Anti-regime action and geopolitical polarization: understanding protester dispositions in Belarus

ABSTRACT Do geopolitical orientations distinguish anti-Lukashenka protesters from non-protesters in Belarus? Employing data from an original online protest survey among 18+year-old citizens of Belarus residing in the country (MOBILISE 2020, n= 17,174) fielded 18August2020–29January2021, this paper compares protesters (n = 11,719) to non-protesters (n = 5,455) to better understand the dispositions that distinguish them. First, our logistic regression analysis finds robust evidence of polarization along geopolitical lines (with protesters preferring apro-EU and an anti-Russia orientation). Second, we show that pro-EU foreign policy preferences of protesters are neither temporally determined nor driven by the crisis, and are thus foundational among the positions held by anti-regime protesters. Third, we find that pro-EU and anti-Russia attitudes align with liberal democratic dispositions. Our study calls for the more systematic integration of foreign policy preferences into the comparative study of mobilization.


Introduction
At first glance, there seems to be little reason to expect that geopolitical positions should drive or even align with anti-authoritarian regime protest participation in Belarus in 2020. Research on political divides finds that geopolitics and foreign policy preferences have not been drivers of ordinary Belarusians' political behavior and are unlikely to be central dividing foci between supporters and opponents of President Alyaksandr Lukashenka's regime. Lukashenka had played off Russia and the European Union (EU) against each other in the run-up to the presidential elections in August 2020 (Allison, White, and Light 2005;Nice 2012;Marples 2013), opposition politicians have also been divided on policy regarding closer ties with the EU versus Russia (Vanderhill 2008;Bosse 2012;Nice 2012;Korosteleva 2014;Hansbury 2021), and ordinary Belarusians are believed to prefer close ties with both the EU and Russia. Not only has no clear majority coalesced around either foreign policy preference in Belarus -pre-2020 research highlighted that citizens who opposed Lukashenka's regime were also split in their geopolitical orientations (Korosteleva and White 2006;Melyantsou 2014;McAllister and White 2015).
Comparative contentious politics research has not considered foreign policy preferences to be among the main foundations of anti-authoritarian regime mass mobilization, when analyzing protest grievances or claims (with the notable exception of specific anti-war/occupation mobilizations). 1 More generally, concrete foreign policy preferences are rarely thought of as In what follows, we examine the role of geopolitics during the 2020 protests and contextualize foreign policy preferences among Belarusians leading up to 2020. Challenging status quo expectations, we further unpack our puzzle and present descriptive statistics helping us to differentiate among protesters by examining their geopolitical orientations over time. In doing so, we show that pro-EU foreign policy preferences among anti-Lukashenka protesters are not temporally bounded and shaped by the escalating crisis, but rather that they are correlates of protest engagement from the very first week of mobilization. Next, before we further test the robustness of these findings, we frame our analysis in a theoretical expectation of liberal democratic normative value alignment of pro-EU orientations. We then present and operationalize our data and justify our analytical approach. Employing logistic regression analysis, we find robust evidence that not only are antiregime protesters and non-protesters polarized along geopolitical lines (with protesters preferring a pro-EU position), but also that this preference aligns with other value-based dispositions and attitudes (specifically liberal democratic sentiments). This important finding not only counters popular discourses on the crisis in Belarus and the foundations of anti-authoritarian action, but also elucidates the continued role of pro-democracy values as conditioning pro-EU attitudes in Eastern Europe.

Geopolitics and the case of Belarus's 2020 mass mobilization
After a contested election, amidst accusations of fraud, on 9 August 2020Belarus's Central Electoral Commission announced that Lukashenka won more than 80% of the vote and would continue as President, a position he had already held for 26 years. The reaction of the opposition and ordinary Belarusians was immediate. The capital city Minsk was among 10 localities across Belarus that saw protests on that first day (Mateo 2020). Belarusian security forces responded with swift, indiscriminate repression -arresting 7,000 people in the first days after the election (Shotter and Seddon 2020). By the week's end, this miscalculated repression and the images of torture accompanying it brought an estimated 250,000-300,000 protest participants to the streets (Onuch 2020) -with protesters regularly turning out on weekends in over a 100 localities across all regions of Belarus (Mateo 2020).
This mass mobilization was not controlled by a single central activist command structure or social movement organization -and yet, the initial master narrative of the protests was highly managed. As a calculated decision to distance themselves from comparisons with the Ukrainian EuroMaidan, geopolitics did not feature in the election and protest discourses. At first, this was done ad hoc by: opposition candidates (such as the presumed winner of the elections Sviatlana Tsikhanouskaya and her team), activists, journalists and other public figures, and later the Coordination Council of the Opposition. Moreover, in the early phase of the protests prominent members of the opposition, including Tsikhanouskaya and Maria Kalesnikava, even went so far as to warn against framing the protests as a geopolitical choice and called on external actors to let Belarusians decide their political future independently (Barigazzi and Herszenhorn 2020). The master narrative was cautiously focused on a domestic political choice and living "a normal life" in Belarus.
Moreover, the initial reaction from the EU was slow and member states struggled to formulate a coherent response. While the EU condemned fraud and violence, initially only individual EU member states (Lithuania and Poland) imposed sanctions on regime insiders and provided asylum to opposition actors (Dapkus 2020). Other EU members feared provoking Russian reaction and -via the OSCE -attempted to broker a dialogue asking Russian President Vladimir Putin to bring Lukashenka to the table (Posaner 2020). Russia also adopted a "wait and see" approach -with the Kremlin refraining from commenting on the "internal affairs" of Belarus and guiding media to stress that the protests were not directed against Moscow (RT 2020).
Foreign policy preferences also did not feature on the protest posters carried or in the slogans shouted by ordinary citizens. Unlike the red and white national flag of independent Belarus, EU flags were mostly absent from the visual esthetic of contention on the streets or online on Telegram. And protesters were overheard telling others to "put away the blue (EU) flag . . . " (RT 2020) when they did see it. Some suggested that this was a reflection of the country's close social, political, and economic linkages to Russia and respectively weaker links to the EU (Liubakova 2020). Based on these initial observations of protesters and taking elite and activist narratives at face value, we would not expect protesters to distinguish themselves from non-protesters on geopolitical lines. 3 The literature on Belarusian-EU relations (Allison, White, and Light 2005;Nice 2012;Korosteleva 2016) and public opinion towards EU accession (Burant 1995;Melyantsou 2014;McAllister and White 2015) also does not suggest that geopolitical orientations drive political behavior in Belarus. The Lukashenka regime, described as pragmatic and even cynical, has been found to play off the EU against Russia and even distance itself from Moscow in the aftermath of the annexation of Crimea in 2014 and the war in eastern Ukraine (Hansbury 2021; see also Nice 2012). Hansbury (2021) notes that foreign policy was "highly constrained by domestic interest group competition" that prevented a complete distancing from Russia or the EU. Some opposition leaders were more pro-Russia than they were pro-EU and others preferred to maintain a distance from both -a position thought to be supported by a majority of citizens.
Scholars have noted that pro-EU or Russia integration did not from a central cleavage electorally (Melyantsou 2014;Bakanova 2015). Yet, this perspective might not convey the full spectrum of micro-level policy preferences among Belarusians and, as we argue, among those who made up the anti-authoritarian protestorate in 2020. The story has been more mixed than the literature suggests. While a plurality of Belarusians supported Belarus "following its own path" in a set of credible opinion polls (IISEPS 2005(IISEPS , 2006(IISEPS , 2007(IISEPS , 2009a(IISEPS , 2009b(IISEPS , 2015, when asked to choose between Russia and joining the EU, the latter often received a plurality of support (Melyantsou 2014(Melyantsou , 2018. Support for joining the EU hovered around 40% (Melyantsou 2014), and as highlighted by O'Loughlin and Toal (2022) elsewhere in this special issue, this trend continued into early 2020.
In our own review of available data, we find that there had been a steady rise in support for joining the EU (see Figure 1). Between 2006 and 2009 (when the question was asked), the percentage of respondents who said that they would vote in favor of Belarus joining the EU in a referendum increased from 32% to 43% and 40-47% agreed that Belarus was likely to be an EU member in 10 years or less (IISEPS 2006(IISEPS , 2007(IISEPS , 2008(IISEPS , 2009a. At the same time, in the immediate aftermath of the EuroMaidan, 36% of Belarusians surveyed in 2015 said their view of the EU had worsened -while only 7% said it had improved (IISEPS 2015), and few Belarusians prioritized EU acession in their voting. And as reported elsewhere in this special issue, although based on a differently designed survey item -in early 2020, O'Loughlin, Toal, and Kolosov (2021) asked Belarusians to choose between joining the EU, staying in the EAEU, not joining any Union, or having close relations with both -it was determined that only 14.5% of survey respondents chose joining the EU and 40.3%, a plurality, opted for both. Thus, although data show some increase in support for closer relations with the EU, this trend was not linearly consistent over time. Moreover, there is little evidence of alignment of a pro-EU orientation with anti-regime preferences leading up to the 2020 elections and the protest they sparked. If geopolitical orientations were to be among the foundational dispositions distinguishing anti-authoritarian protesters from non-protesters in 2020 -the case would prove to be an unexpected and critical one, highlighting a perplexing puzzle (Gerring 2007).

Geopolitical orientations and protest: a Belarusian puzzle
Taken together, the fact that geopolitics did not feature discursively or visually in the initial protest phase in 2020 and that scholarship on Belarus noted the lack of a clear cleavage positioned along geopolitical lines, we were surprised to find a different pattern in our own protest survey data (MOBILISE 2020/2021). Our data show quite decisively that protesters on the whole were twice as likely to support joining the EU (see Figure 2). 4 Furthermore, protesters were also more than four times less likely to want to join Russia and more than twice as likely to see Russia as the biggest threat facing Belarus. This descriptive pattern suggests that protesters distinguished themselves from non-protesters along geopolitical lines and requires further exploration.
Conducting a quick test 5 for correlation between being a protester and holding these three foreign policy positions, we find that there is a statistically significant relationship between having pro-EU and anti-Russia foreign policy positions and being a protester (see Figure 3). Wanting to join the EU increased the likelihood that someone declared being a protester by 21%, and if a respondent declared wanting to join Russia it decreased the likelihood of being a protester by 44%. In light of the above, this unexpected finding might prompt one to hypothesize that the geopolitical positioning among protesters was the product of shifts in opposition elite narratives, a clearer stance by the EU and Russia vis-à-vis the regime, and the general escalation of the situation over time. As the 2020 protests unfolded over weeks and months, prominent Belarusian politicians and activists, especially those who had to flee to EU capitals in Latvia, Lithuania, or Poland, began to emphasize the alignment of Belarusian and European liberal values. The leaders began to openly call on the EU to issue tougher sanctions on the regime and protesters declared that they wanted to live in a "normal" democratic country, "like in the EU" (for an example see BelsatTV coverage on 12 September 2020). At the same time, Russia and the EU were taking clear sides in the crisis.
Starting in September, EU representatives, and member state politicians and diplomats, increasingly made public overtures of support for the protesters, even attending protest events themselves. After two months of discussion, the EU finally reached a consensus in October and instituted travel bans and asset freezes for individuals responsible for the fraud and repression (European Council 2021). By mid-November Lukashenka himself and 14 other officials were added to the sanctions list, with more sanctions following in December. Therefore, we might expect that those who were motivated to join the protests in September-January would pick up on these signals and be more likely to agree with a pro-EU orientation.
Over the same period, both Lukashenka and top political leaders in Moscow stepped up their rhetoric against EU interference. Putin made his support for the status quo known by telling the protesters and the EU to show restraint and to not cross certain undefined boundaries (Higgins 2020). In September, the Kremlin provided financial assistance, held out the prospect of support by Russian security forces, and praised his regime for showing "restraint" (Higgins 2020). By October, when the head of Russia's SVR foreign intelligence service made a public visit to Minsk (Reuters 2020), the signals about Russia's position as on the side of the regime and not the protesters was reinforced. Thus, we might expect that those who were motivated to join the protests after September would also read these signals and be more likely to agree that, in the language of the survey, Russia represents a major threat.
This geopoliticization hypothesis would lead us to expect that: (a) those interviewed in the later stages of the protests, (b) those who joined the protests later, and (c) those who protested repeatedly would be more likely to distinguish themselves along foreign policy preferences from other protesters who joined earlier or were interviewed earlier.
Following on from Saunders et al. (2012), we need to differentiate between protesters to see if those who joined at earlier and later stages of the protest were more or less likely to hold certain geopolitical positions. While Saunders et. al differentiate between novices, returners, repeaters, and stalwarts, here we think it is useful to distinguish between early risers (those who joined in the protests in the first week until the end of August), late joiners (those who joined in the latter five months September-January), and repeaters/stalwarts (those who reported repeat participation, participating in at least two weekend protests or in two months). If the geopoliticization hypothesis is correct, we would expect to see that early risers, late joiners, and repeaters/stalwarts distinguish themselves from one another along foreign policy orientations. If the geopolitical positionality of protesters shifted over the course of the mobilization, we would expect early risers (and especially those who protested in the first week of the first month) to be less likely to distinguish themselves from other protesters along geopolitical lines. If the shift in elite discourses and the more proactive stance of the EU and Russia pushed protesters to adopt certain geopolitical orientations, we would expect late joiners and repeaters/stalwarts to be more likely to be pro-EU and anti-Russia, when compared to other protesters.
Again, we find a puzzling result. When we run simple logistic regressions (Figure 4), controlling only for month of survey, we do not find support for the geopoliticization hypothesis. We instead find that early risers were more and not less likely to agree with the statement that Belarus should join the EU (Figure 4(a)), that there was no significant difference between early risers and late joiners with regard to the statement that Belarus should join Russia (Figures 4(d,e)). Early risers were more likely to view Russia as a threat to Belarus while late joiners were less likely to hold this view. Almost all repeaters/stalwarts reported having joined the protests in the first weeks and distinguished themselves along all three geopolitical questions, thereby highlighting that those who rose early and were more committed to the protests were equally more likely to hold pro-EU and anti-Russia views. Lastly, when we include all three geopolitical positions in one regression model, we see that early risers only distinguish themselves from late joiners when it comes to seeing Russia as a threat (see Figure 5(a-c)), while repeaters/stalwarts distinguish themselves from all other protesters along each geopolitical orientation ( Figure 5(d)). Thus, we can reject the geopoliticization hypothesis, but our central puzzle remains. This initial analysis makes clear that we need to: (a) further explore the robustness of whether geopolitical orientations distinguish anti-Lukashenka protesters from non-protesters in 2020, and (b) we also need to seek an explanation for why this might be the case. In what follows we propose a theory suggesting that these preferences align with other pro-democracy attitudes that jointly make up the foundation of antiauthoritarian mobilization.

Framing the analysis: geopolitical foundation and value alignment hypotheses
Before we move to our analysis unpacking whether and why anti-Lukashenka protesters distinguished themselves from non-protesters along foreign policy lines, we first consider the known drivers of pro-EU (as well as anti-and pro-Russia) policy dispositions among ordinary citizens and whether any of them overlap with known drivers of protest behavior. To do so, our paper first builds on studies of public opinion of foreign policy -specifically towards EU accession -in postcommunist states.
When investigating the factors behind a pro-EU sentiment in Eastern Europe, scholars have tended to distinguish between two sets of factors shaping these attitudes: (1) economic experiences, more specifically grievances and expectations, and (2) normative political value claims and value alignment (Rohrschneider 2002;Tucker, Pacek, and Berinsky 2002;Tverdova and Anderson 2004;Müller 2011). The first set centers around an instrumentalist thesis, focusing on individual-level expectations of socio-economic benefits perceived to result from closer relations with (or accession to) the EU (e.g. visa-free travel, employment, higher standards of living). This research has been criticized for not acknowledging that economic deprivation and negative ego-tropic economic experiences specifically are interlinked with partisanship, as well as socio-economic status and thus might correlate with divergent foreign policy preferences among diverse constituencies. For instance, middle class or highly educated citizens might be equally motivated by their economic deprivation as working class or less educated citizens but this might result in differential blame attribution and/or in a different calculus about the benefits of closer relations with the EU. Voters of nationalist parties might view the EU as the cause of economic deprivation in their own country and their economic grievances would therefore not be associated with support for EU accession. Thus, what it means to be economically deprived, dissatisfied, or why one comes to view an instrumental benefit from a certain foreign policy orientation is not a matter of simple causality but rather one of mediated meaning and normative imaginaries.
Research on the second set of factors thought to influence pro-EU (and anti/pro-Russia) orientations, is thus rooted in notions of values and normative power and focuses on the democratic benefits of "Europeanness" at home -both for the nation and the individual. The EU is perceived as a regional liberal democratic hegemon that can both pull democrats towards it and push away authoritarians (Manners 2006;Sasse 2008a;Niemann and De Wekker 2010;Zielonka 2013). Our own hypothesis as to what can explain why anti-regime protesters distinguish themselves from nonprotesters along geopolitical lines in Belarus 2020 is closer to this line of reasoning.
While these two sets of factors are not necessarily mutually exclusive, they do tend to be connected to two different types of deprivation, grievances, and claims -economic or political. And although both types involve some individual-level calculi, the latter is arguably more amenable to claims related to a collective or communitarian benefit, as the individual benefits of free elections are not as tangible as a higher income. Given that neither the EU nor Russia initially responded to the crisis in Belarus by offering tangible benefits at the individual-citizen or national-community levels, we hypothesize that the symbolic normative power of the EU as representing liberal democratic norms, values, and rights (Manners 2001;Pace 2007;Slobodchikoff 2010;Zielonka 2013) is likely to have played a role in protester perceptions.
We do not suggest that the actual practice and achievements of the EU correspond with this understanding of its normative power, nor that its attempts at democracy promotion through norm dissemination have been successful (Schimmelfennig 2005;Sasse 2008b;Flockhart 2010). But we do know from past research that citizens of neighboring countries often cite the EU when referencing their wish to live in a "normal" or democratic European country (Chaban, Elgstrom, and Holland 2006;Chaban and Holland 2014;Elgström and Chaban 2015). Moreover, we also know that citizens have done so explicitly when explaining their reason for engaging in protest (Galbraith 2003;Herron 2009;Onuch 2014a;Onuch and Sasse 2016;Resnick 2017). Thus, there is evidence of a popular perception of the EU as a symbol of a more "normal," stable, democratic, and liberal way of life; and in the process of their country developing closer ties to (or even joining) the EU, citizens hope that they too will get to live in a more normal, stable, democratic, and liberal society. Research also shows that Russia represents and disseminates an alternative set of norms (Lukin 2009;Østbø 2017;Roberts and Ziemer 2018) centered on strong authoritarian leadership, political stability, predictability, and "traditional" values. Jointly these factors are presented as having the potential to increase the standard of living for the median citizen -and specifically the growing middle classes, who tend to come out in support of the regime ( (Lukin 2009;Cameron and Orenstein 2012;Rosenfeld 2017). But clearly, this model is not one where meaningful pluralism, dissent, and opposition are valued or even allowed. This distinction between the two different models' norms, values, and governance also seems to filter into individual preferences. When looking for correlates of pro-EU and pro-Russia foreign policy positions in Eastern Europe, scholars have identified a significant association between supporting liberal democracy or accepting authoritarianism and support for EU accession or close relations with Russia, respectively (Müller 2011;O'Loughlin, Toal, and Kolosov 2021;Onuch and Arkwright 2021). Furthermore, skepticism of Russia or viewing Russia as a threat has also been shown to correlate to holding democratic views, while wanting closer ties or integration with Russia is linked to preferring a return to a centralized system of governance (Giuliano 2015(Giuliano , 2018O'Loughlin, Toal, and Kolosov 2017).
In line with this thinking, we do not expect to see a developing geopolitical divide over time but rather (H1) a geopolitical orientation undercurrent among protesters as a whole when compared to nonprotesters overall. We call this the geopolitical foundations hypothesis.
Following Chaban and Chaban (2018) who note positive value orientations towards the EU include a preference for good governance, the rule of law, anti-discrimination, liberty, social solidarity, and meritocracy, we would (H2) expect that there is some alignment between geopolitical orientation and support for democracy, and that together these would be important distinguishing factors between anti-authoritarian protesters and non-protesters. We call this the value alignment hypothesis.
To test the strength of our expectations around geopolitical orientation and democratic value alignment, we must consider other known correlates of protest engagement and control for them in our statistical analysis.

Controlling for drivers of protest
First, protesters tend to have a distinctive socio-demographic profile. For instance, enrollment in or attainment of higher education can offer informational resources; being younger might provide an individual with free time; being financially secure might offer security from repercussions, repressions, or lost wages; as might being younger or male increase one's willingness to engage in risky contention in the first place (McAdam 1986;Alan and Soule 2005). We would expect anti-regime protesters in Belarus to distinguish themselves from non-protesters along these factors and thus, we control for education, age, gender, and family financial situation.
Contentious politics research has also established that we have to account for the motivating grievances or claims driving protest (Iain and Pettigrew 1984;Opp 1988;Gurr 2010;Onuch 2010). Because the Belarusian context provides both economic decline (Cojocaru and Matytsin 2017; Natallia and David Cameron 2019) as well as political disaffection and a fraudulent election as a trigger, we control for: (1) economic (sociotropic and egotropic) evaluations capturing economic grievances; and (2) a reported vote for Tsikhanouskaya.

Belarus-specific controls
In the case of Belarus, although there is little evidence that ethnicity or nationality are key factors in Belarusian political behavior, some observers suggest that a national and an ethno-linguistic awakening occurred in the country in 2020 (Kulakevich 2020;Maxwell 2020). For this reason, we feel it is warranted to control for Belarusian language (native language, capturing ethnic identity, and language of the survey, capturing practice and preference), and nationality (capturing civic identity 6 ) in our model (following Olga and Hale 2018).
Lastly, because our survey was collected over a six-month period (August 2020-January 2021) we also control for the month the survey was completed.

Operationalization and analytical strategy
In our main analysis, we compare self-declared anti-Lukashenka protesters (n = 11,719) to nonprotesters (n = 5,455). We count those who self-declared having participated in the anti-Lukashenka protests in person as our "protester" category -this amounts to 68% of our survey sample (see Figure 6). 7 To investigate our theoretical and empirical propositions, we run two sets of models. The first set of models seeks to differentiate between anti-regime protesters and non-protesters in order to elucidate whether geopolitical orientations are explanatory factors (but also to establish that these positions align with democratic value dispositions). 8 Thus, our first dependent variable (DV1) is a binary variable whereby 1 denotes being an anti-Lukashenka protest participant (in-person) and 0 denotes all other responses (see Table 1). This allows us to interrogate the different possible factors correlated with anti-authoritarian protest mobilization, among them our three (binary) explanatory variables of interest, namely the respondents' agreement with the statements that: (a) Belarus should join the EU (agree fully or somewhat agree denoted by 1, all other responses denoted by 0); (b) Belarus should join Russia (agree fully or somewhat agree denoted by 1, all other responses denoted by 0); and (c) the biggest threat to Belarus is Russia (agree fully or somewhat agree denoted by 1, all other responses denoted by 0) (see Table 1).
The second set of models provides a check on how robustly being a protester and holding prodemocratic views aligns with our geopolitical orientations of interest -agreeing that Belarus should join the EU (DV2), agreeing that Belarus should join Russia (DV3), and believing that Russia is the biggest threat to Belarus (DV4). We find that the three variables are, surprisingly, correlated only at very low levels and provide details of Pearson correlation coefficients and results from Cronbach's alpha test in appendix Tables B1 and B2.
As our dependent variables are binary, our main analytical approach is to run our models as a series of logistic multivariate regressions. 11 We report the estimated full effects of each factor on a dependent variable. Full effects can be interpreted as an average marginal effect of a factor when all variables are scaled from 0 to 1. We report the change a given variable produces in a respondent's estimated likelihood of being an anti-regime protester when one increases any given factor from its minimum to its maximum value while holding all else constant. For our main analysis (DV1), we present results from a stepwise model accounting for causal ordering -in line with work by (Campbell 1980). Even if our main independent variables (IVs) are attitudinal, because we are primarily interested in the effect of our geopolitical orientation variables, and how well their effect holds up controlling for other known correlates, we present analyses where these variables were included from the first stage. In the appendix we provide the models without the immediate inclusion of these variables for comparison (Tables B3 and  B4). We do not find that the results vary in any significant way. Thus, in Stage 1, we include first our key variables of interest only -foreign policy orientation variables. In Stage 2, we add demographic controls unlikely to be driven by other factors: (a) Age in years; (b) binary variable denoting Female Gender; (c) Education as six categories; (d) a seven-point scale of the respondent's Family Financial Situation, with 1 denoting financially worst off and 7 denoting best off; and (e) the Language of the Survey (a personal language preference where 1 denotes having completed the survey in Belarusian, with English and Russian coded as 0). In Stage 3, we include identity variables: both are binary, with 1 denoting the respondent having selected Belarusian as their Nationality (civic ID) and as their Native Language (ethnic ID). In Stage 4, we include two economic evaluation variables capturing the respondent's evaluation of the economic situation in Belarus (sociotropic) and their evaluation of their household economic situation (egotropic) over the last 12 months. Following theoretical expectations about the role of deprivation and instrumentalist calculations driving both geopolitical orientation and protest engagement, we include two binary variables (sociotropic and egotropic evaluation), where 1 denotes the respective situation having "deteriorated somewhat" or being "much worse," and 0 denotes all other responses. 9 In Stage 5, we include a variable capturing a democratic disposition. Here, we employ an item asking respondents which system is most preferable for Belarus. 10 We create a binary variable whereby those respondents who stated that democracy is most preferable are denoted with a 1 and coded as being "Democrats." In Stage 6, because it is likely to be influenced by all the above included variables, we include a dummy capturing the respondent's reported vote choice for Tsikhanouskaya. 11 Finally, all of our models in all stages control for the month of survey completion.
For our secondary analysis (DV 3-4) we run the final sixth-stage model for each dependent variable and include protest participant in lieu of the respective geopolitical orientation.

Findings
As can be seen from Table 2 and Figure 7, our main analyses provide robust findings supporting our hypothesis 1 (H1) that geopolitical preferences form part of the foundational basis of anti-authoritarian protest engagement in Belarus in 2020 and allow us to reject the null hypothesis. Our evidence is clear, anti-regime protesters and non-protesters are polarized along foreign policy lines. Our findings  from this main analysis also show that these geopolitical orientations align with other key factors associated with protest mobilization (such as age and education), supporting (H2). And most importantly, in line with our own thinking described above, these dispositions also align with holding liberal democratic values. In fact, it is only after we control for political dispositions -specifically believing that the best political system is democracy and having reported voting for Tsikhanouskaya -that the full effect of holding key geopolitical positions on the estimated likelihood of being an anti-regime protester declines (while still remaining significant at the 99% confidence level) (see Figures 7(e,f)).
In our secondary analysis, we also find that, controlling for other key factors, the inverse relationship is true (see Table 3 and Figure 8). Being an anti-regime protest participant increases the estimated likelihood that someone reports holding pro-EU membership views, is against Belarus joining Russia, and would consider Russia a major threat to Belarus (at the 99% significance level). In this second-order analysis we see clear evidence of robust patterns of alignment between the three geopolitical dispositions, being a protester, and holding democratic values. As shown below, we do not see the same consistency of evidence for economic factor alignment -further confirming our proposed theory on value alignment (H2) -allowing us to reject the secondary null hypothesis.

Discussion
Not only can anti-Lukashenka regime protesters be clearly distinguished from non-protesters along geopolitical lines, but this polarization remains statistically significant even when controlling for a variety of key confounding factors. In Figure 7(b) we see that, when holding for basic demographic controls (Gender, Age, Education, Family Financial Situation, and Language of Survey), agreeing with the position that Belarus should join the EU still increases the estimated likelihood that an individual is an anti-regime protester by 12% (at the 99% significance level). In the same figure, we see that an individual is 30% less likely to be an anti-regime protester if they support Belarus joining Russia (again, at the 99% statistical significance level). And finally, holding for demographic factors, the estimated likelihood that an individual is an anti-regime protester is raised by 18% if they believe Russia is the biggest threat to Belarus.
When we include our ethnicity and economic evaluation controls, we see that the effect of holding these geopolitical positions remains strong (statistically significant at the 99% level) and quite sizable. In this stage of our analysis we find that the full effect of agreeing that Belarus should join the EU and believing that Russia is the biggest threat to Belarus increases the likelihood of being a protester by 9-7% and 15-13%, respectively, and decreases this likelihood by 30% if the respondent agreed that Belarus should join Russia. Finally, when we include our two political dispositions (having voted for Tsikhanouskaya and believing democracy is the best system) while the full effect of our geopolitical positions of interest on the estimated likelihood that a respondent would be protester shrinks (+3%, −25%, +10%), these factors remain statistically significant at the 99% level.
In this last stage we can clearly see that the effect of these geopolitical positions is absorbed or even mediated by also holding normative democratic value dispositions. This is in line with our thinking outlined above -having a preference for joining the EU is inasmuch a potential foundational factor of anti-regime protest participation, as it subsumes a particular normative democratic value or meaning. Again, when we reverse our analysis to examine the drivers of a particular geopolitical orientation (see Figure 8), while controlling for all possible confounding factors (Stage 6 model), we find that being a protester increases the estimated likelihood that an individual holds pro-EU preferences and believes Russia is major threat to Belarus by 6% and 13%, respectively (again at the 99% significance level). We also find that being an anti-regime protester decreases the full effect that a respondent would like to see Belarus join Russia by 7%. This inverted analysis helps us to better understand whether holding these three geopolitical positions aligns with liberal democratic value dispositions as opposed to instrumentalist economic deprivation grievances we discuss above.
According to our thinking, the surprising finding (and the puzzle driving our analysis) that geopolitical preferences are likely to be distinguishing factors between anti-regime protesters and non-protesters is related to the fact that these geopolitical positions go hand in hand with, and represent, liberal democratic normative values or claims. As we explained, these positions are unlikely to align with economic grievances and socio-economic instrumentalist drivers of protest. Our analyses provide empirical evidence for four things. First, Figure 7(b) shows that at its inclusion the family financial situation control is not statistically significant in driving protest engagement on its own. Thus, personal deprivation is not a foundational correlate of the anti-authoritarian protest engagement in 2020. Second, Table 3 with the results from our secondary analysis demonstrates that the full effect of going from reporting having the worst possible family financial situation (i.e. not having enough money even for the most basic things such as food) to the best family financial situation (i.e. being able to afford anything they might want) decreases the estimated likelihood that a respondent would agree with the statement that Belarus should join the EU and, instead, increases the likelihood that they would agree with the statement that Belarus should join Russia.
Similarly, and importantly for our own thinking, in our secondary model (see Table 3 and Figure 8(b)) we find that while evaluating the national as well as one's own household economic situation negatively (i.e. noting that the situation has gotten worse in the last 12 months) increases the estimated likelihood that a respondent agreed that Belarus should join the EU by 13%, it also increases the likelihood that a respondent agreed that Belarus should join Russia by 3%. Thus, as can be seen in Figure 8(a-c), experiencing economic deprivation on its own does not consistently align with either holding democratic values in these models or with clear pro-EU and anti-Russia geopolitical positions. This finding suggests that a large portion of Belarusian society has experienced some economic deprivation -and that this factor on its own does not polarize Belarusians in the same way that geopolitical positioning and holding democratic values does.

Conclusion
Based on original survey data collected among anti-Lukashenka protesters and non-protesters over a six-month period following the presidential elections in August 2020, this paper contributes to the as yet underdeveloped discussion of the role of foreign policy orientations in shaping protest behavior. Based on limited empirical evidence to date, the scholarly consensus has been that such preferences are of secondary importance as factors behind protest participation. The Belarusian case of mass mobilization serves as a powerful test case in this respect. At the outset, the country's economic dependence on Russia, the weak links to the EU, and the deliberate initial framing of the protests as not being about geopolitics by the leaders of the political opposition made us expect no pronounced differences in the geopolitical orientations of protesters and non-protesters. However, our research reveals the existence of clear differences in the geopolitical orientations of antiauthoritarian protesters and non-protesters. This observed distinction informed our hypothesis that potential distinctions between the foreign policy preferences of protesters and non-protesters align with other political preferences and identities.
Our findings are significant in that they, in the first instance, unearth a geopolitical polarization in Belarus that, akin to what we have seen in neighboring countries in the region such as Poland (in the 1990s) and Ukraine (in 2014), shapes oppositional political behavior independent of opposition elite discourses. In the second instance, we find that these patterns in geopolitical dispositions strongly correlate with liberal democratic values, highlighting the alignment of geopolitical orientations with normative value dispositions. Our research also calls into question the primacy of economic and instrumental factors explaining protest behavior.
We present robust evidence for two theoretical propositions: (1) that geopolitical orientations are (in the case of Belarus 2020) and can be (elsewhere) foundational to protest participation; and (2) that these geopolitical orientations align with normative values and meanings attributed to them (for the case of the EU and Belarusian protesters specifically, normative liberal democratic values). More broadly, building on these findings, we propose that context-specific geopolitical orientations form part of a bundle of values and preferences that distinguish anti-authoritarian protesters from nonprotesters more generally -be it in Belarus in 2020, in Ukraine in 2014, in Eastern Europe more broadly in 1989, or in other cases of mass mobilization beyond the region. As part of a set of attitudes, geopolitical orientations may play an indirect role in motivating protests. Thus, our research calls for a more nuanced understanding of the role of foreign policy orientations in antiauthoritarian protest mobilization and further exploration into how these positions align with other attitudes at the heart of anti-authoritarian mobilization more generally. Countries such as Canada, the United States, or China might also have normative symbolic power in different regions and our work calls for further comparative research focusing on the role of regional hegemons and norm entrepreneurs in protest mobilization through the lens of value alignment.
Lastly, our research revisits established research about the EU's normative power in its neighborhood -and highlights its significant import. We do not suggest that the EU has been successful in its democracy promotion through its attempts at norms dissemination nor do we mean to imply that the EU and its member states are free from anti-democratic and illiberal tendencies themselves -we simply highlight that this perception remains among citizens of neighboring countries who oppose authoritarian regimes. One logical consequence of this value alignment would be the expectation that a protest that overturns the authoritarian regime results in closer ties with the EU, but this expectation could also turn into disappointment or apathy. (2002), Schreiber (1976), and Vasi (2006). 2. The survey specifically targeted protesters using social media adverts; for full details about the recruitment, sampling, and design of the survey please see Appendix A. 3. Some might argue that it is possible that protesters did hold pro-EU orientations but hid these for strategic reasons, following elite cues. But again, if this were true we would also expect that protesters, and especially early risers, would also not disclose this disposition in our survey-for the same reason. There is little reason to believe that this falsification of preferences would occur in anonymous online posts about the protests on Telegram and not in an online survey. 4. Questions asked were: (a) "Do you agree or disagree with the following statements: Belarus should join the European Union; Belarus should join Russia; The biggest threat to Belarus is Russia." Answer options: 1 -Completely agree; 2 -Somewhat agree; 3 -Somewhat disagree; 4 -Completely disagree; 97 -Hard to say; 98 -Refuse to answer. (b) "Have you participated in any of the ongoing protests in Belarus (or abroad if you are based outside of Belarus) that began following the Presidential Elections of August 9, 2020?" 1 -Yes, in person only; 2 -Yes, online only; 3 -Yes, both in person and online; 4 -No; 98 -Refuse to Answer. (c) "Were these protests": 1 -Against Lukashenka and against the official August 9, 2020 election results; 2 -In support of Lukashenka and his government; 97 -Hard to Say; 98 -Refuse to Answer. 5. We employ logistic regression analysis with a binary dependent variable, where 1 means a respondent was a protester. Each of the foreign policy positions are coded as 0 or 1, with 1 denoting agreement with the statements Belarus should join the EU, Belarus should join Russia, and Russia is the biggest threat to Belarus. All other answers were coded as 0. We control only for month of survey. 6. Researchers working on ethnic and civic identity in the post-Soviet region have repeatedly noted that this forced-choice survey item, a harbinger of Soviet policy and practice in censuses, captures civic and/or state identification and communitarian belonging (as opposed to ethnic or linguistic identity) in the post-Soviet period of independence. Scholars have also cited that a five-point scale measure of belonging to an ethnic group and native language declarations are more likely to capture ethnic identity (see Charnysh 2013;Kulyk 2001Kulyk , 2011Kulyk , 2014Onuch and Hale 2018;Woolhiser 2001). 7. Those who declared protesting only online are not included in our main protester variable. 8. Noting recent methodological discussions, as a robustness check, we also rerun all of our models employing ordinary least squares multivariate regressions. Our results hold employing both approaches. 9. We ran robustness checks with the variables coded as a five-point scale and found the results unchanged. 10. Options include: 1 -Democracy is preferable to any other kind of system; 2 -Under some circumstances, an authoritarian government is preferable; 3 -For people like me, it doesn't matter; 97 -Hard to say; 98 -Refuse to answer. 11. We also run robustness checks with vote for Lukashenka, or Hanna Kanapatskaya (thought to have been prominently pro-EU). 12. Because our project is also interested in migrant protest engagement, we ran a separate advert for Belarusians Citizens residing abroad. We do not include this data in our analysis. 13. For a small sample of cases this was cross-referenced with Survey Monkey's sandard IP address monitoring. complete in full (making it a rather long online survey). As is typical with online surveys more than 10 questions long, the drop-off (attrition) rate is higher than in face-to-face surveys. We note though that most respondents drop off on the first page of the survey.

For some excellent examples see McAdam and Su
The completion rate for the survey is over 40% -a very respectable outcome when compared to other online surveys of this kind. In total 40,344 (37,785 Russian, 2,486 Belarusian, and 73 English) respondents began the survey. For this analysis, because we are not interested in comparisons with migrants -we drop all those who were not residing in Belarus at the time of the survey and all those who reported not being citizens of Belarus (remaining n = 32,900). Of these, 31,221 respondents answered the question about protest participation -the first post-inclusion/exclusion criteria question (21,966 were self-declared in-person protest participants). Of those, 17,174 completed the survey in full (which we consider as having provided at least one response among the final battery of questions in the survey -those asking basic demographic questions). Outside of age, gender, citizenship, and location of residence (all asked at the very beginning of the survey), the majority of demographic characteristics were collected at the last stage of the survey, as is known best practice when surveying populations and especially when the topic is considered sensitive. For this reason, we only analyze those respondents who completed the survey in full as to allow for better comparison. This is not problematic for our analysis as we have no reason to believe that protesters who did not complete this survey in full were any different from non-protesters who did the same. Table B1 presents Pearson correlation coefficients for each pair of these geopolitical orientation measures, showing some but overall a low degree of intercorrelation ranging from a minimum of −0.1064 for our binary measure of Joining the EU and Joining Russia to a maximum of 0.4032 for the wanting to Join the EU and seeing Russia as a threat. Table B2 reports the findings from a Cronbach's alpha test that estimates the degree to which these variables vary together, yielding an intercorrelation statistic of 0.46. This is significantly below the threshold of 0.70 that is commonly interpreted as the minimum level of mutual correlation justifying treating a set of variables as capturing a single underlying concept. Thus, it is unlikely that these indicators are capturing a generalized phenomenon of a single pro-or anti-Russia/EU/West orientation.