Abstract
This experimental research focuses on four distinct, but related theoretical interests regarding the study of emotion and media effects: (a) emotion serving as a potential mediator of the relationship between the consumption of multiple media messages, (b) media messages sparking emotional reactions to attitude objects not specifically referenced or identified in a message, (c) the assessment of inter-emotion relationships across distinct attitude objects, and (d) varied processes of emotion-based influence based on individual-difference attitude object preferences. This work focuses on Michael Moore's Fahrenheit 9/11, its effects on anger felt toward President George W. Bush and Senator John F. Kerry, and subsequent perceptions of candidate debate superiority across Bush versus Kerry supporters. Moore's film was found to influence levels of anger felt toward both Bush and Kerry and these emotions were related to subsequent perceptions of debate superiority. Distinct processes of influence were found for previously identified Bush supporters versus Kerry supporters, with Bush anger serving as a mediator for Bush supporters and Kerry anger serving as a mediator for Kerry supporters. Ramifications of this study's findings are outlined and future lines of research are summarized.
Notes
1. Although our study focuses on political communication, we wish to stress that the four primary theoretical points addressed in this work are not context specific. In fact, we would encourage future research that would address these same issues in other contexts (e.g., health communication, science communication) in order to offer an empirical assessment of the generalizability of the study's findings.
2. The original Kerry anger indicator variables retained a strong positive skew. As a result, we used a square-root transformation in order to deal with this lack of normality in each variable. Also, reliability estimates are provided for all latent variables, but the actual indices were not used for this study. This study retains the use of latent variables with separate multiple indicators.
3. The following are results from the series of OLS-based multiple regression equations involving the two control variables (i.e., prior F 9/11 viewing and prior Bush-Kerry debate viewing) as independent variables and respective endogenous model observable variable indicators (standardized (values reported): Kerry-anger: prior F 9/11 = −.119∗, prior debate = .120∗; Kerry-disgust: prior F 9/11 = −.130∗, prior debate = .085; Bush-anger: prior F 9/11 = .268∗∗∗, prior debate = −.090; Bush-disgust: prior F 9/11 = .264∗∗∗, prior debate = −.125∗; won debate: prior F 9/11 = −.239∗∗∗, prior debate = .109∗; presidential: prior F 9/11 = −.238∗∗∗, prior debate = .122∗; answered questions: prior F 9/11 = −.158∗∗, prior debate = .094. ∗p < .05, ∗∗p < .01, ∗∗∗p < .001.