Mobilizing opposition voters under electoral authoritarianism: A field experiment in Russia

Abstract

Under electoral authoritarianism opposition supporters often abstain from voting because they think that their votes will not make a difference. Opposition parties try to counteract this apathy by informational campaigns that stress how voting can impact the outcome of the election and policy. Evidence from established democracies suggests that such campaigns are generally ineffective, but it remains an open question whether the same holds in elections under authoritarianism where information is scarce. We follow a large-scale campaign experiment by an opposition candidate in Russia’s 2016 parliamentary election, which distributed 240,000 fliers to 75% of the district’s households. Relative to a control flier, priming voters about the closeness of the election or the link between voting and policy outcomes had no practically meaningful impact on turnout or votes. Contrary to some existing theories and the stated expectations of politicians, information about the value of voting appears as ineffective in uncompetitive electoral autocracies as it is in democracies.

Keywords

Authoritarianism hybrid regimes voting mobilization experimental methods

Introduction

Most of today’s non-democracies are “electoral authoritarian” regimes, which permit opposition and hold regular elections but also use state funds, patronage, coercion, and fraud to undermine the opposition at the ballot-box (Levitsky and Way, 2010). Nonetheless, the opposition can sometimes deliver surprising victories that bring about a change in government or even regime type, as was the case in Poland in 1989, Mexico in 2000, Kenya in 2002, or Malaysia in 2018.

A common explanation for why the opposition under electoral authoritarianism succeeds so rarely – even when the incumbent is weak – is voter apathy. When the political system is stacked in favor of the incumbent who appears to be invincible, supporters of the opposition do not expect their votes to make a difference (Magaloni, 2006; Simpser, 2013). Opposition parties often try to counteract this apathy by arguing that the ruling party is more vulnerable than perceived, and that a respectable performance by the opposition at the ballot box – even if it falls short of winning – will strengthen its position in the future and force the incumbent to make policy concessions (Bunce and Wolchik, 2010; Howard and Roessler, 2006). Russia’s opposition leader Alexey Navalny has suggested that opposition voters might learn from past electoral successes that “they can actually win,” and that this information should be tactically highlighted by political campaigns (Smyth and Soboleva, 2016).

The hypothesis that priming the value of voting should mobilize turnout is advanced by the classical theory of turnout (Riker and Ordeshook, 1968) and studies on the importance of vote pivotality to voters’ decision-making (Palfrey and Rosenthal, 1985). But empirical evidence from established democracies has consistently falsified this prediction: information about the pivotality of elections does not increase turnout (Enos and Fowler, 2014), and, more generally, informational mobilization campaigns are not very effective unless they activate a community-level social shaming mechanism (Green and Gerber, 2019).

It remains an open question, however, whether partisan mobilization campaigns targeting vote pivotality and the link between voting and policy are equally ineffective in elections under authoritarianism. On the one hand, due to generally lower levels of competition, one would reasonably expect such campaigns under authoritarianism to be even less effective than in democracies. On the other hand, even though small informational nudges do not work in developed democracies where voters are overwhelmed by political information, they could have qualitatively different impact in places where political information – especially from the opposition – is scarce. That is perhaps one reason why opposition politicians actually deploy basic partisan information campaigns. Arguably, political information from non-regime sources, and, therefore, especially information about politics from opposition candidates, should be a lot more effective under electoral authoritarianism than in a democracy. This is because authoritarian governments control information flows and mold political information. In this environment of information paucity, genuine data on the closeness of elections or even a basic reminder about the fact that electoral institutions are supposed to be accountable to voters by design should be a lot more impactful than in democracies where information is disseminated freely and there are functional independent media.¹

In this study, we quantify the effectiveness of an opposition-led mobilization campaign using data from a large-scale field experiment implemented by an opposition candidate in Russia’s 2016 parliamentary election. The campaign distributed 240,000 partisan campaign fliers to 75% of the households in the candidate’s district. Campaign staff designed two types of “treatment” fliers that highlighted different reasons for why it is important to vote: one flier primed the closeness of the forthcoming election in the district and another primed the connection between voting and policy-making. The “control” flier contained only information about the election date and the candidate but no messages priming the value of voting. The three messages were cluster-randomized at the level of electoral precincts within the district.

Our analyses of the data from this experiment show that both informational treatments had substantively negligible and statistically insignificant effects relative to the control flier. In this paradigmatic case of uncompetitive electoral authoritarianism, neither aggregate turnout nor the distribution of votes across candidates and parties were meaningfully affected by the messages that voting can make a difference for the success of opposition politicians. These results do not imply that partisan campaigning via fliers as such is ineffective, because the control group received information about the candidate.² Rather, this experiment addresses a narrower but, arguably, theoretically more important question of whether information about the importance of voting adds value beyond generic partisan campaigning.

Experimental design

Context

The experiment was implemented in Moscow’s district 198 (Leningradsky), which is subdivided into 229 electoral precincts with a total of 507,000 registered voters. This district is one of 15 in the city and is broadly representative of Moscow at large. We followed the campaign by Yulia Galiamina – a candidate for a seat in the national parliament in the district. Galiamina, aged 43 at the time of the election, is a local activist on issues of environmental protection and government accountability. In 2014, she ran for Moscow city council, coming second after the candidate for the ruling United Russia party. In the 2016 parliamentary race, Galiamina was running under the aegis of the Yabloko party, a well-established opposition party. In her district, the pro-regime coalition was represented by the candidate from Just Russia. The average turnout rate in 2016 in the district was 34%, which is typical of Moscow. At the conclusion of the race, Galiamina came second with 14% of the vote behind the pro-government candidate who received 31%.

Treatments

Russian opposition politicians generally believe that many of their supporters do not vote because they think that their participation would make no difference.³ In an attempt to counteract these negative expectations, Galiamina’s campaign distributed three types of fliers: a control flier containing the candidate’s photograph, her name, the campaign slogan, her party’s logo, and the election date; and two treatment fliers containing the same information as the control, but with mobilizational appeals. The fliers did not have detailed information about Galiamina’s policy platforms; only her status as an opposition politician was signaled by Yabloko’s party slogan and the candidate’s election slogan exhorting voters to support that which is vibrant/living. Fliers – along with other printed matter, like campaign newspapers and posters – are commonly used in Russian elections, and especially by opposition politicians (Kynev et al., 2017). Printed materials were particularly commonplace in the 2016 race because campaigns had to find an inexpensive way to engage with voters over a long summer break that preceded the polling day in mid-September.⁴

The wording of the two mobilizational appeals is given in Table 1.⁵ In the “closeness of election” appeal, recipients were told factually correct information about the 2013 mayoral election where the independent candidate Alexey Navalny (who was not mentioned by name on the flier) lost by only 208 votes in one of the district’s rayons. Nationally, the ruling United Russia Party and its parliamentary allies have had an unassailable advantage for decades, and at that level the appeal to voter pivotality is not credible. However, at the local level – and specifically in Moscow – opposition candidates have been able to do well and seats have been won by small margins. In the 2013 mayoral election, Navalny came second with 27% of the vote after the pro-government candidate. In the 2019 Moscow municipal elections, Yabloko’s candidates won four seats in city assembly (of 45), and in that race the difference between winning candidates and those coming second was only several dozen votes in several districts. In the 2016 parliamentary race, Yabloko won a respectable 10% of votes in the capital but only 2% of the national vote. All in all, in the local context of Moscow, the candidate’s campaign considered the closeness of the election appeal to be credible. The “policy-impact” appeal primed voters about the connection between voting and policy outcomes. The message is formulated in a colloquial manner and triggers the ex post regret of not taking part in elections. This design allows us to isolate the effectiveness of two mobilizational messages net of the basic factual information about the election and the candidate.

Table 1.

Mobilizational appeals in the two treatment conditions.

	Flier’s front	Inside
Closeness of election	Do you support an independent^a candidate? We have a chance of winning!	In the 2013 Moscow mayoral election, the opposition candidate was just 208 votes short of winning in one of our rayons. In the upcoming Duma election, your vote could be decisive!
Policy-impact	Planning not to vote again?	Then do not complain about the laws passed by the Duma. Do not let others decide what the future parliament will look like!

In the Russian context, “independent” refers to someone from outside of the regime, and the term is used interchangeably with “opposition.”

The campaign was hoping to mobilize pro-opposition voters, but it was also possible that pro-regime supporters would have been mobilized too if they perceived Galiamina to be a threat to the pro-regime candidate. Thus, we study the effects of the treatments on turnout and votes.

Implementation

Treatments were administered and outcomes measured at the level of electoral precincts. In total, there are 229 precincts in the electoral district under study. Of these, 17 service organizations (military facilities, hospitals, etc.), and the campaign did not attempt to deliver fliers there. It did set out to distribute fliers to mailboxes at all residential addresses in the remaining 212 precincts.⁶ On average, precincts have about 2,000 voters. The intent was to treat all 500,000 registered voters in the district. A total of 240,000 fliers were printed for hand-delivery to individual household mailboxes in 2,727 residential buildings. Fliers were distributed by a marketing firm contracted by the candidate’s campaign. The distribution was scheduled for September 12–16, with the election taking place on September 18. Fliers were successfully distributed in 77% of buildings in the control and policy-impact conditions and in 74% of buildings in the closeness of election treatment. These are relatively high rates of coverage in a security-conscious city.

Balance and power

In Table 2, we report how the three experimental groups are balanced with respect to pre-treatment characteristics. Group averages for each of the interventions are reported in columns 1–3, and difference of means statistics for comparison between each of the treatments and the control are in the last two columns. Observations are weighted by a precinct’s vote-eligible population to reduce the risk of small precincts driving the overall results. The balance is generally good, with the exception of turnout in the 2013 mayoral election where precincts in “closeness of election” group have slightly higher turnout than those in control. To address this problem, we control for turnout in 2013 in a subset of subsequent analyses.

Table 2.

Balance tests.

	Group means			Treatment – control
	Control(N = 70)	Closeness(N = 70)	Impact(N = 70)	Closeness	Impact
Number of apartments	1326	1404	1357	77.96 (0.11)	30.82 (0.52)
Number of buildings	14.01	13.02	12.04	−0.99 (0.53)	−1.98 (0.21)
Number of voters	2469	2422	2444	−46.93 (0.33)	−24.65 (0.61)
Turnout in 2011	64.58	64.85	65.19	0.27 (0.89)	0.61 (0.75)
United Russia vote in 2011	45.74	45.34	46.58	−0.40 (0.86)	0.84 (0.71)
Turnout in 2012	56.26	56.64	57.28	0.38 (0.62)	1.02 (0.17)
Putin vote in 2012	46.34	45.22	45.65	−1.11 (0.21)	−0.69 (0.44)
Turnout in 2013 elections	31.52	32.64	31.95	1.12 (0.10)	0.43 (0.52)
Navalny vote in 2013	27.99	29.19	28.46	1.19 (0.35)	0.47 (0.71)

Observations weighted by precinct voting population. P-values for two-tailed t-test in parentheses.

The number of units in this study (N = 210) compares favorably with other field experiments that manipulate treatments at the level of lowest possible electoral or administrative units. For instance, an experiment examining the effects of democracy promotion in Georgia divided 84 precincts across two interventions (Driscoll and Hidalgo, 2014), and the study by Kendall et al. (2015) randomized four interventions across 95 precincts. Experiments by Adida et al. (2017) and by Mvukiyehe and Samii (2017) randomized over 137 and 142 units, respectively. The power of this experiment is between 0.74 and 0.92, depending on exact formulation of the null hypothesis, to detect a two-percentage-point treatment effect under the 95% confidence level (see Online Appendix 3 for calculations). In the US, civic duty mailers raise turnout by 1.8 percentage points (Gerber et al., 2008), and there is reasonable expectation that the effects of mobilizational treatments ought to be larger in a setting where there is a lot less publicity delivered by mail (Gerber and Green, 2017). For instance, in West Africa, leaflets about the illegality of vote buying were found to decrease turnout by 3–6 percentage points (Vicente, 2014), and in Georgia – in the context closest to Russia – leaflets encouraging voters to report fraud suppressed turnout by five percentage points (Driscoll and Hidalgo, 2014). Even more importantly, a treatment effect of less than two percentage points would seem to be trivial, and, therefore, not very interesting substantively, in the Russian context where the opposition often trails regime candidates by dozens of percentage points.

Results

In Table 3 we report the estimated effects of a precinct being assigned to one of the two mobilizational treatments relative to control on turnout, support for the candidate (Galiamina), and support for the candidate’s party Yabloko.⁷ Outcomes are measured in percentage points, and coefficients are estimated using weighted least squares by weighting each precinct according to the number of registered voters.⁸

Table 3.

Campaign effects.

	Turnout		Candidate vote		Party vote
	(1)	(2)	(3)	(4)	(5)	(6)
Closeness	0.74	−0.02	0.29	−0.48	1.11	0.24
	(0.59)	(0.38)	(0.83)	(0.69)	(0.84)	(0.66)
Impact	0.33	0.04	−0.53	−0.83	0.61	0.28
	(0.59)	(0.37)	(0.83)	(0.68)	(0.84)	(0.66)
Turnout in 2013		0.67^*		0.69^*		0.77^*
		(0.04)		(0.07)		(0.07)
Observations	210	210	210	210	210	210
Adjusted R²	−0.002	0.60	−0.005	0.31	−0.001	0.39

p < 0.01.

Dependent variables are measured in percentages.

The coefficients in Table 3 are very small in magnitude across all specifications and not significant. After adjusting for prior turnout, the coefficients for the closeness of election and policy impact primes are −0.02 and 0.04, respectively. Since outcomes are measured on a percentage-point scale (0 to 100), these are substantively negligible effects, irrespective of their statistical significance. Similarly, the estimated coefficients for the percentage of votes cast for the candidate and her party are also small (varying from −0.83 to 0.28) and directionally inconsistent across specifications. The large p-values for one-sided hypotheses tests show that, across all specifications and for all treatments, we cannot reject the null hypothesis that the treatment effect is smaller than two or even one percentage points.

The estimated effects remain substantively small and statistically insignificant, irrespective of the included pre-treatments covariates (see Online Appendix 5). The results are also very similar if we use votes for all opposition candidates as the outcome variable (see Online Appendix 6). Finally, we also find that the informational primes had equally weak effects for all levels of prior opposition support (see Online Appendix 7). In sum, both informational primes appear to have had, at best, none or only negligible effects on election outcomes.

Possible explanations for the observed effects

Lack of statistical power

This study has 92% power of detecting a two-percentage-point effect of at least one of the two treatments with 95% confidence. It is possible that the treatments had much smaller effects that could not be detected given our sample size. We cannot claim that the effects of both treatments are identical to zero, but the evidence does strongly suggest that neither informational prime meaningfully helped the opposition to mobilize its supporters relative to the control flier. We draw this conclusion not so much on the basis of statistical significance as on the basis of substantively very small point estimates. The lack of statistical power would put in doubt our conclusions if the point estimates were large but noisily estimated. However, they are small and quite precisely estimated. After adjusting for prior turnout, the impact prime increases turnout by 0.04 percentage points, which means that one would need a 50-times-stronger treatment to mobilize two percentage points more voters. While we cannot rule out the possibility that the treatments might have affected turnout or candidate support by less than one percentage point, an effect of such magnitude is not substantively relevant in the context of Russian elections where opposition candidates mostly lose by large margins.

Low delivery or take-up rates

The reported estimates could suffer from attenuation biases due to incomplete delivery of the leaflets and potentially low take-up rates. To address this concern, we check whether treatment effects are higher in precincts with higher delivery rates. In places where more fliers are delivered, more fliers should have been read. For each precinct, we calculate the share of buildings that received the leaflets. We then estimate treatment effects at different saturation thresholds. In Figure 1, we report the effect of the two treatments on turnout at various levels of treatment saturation, varying from precincts where at least 0% of buildings received fliers (N = 210) to those where at least 80% of buildings received them (N = 100). Treatment effects consistently hover around zero and do not increase as the saturation rate goes up. This pattern also holds for other dependent variables, as shown in Online Appendix 8. This evidence is inconsistent with the idea that null results are due to incomplete delivery of fliers or their low take-up rates.

Figure 1.

The effects of treatments on turnout for different saturation thresholds.

Election fraud

Another possible explanation for substantively negligible effects of the treatments might have to do with the fact that elections in Russia are at times fraudulent. As discussed in Online Appendix 9, three facts negate this explanation. First, although election fraud was widespread across Russia, there was little evidence of large-scale fraud in Moscow. Second, to negate possible treatment effects the regime would have had to inflate turnout in precincts in the control condition while not inflating support for pro-regime parties there, as we didn’t find any treatment effects on support for candidates. This would defy the logic of electoral fraud. Third, our simulations show that to attenuate treatment effects to the point of our results, the level of electoral fraud would have had to be considerably higher than any existing estimates.

Voter apathy and weakness of mobilizational appeals

A more disturbing possibility from the perspective of potential for political change is that opposition supporters are too apathetic to turn out in a context where the regime seems destined to win either because it enjoys genuine majority support or because it is able to keep out competitors, manipulate electoral results, and suppress dissent effectively. Survey evidence suggests that only 14% of respondents around the time of the 2016 election thought that they could effectively express their political preferences through voting. Eighty-seven percent said that they had no influence over policy decisions.⁹

If the baseline intrinsic benefits of voting are perceived to be extremely low under electoral authoritarianism, then even highly persuasive mobilizational appeals would be insufficient to overcome the costs of voting and deep voter apathy. Given the ruling party’s dominance, it might make little sense to vote locally for opposition candidates in the absence of a plausible signal that change at the national level might be within reach.

Voter apathy would render even the most theoretically persuasive mobilizational appeals ineffective, and it is possible that the specific mobilizational appeals used by the campaign were simply unpersuasive to Russian voters. While information interventions are known to be effective in developing democracies (Adida et al., 2017; Chong et al., 2015; Driscoll and Hidalgo, 2014; Vicente, 2014) and are commonly used by opposition parties in Russia (Kynev et al., 2017), it might be that standard appeals to closeness of the election or policy impact of voting are not credible under authoritarianism.¹⁰ Potential voters might have been skeptical about the closeness of the race in the precinct at large beyond a specific rayon. Likewise, the policy impact prime might have fallen short because of the perception that the legislature is little more than a rubber stamp for decisions made by the executive and, therefore, not an effective agent of political change.

Conclusion

The experiment suggests that information paucity is not the only factor that is keeping opposition voters from the polls. In an environment where the gap between the regime and opposition candidates is large, it seems that voter apathy is hard to overcome through provision of information on the closeness of elections or their impact on policy.

This study is only a single data point, albeit in a paradigmatic case. The effects of identical campaigns would likely be different in other contexts – like, for example, in Algeria, Turkey, Venezuela, or Zimbabwe – where the political landscape is a lot more competitive, the baseline levels of political apathy are lower, or where there is a commonly observed signal (e.g. economic crisis, protest, divisions within ruling elites) that regime change is possible or imminent. In closer elections, a campaign such as the one described in this study is likely to be more effective.

Supplemental Material

Online_Appendix – Supplemental material for Mobilizing opposition voters under electoral authoritarianism: A field experiment in Russia

Supplemental material, Online_Appendix for Mobilizing opposition voters under electoral authoritarianism: A field experiment in Russia by Leonid Peisakhin, Arturas Rozenas and Sergey Sanovich in Research & Politics

Footnotes

Acknowledgements

We thank Yulia Galiamina, her campaign manager Natalia Shavshukova, and campaign staff for allowing us to observe the campaign. We are also grateful to Jenn Larson, Elena Llaudet, and Peter van der Windt for their helpful comments, and to Dmitry Dobrovolskiy for research assistance. We are grateful to New York University-Abu Dhabi for research support.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The open access article processing charge (APC) for this article was waived due to a grant awarded to Research & Politics from Carnegie Corporation of New York under its ‘Bridging the Gap’ initiative.

ORCID iD

Arturas Rozenas

Supplemental materials

The supplemental files are available at .

The replication files are available at

Notes

References

Adida

Gottlieb

Kramon

, et al. (2017) Reducing or reinforcing in-group preferences? An experiment on information and ethnic voting. Quarterly Journal of Political Science 12(4): 437–477.

Bunce

Wolchik

(2010) Defeating dictators: Electoral change and stability in competitive authoritarian regimes. World Politics 62(1): 43–86.

Chong

De La O

Karlan

, et al. (2015) Does corruption information inspire the fight or quash the hope? A field experiment in Mexico on voter turnout, choice, and party identification. The Journal of Politics 77(1): 55–71.

Driscoll

Hidalgo

(2014) Intended and unintended consequences of democracy promotion assistance to Georgia after the Rose Revolution. Research & Politics 1(1): 1–13.

Enos

Fowler

(2014) Pivotality and turnout: Evidence from a field experiment in the aftermath of a tied election. Political Science Research and Methods 2(2): 309–319.

Gerber

Green

(2017) Field experiments on voter mobilization: An overview of a burgeoning literature. In: Banerjee

Duflo

(eds) Handbook of Field Experiments, Vol. 1 of Handbook of Economic Field Experiments. Amsterdam: North-Holland, 395–438.

Gerber

Green

Larimer

(2008) Social pressure and voter turnout: evidence from a large-scale field experiment. American Political Science Review 102(1): 33–48.

Green

Gerber

(2019) Get Out the Vote! How to Increase Voter Turnout. 4th ed. Washington, DC: Brookings Institution Press.

Guan

Green

(2006) Noncoercive mobilization in state-controlled elections: An experimental study in Beijing. Comparative Political Studies 39(10): 1175–1193.

10.

Howard

Roessler

(2006) Liberalizing electoral outcomes in competitive authoritarian regimes. American Journal of Political Science 50(2): 365–381.

11.

Kendall

Nannicini

Trebbi

(2015) How do voters respond to information? Evidence from a randomized campaign. American Economic Review 105(1): 322–353.

12.

Kynev

Lyubarev

Maksimov

(2017) Kak vybirala Rossiia – 2016. Moscow: Liberal Mission Foundation.

13.

Levitsky

Way

(2010) Competitive Authoritarianism. Hybrid Regimes After the Cold War. Cambridge: Cambridge University Press.

14.

Magaloni

(2006) Voting for Autocracy: Hegemonic Party Survival and its Demise in Mexico. New York: Cambridge University Press.

15.

Mvukiyehe

Samii

(2017) Promoting democracy in fragile states: Field experimental evidence from Liberia. World Development 95: 254–267.

16.

Palfrey

Rosenthal

(1985) Voter participation and strategic uncertainty. The American Political Science Review 79(1): 62–78.

17.

Riker

Ordeshook

(1968) A theory of the calculus of voting. American Political Science Review 62(1): 25–42.

18.

Simpser

(2013) Why Governments and Parties Manipulate Elections: Theory, Practice, and Implications. New York: Cambridge University Press.

19.

Smyth

Soboleva

(2016) Navalny’s gamesters: Protest, opposition innovation, and authoritarian stability in Russia. Russian Politics 1(4): 347–371.

20.

Vicente

(2014) Is vote buying effective? Evidence from a field experiment in West Africa. The Economic Journal 124(574): F356–F387.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.14 MB