The Influence of Exposure to Depictions of Race and Crime in TV News on Viewer's Social Judgments

This two-study experimental design utilizes a group-based priming framework to investigate the relationship between exposure to television news portrayals that intersect race with violent crime and viewers' real-world racial judgments. Results from Study 1 reveal that both the gender of the viewer and the race of the TV news suspect influence subsequent judgments, including attributions about the perpetrator and victim. Findings from Study 2 offer somewhat consistent results, additionally indicating that the race of the depicted suspect has a significant effect on attitudes toward Blacks in greater society, beyond the mediated context. Notes aMeans with the same superscript differ significantly at p < .05. bMeans with the same superscript differ significantly at p < .05. cMeans with the same superscript differ significantly at p < .05. dMeans with the same superscript differ significantly at p < .05. eMeans with the same superscript differ significantly at p < .05. fMeans with the same superscript differ significantly at p < .05. gMeans with the same superscript differ significantly at p < .05. hMeans with the same superscript differ significantly at p < .05. iMeans with the same superscript differ significantly at p < .05. aMeans with the same superscript differ significantly at p < .01. aMeans with the same superscript differ significantly at p < .05. bMeans with the same superscript differ significantly at p < .05. dMeans with the same superscript differ significantly at p < .05. hMeans with the same superscript differ significantly at p < .05. iMeans with the same superscript differ significantly at p < .05. jMeans with the same superscript differ significantly at p < .05. eMeans with the same superscript differ significantly at p < .01. gMeans with the same superscript differ significantly at p < .01. aMeans with the same superscript differ significantly at p < .01. bMeans with the same superscript differ significantly at p < .01. cMeans with the same superscript differ significantly at p < .05. *See CitationGreenwald et al. (2003) for D calculations. aMeans with the same superscript differ significantly at p < .01. 1Although logistic regression would have been the preferred statistical test for use with the dichotomous outcome measure of guilt, an attempt to run this model was unsuccessful due to empty/small cells. Under such circumstances the logistic regression model will either continue to iterate and not provide a solution at all and/or will cause the estimation procedure to fail, leading to computation errors such as missing info, skyrocketing odds ratios, and the like. This was the case for the present study. Given this, the authors selected ANOVA due to the fact that empirical tests using Monte Carlo simulations (e.g., Lunney, 1970, Citation2005) have revealed that ANOVA is robust to the violations associated with use of a dichotomous outcome (e.g., normality). Thus, although this procedure is certainly less than ideal, because of the special nature of dichotomous variables such as this one, it is not seen as an altogether unacceptable practice given that meaningful inferences can still be drawn from the results (CitationLeech, Barrett, & Morgan, 2004).