Breaking it is one thing, fixing it is another: Responsibility attributions and support for the U.S. Presidency, Congress, and the Supreme Court

Abstract

When do attributions of responsibility predict support for governmental institutions? Data from an online survey of 508 United States citizens examining attributions for their most salient national problem revealed associations between attributions and evaluations of the Presidency, Supreme Court, and Congress. With respect to the Presidency, causal attributions were related to more negative evaluations, while resolution attributions were related to more positive evaluations. This was especially true for conservative participants. With respect to the Supreme Court, causal attributions were related to more negative evaluations. With respect to Congress, there was an association between causal attributions and negative evaluations, but only for those with a low level of resolution attributions. Results are discussed in terms of the political climate and differences in institutional expectations.

Keywords:

• Attributions
• In-group bias
• Ideology
• Congress
• Presidency
• Supreme Court

Notes

1 Since we analyze a convenience sample rather than a national probability sample, we utilized procedures recommended by Cochran (1977) and Sousa, Zauszniewski, and Musil (2004) to assess whether our sample falls within the average variability range of standard socio-demographic measures in the general population as approximated by the 2006 GSS. Analysis revealed that while our sample does fall within the average variability range for gender, it contains more than the average variability range of Caucasians and college graduates, and Liberals. Additional analyses revealed that level of education and level of political interest did not significantly moderate the results, with one exception: Result patterns regarding support for the Supreme Court were more pronounced among participants with higher levels of education.

2 It should be noted, however, that this effect may be attenuated for Congress due to the party control change from Republicans to Democrats in January 2007, only two months before our survey. Thus, our study may underestimate the role of political attitude regarding Congressional attributions of responsibility. The Presidency, however, should clearly be seen as controlled by Conservatives, while studies of editorials on the Supreme Court indicate that the institution currently has a clearly Conservative bent (Segal, 2006).

3 To determine whether this was the case in our data, we conducted follow-up analyses to test for the possibility that participants’ levels of support and attributions of causal responsibility to the Presidency were related to which most important issue participants chose. We suspected that among more conservatives (though not liberal respondents) levels of support and attributions of causal responsibility would vary depending upon which most important issue participants chose. Confirming our suspicion, the data showed no significant association between these variables for liberal and slightly liberal participants. Among moderates, though, there was a significant association with the Presidential feeling thermometer (F(8, 137) = 2.13, p < .05) and Presidential responsibility attributions (F(8, 137) = 2.63, p < .05). Follow-up Tukey multiple comparison tests revealed that the association with the Presidential feeling thermometer was due to differences between people choosing terrorism (M = 73.75, SD = 26.69) on the one hand, and Iraq (M = 30.00, SD = 30.4) or dissatisfaction with government (M = 28.85, SD = 26.43) on the other hand. This suggests that what we observe is not the difference between foreign policy and domestic policy issues, but one that centers around disenchantment with the Presidency stemming from President Bush's handling of the Iraq war and how he conducted government business more generally, contrasted with continued support among moderates for whom terrorism the most important issue facing the country.

4 We have mitigated these representativeness concerns about sample by controlling for socio-demographic background data and zero-centering the main independent variables. Furthermore, we conducted additional analyses to examine whether education and interest in politics might moderate the pattern of results in the current study. We found that the pattern of results with respect to Congress was especially pronounced among more educated participants. However, we did not find that political attitude or education moderated any of the other results. Finally, we utilized a conservative approach in interpreting the data by focusing merely on patterns of (non)significance of responsibility attributions, rather than absolute effect sizes.