Shocks to the System: Electoral Manipulation, Protests and the Evolution of Political Trust in Russia

Abstract

How do electoral manipulation and resulting anti-fraud protests influence political trust in non-democratic contexts? I leverage the plausibly exogenous variation in the timing of a series of original surveys fielded on nationally representative samples in Russia to understand the impact of political shocks – particularly allegations of electoral fraud and post-election protests – on the evolution of trust in political institutions and individuals. This study demonstrates that allegations of excessive, blatant electoral fraud decrease trust in the autocrat. However, trust rebounds following attendant post-election protests. Finally, I examine the conditional impacts of fraud and protest on trust, finding that updating occurs primarily among those with weak political affiliation.

Russia frequently leads the world in lack of institutional trust (Edelman 2019). Lack of trust in political institutions can have important consequences for rules compliance (Murphy 2004), economic growth (Dasgupta 2010), and regime support and approval (Chen 2017; Hetherington and Rudolph 2015) and can contribute to protest and other forms of contention against the regime (Jian-Hu Zhu and Rosen 1993; Li 2004). Research to date has examined how economic performance (Colton and Hale 2009; McAllister and White 2008; Mishler and Rose 2001; Rose et al. 2011; Treisman 2011) and institutional and cultural factors influence views of political leaders and institutions (e.g. Citrin 1974; Hetherington 1998, 2004; Hetherington and Rudolph 2015; Mishler and Rose 2001; Shi 2001). However, political trust is not static and may be susceptible to political shocks, such as allegations of electoral manipulation and protest. How do electoral manipulation and resulting anti-fraud protests influence political trust in non-democracies?

I leverage the plausibly exogenous variation in the timing of survey implementation during the 2011–2012 Russian election season and subsequent protests to examine the ways in which allegations of electoral manipulation and resulting anti-fraud protests affect trust in political leaders and institutions. Contrary to previous research, I find that allegations of electoral fraud negatively impact trust primarily for the autocrat; trust in other government institutions remains consistent over the periods under examination. Furthermore, I find that trust in the autocrat rebounds following the subsequent anti-fraud protests. Finally, I examine the conditional impact of fraud allegations and protest on political trust, most notably for political affiliation. Here I find that electoral manipulation and protests lead to updating mainly among individuals with weak political affiliation – not regime opponents or supporters as prior research has shown (Aarslew 2023; Frye and Borisova 2019; Reuter and Szakonyi 2021). I also find that protests that occur as a result of fraud allegations, rather than further injuring the autocrat, can actually increase trust. However, as with allegations of electoral fraud, protests improve trust primarily among those with weak political affiliations.

This study has important implications for scholarly understanding of fraud allegations, protests and political trust in non-democracies. Why, when and to what extent non-democracies engage in electoral fraud has long been a question of concern for political scientists. However, the impact of allegations of fraud and protests on public opinion is difficult to untangle and, as a result, has been relatively understudied. First, this study demonstrates that even in contexts where electoral manipulation is common, particularly egregious, blatant fraud can negatively impact perceptions of the autocrat. Although some level of fraud may be seen as ‘business as usual’, autocrats must take care to avoid violating the rules of the game if they wish to not incite public ire.

Authoritarian regimes, particularly personalist ones such as Russia, frequently justify their rule based on the views of the leading authority (Buckley et al. 2022; Frye 2021). Therefore, it is essential to understand how different political shocks can shape trust in the incumbent. Political trust is often assumed to be slow to change, particularly in comparison to other attitudinal measures such as approval and support based primarily on evaluations of government performance (Easton 1965). This study shows that this is not necessarily the case and that political shocks – such as widespread allegations of electoral manipulation and resulting anti-fraud protests – can lead to rapid updating on the part of citizens. Moreover, attitudes across authorities and institutions are not uniform. Rather than assessing attitudes towards the government or political institutions generally, this study shows that political shocks impact trust in political entities in different ways and to different extents.

Theoretical expectations and hypotheses

This article examines the variable effects of: (1) allegations of electoral manipulation and (2) protest on trust in different political leaders and institutions in a particular non-democratic context. I lay out my theoretical expectations and hypotheses below.

Elections and allegations of electoral fraud

Elections in non-democracies provide information to the public about the relative strength and popularity of leaders and can influence decisions to participate in mass unrest (Blaydes 2011; Chernykh and Svolik 2015; Cox 2009; Egorov and Sonin 2018; Gandhi and Lust-Okar 2009; Gehlbach and Simpser 2015; Little 2012; Londregan and Vindigni 2006; Magaloni 2006; Rozenas 2012). It stands to reason that electoral outcomes that differ from general public expectations – those that result in an either better or worse performance by the regime – will lead citizens to update their beliefs about the popularity of the incumbent and the strength of the regime. As a result, non-democratic leaders are incentivized to manipulate electoral results to decrease the possibility of an unfavourable outcome.

Electoral manipulation can itself impact views of political incumbents. On the one hand, electoral manipulation can be seen as a signal of strength: by engineering strong victories, autocrats can deter potential challenges to the regime (Egorov and Sonin 2018; Geddes et al. 2018; Gehlbach and Simpser 2015; Little 2015; Magaloni 2006). According to Alberto Simpser (2013: xv), ‘Manipulating elections excessively and blatantly can make the manipulating party appear strong, while failing to manipulate in this manner can convey weakness.’ On the other hand, excessive manipulation may be viewed as a signal of elite weakness: it can demonstrate to the population that support is low and that the regime was forced to manipulate the election to artificially inflate support – particularly if regime performance is worse than expected. Non-democracies are more likely to commit fraud when they are vulnerable and do not expect to win elections without manipulation (Diamond 2010; Magaloni 2010).

Yet a third outcome is feasible. Instead of signalling regime strength or weakness, electoral manipulation may be viewed simply as ‘business as usual’ and have little impact on citizen attitudes towards the regime. Andrew Little (2015) found that electoral manipulation may have limited influence on perceptions of the government when it is viewed to be common practice. According to this view, when citizens have seen fraud in past elections, they will include fraud in their evaluations of political incumbents. However, all fraud is not created equal. While some citizens may believe that electoral manipulation is a given in elections (‘business as usual’), authorities must still abide by some pretence of rules. Blatant disregard for these implicit rules should result in a decrease in trust in political decision-makers, particularly when the regime performs worse than expected, even with manipulation. Indeed, as Little (2012: 254) argues, ‘Beliefs that there was a high level of fraud – all else equal – will lead to more anti-regime action.’¹

Existing research focuses primarily on how allegations of fraud influence prospects for anti-regime action (Bunce and Wolchik 2011; Daxecker et al. 2019; Lankina 2015; Lankina and Skovoroda 2017; Little 2012; Tucker 2007); this article focuses not on the likelihood of anti-regime protest, but rather on how allegations of fraud impact anti-regime sentiment. As Ora John Reuter and David Szakonyi (2021) show, overt use of electoral fraud can undermine the regime's core base of electoral support.² While anti-fraud protest may be more immediately detrimental for regimes, anti-regime sentiment can have important consequences for regime legitimacy and long-term stability (Chapman 2021, 2024; Gerschewski 2013; Reuter and Szakonyi 2021).

However, we would not expect to see a decrease in trust for all leaders and institutions.³ Rather, I expect that trust will decrease specifically for the parties deemed responsible for the fraud.⁴ Research has shown that when responsibility clearly lies with particular actors, they are more likely to be evaluated negatively (Iyengar 1991). Other studies, however, have argued that politically charged events impact trust in the government broadly, not just particular actors or institutions (e.g. Frye and Borisova 2019).⁵ It is not always a simple task to identify the party responsible for particular outcomes; rather, this calculus will vary based on the particular political context and issue area.⁶

As Henar Criado and Francisco Herreros (2007) argue, public opinion is most susceptible in systems where political power is highly concentrated in a few individuals or institutions, allowing people to more clearly assign responsibility for political and economic outcomes (see also Hobolt and Tilly 2014). In centralized non-democracies with more personalistic elements, such as Russia, responsibility is typically more clearly attributable to the autocrat. This is not likely to be the case for all issue areas: research has shown that poor economic performance is vulnerable to scapegoating (Sirotkina and Zavadskaya 2020), particularly when local leaders are elected instead of appointed (Beazer and Reuter 2019). However, poor economic performance does not clearly benefit any particular party; electoral manipulation, on the other hand, is specifically designed to benefit those in power. Consequently, I would expect allegations of blatant, excessive electoral fraud to lead to a decrease in trust in the autocrat – the individual seen as the primary political decision-maker and, thus, the one responsible for the occurrence of electoral manipulation – not all political institutions.

Hypothesis 1: Awareness of blatant electoral manipulation will reduce trust in the autocrat.

Anti-fraud protests

Protests that challenge the regime, such as those that emerge in reaction to allegations of electoral malpractice, may also shape political attitudes towards political institutions and leaders in non-democracies. On the one hand, citizens are more likely to hold positive attitudes of the regime if they believe that it enjoys widespread support (Buckley et al. 2022; Little 2017). Protests, therefore, may demonstrate to the public the malign nature of the regime and act as a signal regarding the true level of support in society (Hale and Colton 2017; Lohmann 1994; Tertytchnaya 2020), leading citizens who are aware of these protests to revise their views of the country's leaders. Furthermore, protests may reveal information to the public about ‘regime malpractices and abuses’, which may in turn suppress support for the regime (Tertytchnaya 2020: 1930). Moreover, how the regime responds to the protests – either through tolerance or repression – can lead to further updating: government repression in response to protests can further lower people's trust in government institutions and leaders (El-Mallakh 2020; Sangnier and Zylberberg 2017; Tertytchnaya and Lankina 2020).

On the other hand, some scholars argue that protests may improve attitudes towards the regime. Timothy Frye and Ekaterina Borisova (2019: 822) argue that protests that occur unexpectedly can reveal information not only about attitudes towards the regime but also ‘the willingness of the ruling elite to tolerate public opposition’. If citizens expect the government to react to protests with violence and the government represses the protests, attitudes towards the regime should either worsen or stay the same, depending on the expected and true level of repression. However, if citizens expect the government to react negatively towards the protest, but it does not, then trust may increase as citizens update their beliefs regarding the tolerance of the regime to dissent. In this framework, it is the government's reaction to unexpected protests that sends a meaningful signal to citizens, not the content of the protests themselves.

Alternatively, an increase in regime trust following protests may be evidence of ‘voting against disorder’ (Pepinsky 2017). Under this framework, the occurrence of protests may increase support for the existing order due to fear of instability. In a study of protests in Egypt, Nelly El-Mallakh (2020: 320) finds that greater exposure to protests ‘makes individuals more willing to give up rights in exchange for security and political stability’ and to vote for pro-regime candidates (see also Tertytchnaya and Lankina 2020). Under either mechanism, trust in the regime should increase when people are made aware of protest events. While the goal of this article is not to adjudicate between these two potential mechanisms, I include preliminary analyses in the Supplementary Material, Appendix C, finding greater but mixed support for the authority response mechanism than the fear of disorder argument. As laid out in the previous section, I would expect this positive updating to occur primarily for the autocrat.

Hypothesis 2: Awareness of anti-fraud protests will increase (decrease) trust in the autocrat.

Setting the stage: The Russian election season of 2011–2012

In September 2011, then Russian President Dmitri Medvedev announced in a televised interview that he planned to step down as president and not run for a second term in office. Medvedev's decision to step down, he said, would allow Prime Minister Vladimir Putin to run for the presidency – a position he had vacated in 2008. While Medvedev, in his own words, enjoyed ‘fairly high levels of trust … Prime Minister Putin is the most authoritative politician in our country, and his approval ratings are somewhat higher’ (Barry 2011).

In fact, this so-called ‘castling’ move angered many Russians, who believed that this decision demonstrated pre-planning on the part of authorities and robbed them of a true voice in the political process (McFaul 2014: 244). Medvedev himself stated that the castling move had been decided years in advance. Following the announcement, Putin was booed at a boxing match, and Communist Party delegates refused to stand when he entered the Duma (parliament). The lead-up to the December 2011 Duma elections saw a decrease in support for Putin, Medvedev and United Russia, the ruling party, according to public opinion polls.

The Duma election on 4 December was marred by widespread allegations of manipulation and fraud (OSCE 2012) beyond that previously seen in Russian elections. Local and international media reported serious irregularities in the voting booth before and after the election. Numerous cases of ballot stuffing and carousel voting were caught on camera and posted on YouTube (Ananyev and Poyker 2021). Election observers were disallowed from witnessing the sealing of ballot boxes, activists were harassed and local websites that attempted to expose the fraud were subject to cyberattacks. These, along with other forms of manipulation, led the Organization for Security and Co-operation in Europe (OSCE) to conclude that the election was subject to ‘undue influence of state authorities’ and that the state ‘did not provide the necessary conditions for fair electoral competition’ (OSCE 2012).

Although electoral manipulation is part and parcel of the Kremlin's menu of manipulation, the 2011 Duma electoral fraud was particularly egregious. Even with these administrative manipulations, United Russia received slightly less than 50% of the vote, down from 64% in 2007. This figure is doubtless inflated: Ruben Enikolopov et al. (2013) found that the vote total for United Russia was approximately 11 percentage points lower in areas in Moscow that did not have independent election monitors than in areas that had election monitors (see also Buzin and Lyubarev 2008; Kalinin and Mebane 2011; Mebane and Kalinin 2009; Myagkov et al. 2005; Shpilkin 2009; Vorobyev 2010).

More importantly, the election was perceived as being more unfair than previous Duma elections; 45% of Russians viewed the December 2011 Duma elections as being not fair or very unfair, compared to 20% for the 2007 Duma elections and 31% for the September 2016 Duma elections, according to public opinion polls (Figure 1). Similarly, 34% of Russians thought there were significant irregularities in the counting of votes in the 2011 Duma elections, compared to 9% in 2007 and 19% in 2016 (Figure 2). The 2011 Duma election was also considered to be more dishonest than the subsequent 2012 presidential election. According to estimates, approximately 60% of respondents thought that the presidential election was fair, while only 27% believed it to be partially or completely dishonest (Levada Analytical Center 2012a).

Figure 1. Fairness of Duma Elections.

Source: https://www.levada.ru/en/2016/11/09/parliamentary-election-results/.

Figure 2. Irregularities in Duma Elections.

Source: https://www.levada.ru/en/2016/11/09/parliamentary-election-results/.

While electoral manipulation allowed United Russia to maintain a narrow majority in the Duma, it ultimately led to widespread protests. On 10 December, approximately 50,000 protesters took to the streets in Moscow to denounce the flawed elections in the largest demonstration since the fall of the Soviet Union. Protests during this period ‘featured a clear target of blame’ (Tertytchnaya and Lankina 2020: 288): while the protests denounced the national government as a whole and, despite attempts by Putin and his government to shift the blame to lower-level actors, much of the ire was directed specifically at Vladimir Putin. Protesters called for Putin's resignation, rallying around cries of ‘Russia without Putin’, ‘Putin, go away’ and ‘Putin is a thief’. Despite not holding the presidency at the time of the protests, there was little doubt that Putin was still the dominant political actor and decision-maker, a belief reinforced by the castling move announced by Medvedev three months prior. Consequently, Putin was the primary target for blame surrounding the electoral manipulation and fraud.

Despite expectations of ‘an extreme state response’ (Smyth 2020: 174), the Kremlin's initial reaction was surprisingly mild. Authorities staged counter-demonstrations and downplayed the protests in the state media, but the protests were nevertheless permitted to continue relatively undisturbed. The Kremlin's rhetoric even began to soften over time: during Putin's annual widely viewed question-and-answer session on 15 December, he acknowledged that there was some dissatisfaction stemming from legitimate grievances about the political system and stated that he was glad that protesters were able to legally and peacefully voice their opinions. For his part, Medvedev proclaimed, ‘I treat any criticism of state institutions and individual officials with the utmost attention and respect’ in his 22 December state-of-the-union address. More importantly, the authorities' response to the protests was perceived as being appropriate and fair: 55% of survey respondents in the Democracy International Russian Election Surveys (DIRES) stated that they believed the authorities' response to the protest was just right while less than 10% thought the response was too harsh.

Large-scale protests continued well into the summer months of 2012, growing beyond Moscow and spreading throughout Russia. However, initial widespread support for the protests began to dwindle and opinions of the protesters themselves soured over the months following the initial protests of 10 December. In December, a plurality of respondents (44%) supported the protests. By the time the presidential election took place on 4 March, successfully restoring Putin to the presidency with 63% of the vote, support for the protests had decreased to 32% with the majority of respondents (54%) stating that no good had resulted from the protests (Levada Analytical Center 2012b).

Data and methodology

The Democracy International Russian Election Surveys (DIRES) were fielded by the Levada Analytic Center, Russia's leading academic survey organization, using a standard multi-stage cluster sampling approach to produce probability samples of respondents aged 18 and over. The surveys took place in four separate waves during the 2011–2012 Duma and presidential election seasons. This study focuses on the first two: the first wave was conducted directly prior to the 2011 Duma election (on 4 December) from 17 to 30 November while the second was conducted after the Duma elections from 9 to 22 December.

The analyses examine trust in institutions over three distinct periods: the ‘pre-Duma’ period corresponds to the first survey wave, which took place prior to the 4 December Duma elections; 50% of respondents (n = 1,202) were interviewed during this period. The second survey wave is divided into two periods for the sake of analysis: the ‘post-Duma’ period and the ‘post-protest’ period. Respondents in the ‘post-Duma’ period were interviewed in the short interval between the 4 December elections and the protests on 10 December (including the day of the protest); 6% (n = 141) were interviewed during this period.⁷ Respondents in the ‘post-protest’ period were interviewed in the first few days of the protests (11–22 December); 44% (n = 1,060) of respondents were interviewed during this period.⁸ This short period is particularly useful because it captures initial responses to the protests and the government's response while avoiding potential confounding factors related to the different phases of the protest movement over time or the government's changing response to the protests over time. Table 1 lists the time periods under examination and the number of observations in each time period. See the Appendices for more information.

Table 1. Information on Time Periods

Because the assignment of the day of the interview is plausibly exogenous to trust in political institutions and leaders, I attribute the differences in responses to the impact of the election and the protests. I assume that without reports of the extensive fraud in the Duma elections and without protests, the underlying baseline levels of political trust would have remained constant over all periods under examination. Reports of fraud and the protests were easily the most important events that took place in Russia during this period; furthermore, I am unaware of any other external shocks that could account for the variation in attitudes documented in this study.

Assumptions

To conduct the analyses, I assume that respondents interviewed after the Duma elections (‘post-Duma’) were aware of the results and allegations of fraud. The electoral manipulation was widely covered in domestic and international media and posted on social media; government officials, including President Medvedev, the prosecutor general, the first deputy interior minister and Putin's press secretary, all acknowledged the allegations in the days following the election. Therefore, it is reasonable to assume that respondents would have heard of the allegations.

I also assume that respondents interviewed directly after the protests had heard of their occurrence. Once again, the protests were widely covered by the popular media, and DIRES survey results in early 2012 show that the overwhelming majority of Russians had heard of the protests. However, it may be the case that some respondents were not aware of the allegations of fraud or the protests at the time of the survey. For example, some of the largest protests took place after the post-protest period covered in the survey. If some respondents were unaware of these events, it would artificially reduce the difference in responses between the groups. Consequently, if these assumptions are violated, the results would provide an overly conservative measure of the actual impact of these events on attitudes.

I also confirm that respondents across the different time periods are not biased in covariates that could affect the variables of interest. This is particularly important given the imbalance in the number of observations for the post-Duma period; because of the short time between the Duma elections and initial protests, fewer respondents were interviewed during this period than in the other periods under examination. I assessed balance with respect to respondent socioeconomic, demographic and location variables such as gender, age, income, education and locality. On average, the respondents were relatively similar across time periods, although there was some variation in income and education. As a result, I control for these two variables in the analyses; results are consistent with and without the inclusion of controls. See the Appendices in the Supplementary Material for results of randomization checks.

Finally, to explore how electoral manipulation and protest influence trust in different political actors and institutions, I assume that respondents have distinct and separate attitudes towards Putin, Duma and other political institutions, including the army, courts, local government and police.⁹ If respondents are unable to disentangle their attitudes towards Putin and other political entities, I would be unable to make any conclusions about the differential influence of events on political trust. Indeed, correlation coefficients show a weak to moderate correlation in trust in various institutions, suggesting that it is appropriate to examine views of these institutions separately. See Table A2 in the Supplementary Material for correlation coefficients.

Results and discussion

In this section, I examine whether there are any differences in trust in political leaders and institutions following the 2011 parliamentary election and the resulting protests. As argued above, I would expect the elections to lead to a decrease in trust for Putin specifically, followed by an increase in trust for Putin directly after the protests. I do not expect either event to influence views of political institutions. Table 2 depicts the mean and standard deviation for the pre-election, post-election and post-protest periods under examination for trust in Putin, the Duma, army, courts, local government and police. It also includes the result of t-tests between the three periods.¹⁰ Results are robust to ordinary least squares with fixed effects for income and education, robust standard errors and ordinal logistic regressions with fixed effects for income and education. See the Appendices for results.

Table 2. Trust in …

Notes: Table reports means for each period with standard errors in parentheses.

^a Difference in means between pre-Duma and post-Duma are statistically significant at p < 0.05.

^b Difference in means between post-Duma and post-protest are statistically significant at p < 0.05.

Difference in means between pre-Duma and post-protest are statistically significant at p < 0.05.

There is no evidence to suggest that the parliamentary elections and subsequent protests influenced public trust in Russia's institutions. Levels of trust for the Duma, army, courts, local government and police remain relatively stable before the Duma election (pre-Duma period), after the Duma election and before the protests (post-Duma period), and after the initial wave of protests (post-protest period). The results for Putin tell a different story. Following the fraudulent elections, Putin's rating dropped by nearly 0.4 points on a five-point scale – more than a 10-percentage-point decrease. However, his ratings rebounded in the immediate post-protest period to pre-election levels (p = 0.11).

Why do we see this difference for Putin but not for any of the other institutions, including the Duma, whose fraudulent elections sparked the protest movement? As demonstrated above, protests were not just critical of the regime as a whole, but specifically of Putin. Despite not holding the office of the presidency, there was little doubt among Russians that Putin was the key decision-maker in the Russian government – a belief doubtlessly reinforced by the castling move and evidenced by the specifically anti-Putin rhetoric of the protesters.¹¹ The Duma itself had little independence and was seen as a rubber-stamp for Kremlin directives – frequently being called the Kremlin's ‘Approval Ministry’ by critics. Moreover, Putin was then head of United Russia, the ruling party, and the party is closely aligned with Putin and his policies. As time went on, the regime explicitly engaged in blame-shifting surrounding the electoral fraud, putting the blame on local and other officials who would be punished by Putin. While this blame-shifting primarily occurred leading up to the presidential election, results in Table 2 suggest that trust in local government did not change as a result of the allegations of fraud, at least during the period under examination. However, just as Putin received much of the blame for manipulating the elections, so too did he reap the benefits of the post-election protests.

Alternative explanations

It may be the case that the data are vulnerable to preference falsification. Polling data in Russia have been shown to have an authoritarian bias that distorts how the public expresses its opinions (Rogov 2017). There is also evidence to suggest that reported electoral turnout is inflated in opinion polls (Kalinin 2016). However, scholarship during this period shows limited evidence of self-censorship in political attitudes in Russia. In a series of list experiments, Frye et al. (2017) found little to no evidence of preference falsification when asking about sensitive attitudinal topics, such as approval of Putin (see also Greene and Robertson 2019).

Even if bias does exist in the data, as long as the bias is consistent across time, the results would still be meaningful. That being said, it is plausible that social desirability bias may drive respondents to refuse to answer before the election when the regime's position was more secure; it held a majority of seats in the Duma and no elections had taken place to show people that support for the authorities was lower than expected. If that is the case, I would expect to see higher non-response before the Duma election.

In fact, the data show higher non-response after the Duma election (Table 3). Before the election, approximately 9% of respondents refused to answer the question about trust in Putin; this figure increases to 14% after the election (p = 0.04). It may be the case that respondents were more fearful to answer after the election due to potential instability, but in that case, I would expect non-response to be higher specifically after the protests. In reality, the non-response rate after the protests (post-protest period) is statistically indistinguishable from non-response before the election (pre-Duma period) and after the elections but before the protests (post-Duma period).

Table 3. Non-Response for Trust in Putin

Notes: Mean non-response (0 = non-response) by time period. Standard errors in parentheses.

^a Difference in means between the pre-Duma and post-Duma periods are statistically significant at p < 0.05. The difference between other period dyads do not reach conventional standards of statistical significance.

I have no reason to believe that supporters of the regime would have a higher non-response rate than non-supporters. Indeed, non-responders appear to be more opposed to the regime than the average respondent: of the non-responders, 17% voted for United Russia compared to 35% of the total respondents; 36% of non-responders stated that they supported United Russia as opposed to 58% of the total respondents. Because non-responders are predominantly non-supporters, the results should be a conservative estimate of the true impact of perceived electoral manipulation on trust.

It may also be possible that the increase in support for Putin following the protests is simply a demonstration of reversion to the mean. In other words, respondents may not be responding to the protest signal and are simply reverting to their long-term level of trust. However, this is unlikely given the short time frame under examination: the post-protest period covers less than a two-week period following the start of protests. Reversion to the mean typically occurs over longer time horizons and is unlikely to explain changes over the short term. To further address this concern, I reduce the post-protest period to cover only the first week of protests (11–17 December); results hold to this alternative specification.

Political affiliation and trust in institutions

Do electoral manipulation and resulting protests influence all individuals in the same way? In this section, I examine a key potential mechanism that may underpin the relationship between fraud allegations and protest and trust in the regime: political affiliation.¹² Of course, this is not the only factor that may condition trust in authorities; future research should examine additional pathways. See Appendix D in the Supplementary Material for a consideration of additional mechanisms, including residency, interest in politics and media exposure. As always, conditional results should be viewed with caution. Political affiliation is not randomized, so results are suggestive.

Existing research has argued that attitudes towards the government should be conditioned on individuals' prior political beliefs: attribution of responsibility for electoral fraud has been shown to be conditioned on partisanship (Cantú and García-Ponce 2015; Frye 2019; Reuter and Szakonyi 2021; Sirotkina and Zavadskaya 2020). In their study of the 2016 elections in Russia, Reuter and Szakonyi (2021) show that providing respondents with information about electoral fraud reduces support for ruling party candidates primarily among regime supporters. Similarly, Laurits Aarslew (2023) finds that revelations of fraud primarily erode support among the regime's supporters who expect relatively fair elections. According to this logic, ‘core regime supporters are more likely to have ex ante beliefs that elections are free and fair’ (Reuter and Szakonyi 2021: 275) and are therefore most likely to update their beliefs when presented with information that challenges these preconceptions.

Conversely, Frye and Borisova (2019) argue that respondents update their views about the regime based on authorities' response to protest. Their theory implies that the updating should be mainly among regime opponents who have low estimations of the regime before the protests. When regime opponents see that the government is allowing the protests to occur with limited repression, trust should increase specifically among this group. In other words, individuals whose prior beliefs are challenged by new information are more likely to update their views than individuals whose beliefs are reinforced by new information.

However, research on information updating has long shown that the strength – not just the direction – of political attitudes has important consequences for public opinion. Individuals with strongly identified beliefs are less likely to be swayed by information that is contrary to their belief set than individuals with weak political priors (Lodge and Taber 2000, 2013; Muñoz and Anduiza 2019; Reuter and Szakonyi 2021; Strickland et al. 2011). As John Zaller (1992) argues, individuals are more likely to be receptive to particular political messaging only if the message is consistent with their pre-existing political orientation. This implies that, conversely, people are more likely to disregard arguments that are perceived to be in conflict with their underlying values and partisan identities. The stronger the belief, the less likely that updating will occur. Therefore, it may be the case that updating will occur mainly for individuals who have weak or indifferent attitudes of the regime, not strong proponents or opponents.¹³

Following Frye and Borisova (2019) and Reuter and Szakonyi (2021), I measure political affiliation using support for United Russia, the ruling party. The question asks, ‘Please tell me the extent to which you support the party “United Russia”.’ To examine whether the influence of protest on trust is contingent upon the strength of support for United Russia (UR), I include an interaction term for individuals who state that they ‘support United Russia more than any other party’ after the elections (Strong UR supporter × Post-Duma period) and after the protests (Strong UR supporter × Post-protest period). I also create an interaction term for individuals who state that they are ‘strongly opposed’ to United Russia by time period (Strong UR opponent × Post-Duma period, Strong UR opponent × Post-protest period). Finally, I create an interaction term for individuals who are indifferent to United Russia or hold weak opinions (who state that they are ‘somewhat’ opposed or supportive) by time period (Weak opinions × Post-Duma period, Weak opinions × Post-protest period). Individuals who state that they have never heard of United Russia or who did not respond are excluded from the analysis.

Figure 3 shows predicted margins on trust in Putin for strong support of, strong opposition to and weak views of United Russia over time.¹⁴ Unsurprisingly, strong support for United Russia is positively correlated with trust for Putin generally; conversely, strong opposition to United Russia has a negative correlation with trust. However, it is partisanship over time that proves most interesting. Contrary to previous results, strongly opposing United Russia does not appear to be correlated with trust for any time period.¹⁵ Similarly, strong ruling party support does not appear to mediate the impact of electoral manipulation on regime attitudes. Rather, results suggest that updating occurs primarily for individuals who are indifferent to or hold weak opinions of United Russia.

Figure 3. Predicted Margins for Trust in Putin by Views of United Russia.

Note: Predicted trust in Vladimir Putin by views of United Russia with 95% confidence intervals. Trust in Putin is measured on a 1–5 scale with higher numbers corresponding to greater levels of trust. Ordinary least squares with robust standard errors. Fixed effects for income and education.

Allegations of electoral fraud appear to influence trust among those who are indifferent to the regime or who hold weak opinions. The coefficient for the post-Duma period is negative and significant, which indicates that the electoral fraud allegations are associated with decreased trust in Putin for those who hold neither strongly positive nor strongly negative opinions of the ruling party. Individuals who hold weaker opinions of the regime are more likely to update their views based on new information – in this case, the occurrence of egregious electoral manipulation. However, contrary to expectations, there is limited evidence that the resulting protests improved trust in Putin: the coefficient for the post-protest period is also negative and significant, indicating that trust in Putin after the protests was lower than in the pre-Duma period, but the f-statistic (1.07) indicates that the difference between these periods is negligible. Furthermore, I find that the influence of electoral fraud allegations on trust in Putin does not appear to be conditional simply on political affiliation. Trust in Putin among opponents or supporters of United Russia does not change over time (see Supplementary Material, Table A6).

Thus, in comparison to the average effects examined above, these results tell a more nuanced story about political affiliation and trust. In line with research on motivated reasoning. I find that new information appears to contribute to updating specifically for individuals who have weak beliefs – but mainly in regard to electoral manipulation, not the resulting protests. Strong opponents are more likely to already hold beliefs that manipulation occurs during elections. This finding has important implications for trust as individuals with weak political affiliation constitute 58% of the sample and highlights the importance of examining the conditional impacts of fraud allegations and protest on trust.

However, individual support for particular political parties, like trust in institutions or political figures, is also likely to be influenced by political shocks. Individuals who update their trust of Putin in response to electoral manipulation or protests are likely to also update their support of United Russia, the party of power and the party that benefited from electoral manipulation in the 2011 Duma election. Indeed, overall support of United Russia decreased after the election (from 57% in the pre-Duma period to 46% in the post-Duma period) only to rebound in the post-protest period (to 53%). I therefore examine two additional variables that help capture the direction and strength of attitudes towards the government: regime preference and beliefs that the Russian government is ‘on the right path’. For both, respondents are divided into three categories: two categories representing strong attitudes and one representing weak attitudes. See the Appendices for more information on the variables and results. As with support for United Russia, trust in Putin decreases in the post-Duma period primarily for those who have weak political attitudes: those who have no preference for regime type and those who hold weak opinions about whether Russia is on the right path.

Conclusion

Whether and when authorities use repression remains one of the key questions in the study of authoritarian politics. As the number of authoritarian regimes around the world continues to grow, the conditions under which particular strategies of societal control benefit – or harm – authoritarian regimes will continue to be of central import. The 2011 Duma elections and resulting protests were a turning point for the Putin regime, one which has resulted in the increase in repressive tactics to limit opposition and alternatives to the ruling order.

This study implies that the mere existence of electoral fraud allegations in contexts where manipulation is often considered ‘business as usual’ does not lead the public to update their beliefs about authorities. The 2011 Duma elections were perceived by the Russian public as being particularly unfair and fraudulent compared to similar elections. Indeed, the high levels of perceived fraud were the driving factor behind one of the largest protest movements in Russian history, a movement that would have been unlikely to occur after an election with ‘average’ levels of fraud. This suggests that studies of electoral fraud on public opinion should consider the relative levels of perceived manipulation. Trust in authorities should decrease primarily when fraud is seen as being particularly egregious – when the visibility and extent of fraud is higher than expected – trust in authorities decreases.¹⁶ This finding has important implications on the future study of public attitudes and elections in non-democracies: instead of focusing on the mere existence or level of fraud, research needs to interrogate how people view the extent of manipulation. Furthermore, citizens do not punish the entire system for alleged manipulation, as is often assumed. Rather, the public can and do differentiate among different political institutions and leaders and direct their ire towards those deemed to be responsible for decision-making. In highly centralized, personalistic regimes such as Russia, this leads to a decrease in trust specifically for the autocrat.

Yet, allegations of electoral fraud do not sway all groups equally. Those with strongly identified priors, who hold either strongly favourable or strongly opposed views of the regime, are not swayed by allegations of egregious electoral manipulation. This finding provides further evidence that individuals with strong political priors are less likely to update their beliefs when presented with new information. Rather, it is the people in the middle, those with weaker political beliefs, who are most susceptible to new information: trust in Putin decreases after allegations of electoral fraud specifically for individuals who are weakly opposed or weakly supportive of Russia's ruling party.

These findings have important implications for understanding not just the level but also the type of electoral manipulation that authorities may pursue. In non-democracies, authorities frequently rely on electoral manipulation to secure victory and achieve their desired results, particularly when they do not expect to win elections without manipulation. But, as this study demonstrates, reliance on manipulation can further erode support for authorities, which may in turn require the use of even more manipulation to secure victory. However, fraud allegations are only likely to depress views of the regime if citizens view the manipulation as being worse than usual. Ballot stuffing and harassment of election monitors, a few of the common manipulation tactics, are highly visible to the public and are therefore more likely to harm attitudes towards the regime. But all forms of manipulation are not equally visible to the public or even viewed as a form of manipulation in the first place. Instead of relying on these tactics, authoritarian regimes can rely on administrative tools – such as increasing barriers to entry, barring candidates or online elections – to manipulate elections away from the public eye, thus limiting the potential costs of manipulation (Szakonyi 2022). A more recent Duma election in Russia in 2021 exemplifies this shift to less visible forms of manipulation. Authorities extended the voting period and introduced online voting, decisions that contributed to what experts have argued were the most falsified in Russia's history (Cordell 2021); yet the elections were seen as no more fraudulent than prior elections, in part because the majority of Russians do not view these tactics as increasing the potential for manipulation (Levada Analytical Center 2021).

This study also shows that anti-fraud protests may benefit authorities. Following the December 2011 election protests, trust in Putin increased – though, once again, this updating occurs primarily among those with weaker prior political beliefs. These results demonstrate the speed by which individuals can and do update their views of political trust when presented with new information. Political trust is often seen as slow-moving and difficult to change, but this study has shown that political shocks can and do lead individuals to update their confidence in political leaders. Research has shown that people have limited political knowledge and do not take into account their complete knowledge set when making political judgements; instead, individuals make judgements based on information that readily comes to mind (Iyengar et al. 1982). In this instance, the protest events all but erased the sins of electoral manipulation in the minds of the public, restoring trust to roughly its pre-election levels. These results imply that while specific events can and do influence attitudes towards the government, these attitudes are difficult to sustain in the long term.

While results suggest that protests can improve trust in the autocrat, future research must pay greater attention to the mechanisms by which protest is linked to regime attitudes. On the one hand, protests may lead individuals to revise their attitudes based on authorities' response to protests. If authorities pursue a mild response to protests (as they did initially during the December 2011 protest movement), individuals may update their beliefs about the freedoms afforded to them and, consequently, their attitudes towards the regime. However, it may also be the case that protests signal the potential for future instability, leading individuals to double down in their support for the regime, essentially ‘voting against disorder’ (Pepinsky 2017). Future research should further consider these potential mechanisms to better understand the link between protests and political attitudes in non-democracies.