Unhealthy Election Skepticism and Conspiracy Theories

In recent years, scholars of democratic politics have expressed concerns about the lacking depth of support among the mass public for democratic institutions and processes in Western democracies. First, studies have long shown that public trust is in short supply (Hetherington Reference Hetherington2005; Pew Research Center 2017). Second, a growing scholarship suggests that mistrust has transformed from generalized skepticism of government to noxious suspicions over political institutions that undergird democracy, such as the fairness of election systems (e.g., Norris, Garnett, and Grömping Reference Norris, Garnett and Grömping2020). These doubts are often sowed and magnified by political losers who spin conspiracy theories as a way to rationalize the electoral outcome, soothe their followers’ political anxieties and feelings of powerlessness, and stoke outrage (Douglas et al. Reference Douglas, Uscinski, Sutton, Cichocka, Nefes, Ang and Deravi2019; Federico, Williams, and Vitriol Reference Federico, Williams and Vitriol2018; Papaioannou, Pantazi, and van Prooijen Reference Papaioannou, Pantazi and van Prooijen2023). As partisan identities have aligned with other social identities and their effect on attitudes and behavior has strengthened (Mason Reference Mason2018), and the media environment has become more fragmented, insular, and partisan (Banducci and Karp Reference Banducci and Karp2003), skepticism, doubt, and conspiracy theories are easy to spread.

For the mass public, support for conspiracy theories satisfies important psychological needs such as a desire to understand and control the social world, maintain a positive image of one’s group, and account for beliefs that one’s group is victimized and undervalued (Douglas, Sutton, and Cichocka Reference Douglas, Sutton and Cichocka2017; Mashuri and Zaduqisti Reference Mashuri and Zaduqisti2014; Uscinski and Parent Reference Uscinski and Parent2014). Political elites can use conspiracy theories to mobilize their supporters and retain political power. As a result, after a political loss, the emergence of elite-promoted conspiracy theories can transform mild anxieties into deep hostility against the system of elections (Fahey Reference Fahey2022; Vail et al. Reference Vail, Harvell-Bowman, Lockett, Pyszczynski and Gilmore2022). This hostility can persist over time, especially if one’s side experiences repeated defeats at the polls (Daniller and Mutz Reference Daniller and Mutz2019).

Scholars warn that unwarranted skepticism of democratic institutions can have deleterious effects (Norris Reference Norris2014, Reference Norris2022). Democracy depends on the losing side’s consent; if losers do not accept their defeat or don’t believe that they have a fair chance to prevail in future elections, democratic systems may collapse (Anderson et al. Reference Anderson, Blais, Bowler, Donovan and Listhaug2005; Anderson and LoTempio Reference Anderson and LoTempio2002; Cantú and García-Ponce Reference Cantú and García-Ponce2015; Flesken and Hartl Reference Flesken and Hartl2018). Mistrust in elections can depress voter turnout and alienate citizens from the political system (Birch Reference Birch2010). Such doubts may also incentivize people to support authoritarian leaders and parties or endorse political violence (Kalmoe and Mason Reference Kalmoe and Mason2022; Rose and Mishler Reference Rose and Mishler2009).

The United States of the 21st century is not a stranger to election-related conspiracy theories and many of them have been pushed by right-wing elites and draw on White Americans’ racial anxieties. The elevation of the first African American to the U.S. presidency in 2008 was accompanied by conspiratorial theories that Obama was not a “natural born” American citizen, he lacked a U.S. birth certificate, and thus could not be the legitimate president of the country. Others falsely claimed that Obama was a secret Muslim which was meant to further impugn his legitimacy (Jardina and Traugott Reference Jardina and Traugott2019; Pasek et al. Reference Pasek, Stark, Krosnick and Tompson2015). Other conspiracy theories popular in 2016 and 2020 drew on the country’s changing racial dynamics. Some claimed that sinister elites are orchestrating the extinction of the White population and its replacement with racial and ethnic minorities and only Trump could stop this plan (Obaidi et al. Reference Obaidi, Kunst, Ozer and Kimel2022). Others held that millions of non-citizens had been allowed to participate in presidential elections skewing outcomes in favor of the Democratic party. When Trump lost to Joe Biden in the 2020 election, the former president promoted a conspiracy theory known as “Stop the Steal.” Promoters of this fiction, including popular media elites, argued that the election results in several states were illegitimate because various bad actors—from individual “illegal” voters to local election officials to companies that build voting machines—had committed fraud at the expense of (mostly White) Trump voters (Herron Reference Herron2023).

White Ingroup Processes and White Grievance

Scholars have long documented Whites’ mistrust in institutions. Historians and social scientists have argued that after the Civil Rights Revolution of the 1960s that included programs meant to ensure social and economic parity for people of color and women, White trust in government declined precipitously (Filindra, Kaplan, and Buyuker Reference Filindra, Kaplan and Buyuker2022). The political changes of the era gave rise to “an increasingly angry, bitter and frightened group of White people who feel persecuted and unrepresented [by institutions],” suggested analysts at the time (Aberbach and Walker Reference Aberbach and Walker1970, 64). The result was the racialization of public trust in institutions as more prejudiced and nativist White Americans came to view the government with unhealthy suspicion (Filindra, Kaplan, and Buyuker Reference Filindra, Kaplan and Buyuker2022; Macdonald Reference Macdonald2020). The Obama election not only made racial considerations chronic (Tesler Reference Tesler2016), but through conspiracy theories such as “birtherism,” it linked election-related beliefs, such as perceptions of election integrity, to racial priors (Appleby and Federico Reference Appleby and Federico2018; Jardina and Traugott Reference Jardina and Traugott2019).

The 1960s and 1970s also transformed how White Americans understood their ingroup. The ascendancy of people of color to positions of political power and the development of racial policies such as affirmative action increased feelings of victimhood and racial grievance among many Whites. This victimization tendency was expressed in terms of beliefs in “anti-White” or “reverse discrimination,” a term popularized by Ronald Reagan. To them, racial policies were not meant to produce racial parity, but instead a form of discrimination against their group (Fraser and Kick Reference Fraser and Kick2000; Smith and King Reference Smith and King2021). Many White people came to see themselves as the “silent majority” of virtuous citizens whose identities and interests were consistently and intentionally subordinated to those of “undeserving” others (Filindra and Kaplan Reference Filindra and Kaplan2016; Lowndes Reference Lowndes2016).

Changing population dynamics may have exacerbated White grievance. Since the 1990s, the country is rapidly turning into a majority-minority society. Many White Americans perceive the decline in the White population and the corresponding increase of minority groups as a threat to their social and political status. In response to these threats to the existing social hierarchy, victimhood was expressed as perceptions of discrimination against Whites soared (Norton and Sommers Reference Norton and Sommers2011; Peacock and Biernat Reference Peacock and Biernat2023). According to the Pew Research Center, 65% of White Americans today believe that their group faces discrimination at least some of the time and that such discrimination will persist into the future (Daniller Reference Daniller2021).

Going back to the classic work of Ted Gurr (Reference Gurr1970), scholars have long shown that perceptions of group victimhood are politically consequential, linked to social mobilization and support for violence. Grievances linked to perceptions of victimhood are also easily activated by political elites through inflammatory rhetoric that emphasizes the group’s status as a victim (Armaly, Buckley, and Enders Reference Armaly, Buckley and Enders2022; Armaly and Enders Reference Armaly and Enders2022). Scholars have shown that people can experience political dynamics as group-level trauma that targeted themselves or their ancestors and such perceptions affect their response to the political system (Volkan Reference Volkan2001). Studies show that groups often gravitate towards victimhood because such a status confers the aura of innocence and moral superiority (Gray & Wegner, Reference Gray and Wegner2011). Oftentimes, groups, including dominant groups, use victimhood to make political claims that are presented as standing for principles of justice (Horwitz, Reference Horwitz2018). Yet, such beliefs can have very harmful social consequences because when issued by dominant groups they can undermine the claims to justice of disadvantaged groups (Noor et al. Reference Noor, Shnabel, Halabi and Nadler2012). Claims of discrimination among dominant groups can contribute to intractable social conflict (Volkan Reference Volkan1998), denial of structural inequalities (Taylor Phillips and Lowery Reference Taylor Phillips and Lowery2015), and legitimization of existing social hierarchies (Danbold, Onyeador, and Unzueta Reference Danbold, Onyeador and Unzueta2022; Saguy et al. Reference Saguy, Chernyak-Hai, Andrighetto and Bryson2013).

The grievances rooted in Whites’ beliefs that racial policies and the country’s changing population dynamics victimized their group were further boosted by the Obama election. Obama’s elevation to the Presidency contributed to strengthening White Americans’ understanding of their racial identity in political terms, exhibiting ingroup solidarity and White political consciousness, which are distinct from prejudice (Jardina Reference Jardina2019). At the same time, many Whites have embraced the belief that efforts to address racial inequality are a form of anti-White discrimination and understand themselves as innocent victims (Isom et al. Reference Isom, Boehme, Cann and Wilson2022). Such beliefs have strengthened overtime (Wellman, Liu, and Wilkins Reference Wellman, Liu and Wilkins2016; Wilkins et al. Reference Wilkins, Wellman, Babbitt, Toosi and Schad2015).

Already during the 2016 campaign, anticipating that he may lose to Hillary Clinton, Trump appealed to White grievance while sowing doubt about the integrity of the election process. However, this narrative peaked during and after the 2020 election when the former president and his representatives cast suspicion on Joe Biden’s victory and the legitimacy of the country’s election system. Trump-aligned elites claimed that shadowy elites conspired with companies that make voting machines to enable illegal voters to cast ballots or to alter the votes of Trump supporters. More recently, Trump drew on White grievance when he proclaimed: “I am your warrior. I am your justice. And for those who have been wronged and betrayed: I am your retribution” (The Guardian 2023). As a result of the “Big Lie,” almost half of all White Americans—the country’s largest racial group and the bulk of Trump’s supporters—expressed low confidence in the 2020 election outcome. According to data from the Pew Research Center, in November 2020, 49% of White Americans expressed a lack of confidence in the outcome of the mail-in vote, 24% doubted the in-person vote, and 15% believed that their own vote was not counted accurately.

It is not surprising that election-related conspiracies are more prevalent on the Right. The Republican party has lost the popular vote in every presidential election since 1992 with the single exception of 2004. Its victories are the result of the structural imbalances of the American electoral college than its popularity among the public. Since the Republican party is also more racially homogeneous with the majority of its voters being non-Hispanic Whites, the fear of electoral loss may motivate stronger conspiracist beliefs among White Republicans who feel victimized and powerless, such as people with high levels of White grievance (Douglas et al. Reference Douglas, Uscinski, Sutton, Cichocka, Nefes, Ang and Deravi2019; Papaioannou, Pantazi, and van Prooijen Reference Papaioannou, Pantazi and van Prooijen2023). This may be especially the case in the aftermath of the 2020 election when Trump drew on White grievance to spread conspiracy theories about the integrity of the election outcome. However, even though the conspiracy theories literature emphasizes the role of ingroup identity processes as central to support for political conspiracies such as election denialism (Armaly, Buckley, and Enders Reference Armaly, Buckley and Enders2022; Douglas et al. Reference Douglas, Uscinski, Sutton, Cichocka, Nefes, Ang and Deravi2019), scholars have focused on outgroup attitudes (Appleby and Federico Reference Appleby and Federico2018) and less is known about how White grievance may influence election skepticism.

Measuring White Grievance

White grievance, the belief that Whites are victims of discrimination, is analytically distinct from Jardina’s (Reference Jardina2019) White consciousness and likely constitutes a separate dimension of White ingroup identity, capturing the relational dimension of identity rather than ingroup affect or solidarity (Abdelal et al. Reference Abdelal, Herrera, Johnston and McDermott2006; Filindra, Buyuker, and Kaplan Reference Filindra, Buyuker and Kaplan2023; Fording and Schram Reference Fording and Schram2023). White grievance is rooted in relative deprivation and captures feelings of victimhood that Whites may develop when comparing their group’s social and political status across time and finding it to have declined (Armaly and Enders Reference Armaly and Enders2022; Isom et al. Reference Isom, Boehme, Cann and Wilson2022).

By contrast, White consciousness as designed by Jardina (Reference Jardina2019) captures positive ingroup affect, salience of group membership, solidarity with the ingroup, and to a degree outgroup hostility (Fording and Schram Reference Fording and Schram2023). First, the “White identity importance” item lacks the negative emotion associated with deprivation and grievance. Second, some liberal Whites may express high levels of White identity but combine it with an appreciation for White privilege (Cole Reference Cole2020). As a result, a number of studies have noted that White identity is not always a strong predictor of White conservatism in vote choice and other preferences, and it operates contrary to expectations in some contexts (Buyuker et al. Reference Buyuker, D’Urso, Filindra and Kaplan2021; Filindra, Buyuker, and Kaplan Reference Filindra, Buyuker and Kaplan2023; Fording and Schram Reference Fording and Schram2020, Reference Fording and Schram2023).

A factor analysis based on the 2012–2020 ANES shows that White grievance does not load on the same factor dimension as White identity, the White thermometer, or various outgroup measures such as racial resentment, or minority group thermometers (Appendix Tables A1a–c). The correlation between White identity and White grievance is positive but low (r = 0.13) which further supports the contention that the two tap distinct aspects of ingroup identity processes (Filindra, Buyuker, and Kaplan Reference Filindra, Buyuker and Kaplan2023). This analysis also suggests that White grievance is not another measure of outgroup hostility or bias, as some have suggested (Fording and Schram Reference Fording and Schram2023). Although Whites who express high levels of grievance may be prejudiced against outgroups, this is not necessarily the case, as indicated by the modest correlations between measures of outgroup prejudice and White grievance (on the low correlation, also see: Jardina Reference Jardina2019).Footnote ¹

Although beliefs that White Americans face discrimination are objectively untrue, this perception of victimization is not equivalent to a conspiracy theory nor a component of conspiratorial beliefs. In this case, White Americans are making inferences of intent based on government policy and action, so feelings of White grievance are the result of an interpretation of government policy. This is important because if White grievance was not an attitude but another conspiratorial belief, we would be establishing a circular argument, using one measure of conspiracism (White grievance) to explain another (election skepticism).

Analytically, we can document that conspiracism is distinct from White grievance by leveraging the 2020 ANES. That survey has two items measuring conspiracy beliefs. One item asks: “How well does the following statement describe your view? ‘Most business and politics in this country are secretly controlled by the same few powerful people’.” The item is scored on a five-point scale from “not at all well” to “very well” (M = 0.529; SD=0.284). The second item asks: “How well does the following statement describe your view? ‘Much of what people hear in schools and the media are lies designed to keep people from learning the real truth about those in power’.” This item is scored on the same scale as the one above (M=0.447; SD = 0.313). A factor analysis that includes the two conspiracy theory items and the three White ingroup measures (grievance, thermometer, and identity/consciousness) shows that the White ingroup processes measures do not load on any dimension (Appendix Table A2). This supports our contention that White grievance is not another expression of conspiracist beliefs. This is also the case for the other measures of ingroup identity.

Hypotheses

Our study tests two hypotheses:

Data and Methods

The study utilizes data from two sources: The ANES (2012–2020) and a 2021 YouGov national survey of 1,000 non-Hispanic Whites. For all four datasets, the sample is restricted to non-Hispanic Whites. We have recoded all variables on 0 to 1 scales consistent with the nature of the original scale. This means that an ordinal variable remains ordinal with values that range between 0 and 1. This allows us to interpret the coefficients as “maximum effects,” that is, the change in probability of the dependent variable associated with a switch from the lowest to the highest level of the independent variable. This facilitates the comparison of coefficients. Descriptive statistics for all variables from each dataset are in Appendix Tables A3–A6. All analyses are weighted.

The ANES has asked questions related to election fairness going back to the 1990s; however, it only started asking our key independent variable (White grievance) in 2012. As of 2012, the ANES has included a broader election integrity question, not tied to a given election. Specifically, it asked: “In your view, how often do the following things occur in this country’s elections: Votes are counted fairly?” (5-point scale). In that year, 18% of Whites expressed mistrust in election counts was 18%. In both 2016 and 2020, it stood slightly above 20%, which suggests relative stability over these three cycles. However, given the absence of a measure in 2008 and asking a different question in earlier waves, it is impossible to say whether overtime differences are due to an “Obama effect” (Tesler Reference Tesler2016). What is clear from the ANES data is that about one in five White Americans has had misgivings about the overall administration of American elections since at least 2012.

The 2012 ANES also included a question assessing belief about the fairness of election officials. Specifically, it asked: “In your view, how often do the following things occur in this country’s elections? Very often, fairly often, not often, or not at all often? Election officials are fair.” This was not included in 2016. The 2020 ANES included additional questions asked in the pre-election wave, so respondents’ assessments are based on their expectations, not an actual evaluation of the election itself. Therefore, these results reflect Trump’s efforts to prepare the ground for the “Big Lie.” Specifically, the ANES asked people: “In the November 2020 general election, how accurately do you think the votes will be counted?” (5-pt scale). More than a fourth of White Americans (27%) said that they expected the vote not to be counted accurately (“not at all” or “very little”). Furthermore, 21% said that they had no trust or very little trust in local election officials, and 15% said that people who are eligible to vote are denied the right “very often” or “fairly often” (Fig. 1-denoted as “no confidence”). Therefore, even though two-thirds of White Americans said that they believe votes are counted fairly in American elections overall, substantially fewer (about half) expected this to be the case for the 2020 election.

Figure 1. Election Fairness and Prospective Evaluations of 2020 Election (2020 ANES, Whites only)

Dependent Variables

We used six dependent variables in the analysis. The first is asked in all three ANES datasets in the post-election wave, while one other is specific to the 2012 ANES (but similar to one of the 2020 items), and the other three are specific to the 2020 election and were included in the 2020 pre-election ANES wave. The wording of all ANES-dependent variables is described above (see section: White Americans’ Confidence in Presidential Election Results) and can also be found in the Appendix. The dependent variable from the 2021 YouGov survey asked respondents how strongly they agreed or disagreed with the statement: “The 2020 election was conducted fairly and legitimately” (4-point scale). Unlike the ANES “vote fair” question, this question was specific to the 2020 election, and it was the only one that was fielded after the “Big Lie” was deployed. All dependent variables are coded on 0 to 1 ordinal scales, so that “1” represents the most trustful answer option.

Key Independent Variables

We measure White grievance, using an item from the ANES that asks: “How much discrimination is there in the United States today against each of the following groups? [Whites]” The answer categories are: “a great deal, a lot, a moderate amount, a little, and none at all.” The YouGov item was very similar (In American society, how much discrimination is there against each of the following groups? [White]. The categories are: “a lot, some, a little, very little, none at all.”). Given our argument, this item has strong face validity.

Furthermore, we include measures of outgroup attitudes. Our primary measure is the negative Black stereotype measure. The stereotype measure has strong face validity, and it does not suffer the problems associated with racial resentment, another popular measure of racial prejudice. Specifically, racial resentment is thought to conflate ideological conservatism and racial antipathy (Neblo Reference Neblo2009). That said, in the Appendix we include alternate models using racial resentment instead of the stereotype measure. These models produce very similar results. Given evidence that anti-immigrant attitudes may also be implicated in Whites’ mistrust in government and institutions (Filindra, Buyuker, and Kaplan Reference Filindra, Buyuker and Kaplan2023; Macdonald Reference Macdonald2020), we also specified models that included the Hispanic thermometer,Footnote ² the Muslim thermometer, and the illegal immigrant thermometer (Fording and Schram Reference Fording and Schram2020). The inclusion of these measures does not alter our main finding and it does not produce multicollinearity concerns (Appendix Tables A10–A12). Our controls include trust in people (Appleby & Federico Reference Appleby and Federico2018), authoritarianism (Stenner Reference Stenner2005), partisanship, ideology, Christian Protestant religion, age, income, education, and gender.

Analysis

1. Main effect models

Figure 2 shows the results of OLS regression models for the dependent variable that is common across all three ANES studies. For all models, VIF tests show no evidence of multicollinearity. For the first three models, the dependent variable is whether votes are counted fairly in American elections. This is the item that was asked in all three ANES surveys. It is important to note that the models differ in variance explained, which suggests that the variables in the model have become stronger predictors of mistrust in the vote count in 2020. Specifically, the 2012 model for “votes counted fairly” explains 14% of the total variance, the 2016 model 10%, and the 2020 model explains 33%. (For full results, see: Appendix Table A7).

Figure 2. How often votes are counted fairly (2012–2020 ANES)

As the models show, White grievance is negative and statistically significant in all three years. Racial prejudice (negative Black stereotypes) is significant and negative in 2012 but null in 2016 and 2020. Consistent with expectations, trust in people is positive and significant in all three years. Authoritarianism is negative and significant in 2012 and 2020 but not in 2016. Republicans are significantly less likely to think that votes in America are counted fairly in 2012 and 2020, but not in 2016. This may reflect Republican erroneous beliefs that Barack Obama was an illegitimate president because he was not a “natural born” citizen—also known as “the birther conspiracy”—a rumor spread by Donald Trump (Jardina & Traugott, Reference Jardina and Traugott2019). Conservative ideology is negative and significant only in 2020. College graduates and higher-income Whites are more likely to believe that votes are counted fairly across surveys.

Next, we turn to the relative substantive effects. In 2012, White grievance (−11ppts), racial prejudice (−17ppts), and trust in people (11ppts) are in a top tier of predictors. Tests of coefficients show that the effects of prejudice and White grievance are equivalent in magnitude. By comparison, the effect of partisanship (republicanism) is substantively smaller (−6ppts). In 2016, trust in people (16ppts) is alone in the top tier and White discrimination (−7ppts) is in a second tier. In 2020, White discrimination (−20ppts) is in the top tier along with partisanship (−22ppts) and these effects are statistically similar in magnitude. Trust in people (13ppts) and ideology (−12ppts) are in a second tier.

Figure 3 shows the results of OLS regression models of two similar questions, one from the 2012 ANES and the second from the 2020 ANES that assess perceptions of election officials (Also see Appendix Table A8). The models explain 13% and 16% of the total variance respectively. The 2016 ANES did not include a similar question. In both models, White grievance is negative and significant: the more one believes that White face discrimination the more likely they are to believe than election officials are not fair or trustworthy. Racial prejudice is also negative and significant in both models. In contrast, racial prejudice is not significant in 2020 when it comes to assessments of election fairness (Figure 2.3 above). Trust in people is positive and significant in both years, and conservative ideology is negative and significant as well. Partisanship is negative in both years but significant only in 2020. Authoritarianism is negative and significant only in 2012. Turning to the substantive effects, trust in people has the largest substantive effect in 2012 (14ppts) and it remains equally strong in 2020 (16ppts). The maximum effect of White grievance is in a second tier (−7ppts) in 2012 but in the top tier of effects in 2020 (−14ppts). A comparison of the two coefficients shows that the effect of trust in people and White grievance are equivalent in 2020 but not in 2012. The effect of racial prejudice is −12ppts in 2012 but only −7ppts in 2020. The two models are not directly comparable, but they suggest an overtime decline in the relative importance of racial prejudice and an increase in that of White grievance when it comes to judgments of election officials. The effect of conservative ideology is modest in both years (−8ppts and −6ppts respectively). Partisanship is null in 2012 but significant and with a top-tier effect (−11ppts).

Figure 3. Perceptions of election officials (2012 and 2020 ANES)

Figure 4 shows the results of two additional models that specifically ask about the 2020 election. The results in table format are in Appendix Table A7. The first models the item from the 2020 ANES pre-election wave which assesses respondents’ expectations about the accuracy of the 2020 election count. This model explains 13% of the total variance. The second is for the 2021 YouGov survey and asks whether the 2020 election was conducted fairly and legitimately, so this is a retrospective evaluation. The model explains 62% of the total variance. In both models, White grievance is negative and significant along with Republicanism and conservative ideology. Racial prejudice is significant only in the 2020 ANES model but not in the 2021 YouGov analysis. Trust in people is positive and significant in 2020 and authoritarianism is null (these are not included in the 2021 YouGov study). Turning to the substantive effects, in 2020 White grievance is in a top tier of effects along with partisanship, ideology, trust in people, and racial prejudice. A change from the lowest to the highest levels of White grievance in 2020 corresponded to a decline in beliefs that the elections will be conducted accurately of 7ppts. The substantive effect of the Black stereotype is −9ppts, Republicanism is −9ppts, and conservative ideology is 12ppts. In 2021, partisanship is alone in the top tier with a substantive effect of −50ppts. Ideology (−25ppts) and White grievance (−29ppts) are in a second tier but their effects are substantively quite large, nevertheless. There are differences in data collection methods and sample size, which means that the two models are not directly comparable. However, they suggest that White grievance, partisanship, and ideology may have become a lot more salient in White Americans’ judgments of the 2020 election as the “Big Lie” conspiracy unfolded after the election.

Figure 4. Expectations and evaluations of the 2020 election (2020 ANES and 2021 YouGov)

So far, the results are consistent with the first hypothesis. White grievance correlates with lower levels of trust in the integrity of the vote for the three ANES waves. The relationship is also negative and significant in the models that assess the 2020 election whether prior to the election (2020 ANES)—specifically the models assessing whether the vote in 2020 will be counted fairly and whether local election officials can be trusted—or half a year later (2021 YouGov). Furthermore, our robustness checks show that our findings are very similar even when we include alternate measures of outgroup attitudes, such as racial resentment and anti-immigrant attitudes. The inclusion of these items does not substantively affect the direction or size of the coefficients of White grievance and thus our inferences remain the same (see Appendix Tables A10–A12).

These additional models also show that White identity/consciousness is a positive and significant predictor of election fairness in the 2020 models. This indicates that people who exhibit higher levels of White consciousness are more likely to believe that elections are conducted fairly. This is the case whether it is included as the only ingroup measure in the models or in conjunction with White grievance. The White thermometer is also positive and statistically significant in two 2020 ANES models, but the substantive effect is very small, and it ceases being significant once White grievance is included in the model (Appendix Tables A9–A12). We return to this finding in our discussion.

Interactions with Partisanship

The second hypothesis is that in 2020, the effect of White grievance of beliefs about vote fairness should be stronger among Republicans. Here, we use a three-point version of partisanship for the moderator to facilitate interpretation of the results. Consistent with our expectation, the interaction between White grievance and partisanship is not statistically significant at conventional levels (p < 0.05) in 2012 or 2016. Appendix Table A8 shows the full models. Figure 5 shows the results of the interactions for 2020 and 2021, which are statistically significant, in graphical forms. The graphs compare the perceptions of election fairness among Democrats, Republicans, and independents across the range of White grievance. Our expectation was that among Republicans, expectations, and assessments that the 2020 election was fair would decline as a function of White grievance. For both datasets, we find this to be the case: beliefs in election fairness among Republicans are lower among those who score high on White grievance than those who score low. However, other groups are also affected: in 2020, the perceptions of election fairness decline among both independents and Democrats as a function of White grievance, though these effects are smaller than the effect on Republicans. In 2021, independents and Republicans show the largest effects, while Democrats trend in the same direction but the change remains within the margin of error.

Figure 5. (a) Conditional effect of White Grievance on Election Fairness by Party (2020 ANES) Predictive Margins with 95% CIs. Conditional effect of White Grievance on Election Fairness by Party (2021 YouGov) Predictive Margins with 95% CIs

Notes: Non-Hispanic Whites only. Data are weighted. All variables are recorded as 0–1 scales. Coefficients are maximum effects.

Lagged Dependent Variable Analysis

The 2016 and 2020 ANES formed a panel as 1,750 non-Hispanic White respondents responded to both surveys. Since our dependent variable was asked of these respondents both in 2016 and in 2020, we can specify a lagged dependent variable model by including the 2016 election confidence measure as a predictor of the 2020 election confidence measure. LDV models have the added benefit of helping to account for possible omitted variable bias that can influence the results and thus our interpretations. Therefore, the LDV analysis strengthens our causal inferences. Table 1 below shows the results of the LDV analysis. The first model includes a lagged dependent variable from 2016 along with the White grievance and negative Black stereotypes measure. The results show that even after accounting for the LDV, White grievance continues to be a negative and significant predictor of confidence in the administration of elections. It is also important to note that the Black stereotype measure is not significant in the LDV model, only in the cross-sectional analyses. The remaining models (Models 2–6) include alternate measures of ingroup and outgroup attitudes. These results also show White grievance to be negative and significant regardless of model specification. Furthermore, racial resentment is also negative and significant, but as we saw in earlier models, White identity is positive and significant. Measures of anti-immigrant and anti-Muslim prejudice are significant and consistent with expectations.

Table 1. Lagged DV models with alternate measures of outgroup and ingroup attitudes

These results provide reassurance that the relationship between White grievance and confidence in elections is robust and not the result of a confounder not measured in our models.