Explaining Partisan Gaps in Satisfaction with Democracy after Contentious Elections: Evidence from a US 2020 Election Panel Survey

Abstract

What effects do contentious elections have on partisan appraisals of democracy? We consider the case of the November 2020 US election, a highly polarized partisan contest but also an objectively free and fair election by credible accounting. We conducted a panel study embedded within two nationally representative surveys before and after the election. Results indicate a familiar but underexamined partisan gap, in which satisfaction with democracy decreases among Republicans and increases among Democrats relative to nonpartisans. We find that the gap is fully mediated by partisan shifts in satisfaction with elections and the news media that cover them. Our results underscore how eroding institutional confidence can undermine democratic legitimacy in hitherto consolidated democracies. To overcome partisan divisions following contentious elections, we highlight the need to bolster confidence in democratic institutions to reduce partisan fears and uncertainties—both rational and irrational—that electoral losses may trigger.

Although elections often occur without democracy (Diamond 2002; Levitsky and Way 2002), and democracy sometimes survives even in the absence of elections (Flores and Nooruddin 2012), the critical linkages between elections and democratic legitimacy anchor both democratic political theory and practice (Dahl 1956; Pateman 1970; Pennock 2015; Schumpeter 1950; Weber 2013 [1919]). Whereas democratic legitimacy presupposes free and fair elections, the processes by which public confidence in foundational institutions can strengthen or erode democratic governance require further attention.

We examined the linkages between electoral and democratic legitimacy in a panel study of public opinion before and after the US 2020 election. Consistent with a long line of comparative research on democratic-system support (Almond and Verba 1963; Easton 1965; Lijphart 1977), we observed a significant partisan gap in satisfaction with democracy between winners and losers following the election. This outcome often is amplified by majoritarian electoral institutions (Anderson and Guillory 1997; Bernauer and Vatter 2012; Farrer and Zingher 2019; Wells and Krieckhaus 2006), presidential systems (Anderson and LoTiempo 2002), extreme ideological polarization (Curini, Jou, and Memoli 2012), income inequality (Han and Chang 2016), narrow margins of victory (Howell and Justwan 2013), and polarized media coverage (Banducci and Karp 2003). Building on prior research, we found that the postelection winner–loser gap can be closed effectively by the mediating effects of perceptions of electoral legitimacy and satisfaction with the news media. Losers who have confidence in the electoral system, regard their elections as free and fair, and are more satisfied with the news media showed no partisan gap in appraisals of the quality of democracy. To overcome partisan divisions following contentious elections, we highlight the need to bolster confidence in democratic institutions to reduce partisan fears and uncertainties—both rational and irrational—that electoral losses may trigger.

We examined the linkages between electoral and democratic legitimacy in a panel study of public opinion before and after the US 2020 election.

To overcome partisan divisions following contentious elections, we highlight the need to bolster confidence in democratic institutions to reduce partisan fears and uncertainties—both rational and irrational—that electoral losses may trigger.

THEORY

The existence of a partisan gap in democratic satisfaction between winners and losers is a well-known phenomenon in the aftermath of elections (Anderson et al. 2005; Anderson and Guillory 1997; Blais, Morin-Chassé, and Singh 2017; Dahlberg and Linde 2017; Loveless 2021). However, scholars acknowledge that “the concrete causal mechanism behind the winner/loser gap in democratic satisfaction is not uncovered yet” (Howell and Justwan 2013, 335). The existence of the gap raises concerns about the ability of a democratic political system to achieve legitimate and peaceful transitions of power (Curini, Jou, and Memoli 2012; Easton 1965; Han and Chang 2016; Norris 1999). Closing the gap, especially among losers, is deemed critical to democratic stability (Anderson et al. 2005; Esaiasson 2011). ¹ Anticipating a gap, we tested the following hypothesis:

Hypothesis 1 (Partisan Gaps in Democratic Satisfaction): Satisfaction with democracy increases among partisan winners and decreases among partisan losers after an election.

To explain the partisan gap, scholars typically distinguish between rational/economic and psychological/affective mediators (Howell and Justwan 2013). ² The economic mechanism underscores how election losers fear economic losses as winners redistribute spoils of power to their advantage (Anderson and Guillory 1997). Mediation occurs when a partisan win/loss triggers positive/negative evaluations of future economic prospects that enhance/erode democratic satisfaction. The affective mechanism emphasizes that elections are emotionally driven identity struggles, in which winning boosts positive emotional responses (i.e., pride, self-esteem, and happiness) and losing increases negative affect (i.e., anger, fear, sadness, and anxiety) beyond any rational concerns about benefits or costs accrued because of the election (Campbell et al. 1980; Craig et al. 2006; Singh 2014; Singh, Karakoç, and Blais 2012; Thaler 2012). We tested the following hypotheses regarding the existence of partisan gaps after elections and their potential drivers:

Hypothesis 2 (Economic Mechanism): Increased concern about future economic losses among partisan losers relative to partisan winners mediates the partisan gap.

Hypothesis 3 (Affective Mechanism): Increased negative emotional affect among partisan losers relative to partisan winners mediates the partisan gap.

We also proposed an alternative psychological explanation for the partisan gap based on changes in institutional confidence following partisan electoral wins or losses. In Weber’s (2013 [1919]) framework, institutions are critical to legitimating democratic processes and outcomes. Satisfaction with and confidence in electoral institutions reflect beliefs about the legitimacy of the democratic process, and satisfaction with democracy legitimates democratic outcomes. Indeed, the voter-confidence literature demonstrates that individuals have greater satisfaction with democracy after elections when they trust that elections are conducted freely and fairly (Alvarez et al. 2008; Levy 2020; Norris 2014; Sinclair, Smith, and Tucker 2018). As the primary instrument for conveying electoral outcomes, the news media also can have an important role in facilitating trust and confidence in elections and democratic legitimacy, especially when political elites are not trusted to follow the rules or not perceived as credible conduits of information (Kerr and Lührmann 2017; Page, Shapiro, and Dempsey 1987). However, we view this mechanism as more psychological than institutional because trust and confidence may not necessarily reflect objective institutional performance. ³ Following cognitive dissonance theory (Festinger 1957), partisanship could increase cognitive dissonance between institution confidence and electoral outcomes, in which partisan losers project frustrations onto electoral institutions and democracy itself rather than acknowledge flaws with their preferred candidate. Conversely, partisan winners bestow greater confidence in electoral institutions and satisfaction with democracy because both process and outcome reaffirmed their partisan preferences.

Mediation thus occurs when a partisan win/loss triggers increases/decreases in institutional confidence that enhance/erode democratic satisfaction. We tested the following hypothesis:

Hypothesis 4 (Institutional Confidence Mechanism): Confidence in elections and the news media that cover them mediates the partisan gap.

In summary, we evaluated economic, affective, and institutional-confidence mechanisms for explaining the partisan gap in appraisals of democracy that often occurs following highly contentious elections.

RESEARCH DESIGN

The US 2020 election provides a compelling case for examining the impact of contentious elections on postelection partisan gaps in satisfaction with democracy. This election should have produced a strong partisan gap due to high-stakes presidential selection taking place under majoritarian electoral institutions with strong political polarization (Anderson and LoTiempo 2002; Curini, Jou, and Memoli 2012); rising inequality due to economic disruptions from COVID-19 (Han and Chang 2016); and margins of victory in flux on election night and the days following (Howell and Justwan 2013).

To evaluate partisan gaps in democratic satisfaction, we conducted a panel survey embedded within two nationally representative online surveys conducted before and after the election. Panel data provided advantages for causal inference over pooled cross-sectional data due to the ability to control for potential time-invariant confounders (Hsiao 2014). We used the following step models to estimate the effects of the election on changes in partisan satisfaction with democracy:

(1) $$ {\varDelta Y}_{it2\hbox{-} it 1}={\beta}_0+{\beta}_1\left( Party\;{ID}_i\right)+{X}_i+{e}_i $$

(2) $$ {\varDelta Mediator}_{it2\hbox{-} it 1}={\beta}_0+{\beta}_1\left( Party\;{ID}_i\right)+{X}_i+{e}_i $$

(3) $$ {\varDelta Y}_{it2\hbox{-} it 1}={\beta}_0+{\beta}_1\left( Party\;{ID}_i\right)+{\beta}_2{\left(\varDelta Mediator\right)}_{it2\hbox{-} it 1}+{X}_i+{e}_i $$

where ΔY_it2-it1 is the dependent variable measuring the change in satisfaction with democracy for individual i between time t=2(postelection) and t=1(preelection). We used the individual time-invariant PartyID_i to capture the effect of the election on changes in partisan democratic satisfaction (i.e., the partisan gap). ⁴ ΔMediator_it2-it1 examines the time-variant mediating effects of changes in economic, affective, and/or institutional-confidence variables on changes in democratic satisfaction from before to after the election. X_i is a vector of extended controls (see the online appendix robustness checks). Step 1 examines the direct effect of the changes in satisfaction with democracy by partisanship. Step 2 estimates the relationship between partisanship and the proposed mediator. Step 3 captures the indirect effects of the mediation pathway on the direct effects observed in Step 1, as illustrated in figure 1.

Figure 1 Mediator Model of Partisan Electoral Effects on Satisfaction with Democracy

Finally, due to the observational nature of our data, causal inferences regarding pathways and mechanisms should be considered as exploratory guideposts for future research. The online appendix further discusses mediation analysis.

SAMPLING AND DATA COLLECTION

Our design received Institutional Review Board approval as an online survey to avoid the risk of spreading COVID-19. We commissioned Dynata for online data collection for our project. We requested that Dynata target several demographic characteristics to ensure that the sample was representative of the population of interest. However, we did not impose quotas, and subjects within each demographic stratum were selected randomly for the panel. The resulting sample is representative of US national-level diversity in gender, education, age, race, and ethnicity, as well as urban–rural demographics.

Data were collected between October 27 and November 1, 2020, for the preelection survey and between November 10 and 23, 2020, for the postelection study. A total of 1,564 respondents completed the study (i.e., 955 in wave 1 and 609 in wave 2), of which 504 completed both the preelection and postelection waves (Whitt et al. 2023). The resulting samples were well balanced across demographics and partisanship over time; however, voters were overrepresented in both waves. The panel sample is balanced on time-invariant controls, and panel attrition effects are minimal regarding democratic satisfaction. ⁵

RESULTS

We measured satisfaction with democracy, our dependent variable, using the following categorical survey item: “On the whole, are you very satisfied, somewhat satisfied, not very satisfied, or not at all satisfied with the quality of democracy in the United States?” Response options ranged from 1=not at all satisfied, 2=not very satisfied, 3=somewhat satisfied, to 4=very satisfied. Figure 2a-b lists mean and frequency distributions in democratic satisfaction change by party identification for panel-survey respondents from before to after the election. ⁶ Our panel sample consisted of 41.5% Democrats, 29.8% Republicans, and 24.4% Independents. During the US 2020 election, Democrats achieved a plurality of the popular and Electoral College votes, retained their majority in the US House of Representatives, and ultimately won control in the US Senate in Georgia’s runoff election. Among Democrats, 55.0% were somewhat or very satisfied before the election, and this percentage increased to 66.8% after the election, whereas Republican satisfaction with democracy decreased from 66.0% to 56.7.3% after the election. In contrast, Independents—whose satisfaction with democracy was lower than either Democrats or Republicans—did not change following the election (i.e., 43.4% before, 43.9% after). Comparing means using matched-pair t-tests, Democratic satisfaction increased (t=4.03, p<0.0000), Republican satisfaction decreased (t=-3.62, p<0.0002), and Independents did not change (t=0.12, p<0.4543) from before and after the election.

Figure 2 Changes in Democratic Satisfaction by Party Identification

Consistent with Hypothesis 1, we found a clear partisan gap between electoral winners and losers and a reversal of the preelection partisan gap that existed when Republicans controlled the presidency and the Senate. The preelection gap favored Republicans with +11.0% more democratic satisfaction, whereas the postelection gap reversed so that Democrats had +10.1% more satisfaction—an overall swing of 21.1% from before and after the election.

We now discuss evaluations of our mediator hypotheses in table 1, which reports direct partisan electoral effects (Step 1) on changes in democratic satisfaction and other plausible indirect mediator variables using ordinary least squares (OLS) regression. Model 1 indicates the direct effect of how Democratic satisfaction with democracy increased and Republican satisfaction declined following the election relative to Independents, the constant comparison group. We then consider whether the election produced comparable changes in other mediator variables that might explain the partisan gap (Step 2). We examine economic explanations of the partisan gap (Hypothesis 2) by measuring how an electoral win or loss affects changing concerns about economic self-interest. We asked respondents before and after the election whether they expected the economy to 1=get better, 2=stay the same, or 3=get worse during the next 12 months. We also asked whether they thought their and their family’s income would get much or somewhat better, stay the same, or get much or somewhat worse (1–5 range). ⁷ Models 2a and 2b show limited support for Hypothesis 2. Democrats appeared less concerned about the general economy after the election but not their personal income, whereas Republicans and Independents were unchanged following the election. The modal response for both items was that the economy and personal income would remain the same. Panel fixed effects control for different levels of personal income among respondents.

Table 1 Partisan Satisfaction with Democracy and Plausible Mediators (OLS Regression)

Notes: Robust standard errors are in parentheses. ***p<0.01, **p<0.05, *p<0.1.

We evaluated Hypothesis 3 using an emotional-inventory battery as the dependent variable. We asked respondents: “When you think about the direction the country is headed today, how does it make you feel?” Emotional items included negative affect (i.e., angry, afraid, sad, and worried) as well as positive affect (i.e., happy and proud), with response options ranging from 1=not at all to 5=extremely. We combined the four negative affect responses into an index, which factor analysis showed to align on a single latent negative-affect dimension (Cronbach’s alpha=0.87). Model 3 indicates that negative affect increased significantly after the election for Republicans and decreased for Democrats relative to Independents. Results from the positive affect variables (not shown) indicate that Republicans became less happy after the election, with no change for Democrats and Independents. Compared to Hypothesis 2, there is stronger evidence of a potential mediating effect of affective psychological mechanisms considered by Hypothesis 3.

Finally, we evaluated changes in institutional confidence from before and after the election as dependent variables in Models 4a-4c. We measured confidence in elections and approval of the media using two items. The first was a satisfaction index in which respondents were asked about their satisfaction with US elections and the news media, with response options ranging from 1=not at all satisfied to 4=very satisfied. “Somewhat satisfied” was the modal response to both items. We also included a second item for electoral confidence in which we asked respondents if they agreed that “American elections are free and fair,” with response options ranging from 1=strongly disagree to 4=strongly agree. “Somewhat agree” was the modal response. Models 4a-4c show increased partisan gaps in satisfaction with American elections and the belief that American elections are free and fair, as well as decreasing satisfaction with the news media only among Republicans. Consistent with Hypothesis 4, these items provide a plausible explanation for changes in the partisan gap in democratic satisfaction following the election.

The mediator models (Step 3) in table 2 examine how the inclusion of our economic, affective, and instutional-confidence mediator variables impact winning and losing partisians’ appraisals of democracy. The base model in column 1 reports the direct (Step 1) effect of the election on changes in democratic satisfaction by party identification. Models 2 and 3 indicate that economic or affective mediators minimally account for partisan gaps in democratic satisfaction. Models 4a-4c and the combined Model 5, however, show that the partisan gap was reduced significantly by the inclusion of institutional mediators. The formal mediation analysis results in figure 3 reveal that the mediating effects are stronger for electoral than for media satisfaction. Further robustness checks are provided in the online appendix, including formal and implicit mediation analysis; structural equation modeling; and controls for the effects of COVID-19, voting behavior, candidate approval, and other time-variant and invariant controls, which we found to be limited.

Table 2 Mediators of the Postelection Partisan Gap (OLS Regression)

Notes: Robust standard errors are in parentheses. ***p<0.01, **p<0.05, *p<0.1.

Figure 3 Percentage of Partisan Gap That Is Mediated by Institutional Confidence

CONCLUSION

Our study finds a clear partisan gap in satisfaction with democracy, raising concerns about partisan divides over the democratic legitimacy of the US 2020 election. These gaps reflect an emerging trend in American political polarization compared to previous eras. ⁸ Our results indicate that this gap is not driven primarily by concerns about economic self-interest or affective responses to the election’s outcome. Elections like the US 2020 case do not appear to be simple economic referenda or rousing “horse races”. Instead, we find that polarization is more foundational: the partisan gap is explained most directly by changing confidence in the electoral system and not simply a decline in confidence in how news media covers those elections. Partisan Republicans showed marked declines in satisfaction with democracy and the electoral process following the 2020 election, including increasing doubt that American elections are free and fair. This finding reinforces the broader literature concerning the growth and consequences of partisan political polarization in the United States (Graham and Svolik 2020; Levitsky and Ziblatt 2019). Such polarization poses challenges if partisan losers project frustrations onto electoral institutions and democracy itself, consistent with cognitive dissonance theory. However, this projection also may underscore the existential fear that winners in a polarized landscape will reshape democratic processes and outcomes to their permanent advantage. At the same time, our research suggests tentative paths to improving democratic satisfaction by increasing confidence in electoral institutions. Although this may not be accomplished easily in the short term, this path offers encouragement that partisan gaps in support for democracy potentially can be overcome through unity-building efforts after a contentious election. It also presages potential threats to democratic legitimacy if partisan divisions in public confidence in the US electoral system go unheeded.

Partisan Republicans showed marked declines in satisfaction with democracy and the electoral process following the 2020 election, including increasing doubt that American elections are free and fair.