This post was a Judges' Choice winner in the Jordan Center Blog's fourth annual Graduate Student Essay Competition.
Kirill Chmel is a third-year graduate student in Political Science at Columbia University, focusing on media exposure, identity politics, and ethnic conflict. Prior to Columbia, he was a research fellow at the Ronald F. Inglehart Laboratory for Comparative Social Research and a lecturer at HSE University.
Public support for the Russian invasion of Ukraine has remained high and relatively stable since the conflict began in February 2022. In a Levada Center poll conducted in March 2024, 76% of respondents expressed support (either definite or general) for the invasion when asked about it directly. However, despite apparently overwhelming majority support, there is ongoing debate about its sincerity. The research my colleague, Nikita Savin, and I conducted in February-August 2022 offers suggestive evidence that a significant portion of the Russian public does not actually support the invasion, despite indicating otherwise when directly questioned.
Why do people conceal their genuine opinions?
The disparity between individuals’ authentic attitudes regarding the conflict and those they articulate in public can stem from various factors. Indeed, the phenomenon of preferences falsification—as coined by Timur Kuran in his seminal work—has been thoroughly investigated by political scientists, sociologists, and psychologists. One possible reason why respondents may refrain from providing truthful responses to direct questions about the war may be that they perceive widespread societal approval of military activities. This perception may cause them to fear social stigma or pressure associated with expressing dissenting opinions.
Respondents may additionally hesitate to express their genuine sentiments out of fear of facing official sanctions. On 4 March 2022, the Russian government enacted legislation penalizing individuals for disseminating “false information” regarding the “special military operation” with potential sentences of up to 15 years in prison. This law, and others like it, which target the spread of “fake” information about the conflict or “discrediting” the Russian armed forces, may deter opposition.
Does preferences falsification matter in public opinion surveys?
The impact of fears related to either social or official sanctions on survey responses is collectively termed “social desirability bias” and can significantly influence our interpretation of public opinion data. In the Russian context, respondents’ concerns about social pressure and potential repercussions for expressing dissent may lead to biases in the results of direct questions, ultimately causing pollsters to overestimate support for the war.
Social scientists have developed many techniques to account for and measure these biases. These methods are employed to examine various social taboos (like racism, antisemitism, sexual orientation, and issues related to “family values”) as well as criminal behaviors (including sex work, drug use, tax evasion, and participation in civil wars) that are sensitive yet crucial for scientists, policymakers, and the public to better understand. Typically, these techniques involve enabling individual respondents to respond to sensitive questions indirectly, thereby allowing them to mask their true attitudes and reducing their fears of answering truthfully.
While scientists cannot say for certain how any given individual responded, they can examine the aggregate data to understand trends and make inferences about the true prevalence of sensitive attitudes or behaviors.
How we measured support for Russia’s war in Ukraine
To study the social desirability bias associated with the conflict in Ukraine, we employed a widely utilized technique known as a list experiment. In this method, individual survey participants are randomly assigned to two groups. One group is presented with a list of innocuous options, while the second group is presented with the same list along with an additional sensitive item. Respondents in both surveys are then asked how many of the listed items they support.
If the groups are truly random, then respondents in both should support the same number of non-sensitive options on average. Any difference in the count between the two groups is attributable to the sensitive item and therefore shows the prevalence of support for it. Properly designed, this setup provides individuals plausible deniability about the items on the list: they never say which items they support—only how many.
We designed the following list experiment and compared it with the direct question asking respondents whether they supported the invasion of Ukraine. The central question read as follows:
“Now I will list several decisions the government of the Russian Federation has made in the last few years. Don’t tell me which specific decisions you personally support. Just tell me how many of these decisions you support.
1. 2018 Pension Reform
2. Initiating a lockdown in response to Coronavirus outbreaks
3. 2020 Amendments to the Russian Constitution
4. Invasion of Ukraine by Russian troops”
What we found
We conducted our list experiment repeatedly in five waves, which we carried out in collaboration with the independent pollster Levada Center in Russia in 2022. The results of our experiments are shown graphically in the figure below. In the top panel, the blue line represents the trend for direct questioning, while the yellow line represents the trend for the list experiment measure. The bottom panel illustrates the size of the social desirability bias.
Based on the results of our list experiment, approximately 70% of Russians supported military activities in Ukraine at the end of February, just a few days after the invasion started. We found no evidence of social desirability bias in the first survey wave, indicating that figures from polls conducted during the initial days of the war can potentially be trusted and used as an indicator of public opinion.
However, we observed a noticeable difference between the direct question and the list experiment in all waves since March—precisely after the Russian government passed the law on “discrediting the Russian army.” In all instances, we observed a significantly lower percentage of respondents supporting military activities in Ukraine when asked indirectly. This difference was approximately 24 percentage points at the end of March, 13 percentage points at the end of April, and roughly 30 percentage points at the end of June and August.
Although these results imply a decline in support for military activities, it is crucial to note that the margin of error for indirect questions is greater than for direct ones. Therefore, differences across months are likely to be smaller than reported. Nevertheless, this trend largely aligns with results from direct questions, which also indicate a modest decline in relatively stable support over the same period.
What do our findings mean for understanding public opinion in Russia?
Taken together, our results suggest that a significant portion of respondents hide their disapproval when questioned directly about support for military activities. Identifying the source of this concealment—whether it be fear of social stigma or official sanctions—will require further research. Our results only provide suggestive evidence of the impact of the law and escalating repercussions for publicly expressing dissent. What we can conclude with certainty is that the figures from public opinion polls indicating overwhelming support for the invasion since March 2022 cannot be taken at face value and unquestionably trusted.
