Abstract
Taking advantage of a panel survey in Ukraine before and after the Euromaidan, we analyze the relationship between ethnicity, language practice, and civic identities on the one hand and political attitudes on the other. We find that while ethnic identities and language practices change little on the aggregate level over the period, there has been a significant increase in the proportion of people thinking of Ukraine as their homeland. There has also been a large fall in support for a close political and economic relationship with Russia and some increase in support for joining the European Union. Nevertheless, we find that identities in general, and language practice in particular, remain powerful predictors of political attitudes and that people are more likely to shift attitudes to reflect their identities rather than modify their identities to match their politics.
Disclosure statement
No potential conflict of interest was reported by the authors.
Supplemental data
Supplemental data for this article can be accessed here https://doi.org/10.1080/1060586X.2018.1452181.
Notes
1. Important work on the effect of regional-level variables includes Katchanovski (Citation2006), Peisakhin (Citation2013), and Darden (Citation2014). Other studies, more in the tradition of political or economic geography, emphasized the structures of economies and trade as being crucial in shaping regional political identities, relegating ethnicity and language to a less important role (Barrington Citation2002). While we do not deny the potential importance of regional-level variables as a distal cause of the division, the effects of these regional-level variables should be visible at the individual level.
2. In the 2012 survey that included Crimea and all of the Donbas, 85% of respondents gave their nationality as Ukrainian.
3. Of course, this does not preclude the possibility that other political preferences, such as about economic policy, could be intertwined with identity politics. However, given the space constraints in this paper, we decided to focus on political issues with clear links to ethnolinguistic identity.
4. For the purpose of our statistical analysis we coded “hard to say” as an intermediate category, but our results are not sensitive to alternative coding (e.g. as two sets of dummy variables).
5. We address the processes driving these changes in a separate paper.
6. The marginal effects were calculated using the margins command in Stata 13.1 on the basis of separate regressions for each survey wave. In addition to the three sets of identity variables (nationality, home language practice, and homeland), the regressions included a set of standard demographic controls (see Table A1 of the online appendix for full results).
7. Note that for each attitude we coded separately whether a respondent became more or less likely to embrace a particular attitude (e.g. Russian as a state language.) We did so in order to avoid the implicit assumption that effects are completely symmetric. See Table A2 of the online appendix for full regression results).
8. However, it should be noted that the overall increase in support for having public officials speak Russian in addition to Ukrainian (see Figure ) was not just driven by a hardening of attitudes among Russian-speakers. In fact, of the respondents who changed toward allowing for greater use of Russian, 70% reported speaking only Ukrainian at home in 2012. In other words, while Ukrainian speakers may have been somewhat more reluctant to embrace the national trend toward greater support for bilingualism among public officials, many of them nevertheless did so. This is in line with our argument about the growth in civic rather than ethnic nationalism.
9. The remaining language effects on attitude change in Figure were also consistently in the predicted direction, but were somewhat less statistically conclusive.
10. The statistical approach is identical to the analyses of identity effects on attitude change discussed above. See Table A3 of the online appendix for full regression results.