The political crisis following the fraudulent 2020 presidential election and subsequent mass protests in Belarus renewed international attention to the country and, in particular, to Belarusian public opinion. Demand for systematic data on societal attitudes increased further after Russia’s full-scale invasion of Ukraine, in which Lukashenka’s regime played an instrumental role. Despite this renewed interest, post-2020 public opinion data on Belarus remain insufficient for drawing valid conclusions about Belarusian society. At best illustrative, these data suffer from severe methodological problems common to opinion polling in authoritarian contexts.
Until 2006, Belarus had regular in-person surveys based on representative sampling, conducted by the Independent Institute of Socio-Economic and Political Studies (IISEPS). In that year, however, IISEPS was forcibly shut down (most likely, for reporting figures unfavorable to the authorities), and its founder and director eventually emigrated. Since then, Belarusian authorities have effectively prohibited independent sociological research, including by requiring special licenses for political polling that are unattainable without state affiliation. Some limited polling on sensitive political issues (e.g., electoral fairness) did occur after 2006 within international projects such as the World Values Survey (2011) and the European Values Study (2017), though these surveys relied on state-affiliated institutions (this fact alone, however, does not suggest that the results of these surveys are inaccurate).
Independent polling re-emerged during the 2020 protests. While some researchers conducted ad hoc surveys via social media (see, e.g., the 2021 ZOiS report by Douglas et al.), the most visible and enduring initiative has been the Belarus Initiative, launched under the auspices of Chatham House. Since late 2020, it has conducted more than 20 online surveys. Given the uneven distribution of internet access, these surveys effectively cover only urban residents, who constitute roughly 80 per cent of the population (according to the World Bank) and tend to be younger and more educated than those living in rural areas. Other initiatives emerged after 2020, but they suffer from similar methodological problems. The Belarus Initiative nevertheless stands out for its relatively transparent data sharing and regularity, which is why the discussion below focuses primarily on its surveys.
Who Dares to Answer: Sampling Bias under Repression
A key issue concerns sample recruitment. The Belarus Initiative provides little or no information about its recruitment procedures. While this opacity may reflect security concerns, even limited disclosure–such as the platforms used for recruitment–would improve understanding of whom the data actually represent.
More fundamentally, online surveys in contemporary Belarus almost certainly produce systematic over- and under-representation. This problem goes well beyond the usual bias towards more educated or politically interested respondents. In a context where consuming independent media or engaging in political activity can trigger repression, participation in political surveys is inherently risky. Respondents are therefore likely to be less risk-averse and to hold more salient or extreme political views than the average urban resident – characteristics that are largely unobservable and cannot be corrected through weighting.
This bias affects not only regime opponents. Some regime supporters may also avoid surveys, particularly if they are aware of repression or know individuals who have been affected. Estimates by the human rights organisation Viasna suggest that between 300,000 and 1,000,000 Belarusians (up to around 9 per cent of the population) have left the country after 2020, making indirect exposure to repression common even among loyalists, who are likely to have friends or relatives who have emigrated due to political risks. As a result, both opposition-leaning and pro-regime groups may be systematically misrepresented.
Despite these risks, the Belarus Initiative has not assessed whether its samples under-represent respondents with particular political orientations (unlike efforts undertaken by organisations such as the Levada Center in Russia after 2022). Increasing sample sizes could partially alleviate these problems, yet most survey waves include only around 800 respondents.
Taken together, these issues imply that the Belarus Initiative’s samples do not merely exclude rural residents, who may be especially supportive of Lukashenka given their reliance on state-provided jobs and state TV for information, but may also seriously misrepresent urban Belarusian society itself. Post-stratification weighting employed by the Belarus Initiative does not resolve this problem. Where the underlying sample is not plausibly representative because of likely unobserved characteristics, census-based weighting is unlikely to improve accuracy and may instead amplify bias, especially in authoritarian contexts where census data are themselves unreliable (the last Belarusian census was conducted in 2019).
What Respondents Say – and What Fear Distorts
Data validity poses an additional challenge. In authoritarian settings, respondents may conceal their true preferences due to fear or social desirability pressures (see, e.g. Hale 2022 and Buckley et al. 2024on support for Putin). The Belarus Initiative has attempted to assess preference falsification, but only in a limited way. Its analysis from mid-2021 suggested that, at the time, at least 10 percent of urban Belarusians, if not substantially more, concealed their support for Sviatlana Tsikhanouskaya in the 2020 election. Adjusting for this estimated concealment, her actual vote share in urban areas may have been no lower than 60 per cent. Then, a late 2023 probe indicated that disapproval of Russia’s actions in Ukraine may be understated by roughly 16 per cent, suggesting that more than 50 per cent of urban Belarusians may disapprove of the Kremlin’s aggression.
Whether these discrepancies are large or small is open to debate. More importantly, severe sampling deficiencies complicate interpretation of such adjustments and limit the data’s suitability for rigorous research. Their frequent use in media reporting is especially problematic, as it may mislead audiences about the distribution of political attitudes in Belarusian society.
More Surveys, Same Problems
Other survey initiatives include those by the International Strategic Action Network for Security (iSANS), Narodnae Apytanne, and the Belarusian Analytical Workshop. iSANS and Narodnae Apytanne rely on online surveys, while the Belarusian Analytical Workshop claims to conduct telephone surveys. Since late 2023, iSANS has fielded five surveys on foreign policy attitudes. Although its samples are larger than those of the Belarus Initiative (more than 1,000 respondents), the same methodological problems apply, and the data are not publicly accessible. It is also possible that iSANS and the Belarus Initiative draw from the same respondent pools, though this cannot be verified.
Narodnae Apytanne, active mainly in 2020–2021, focused explicitly on opposition-minded Belarusians recruited through social media. While valuable for studying this group retrospectively, its data are unsuitable for drawing conclusions about Belarusian society as a whole. While its partial inclusion of Belarusians abroad is a relative advantage, this coverage remains limited in scope and insufficient for systematic research on Belarusian political emigrants.
Finally, the Belarusian Analytical Workshop is the most opaque initiative. Founded by a prominent Belarusian sociologist in exile, it claims to conduct telephone surveys of Belarusians in Poland yet provides no information on sampling, instruments, or raw data, raising serious concerns about data quality.
In sum, while post-2020 polling initiatives have partially filled the informational void created by the repression of independent sociology in Belarus, their methodological limitations severely constrain their analytical value. Without greater transparency, systematic assessment of selection bias, and more credible sampling strategies, these data should be used with extreme caution in both academic research and public discourse.
About the author
Stas Gorelik, originally from Belarus, holds a PhD in political science from George Washington University. His research relies extensively on public opinion data.
Further reading
Greene, Samuel (2022): You are what you read: media, identity, and community in the 2020 Belarusian uprising. Post-Soviet Affairs, 38(1–2), 88–106. https://doi.org/10.1080/1060586X.2022.2031843.
Kazharski, Aliaksei (ed.) (2025): Routledge Handbook of Contemporary Belarus, Routledge, London and New York.
Rosenfeld, Bryn (2023): Survey research in Russia: in the shadow of war. Post-Soviet Affairs, 39(1–2), 38–48. https://doi.org/10.1080/1060586X.2022.2151767.