Who Moves Presidential Approval?
Download data and study materials
Sample size: 1255
Field period: 4/28/2005 - 12/7/2005
Without question, the most politically important surveyed opinion trend in the United States comes from the presidential job approval series. Presidential approval is an important source of bargaining leverage with Congress, a widely-used indicator of public support for the president’s policy initiatives, and an implicit gauge of the president’s electoral vulnerability. Besides being politically potent, movement in presidential approval ratings is also among the most thoroughly researched political phenomena in the social sciences. But despite the voluminous research into the aggregate movement of presidential job approval, social scientists still know very little about why individuals change their approval of the president over time, and how these individual changes drive aggregate shifts in job approval. The main reason we know so little about who is responsible for stabilizing or shifting the president’s approval rating is the lack of individual-level panel data designed to capture repeated measures of approval over periods when presidential fortunes are changing. This project collected a unique set of data using a quasi-experimental, interrupted time-series design that allow a definitive analysis of who moves presidential approval over the short run.
There are three research questions:
RQ1. Are the dynamics of presidential job approval driven by changes in opinion among in-partisans, out-partisans, or independents?
RQ2. Are the dynamics of presidential job approval driven by changes in opinion among well-informed or ill-informed respondents?
RQ3. Are the dynamics of presidential job approval driven by information conveyed through “routine” news coverage, by the occurrence of high-profile events associated with the president, or by something else?
H1. During periods of low-intensity news coverage, attentive citizens will be more likely to change their approval of the president than inattentive citizens.
H2. During periods of high-intensity news coverage, attentive citizens will be less likely to change their approval of the president than inattentive citizens, since even inattentive citizens will be exposed to news about high-profile events or controversies.
H3. Out-partisans and independents will be more likely to change their approval of the president than in-partisans.
H4. Strong partisans will be less likely to change their approval of the president than weak partisans, who in turn will be less likely to change their approval of the president than Independents.
None (this project used a quasi-experimental interrupted time-series design).
[PLEASE NOTE: The Codebook link at the top, as well as the information about sample size and field period, refer only to the PRE-stage. The POST codebook may be accessed by clicking on "Download all material relevant to this study" at the top. The sample size for the POST stage was 1077, and the field period was 6/9/2005 to 1/15/2006].
Presidential job approval, party identification.Additional Information:
The survey data collected for this study employed a rolling panel, interrupted time-series design to continuously track naturally-occurring, individual-level changes in presidential approval over an eight-month period. Beginning in late April of 2005, small random samples of 100 American adults were drawn every two weeks from the Knowledge Networks panel and prompted to complete a short “pre” wave survey instrument. In each of the pre-wave surveys, subjects were asked six questions (in this order): the standard Gallup presidential approval question, the standard Gallup partisanship question, the ANES “branching” question to gauge strength of partisanship, and three questions assessing levels of factual knowledge about national politics. Six weeks after each pre-test interview, a post-test interview was administered with the same mix of questions, except that three different knowledge questions were used (the complete survey questionnaire is reproduced in the appendix). A total of 16 panels were drawn in this way, each begun two weeks after the previous panel was started. Panels were staggered at two-week intervals so that at nearly any point in this eight-month period, one panel would have just started, one would be in the middle of its pre- and post-wave surveys, and one would be about to complete its post-wave questionnaire. Each of the 16 panel replicates was designed to be a random sample of adults in the United States.Summary of Findings:
(A) Independents no more likely to change than partisans; Democrats no more likely to change than Republicans
But...(1) Republicans more likely to consistently switch from approve to disapprove than either Independents or Democrats; (2)Partisan “defectors” more likely to change their approval than stable partisans, and this increases the apparent stability and extremity of partisan approval trends; (3)Democrats and Republicans changed at different times and in response to different stimuli; and (4) No news variable or major event in any test had a significant effect on Independents, whose change pattern appeared largely unrelated to ongoing developments
(B) Attentive citizens no more likely to change than inattentive citizens; both groups tended to become more disapproving
But...Attentive and inattentive citizens changed at different times and in response to different stimuli
1. Individual-level approval dynamics are more complex and conditional than suggested by any of the existing models of opinion change
2. No single stimulus seems to have a uniform effect on presidential approval across subgroups
3. Change is meaningful, but the meaning of change requires disaggregating the data
4. Subgroup trends can be deceptively smooth, due to partisan defectors
5. Offsetting changes can mask substantial amounts of movement underlying aggregate approval, often occurring within rather than merely across subgroups
Althaus, Scott. 2007. “Who Moves Presidential Approval? The Impact of News Coverage on Individual-Level Opinion Dynamics.” Paper presented at the 2007 annual meeting of the Midwest Political Science Association in Chicago, IL, April 12–15.