Communication Mediation Model of Late-Night Comedy: The Mediating Role of Structural Features of Interpersonal Talk Between Comedy Viewing and Political Participation

Abstract

This study advances a communication mediation model of late-night comedy in an effort to understand the process wherein consuming satirical humor indirectly spurs political participation via the conduit of interpersonal talk about politics. The theoretical model was tested utilizing two different research designs. The findings from the experiment and the survey provide considerable support for the model, demonstrating that various structural features of interpersonal talk (e.g., discussion frequency, online interaction, and network size) positively mediate the association between late-night comedy viewing and political participation. Meanwhile, the assessments concerning the mediating role of heterogeneous discussion illustrate that late-night comedy can draw a higher level of political involvement from those who are highly educated. The present study urges the field to extend the scope of the communication mediation model by incorporating a greater number of media channels and more diverse aspects of interpersonal talk.

Notes

¹Unique characteristics of late-night comedy could be summarized by the use of satire to convey a coherent political message (Caufield, 2008). Because the primary goal of this study taps into the effects of political stories with ridiculing twists, late-night comedy was rather narrowly defined to incorporate only those programs belonging to the political satire genre (e.g., The Daily Show with Jon Stewart and The Colbert Report) while excluding standard nighttime talk shows (e.g., Late Show with David Letterman and The Tonight Show with Jay Leno).

²Eveland and Hively (2009) succinctly summarized the inconsistencies in past literatures surrounding the conceptualization and operationalization of network heterogeneity. The current study attends to the most commonly employed conceptual and operational definition of heterogeneous discussion, that is, level of discussion with individuals holding dissimilar views (E. Kim et al., 2011; Kwak et al., 2005; Mutz, 2002, 2006; Scheufele et al., 2004).

³One-way analysis of variance was used to check the effectiveness of the random assignment in terms of age, gender, ethnicity, household income, education, and party affiliation (all ps > .46). Further, chi-square tests showed no significant association between demographic measures and group assignment (all ps > .36). Thus, no control variables were included in analyzing the experimental data.

⁴The specific video clips as well as the complete question wording are available at http://www-personal.umich.edu/~hoonlz/latenightcomedy2011.

Note. Entries are ordinary least-squares regression coefficients along with corresponding standard errors in parentheses.

***p < .001.

⁵As Lee (2012) recently reviewed, we could either examine the separate main effects of discussion frequency and heterogeneous talk or assess interactive workings of these two. The present research adopts the former approach in that one of the primary goals involves estimating a unique mediating effect of each dimension while controlling for the others. Although it is beyond the scope of the current study, this study acknowledged possible interactions among these features.

⁶For a categorical independent variable with k categories, Preacher and Hayes (2008) recommend to construct k-1 dummy variables and then run analyses k-1 times. With each run, one dummy variable is the independent variable and the other one(s) is (are) the covariate(s). Because the independent variable in the present study comprised three categories (late-night comedy, hard news, and control), I created two dummy variables (late-night dummy and hard news dummy) and ran an analysis twice.

Note. CIs are bias-corrected and bias-accelerated 95% confidence intervals (bootstrap N = 5,000). L = late-night comedy exposure; D = discussion frequency; O = online interaction; H = heterogeneous discussion; P = political participation.

†p < .10. *p < .05. **p < .01. ***p < .001.

⁷Different analytical approaches in the two research designs reflect distinct foci. The purpose of the experiment involves examining, under a controlled setting, every detailed consequence of manipulation. Toward this goal, it seems more reasonable to look at all the possible casual links, including those failing to reach statistical significance. The exploited macro by Preacher and Hayes (2008) allows for more exhaustive assessments of causal paths. Meanwhile, the goal of the survey is to identify mundane relationships between media uses, interpersonal talk, and political participation. AMOS enables us to remove insignificant paths that may suggest null relationships in the real world and investigate whether the proposed model as a whole fits the data well.

⁸Although some (e.g., Huckfeldt, 2001; Krosnick, 1990) argued for more nuanced operationalization of political expertise, the current study employed education as a proxy measure, as education affords more advanced cognitive capacity to make sense of complex information (Rosenberg, 1988), which often characterizes heterogeneous discussion.

Note. CIs are bias-corrected 95% confidence intervals (Bootstrap N = 5,000). L = late-night comedy exposure; D = discussion frequency; N = network size; P = political participation.

†p < .10; *p < .05.

Additional information

Notes on contributors

Hoon Lee

Hoon Lee (Ph.D., University of Michigan, 2012) is a postdoctoral research fellow at the Nam Center for Korean Studies at the University of Michigan. His current research agenda encompass social impacts of entertainment media and new communication technologies in cross-national and cross-cultural contexts.