RESULTS
The first set of analyses will examine potential personality differences between comic and non-comic readers and consider whether or not comic readership has any effect on the development of a Hostile Attribution Bias. After this, following the procedures of Kirsch & Olczak (2002b), analyses were conducted examining the correlations between trait hostility, personality factors and perceptions of the stimulus comic books. In the main analysis the influence of situational (comic book) and trait variables (personality and innate aggression) on processing of social information, as measured by aggressive responses to the ambiguous provocation scenarios is examined.
Preliminary Analyses
TEST TYPE
The first analysis will check for any potential differences between the paper and e-versions of the test. A simple 2-way t-test compares the differences between the two populations three aggression response variables; intent, retaliation and emotion. The subsequent t-values are p=0.090, p=0.723, p=0.370 respectively. The intent variable is the only variable whose result approaches significance. A closer analysis of the data reveals that this is due to the action of gender and not to any systematic difference between the 2 tests (p<0.05). It’s safe to assume that the data gathered by both tests is equally valid and that the test type is not a confounding factor in the results.
COMIC BOOK HISTORY
The first hypothesis to be tested concerns the personality of comic readers. Since 73.08% of non-comic readers are female and 81.23% of comic readers are male, gender is a potential factor that may interact with comic readership. In order to determine whether or not Comic readership is a factor in this study a 2(Reader) X 2 (Gender) ANOVA was conducted for the Total Aggression score from the Aggression Questionnaire. Results indicate a significant Reader X Gender Interaction, F (1,123) = 5.88, p=0.017. There are no main effects for Gender or Reader, F (1,123) =2.15, p=0.145; F (1,123) =0.15, p=0.698. Follow up Pairwise comparisons reveals significant differences between female and male non-comic readers. The Female non-comic readers are significantly higher in Trait Hostility than their male counterparts (p<0.005). There was no difference between Trait Hostility in male and female comic readers or between male readers and male non-readers. An examination of the means indicates that this result is due to the actions of a few females with high Trait Hostility.
The data gathered from the NEO-PI questionnaire is examined. To examine any potential personality differences between Comic readers and non-comic readers a series of ANOVAs was conducted. For Neuroticism, there were no significant main effects for Gender and Reader or interaction effects, F (1, 123) =0.68, p=0.410; F (1,123) = 0.12, p=0.624; F (1,123) = 0.12, p=0.729. For Extraversion, there was no significant effect for Reader, F (1,123) =0.16, p=0.688 or for the Gender X Reader interaction, F (1,123) = 4.51, p<0.05. There was a significant main effect for Gender F (1,123) = 4.46, p<0.05. The Females in this study are more extraverted than the males. Pairwise comparisons reveal that female non-comic readers are significantly more extraverted than male non-comic readers (p<0.05). For Openness there was almost a Significant main effect for Gender F (1, 123) =3.33, p=0.007. Males are almost significantly more open than females. The Reader main effect and the Gender X Reader interaction were both non significant, F (1,123) =1.53, p=0.218; F (1, 123) =1.52, p =0.220. For Agreeableness, there is a significant main effect for Gender. Females are significantly more agreeable than males F (1,123) =18.05, p<0.001. There is no significant main effect for Reader, F (1,123) = 1.22, p=0.271 and for the Gender X Reader interaction, F (1,123) =0.15, p=0.697. Finally for Conscientiousness, there were no main effects for Gender, F (1,123) =1.35, p= 0.248, for Reader F (1,123) =0.64, p=0.426 or for the Gender X Reader interaction F (1,123) =1.23, p=0.270. It would seem as if there are no obvious personality differences between comic readers and non-comic readers. Any differences that occur in this sample group are due to inherent gender differences.
According to the GAM the long-term effects of exposure to violent media is primarily the development, rehearsal and eventual automatization of aggressive knowledge structures such as perceptual schema (Anderson & Bushman 2001). Viewers of media violence may not go on to commit murders but they will be more aggressive and will instead use indirect aggression that does not result in societal sanctions (Coyne & Elsea, 2003). It’s proposed that years of exposure may build up a HAB and that those who have read comics the longest will respond most negatively to the ambiguous provocation scenarios. To test this, a 2 (Gender) X 2(Years Reading) ANOVA is applied to the three aggression variables; Intent, retaliation and emotion. For retaliation the Gender and Years Reading main effects are non-significant F (1,123) = 0.00, p=0.992; F (1,123) = 0.53, p=0.65. The interaction between the 2 variables is also non-significant, F (1, 123) =1.00, p=0.397. For intent the Gender main effect is significant, F (1,118) = 4.32, p=0.040. Females ascribe more negative intent than males. The Years Reading main effect and interaction of Gender X Years Reading are both non- significant, F (1,118) = 1.65, p=0.181; F (1,118) = 0.25, p=0.859. For emotion the Gender main effect is significant, F (1, 123) = 4.96, p=0.028. Females ascribe more negative emotion to the provocateur than males. The Years Reading and Gender X Years Reading interaction are both non-significant. F (1, 123) = 0.69, p=0.562; F (1,123) = 0.25, p=0.862. Examination of the means suggests that the lowest responses are coming from male comic readers who have read for 10+ years. The highest responses are coming from female non-comic readers. This would seem to suggest that years reading comics does not raise HAB, if anything it lowers it.
TRAIT HOSTILITY
Analyses were conducted to asses the relationship between Trait Hostility and various personality variables. The results indicated a moderately negative relationship between Trait Hostility and Neuroticism (r= -0.349, p=0.001). As hostility increases emotional stability decreases. There is a weak negative relationship between Trait Hostility and Agreeableness, (r=-0.037, p<0.005). As hostility increases Agreeableness decreases. There were also intercorrelations between Openness and Conscientiousness (r=0.209, p=0.018), between Extraversion and Agreeableness (r=0.46, p<0.001) and between Neuroticism and Conscientiousness (r= 0.205, p=0.021). Because of the strength of the relationship between hostility and Neuroticism, Neuroticism will be entered as a covariate for later analysis.
Analyses were also conducted to asses the relationship between Trait Hostility and between the scores on the composite variables. The results indicate significant Pearson correlations between Trait Hostility and Intent (r=.212, p=0.0117). Between Trait Hostility and Retaliation (r=.408, p<0.001) and Trait Hostility and Emotion (r=.370, p<0.001). Intent, Retaliation and Emotion were also all intercorrelated. With strong to medium correlations between Intent and Emotion (r=0.709, p<0.05), Intent and Retaliation (r=0.447, p<0.001) and Emotion and Retaliation (r=0.532 p<0.01)
COMIC BOOK RATINGS
To asses the relations between the comic book ratings a series of Pearson correlations are performed. For like and interest there is a strong positive relationship (r=0.894, p<0.001). There are strong relationships between humour and like (r=0.728, p<0.001) and between humour and interest (r=0.625, p<0.001). There are no correlations between Perceived aggression and like and interest. Between humour and aggression/violence there is a weak negative relationship (r=-0.268, <0.001). The more violence there is the unfunnier the comic is.
To verify that the comics are correctly categorised a Gender X Stimulus ANCOVA was conducted for the Perceived Aggression of the comics. Because of the influence of innate aggression Trait Hostility is entered as a co-variable. Results indicate a main effect for Stimulus, F (1, 126) =149.74, p<0.001. Extremely Violent comic books are rated as significantly more violent than NVCB. There is no main effect for Gender F (1,126) = 0.68, p=0.411. The interaction between Stimulus X Gender approaches significance F (1,126) =3.02, p=0.085. This difference comes from the high perceived aggression ratings of male readers of EVCB compared with low perceived aggression ratings from male readers of NVCB.
To test for differences in participants’ perceptions a Gender X Stimulus ANCOVA was conducted with Trait Hostility being entered as a covariate. For like, there is a significant main effect for Gender. There’s hope for nerds everywhere, it seems that Females like comic books more than males, F (1,126) =5.31, p<0.05. There is no main effect for Stimulus, F (1,126) = 0.61, p=0.437. Subjects express no preference for any particular comic book type. There is no significant Gender X Stimulus interaction, F (1,126) = 0.00, p<0.973. For interest there are no main effects for Gender, Stimulus or the Gender X Stimulus interaction, F (1,126) = 2.46, P=0.119; F (1,126) = 0.16, p=0.689; F (1, 126) = 0.08, p= 0.774. There are no differences in subjects’ interest in comics. For humour there are significant main effects for Gender and Stimulus F (1, 126) =13.84, p<0.001; F (1, 126) =16.38, p<0.001. Females found the comics funnier than males and both sexes found the found the NVCB funnier than the EVCB. The interaction effect is not significant, F (1, 126) =05.4, p=0.464. Since perceptions of comic books may influence responses to ambiguous provocation scenarios humour is entered as covariate in subsequent analyses. (Appendix Table 1 presents means and standards errors of comic book ratings)
Main Analyses
In order to test the contention that violent comic books influence processing of social information a 2(Stimulus) X 2 (Gender) multivariate ANCOVA is conducted with Trait Hostility as a co-variable. There are main effects for Stimulus, Gender and the covariate Trait Hostility F (1,122) = 4.61, p 0.034; F (1,122) = 8.04, p=0.005; F (1,122) =10.22, p=0.002. There is no significant interaction between Gender X Stimulus F (1,122) = 0.86, p= 0.355. Follow up ANCOVAs were then conducted for each composite variable. For intent there are main effects for Stimulus F (1,122) = 11.55, p <0.001 and for Gender F (1,122) =16.32, p<0.001. The Stimulus X Gender interaction effect is non-significant, F (1,122) = 11.55 p= 0.167. Readers of EVCB ascribe more hostile intent than readers of NVCB and females ascribe more hostile intent than men. Results for emotion show significant main effects for Stimulus and Gender, F (1,122) = 8.61, p =0.004; F (1,122) =10.51, p=0.002. There is no significant Gender X Stimulus interaction, F (1,122) =, 0.09 p= 0.766. Readers of EVCB ascribe more negative emotion to the provocateur than readers of NVCB. Females also ascribe more negative emotion than males do. In the test examining retaliation a significant main effect is found for Stimulus, F (1,122) = 6.10, p=0.0015. There is no significant result for either Gender or the Gender X Stimulus interaction F (1,122) =0.24, p=0.624; F (1,122) = 0.39 p= 0.533. Readers of EVCB ascribe more retaliation than readers of NVCB. (See Appendix Table 2 for means and standard errors of the three aggression variables).
Factor Analysis
Factor analysis is used to reduce the data into meaningful variables. This technique, used by Dodge et al (1986), has more power in assessing the relationship between aggression and individual SIP steps. Composite variables are formed by summing responses to like questions for like scenarios. For overt scenarios the two intent- related questions were combined and the mean is taken to form a single intent variable. The two retaliation related questions are summed and the mean is taken to form a single overt retaliation variable and the 2 emotion related questions are summed to form a single overt emotion variable. There is only a single relational aggression scenario so the total intent, retaliation and emotion for this question are comparable to the mean values for the overt scenarios. These 6 variables are entered into factor analysis. A varimax rotation produced a 2 factor solution. The 2 factors account for 42% and 25.3% of the total variance results. Taken together these factors account for 67.3% of the total variation. This means 32.7% of the results are unexplained by these common factors. This may be due to specific factors related to the responses to individual questions. Factor I consisted of overt retaliation, overt intent and overt emotion with factor loadings of .76, .84, .86 respectively. Factor II consisted of relational retaliation, relational intent and relational emotion with factor loadings of .73, .82, .83 respectively. Cross factor loadings ranged from 0.1 to 0.3. Based on the results from the FA two summated scales were formed; Overt aggression (consisting of factors from Factor I) and Relational Aggression (consisting of factors from Factor II).
In all of the analyses so far Trait Hostility has been used as a covariate. In order to examine the interaction and to test the full spread of social information a 2(Stimulus) X 2(Gender) X 2(Trait Hostility) multivariate ANCOVA is constructed. Neuroticism, Agreeableness and Humour are entered as covariates as these factors have been implicated in earlier analysis as potential confounds. There are significant main effects for Trait Hostility and Stimulus, F (1,116) = 26.43, p=0.000 and F (1,116) = 11.68, p=0.001. The interactions of Gender X Stimulus, Gender X Trait Hostility, Stimulus X Trait Hostility, Gender X Stimulus X Trait Hostility are all non significant; F(1,16)= 1.69, p= 0.197; F(1,116)= 0.85, p= 0.359; F(1,116)= 0.00, p= 0.980; F(1,116) = 0.06, p=0.814. The co-variables of Neuroticism and Humour are both significant F (1,116) =5.44, p = 0.021 and F (1,116) = 4.79 p= 0.031. Agreeableness is also non significant F (1,116) = 1.04, p= 0.309. There is no main effect for Gender, F (1,116) =2.95, p=0.089. However when the covariates are removed gender is a significant result, F (1, 119) =11.65, p= 0.001. This suggests that Neuroticism co-varies with Gender. In the Kirsch & Olczak 2002b study perceived aggression is added as a covariate here. This examines the affect of the violence in comic books on SIP. In the Kirsch & Olczak study the addition of perceived aggression as co-variate made the stimulus main effect non- significant. When perceived violence is added as a covariate here the main effects for Stimulus, F (1, 118) = 7.54, p= 0.007, and Trait Hostility, F (1, 118) = 25.08, p= 0.000, are still significant. The p-value does alter but the result is still strongly significant. It seems that the violence in comic books is not responsible for any stimulus effects.
Follow-up analyses of covariance were conducted to investigate the significant main and interaction effects. Neuroticism, Agreeableness and Humour are entered as covariates. For Overt Aggression there were main effects for Stimulus, F (1,116) = 5.10, p= 0.026, and Trait Hostility, F (1,116) = 13.98, p< 0.01. The results indicate that participants reading EVCB provide more aggressive responses than those reading NVCB and that those with high Trait Hostility respond more violently than those with low Trait Hostility. There are no Gender main effects, males and females are equally as violent in their responses, F (1,116) = 0.01, p= 0.931. The Comic book X Trait Hostility , Gender X Trait Hostility, Comic Book X Gender and Comic Book X Gender X Trait Hostility interaction effects were all non significant, F(1,126)=0.24, p= 0.623; F(1,116)=1.12, p=0.293; F(1,126)=0.08, p= 0.773; F(1,126)=0.67, p=0.415. Neither Neuroticism, Agreeableness nor Humour played a role in this model, F (1,116) = 2.57, p=0.111; F (1,126) =0.34, p= 0.560; F (1,126) = 3.22, p= 0.076
For Relational Aggression, there were main effects for Stimulus, F (1,126) = 8.06, p=0.005, Gender, F (1,126) = 8.17, p= 0.005 and for Trait Hostility F (1,116) = 14.94, p< 0.01. Subjects high in Trait Hostility respond more violently than those with low Trait Hostility. There was also a significant Gender X Stimulus interaction, F (1,126) = 5.87, p= 0.017. Pair wise comparisons indicate that females reading EVCB respond more violently than females reading NVCB (p=0.0113) and more violently than males reading NVCB (p<0.01). They also indicate that males reading EVCB differ from females reading EVCB (p<0.01). Females are more violent in their responses. Males reading EVCB did not significantly differ from males reading NVCB. The Gender X Trait Hostility, F(1,126)=0.06 ,p= 0.811, Gender X Stimulus X Trait Hostility, F(1,126)=0.34 0.563, and Stimulus X Trait Hostility, F(1,116)=0.29 0.592, interaction effects were all non significant. Agreeableness was significantly covaried within this model, F (1,116) = 5.41, p=0.022. Neuroticism, F (1,116) = 3.47, p=0.065 and humour, F (1,116) =1.98, p=0.162, were not significant. (See Appendix Table 3 for Estimated Mean and Standard Errors of Overt, Relational and Total Aggression as a Function of Comic Book, Gender and Trait Hostility).
The experiment set out to test several hypotheses. Some of these have been proven others have not. The first hypothesis tested whether or not there were any personality differences between comic readers and non-comic readers. It’s been has shown that there are no tangible personality differences that exist between comic readers and non-comic readers. Any differences that exist are gender effects. It was also shown that female non-comic readers are significantly higher in their aggression than both male non-comic readers and female comic readers. The results show that there are no increases in HAB after long-term reading of comics. Thus it would seem that long-term exposure to comics does not increases aggressive tendencies. The second hypothesis examining perceptions of comic books found that subjects rated EVCB as violent and NVCB as non-violent. Females found comics funnier than males and NVCB were rated as funnier than EVCB. There were no interest or like effects. Although subjects disliked/were more uninterested in EVCB than NVCB, the differences were not significant. A third hypothesis, which tested whether or not individuals exposed to EVCB will show evidence of a biased SIP and respond more aggressively than subjects exposed to NVCB do, found a HAB in individuals exposed to EVCB. This means that readers of violent comics ascribe more hostile intent, retaliation and emotion to provocateur than readers of NVCB. Females also ascribe more hostile intent and more negative emotion than males. The final hypothesis examined whether as Trait Hostility increases participants will ascribe more hostile intent, suggest more retaliation and attribute more negative emotion to the provocateur. The data found that individuals high in Trait Hostility responded with consistently high aggression. Females reading EVCB provide the highest responses to scenarios involving relational aggression and those reading EVCB respond most aggressively for the overt scenarios. There were no significant gender differences in the responses to the overt aggression scenarios. A hypothesis involving the interaction of Trait Hostility with the stimulus was not proven. Individuals high in trait hostility do give high aggressive responses after reading EVCB but their responses are not statistically significantly.