Correcting misperceptions of out-partisans decreases American legislators’ support for undemocratic practices

Abstract

There is substantial concern about democratic backsliding in the United States. Evidence includes notably high levels of animosity toward out-partisans and support for undemocratic practices (SUP) among the general public. Much less is known, however, about the views of elected officials—even though they influence democratic outcomes more directly. In a survey experiment conducted with state legislators (N = 534), we show that these officials exhibit less animosity toward the other party, less SUP, and less support for partisan violence (SPV) than the general public. However, legislators vastly overestimate the levels of animosity, SUP, and SPV among voters from the other party (though not among voters from their own party). Further, those legislators randomly assigned to receive accurate information about the views of voters from the other party reported significantly lower SUP and marginally significantly lower partisan animosity toward the other party. This suggests that legislators’ democratic attitudes are causally linked to their perceptions of other-party voters’ democratic attitudes. Our findings highlight the importance of ensuring that office holders have access to reliable information about voters from both parties.

American democracy is in crisis. International indicators of democracy have downgraded the United States, suggesting backsliding (1). Social scientists have documented voters’ high level of animosity toward out-partisans and their acceptance of undemocratic behaviors (2). There also is recognition that backsliding typically occurs via elites (3), or as one author puts it, “democracy erodes from the top” (4). In the United States, state-level elites are crucial: “episodes of democratic collapse at the state level have had profound reverberations for national politics” (5, p. 301). Examples of undemocratic state practices include partisan gerrymandering, voting restrictions, challenging federal court rulings, etc.

We offer the first study of American state legislators’ animosity toward the other party (PA), support for undemocratic practices (SUP), and support for partisan violence (SPV), with a preregistered four-condition survey experiment (N = 534; see Materials and Methods). Legislators answered questions about their PA (i.e., dislike of members of the other party), SUP (e.g., not accepting the results of elections their party loses), and SPV (e.g., justifiability of their party using violence to advance goals), all on 0 to 100 scales. Legislators were assigned to one of four experimental conditions such that prior to providing their own opinions, they estimated either 1) how the average voter from their own party would respond to PA, SUP, and SPV (same-party no correction); 2) how the average voter from the other party would respond (other-party no correction); 3) how the average voter from their own party would respond and then also received a “correction” with accurate information about the average same-party voter (same-party correction); or 4) how the average voter from the other party would respond and then also received a correction with accurate information about the average other-party voter (other-party correction). The correction data came from a nationally representative sample we previously conducted with the same items (6).

Results

We compare legislators’ own scores to the public’s scores, the latter of which came from the same survey as the correction data (6). We merge parties (Democrats and Republicans) since there are no party differences that alter the key results (SI Appendix). Across all three outcomes—PA, SUP, and SPV—legislators report significantly lower scores for themselves than the public’s scores (P < 0.01 for all outcomes), significantly overestimate (i.e., misperceive) the scores of other party voters (P < 0.01 for all outcomes), and either accurately perceive (P = 0.23 for PA) or significantly underestimate (P < 0.01 for SUP, P < 0.05 for SPV) the scores of same-party voters (Fig. 1 and SI Appendix). For instance, for SUP, legislators’ 11.20 is much lower than the public’s 26.40. And, relative to the 26.40 baseline, legislators overestimate the SUP of other party voters to be 48.13 while underestimating voters from their own party to be 19.44. In short, legislators exhibit less animosity toward the other party, less SUP, and less SPV, relative to the public. However, they also exaggerate the extent of these beliefs among voters from the other party (but not among voters of their party).

Fig. 1.

Legislators’ actual, legislators’ perceptions of partisans’ actual, and partisans' actual animosity, support for undemocratic practices, and support for partisan violence. Note: *The public support score merges (same and other) parties since differences between the parties are minimal (SI Appendix).

Correlational evidence from the no correction conditions suggests that legislators’ views are associated with what they believe voters think. For PA, legislators’ attitudes correlate with both same-party (r = 0.39, P < 0.01) and other-party perceptions (r = 0.46, P < 0.01). For SUP, legislators’ attitudes are significantly correlated with both same-party perceptions (r = 0.53, P < 0.01) and other-party perceptions (r = 0.31, P < 0.01) (P < 0.01 difference in correlations). For SPV, legislators’ attitudes significantly correlate with same-party perceptions (r = 0.42; P < 0.01) but not with other-party perceptions (r = 0.02, P = 0.82).

Highly exaggerated other-party perceptions of PA (5 percentage points) and SUP (more than 20 percentage points), along with their strong correlations with legislators’ own attitudes, raise the question of whether the corrections causally reduce legislators’ own PA and SUP (7, 8). We find that they do (Fig. 2). The other-party perception correction marginally significantly reduces legislators’ PA from 64.39 to 59.61 (P = 0.06), a 7% reduction; the SUP correction significantly lowers legislators’ scores from 11.28 to 7.76, a 31% reduction (P = 0.03). There is no significant reduction for SPV. The correction significantly reduces the correlation between legislators’ own attitudes and other-party perception, respectively (across the out-party conditions) from 0.46 to 0.17 (P < 0.01) and 0.31 to 0.15 (P < 0.01). For SPV, the analogous correlations are 0.02 and 0.12.

Fig. 2.

Impact of correction on legislators’ actual animosity, support for undemocratic practices, and support for partisan violence.

For PA and SUP, the corrections have larger impacts among those with more exaggerated other-party (mis)perceptions (see SI Appendix and ref. 9). In sum, providing accurate information about the other party’s PA and SUP causes legislators to lower their own levels of animosity (PA) (marginally) and SUP.

Discussion

These results may appear reassuring. State legislators report significantly lower scores on each indicator, relative to the public’s scores, and have fairly accurate views of voters from their party. Yet, the results also suggest that exaggerated misperceptions of the other party (as having more animus and being undemocratic than they actually are) endanger democracy by shaping elected officials’ own views, leading to higher levels of PA and SUP. If the same officials publicly state their undemocratic positions, that could increase public SUP (10). This could create a self-perpetuating dynamic where exaggerated perceptions create problematic attitudes among citizens and, in turn, among elites.

There is a clear benefit from elected officials learning the true attitudes of the other party’s voters so that they hold accurate views. Candidates invest substantially to learn about what their (potential) supporters believe to ensure election (11); once elected, it is important officials learn about those who did not support them as well. Otherwise, state legislators’ misperceptions may push them to continue to tolerate undemocratic practices such as partisan gerrymandering or voting restrictions—eroding democracy from the top. Just as the public benefits from bipartisan information (12), so do legislators: Learning accurate information about out-partisans leads to less undemocratic attitudes among elected officials.

Materials and Methods

We preregistered our study at https://aspredicted.org/z4hc5.pdf. We collected the data from June 22 to July 29, 2022, achieving a response rate of 7.8%, in line with our preregistered expectation of 7% (SI Appendix). Demographics of the sample are provided in SI Appendix, which also reports the results with a sample weighted to observables. Each respondent was in the same condition for each outcome, and the survey was always ordered from PA to SUP to SPV, meaning those in the correction conditions received a prior correction(s) in the latter two cases. Respondents answered perception questions, received a correction (or not), and then provided their own responses. Following Voelkel et al. (6), we measured PA with a feeling thermometer, SUP with the average of a four-item battery (α = 0.70), and SPV with the average of a four-item battery (α = 0.86) (on the measurement of SPV, see SI Appendix). Perception measures asked the same items for each battery, but from the perspective of an average member of the other party or of an average member of the same party. Full text of treatments and survey items are provided in SI Appendix. We used “an average member” of the other party since it was the only available data, and we suspect that it highly correlates with perceptions of other partisan populations (e.g., other legislators). Research was approved by the Northwestern University Institutional Review Board.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Data and analysis code files (in Stata) are available at Harvard Dataverse: https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/EJTXLY (13). All other study data are included in the article and/or SI Appendix.