Undecided voters and the myth of punishment for corruption: the case of Serbia on the eve of the elections

(Peščanik.net)

The political scene in Serbia ahead of the next national elections is marked by pronounced volatility of the electorate. After the waves of anti-corruption and student protests that significantly shook the political balance in 2024 and 2025, public opinion polls have recorded two parallel phenomena: the gradual weakening of the dominance of the ruling Serbian Progressive Party (SNS) and, at the same time, an unusually high percentage of voters who do not clearly state their electoral preferences. This text analyzes the findings of two relevant surveys - CRTA/Democracy Action Lab and New Serbian Political Thoughts (NSPM) - and questions one of the most widespread assumptions in commenting on these data: that undecided voters, due to their expressed dissatisfaction with corruption and other accumulated problems in society, will almost automatically side with the regime.

CRTA and NSPM research results

The survey conducted in June 2026 by CRTA, in cooperation with Democracy Action Lab, included a representative sample of 2.324 respondents surveyed face-to-face. The results show that the Student List would win 44,9% of the vote, the SNS 35,7%, and the Socialist Party of Serbia 3,8% - however, these percentages refer exclusively to the so-called "determined" voters, who make up only 70% of the total sample. In other words, as many as 30% of respondents in this survey were not ready to clearly state the party or list they would vote for. In parallel, a survey by an agency affiliated with the NSPM, published at the end of July 2026, presents a somewhat different picture of the distribution of votes - Student List 39,6%, SNS 37,5%, with the SPS independently passing the threshold with 4,9% - but also an explicit statement that approximately a quarter of voters are "still weighing" their decision, i.e. that approximately 25% of respondents cannot be classified into any clearly defined camp. What these two surveys, despite methodological differences and different figures for individual parties, have in common is precisely this finding: a significant part of the electorate - between a quarter and a third of the total number of respondents - does not currently have a formed or at least publicly expressed electoral preference. This is a historically high percentage for Serbian conditions, and it in itself deserves analytical attention regardless of which party is currently leading among those who have made a decision.

The temptation of a simple interpretation

In public discourse, and partly in the analytical comments that accompany the publication of such surveys, the interpretation has become widespread that the high percentage of undecided people represents a kind of anti-regime "reserve army" - voters who are deeply dissatisfied with the ruling structure, primarily due to the perception of corruption, but who, for tactical, generational or identity reasons, are not yet ready to publicly admit this in front of pollsters. According to this logic, when the moment of actual voting comes - in the secrecy of the ballot box, without the social pressure present in a survey situation - these voters will predominantly choose against the ruling party, since surveys consistently show that Serbian citizens assess corruption extremely harshly as a social problem.

This assumption is not unfounded in the sense that there is indeed abundant evidence that Serbian citizens perceive corruption as one of the country's biggest problems - public opinion polls in recent years consistently rank corruption and crime among the three most frequently mentioned problems, along with economic issues. However, the thesis advocated by this text is that lightly translating this general dissatisfaction with corruption into the specific electoral behavior of undecided voters is methodologically unjustified, and that there are serious scientific works that question such a connection.

What does research on corruption and electoral behavior show?

Particularly relevant here is a meta-analysis by Trevor Incerti, published in 2020 in the prestigious journal American Political Science Review, which addresses precisely this gap between stated attitudes toward corruption and actual voting behavior. Incerti analyzed a large number of experimental studies from different countries and divided them into two categories: field experiments, in which real voters are provided with information about the corruption of specific politicians and then their actual voting behavior in elections is monitored, and survey experiments, in which respondents are presented with hypothetical scenarios with an indication of corruption. The central finding of this meta-analysis is striking: while survey experiments show that corrupt candidates would lose on average between 32 and 35 percentage points of support, field experiments - which measure actual behavior in real elections - show an effect close to zero. In other words, the declared willingness to punish corruption in elections hardly materializes at all in real electoral conditions. Incerti explains this discrepancy through two mechanisms. First, field experiments often suffer from weaknesses in the treatment itself - information about corruption does not reach voters in a sufficiently strong and convincing way, or voters simply do not pay enough attention to it in the everyday information noise of the election campaign. Second, and probably more importantly, large effects in survey research are most likely a consequence of the socially desirable response bias: when a respondent in a survey states that he would punish a corrupt candidate, that statement carries no real cost or obligation - it is a hypothetical, low-risk scenario in which the "correct" answer is socially expected. In a real election situation, a voter, faced with a whole range of other factors - economic stability, party identification, fear of the unknown, procedural uncertainty about who would actually form the government - as a rule does not make a decision solely, or even primarily, on the basis of attitudes towards corruption.

Implications for interpreting the Serbian case

If this finding is applied to the context of undecided voters in Serbia, an important methodological caveat emerges. The very fact that surveys consistently show a high level of condemnation of corruption among Serbian citizens does not allow the conclusion that these same citizens, when the time comes to actually vote, will direct their vote against the ruling structure. Undecidedness in a survey does not have to be a disguised anti-regime commitment - it can just as likely represent: (a) genuine indecision that resolves itself only in the final days of the campaign under the influence of factors unrelated to corruption, (b) abstention, i.e. the intention not to vote at all, or (c) rationalized support for the status quo motivated by other priorities - economic stability, fear of institutional uncertainty, or simply a lack of confidence that a change of government would bring real improvement. This is all the more important because surveys such as CRTA's show a pronounced polarization in the perception of democracy along the lines of media consumption - the overwhelming majority of supporters of the Student List support democracy as a system, while this percentage among SNS voters is significantly lower. This polarization suggests that the electorate is not divided exclusively along the axis of "for or against corruption", but along a much more complex axis of value and institutional orientations, in which corruption represents only one, albeit prominent, but not the only or necessarily decisive factor. For all of the above, it is very important to approach the election campaign seriously and not allow unpleasant surprises due to lightly drawn, erroneous conclusions.

Reference: Incerti, T. (2020). Corruption Information and Vote Share: A Meta-Analysis and Lessons for Experimental Design. American Political Science Review, 114(3), 761-774.

The author is a professor at the Faculty of Technical Sciences, University of Novi Sad and a member of the Inquiry Commission for investigating responsibility for the collapse of the canopy at the Novi Sad Railway Station.