In the paper “Beyond total effects: Exploring the interplay of personality and attitudes in affecting turnout in the 2009 German federal election” (Schoen & Steinbrecher, 2013), we analyzed the interplay of personality traits and pro-participation attitudes in affecting electoral turnout. Due to a miscoding of a multiplicative variable, the results on the interaction between agreeableness and party identification reported in Table 4 are wrong. The correct results are reported below in Table 1. They indicate a positive interaction between agreeableness and party identification. At low levels of agreeableness, party identification is not conducive to turnout. Highly agreeable persons, however, respond strongly to party identification when deciding upon turnout. This result suggests that agreeableness conditions citizens' inclination to obey to party attachments.
|Party identification (−2 sd. Agree.)||.02|
|Party identification (−1 sd. Agree.)||.17|
|Party identification (+1 sd. Agree.)||.45**|
|Party identification (+2 sd. Agree.)||.60**|
Given this interaction effect and the sizable effect of agreeableness on the strength of party identification (b = .16; S.E. = .03), jointly analyzing indirect and conditioning effects of agreeableness shows that agreeableness increases the strength of party identification and augments the latter's effect on turnout. Figure 1 includes three lines indicating the effect of party identification on turnout at three different levels of agreeableness. We assume a move from two standard deviations below the mean of agreeableness to two standard deviations above its mean. This change translates—while setting all other variables in the models to their mode, median, or mean, respectively—into an increase of the strength of party identification from −0.43 to 0.22. If the impact of party identification were independent of agreeableness, this increase would result—on the “unconditioned effect” line—in an increase in turnout by less than five percentage points (dashed arrow). Because agreeableness serves as conditioning variable, however, the actual effect is somewhat more sizeable. For those respondents located two standard deviations below the mean of agreeableness, the virtually flat line indicates the impact of party identification on turnout. Given the above results, however, a move from two standard deviations below to two standard deviations above the mean of agreeableness does not only result in an increase in the strength of party identification from −0.43 to 0.22. Rather, capturing the full effect also requires switching from the flat to the steep line (solid arrow). As a result, the combined effect of this change in agreeableness increases the likelihood of turnout by some 17 points.
In effect, the coding error prevented us from fully capturing the interplay of personality traits and pro-participation attitudes on turnout in the 2009 German federal election.