Communication Modality and Biased Processing: A Study on the Occasion of the German 2002 Election TV Debate

Abstract

Participants were presented with a segment of the 2002 Schroeder-Stoiber TV debate, in which the two candidates running for the post of German Chancellor argued for clearly different positions. Prior to presentation of the segment in either audiovisual, audio, or written modality, participants' politician preference was measured. Position agreement and valence of thoughts about the candidates' specific statements were found to be affected by politician preference in all three modalities. However, as shown by a modality by preference interaction, the effect of politician preference on position agreement was stronger in audio and audiovisual modalities than in the written modality. Similar interactions were found regarding the valence of thoughts broadly related to the issue discussed as well as regarding the valence of thoughts related to the two candidates themselves. Taken together with further path analyses, these findings are interpreted as showing biased processing across modalities, with the broadcast modalities intensifying this bias due to the heightened salience of nonverbal politician-related cues.

Notes

∗p < .06

∗∗p < .05

∗∗∗p < .001

¹Given the scale midpoint (4), overall the agreement is obviously stronger for the position advocated by Schroeder. In fact, it has been found that the position advocated by Schroeder (no participation of Germany in a military intervention) as compared to the one advocated by Stoiber (Germany's participation possible) on the Iraq question elicited more agreement (cf. Maier & Faas, 2004). Nonetheless, of major importance here is the difference in agreement between individuals preferring Schroeder and individuals preferring Stoiber.

²Separate analyses showed that participants had more unfavorable thoughts related to Schroeder's (Stoiber's) statements the more they preferred Stoiber (Schroeder), both ps < .01. Further, while participants had more favorable thoughts toward Schroeder's statements the more they preferred Schroeder (p < .04), participants did not have more favorable thoughts toward Stoiber's statements the more they preferred Stoiber (t < 1).

³Separate analyses of the amount of favorable and unfavorable thoughts towards Schroeder and Stoiber were directionally consistent with the effects reported for the index of politician-related thoughts, though neither the preference effect nor the preference by condition interaction effect was found significant for any of the four elements.

⁴Further evidence for the heightened salience of the candidates in the two broadcast modalities is given by the results of an analysis of the number of neutral politician-related thoughts. As revealed by a marginal effect of the contrast of the text condition versus the audio and video conditions, b = .03 (SE = .02), t(103) =1.90, p = .06, such thoughts occurred somewhat more often in audio and video conditions (M = 0.12) than in the text condition (M = 0.02). In comparison, no such difference was found regarding neutral statement-related thoughts and neutral issue-related thoughts.

⁵We also conducted analyses in which the major dependent variables were regressed on each of the following variables in addition to modality and politician preference: need for cognition (M = 5.10, SD = 0.73; response scale from 1 to 7, higher scores indicate higher NC), interest in politics, interest in the issue, prior issue familiarity, and extent of thinking about the politician statements. The only significant effects of these additional variables were main effects such that agreement with Schroeder's position was stronger the higher an individual's need for cognition score, the more an individual was interested in politics, and the more he/she had an interest in the issue. No two-way or three-way interaction involving these additional variables was significant. Effects involving modality and preference remained basically unchanged.

⁶In light of the results on politician-related thoughts reported in the section on Cognitive Responses, we also conducted separate analyses for the audio condition and the video condition. The results were largely similar, the major difference relating to the thoughts that qualified as mediators of the relation between politician preference and position agreement. While all three kinds of thoughts qualified as mediatiors in the audio condition, only specific statement–related and more global issue–related thoughts, but not politician–related thoughts, qualified as mediators in the video condition.

⁷Combining the statement-related thoughts and the issue-related thoughts (i.e., the sum of both indexes) into one index reveals that the correlation between this thought index and position agreement is somewhat higher in the video (r = .71, p < .001) and audio (r = .62, p < .001) conditions than in the text condition (r = .39, p < .05), z = 1.87 and 1.73 (both p < .10), respectively.