“That’s not how it works”: economic indicators and the construction of partisan economic narratives

ABSTRACT

This study examines the processes through which partisans update their (biased) economic judgments during periods of mixed and asymmetric economic performance. I show evidence that citizens express relatively unbiased perceptions of the movement of the stock market, suggesting that partisans do not engage in processes of motivated reasoning when reporting judgments of widely available economic data. Instead, partisans respond to fluctuations in stock market performance by revising their assumptions about the way the economy works: in response to positive or negative developments, the stock market is perceived to be more or less important for the health of the broader US economy depending upon Americans’ partisan worldviews. This form of biased narrative construction has substantial importance in light of a “two-speed” post-Great Recession economy.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes on contributor

Ian G. Anson is Assistant Professor at the Department of Political Science, University of Maryland, Baltimore County (UMBC). His research interests include the study of American public opinion, economic voting, media effects, the politics of inequality, and the scholarship of teaching and learning.

ORCiD

Ian G. Anson http://orcid.org/0000-0002-1545-4270

Notes

1 Following an alternative logic in which partisans’ economic perceptions are assumed to be sincere, the above expectations may seem entirely unsurprising, if not tautological. For example, if a Republican believes the stock market has improved despite weak overall economy, she would almost certainly argue that the stock market is relatively unrelated to overall economic progress – because this belief has been substantiated by objective conditions. However, if we assume that the overall economic perceptions of partisans are mostly expressive and not sincere, it would be possible to observe less logically consistent scenarios across economic and political contexts. For instance, in a pre-Great Recession period overseen by a Republican incumbent, Democrats should be less likely to believe the stock market is important, even if expressing positive perceptions of the stock market and the economy as a whole. These hypotheticals demonstrate the need to assess the current theory across a variety of political and economic contexts.

2 A cursory search of Google News confirms the relative infrequency of unemployment news reports relative to stock market coverage: across the ranges spanning 1 January 2010 to 1 January 2016, the Google archives displayed 63 pages of representative results containing the phrase “stock market” in the headline (each representing thousands of individual headlines), compared to 39 pages of “unemployment” headlines and 33 pages of “wage” headlines.

3 It may be that partisans are also incidentally exposed to other frequently aggregated economic statistics such as prices, though the present study cannot assess this argument.

4 Keyword searches were employed for the phrase “stock market”. The codebook of each survey including this phrasing was inspected for the use of the words “economy” and “party”. This process yielded the full dataset described in-text.

5 Economic voting literature strongly indicates that the temporal nature of such evaluations can affect the ways in which they are both constructed and utilized to form performance evaluations (e.g. Erikson, MacKuen, and Stimson 2000; MacKuen, Erikson, and Stimson 1992), meaning we may expect differences in partisan bias across these measures.

6 These differences in response options should have little effect on the comparisons of partisans within each survey, which is the principal goal of the present analysis. Therefore, the presentation of standardized coefficients is intended to allow for more convenient presentation, not to persuade the reader that the questions are fully comparable from survey to survey.

7 Question wording varies slightly across surveys, but reflects a three-category item in all surveys. Volunteered responses of “don’t know” and “another party” were coded as Independent.

8 The surveys are the CBS News Monthly Poll #1, August 2003; and CBS News National Poll, March #2.

9 Question wording for the CBS News National Poll, March 2013 read as follows: “In general, would you say that the condition of the stock market is a good indicator of the condition of the national economy overall, or not?” Response options were coded as “yes” (1) or “no” (0); responses of “don’t know” were excluded from the analysis. Question wording for the CBS News Monthly Poll, August 2003 was as follows: “How much impact does what happens in the stock market have on the condition of the national economy?” Response options included “major”, “minor”, and “no” impact on the national economy. This coding was collapsed into two levels (major and minor/none) due to the extremely low prevalence of “no impact” responses (4 of 266 respondents).

10 In these models, we can reject the hypothesis that the effect of partisanship was 0 at the p < .01 level.

11 This finding suggests that when provided with a strong party cue, stock market perceptions can become targets for expressive cheerleading. It remains unclear how sensitive these perceptions are to such cueing efforts, which constitutes an important subject for further research given the finding that partisans exhibit such low levels of cheerleading in Figure 2.

12 In fact, around 3% of Democrats went so far as to say the economy was “very good”, whereas a slim majority expressed an evaluation of “fairly good” (51%).

13 The stock market appraisal measure was coded on a similar four-point scale, ranging from “very good” (1) to “very bad” (4).

14 All else equal, the average Republican was around 29% likely to believe the stock market mattered in this model (95% CI: [0.23,0.35]); for the average Democrat, the likelihood was roughly 41% (95% CI: [0.35,0.47]).

15 Using the delta method, this difference in probabilities is significant at the p < .001 level.

16 One potential explanation as to why we might expect Republicans to exhibit this pattern more strongly than Democrats (as evidenced by Figure 2) is that Republicans, as the out-party, may be more attuned to the political consequences of economic developments for the incumbent, leading to a stronger desire to cheerlead (Nicholson 2012).