the prosocial behaviour but more the level of involvement with this type of content. Future research needs to focus on different levels of involvement with superhero content. An experiment with different conditions could be conducted with varying levels of involvement, like passive viewing, imagining a superhero situation and playing a game. An experiment with varying levels of involvement of the participants could provide the field with more information on the present topic.
Anti-social outcomes. It was expected that in line with earlier research (e.g., Benjamin
Jr, Kepes, & Bushman, 2017; Konijn, Bijvank & Bushman, 2007) watching a Supergirl episode with aggression would lead to more aggressive attitudes. However, no effect was found between the violent Supergirl video and aggressive attitudes. Thus, the violent behaviour of Supergirl did not cause viewers to have more aggressive attitudes. The lack of effect between the violent Supergirl video and aggressive attitudes contradicts existing literature, which states that the mere presence of aggression can increase aggressive attitudes
24
in viewers (Benjamin Jr, Kepes, & Bushman, 2017; Konijn, Bijvank & Bushman, 2007). The absence of an effect of condition on aggressive attitudes could be explained by prior exposure to violent superhero content. Aggressive superhero behaviour could be seen as the norm because viewers are so used to seeing violence in superhero media. Since so much superhero content involves violence this may result in desensitisation to the aggression effects. Another reason could be that the violence shown in Supergirl is not that extreme compared to other kinds of violence in entertainment. This episode did not include a lot of blood, gore or extreme violence and can be considered mild violence. For content with more extreme kinds of violent the results may have differed. The lack of research on this topic makes this
explanation just a theoretical one. More research is needed in explaining when superhero violence has an effect on viewers and when not. The last possible explanation as to why no effect was found could be that, in hindsight, the measure for aggression was not ideal for this specific exposure to a more mild form of violence. It is possible the mild form of violence was not adequately measured by the aggressive attitudes scale used in this experiment.
The moderation of character identification and gender. It was expected that in line
with earlier research (e.g., Eyal & Rubin, 2003; Greenwood, 2007) character identification would have either a positive or negative moderating effect depending on the presence or absence of aggression. However, in contrast to current literature on the topic no moderating effect was found for both the violent and non-violent conditions. A possible explanation could be that the exposure to Supergirl was relatively short, i.e. less than 20 minutes. This made it possibly difficult for participants to identify with Supergirl in a way for identification to have a noticeable effect. Adding to this that only 29 participants had any prior knowledge about Supergirl, the participants missed the gradual build of getting to know a character that they would have had when they normally view the show.
25
Apart from character identification, gender was also expected to have a moderating effect. Partially in line with this expectation, a moderating effect was found on the
relationship between the Supergirl video and prosocial outcomes. However, gender did not moderate the main effect for all the positive outcomes, gender only moderated the effect of condition on defending behaviour. It was expected that girls would show higher levels of the positive variables in the non-violent condition but, although girls generally displayed quite high levels of defending behaviour, in the non-violent condition they scored slightly lower compared to the control condition. Girls scoring lower on defending behaviour in the non-violent condition compared to the control condition contradicted the expectation.
Interestingly, boys, who generally had low levels of defending behaviour, scored higher in the non-violent condition compared to the control condition. This clear divide between boys and girls is interesting. It is often believed that violence in media content makes, particularly boys (Konijn, Bijvank & Bushman, 2007), more aggressive but this study shows that boys can have the positive effects too.
Since it was expected that girls would display more prosocial behaviour when viewing a Supergirl video it was also hypothesized that girls would have lower levels of aggression when viewing a Supergirl video. However, no moderating effect was found for gender on the relationship between Supergirl and the negative outcome variables. The lack of a moderating effect may be explained by a combination of factors. To start with girls in general have weaker aggressive attitudes. Greenwood (2007) stated in her study on aggression and
identification that boys generally have higher levels of aggression compared to girls. As was already mentioned before, the violence in the violent Supergirl video was mild compared to a lot of other entertainment content. Girls’ general low levels of aggression combined with the mild forms of violence that the girls were presented with may explain why they were not affected by the violent Supergirl video. It would be beneficial for this field to research the
26
possible predicting effect of gender on character identification to see how and if boys or girls identify differently with aggressive characters.
Implications and limitations
The current study and its finding can be translated into several theoretical and practical implications. Most people have the idea that violence in media content brings more violence and that this is particularly harmful for impressionable adolescents. However, this study found no negative effects of viewing violent superhero content so this often believed ‘fact’ may not be a fact after all. More research needs to be done, but for now this study contributes to the research that disproves that violence in media content leads to more aggression attitudes for all adolescents.
Since some positive effects were found, this study may open up different possibilities of using non-violent superhero content to initiate more prosocial behaviour. Since the effects were only small, no radical changes in aggressive attitudes should be excepted after viewing the non-violent superhero videos. However, youth that suffers from extreme aggression issues, and thus have very strong aggressive attitudes may positively benefit from non-violent superhero content. If their aggressive attitudes are stronger to start with the lessening effect of non-violent superhero content may be bigger for them. This could be combined with training sessions to learn youth suffering from aggression issues how to solve situation in non-violent ways. More research is needed, but since viewing non-violent superhero content does seem to have some kind of positive effect, it is a worthy topic to look into.
Even though this study adds to our understanding of the effects of superhero content on prosocial behaviour and aggressive attitudes of viewers, it is not perfect and thus comes with limitations that need to be discussed in future research. First of all, the sample suffered from some minor flaws. The sample consisted of slightly more girls than boys. The data was mostly gathered at Het Vlietland College in Leiden, except for 21 MBO students from
27
Tilburg. Although the students from Het Vlietland College came from different educational levels, HAVO and VWO, it was just one school. This school is very homogeneous in terms of most students being Christian, white, and from the city area of the Netherlands (Randstad). This homogeneously makes the data not representative of Dutch adolescents and findings are therefore hard to generalise to the entire Dutch population of adolescents.
Second, although the Supergirl episodes were carefully selected and edited, it has to be acknowledged that the episodes did not include Dutch subtitles. Because the videos lacked subtitles, this could have made the videos more difficult to understand and some of the effects of the video may have been lost. However, most students were used to viewing English content without subtitles and generally the level of English was good. Future research could continue to research this type of content with native English speakers or use different dubbed and subbed versions for non-native speakers.
A third limitation is that on average the series Supergirl was not that well known amongst the participants, this can be seen as both a strength and a weakness of the study. It can be seen as a strength because there was, for most respondents, no prior exposure with the series and therefore no influence from prior opinions on the show or storyline. However, since so little respondents knew the series, 20 minutes of Supergirl may not be enough exposure to sufficiently be able to identify with Supergirl, as was mentioned before. In reality when a series is watched the viewers stay with the characters for a longer time and this could influence the level of character identification of the viewers.
The last limitation involves the ecological validity of this study. In reality no movie, TV series or comic book series is totally without violent superhero behaviour. Every movie or series will have moments of violent and of non-violent ways in which the superhero solves the problem at hand, therefore the results of this experiment cannot easily be generalised to real life. Take Supergirl, two episodes from the same series were used in this experiment, one with
28
and one without aggression. When viewing Supergirl in real life the viewers would see both types of content when viewing this series. Future research should explore the long-term effect of combined exposure to violent and non-violent behaviour of superheroes in movies or series. This could be done by first doing a content analysis of several different superhero series or movies, to indicate how much violence is present in each series or movie. The superhero content can then be categorised according to the amount of violence present. For instance low versus high levels of violence or the different superhero content could be placed on a ‘aggression continuum’ with ‘low levels of violence’ on the one end and ‘high levels of violence’ on the other end. By categorising the series according to the amount of violence in these series they can be compared and the effects of these different violence levels can be explored.
Conclusion
In sum, this study on the effects of female superhero content, provided the field with some interesting results. Now that all the results have been discussed, the research question can be answered: Does exposure to female superhero content have positive or negative effects
on late adolescents? Exposure to female superhero content has mixed results depending on
different moderating variables, presence or absence of aggression and individual differences between the viewers. Contradicting the generally held believe that violent media content causes more aggression no results that support this claim were found in the present study. Non-violent superhero content resulted in mixed outcomes, both positive and negative. In the end the effects of content depend on a wide variety of influences from both the content itself and the viewers who consume it.
30 References
Bacon., T., (8 august 2017). New & Upcoming Superhero TV Shows Coming Out In 2017 & Beyond. Retrieved from:
https://moviepilot.com/p/arrow-defenders-guide-superhero-tv-shows/4169345
Benjamin Jr, A. J., Kepes, S., & Bushman, B. J. (2017). Effects of weapons on aggressive thoughts, angry feelings, hostile appraisals, and aggressive behavior: A meta-analytic review of the weapons effect literature. Personality and Social Psychology Review, 1-29.
Boxofficemojo.com (2017). Superhero 1978-present. Retrieved from: http://www.boxofficemojo.com/genres/chart/?id=superhero.htm
Coyne, S. M., Stockdale, L., Linder, J. R., Nelson, D. A., Collier, K. M., & Essig, L. W. (2017). Pow! Boom! Kablam! Effects of viewing superhero programs on aggressive, prosocial, and defending behaviors in preschool children. Journal of abnormal child
psychology, 45(8), 1-13.
Das-Friebel, A., Wadhwa, N., Sanil, M., & Kapoor, H., S., V. (2017). Investigating Altruism and Selfishness Through the Hypothetical Use of Superpowers. Journal of Humanistic
Psychology, 1-28.
De Leeuw, R. N., & Buijzen, M. (2016). Introducing positive media psychology to the field of children, adolescents, and media. Journal of Children and Media, 10(1), 39-46.
Eyal, K., & Rubin, A. M. (2003). Viewer aggression and homophily, identification, and parasocial relationships with television characters. Journal of Broadcasting &
Electronic Media, 47(1), 77-98.
Free, E. (2016). The age of heroes: Why are superhero movies so popular?. Retrieved from: https://filmink.com.au/2016/the-age-of-heroes-why-are-superhero-movies-so-popular/ Greenwood, D. N. (2007). Are female action heroes risky role models? Character
31
identification, idealization, and viewer aggression. Sex Roles, 57, 725-732. Goodman, R. (1997). The Strengths and Difficulties Questionnaire: a research note. Journal
of child psychology and psychiatry, 38(5), 581-586.
Hagley, A., & Harrison, M. (2014). Fighting the Battles We Never Could: The Avengers and Post-September 11 American Political Identities. PS: Political Science Politics, 47(1), 120-124.
Hoffner, C. (1996). Children’s wishful identification and parasocial interaction with favorite television characters. Journal of Broadcasting, 40, 389-402.
Hartmann, T., Toz, E., & Brandon, M. (2010). Just a game? Unjustified virtual violence produces guilt in empathetic players. Media Psychology, 13(4), 339-363.
Hartmann, T., & Vorderer, P. (2010). It's okay to shoot a character: Moral disengagement in violent video games. Journal of Communication, 60(1), 94-119.
Huesmann, L. R., Eron, L. D., Klein, R., Brice, P., & Fischer, P. (1983). Mitigating the imitation of aggressive behaviors by changing children's attitudes about media violence. Journal of personality and social psychology, 44(5), 899.
Huesmann, L. R., Moise-Titus, J., Podolski, C. L., & Eron, L. D. (2003). Longitudinal relations between children's exposure to TV violence and their aggressive and violent behavior in young adulthood: 1977-1992. Developmental Psychology, 39(2), 201-221
Igartua, J. J. (2010). Identification with characters and narrative persuasion through fictional feature films. Communications, 35(4), 347-373.
Konijn, E. A., Nije Bijvank, M., & Bushman, B. J. (2007). I wish I were a warrior: the role of wishful identification in the effects of violent video games on aggression in adolescent boys. Developmental psychology, 43(4), 1038-1044.
32
acceptable? The effects of character motivation and outcome on perceptions, character liking, and moral disengagement. Mass Communication and Society, 16(2), 179-199. Macdonald, H. (2012). ‘Why DC and Marvel will never truly target female readers.’
Retrieved on 30th November 2017 from: http://www.comicsbeat.com/why-dc-and-marvel-will-never-truly-target-female-readers/
Peña, J., & Chen, M. (2017). With great power comes great responsibility: Superhero primes and expansive poses influence prosocial behavior after a motion-controlled game task. Computers in Human Behavior, 76, 378-385.
Peppard, A. F. (2017). This Female Fights Back!. In Yockey, M. (eds.). Make Ours Marvel:
Media Convergence and a Comics Universe. University of Texas Press.
Pozzoli, T., Gini, G., & Thornberg, R. (2016). Bullying and defending behavior: The role of explicit and implicit moral cognition. Journal of school psychology, 59, 67-81. Rosenberg, R. S., Baughman, S. L., & Bailenson, J. N. (2013). Virtual superheroes: Using
superpowers in virtual reality to encourage prosocial behavior. PloS one, 8(1), 1-9 Rushton, J. P., Chrisjohn, R.D., & Fekken, G. C. (1981). The altruistic personality and
self-report altruism sale. Personality and Individual Differences, 1, 292-302.
Valkenburg, P.M., & Piotrowski, J.T. (2017. Plugged in: How media attract and affect youth. New Haven, CT: Yale University Press.
33 Appendix 1
Conceptual model for the possible (moderating) effects of Supergirl on positive and negative outcomes
34 Appendix 2
Description of experimental videos
Below you find a short description all three conditions of this experiment.
Condition 1 Violent superhero
In the violent condition, the respondents were an shown edited version of the 3rd episode of the first season of Supergirl. The video started with the Supergirl intro that gives some information about where Supergirl came from, why she is on earth and explains her
undercover life as Kara and her life as Supergirl. The main storyline dealt with Supergirl that struggled to use her superpowers, since she was new to them. In this episode Reactron, an enemy of Superman, Supergirl’s cousin, tries to kill Supergirl to take revenge on Superman. Reactron’s suit is damaged and he kidnaps a scientist to fix his suit. The first time Supergirl tries to fight Reactron she manages to free the scientist but is beaten by Reactron and
Superman saves her just before she passes out. Later on a gala were Kara is present, Reactron violently disrupts the gala and Kara has to switch into Supergirl mode to fight him. This time she does manage to defy him. She does this by encasing her hand in lead so she can get the radioactive core out of Reactron’s costume without causing a meltdown. The video ends with Supergirl flying away though the city.
Condition 2: Non-violent superhero
In the non-violent condition, the respondents were shown an edited version of the 7th episode of the first season of Supergirl The video started with the Supergirl intro that gives some information about where Supergirl came from, why she is on earth and explains her
undercover life as Kara and her life as Supergirl. In this episode the main conflict deals with Supergirl who temporarily lost her powers after a fight in the previous episode. Without her
35
powers Kara is weaker and therefore unable to help the city as an earthquake hits. She breaks her arm during the earthquake. Despite Supergirl being powerless she does try to help people. She struggles with her identity and sense of self without her powers. This turmoil causes her to try to be a hero without her powers. When she sees a robbery taking place she enters the store as Supergirl but without her powers she is vulnerable and of course unable to stop the robber in any other way than talking to him. She does manage to talk down the robber. After that she regains her powers during an elevator accident and the episode ends. In this episode no violence is used by Supergirl.
Condition 3: Control
In the control condition the respondents first answered questions on defending behaviour, prosocial, altruistic intentions, aggressive attitudes and some demographics. These
demographics include: gender, age, level and year in school, level of English. After these they will see the non-violent episode of Supergirl. This clips is shown after the important effect questions are asked so that this group is the control group but still feels like they have been exposed to something. After the video some questions regarding the episode and character identification will be asked. In this way the groups will take about the same time completing the survey.
