What Do Voters in Ukraine Want?: A Survey Experiment on Candidate Ethnicity, Language, and Policy Orientation

Abstract

Language, ethnicity, and policy orientation toward Europe are key cleavages in Ukrainian society, but there is much debate about their relative importance. Based on a survey experiment of 1000 residents of Ukraine that manipulated three features of a fictional candidate running for parliament, I find that a candidate’s ethnicity and language had little impact on reported vote choice, whereas economic policy orientation toward Europe was strongly associated with vote preference. Despite the intense politicization of both ethnicity and language and the violence in eastern Ukraine vote choice has not been reduced to an ethnic or linguistic census.

ACKNOWLEDGMENTS

Thanks to Yuri Shevchuk, Valery Kuchynsky, and Marko Andryczyk for linguistic advice, Oxana Shevel, Mark Beissinger, Graeme Robertson, Joshua Tucker, and members of a Harriman Institute/ICSID workshop held September 18–19, 2014 for comments.

Notes

1. See Clem (2014) for a rebuttal of this view.

2. This problem should be familiar to voters in U.S. Presidential elections who often complain of having limited choices.

3. Wanner (2014:427) notes: “After 23 years of independence, Ukraine retains significant regional diversity and strong local identities. At the same time, social differences understood in terms of ethnicity, language choice, and religious affiliation have become less defined, as Ukrainians have embraced fluid linguistic and religious practices that defy easy characterization.”

4. Here I focus on candidate ethnicity, language, and policy orientation. Other potential sources of cleavage are beyond the scope of this article. On the impact of religion as a source of division, see Snegovaya (2010); on the influence of education on identity, see Darden and Grzymala-Busse (2007); and Darden (2014); on the relationship between ethnic and political identity, see Shevel (2002).

5. Barrington (2002) finds that both language and ethnicity are less important than one might expect once region is controlled for.

6. See Darden, Keith and Lucan Way, 2014. “Who are the Protesters in Ukraine?” Monkey Cage February 12, 2014. http://www.washingtonpost.com/blogs/monkey-cage/wp/2014/02/12/who-are-the-protesters-in-ukraine/

7. One often-studied cleavage in Ukrainian politics is region (cf. O’Loughlin 2001; Barrington and Herron 2001). Here I do not manipulate the region of the candidate for a number of reasons. First, adding a fourth candidate dimension would make hinder the analysis by reducing degrees of freedom. In addition, capturing region as a concept in a single manipulation is very difficult given debate over the number of “regions,” the political salience of regions, and the precise nature of the region effect. On this last point it is unclear whether regional effect is primarily a compositional effect in that in merely proxies for other factors that are geographically concentrated or exerts an independent effect. See also Birch (2000).

8. Interviewers did not travel at all to Luhansk region and avoided areas of fighting in Donetsk region.

9. See Barrington and Herron (2004) for a discussion of how different definitions of region can shape results.

10. This vignette is unlikely to capture the nuances of language use, ethnicity, or policy orientation. Economic policy orientation toward Russia and Europe has deep cultural and political connotations; ethnicity is more subtle than an individual’s name; and native language does not include the possibility of being bi-lingual. In addition, the question holds constant and does not manipulate region, class, or other potentially relevant dimensions of candidates. Yet, the hope here is that comparing how small changes in information about a candidate shapes vote preferences can help identify the independent impact of these factors that are often highly correlated.

11. Recoding the “hard to say” responses as a middle category produces essentially the same results.

12. These results are similar if we limit the sample to the 81 percent of respondents who identified themselves as ethnic Ukrainian rather than those who self-identify as native Ukrainian speakers. On the problem of relying on self-reports of native language in Ukraine, see Bremmer (1994).

13. Breaking the population into three groups—native Ukrainophone Ukrainians, native Russophone Ukrainians, and native Russophone Russians—produces similar results but the sample size for each treatment is smaller, which makes it more difficult to draw firm conclusions.

14. These results are not shown but are available from the author.

15. It should be noted that Boris is a common Russian name as well, and perhaps Taras would have been a better choice, but the patronymic and last name are clearly Ukrainian and among Ukrainian speakers the Ukrainian spelling, Borys, was used.

16. The results from Figures 1–5 are also borne out using regression analysis.

17. The subgroup results are similar with samples limited to self-identified ethnic Ukrainians as well, but because ethnic Russians make up only about 16 percent of the sample it is hard to analyze their responses to different candidates with a great deal of confidence.

18. Vitali Klitschko. “Pugilist politician aiming to lead Ukraine.” December 6, 2013. http://www.ft.com/intl/cms/s/0/24824ffc-5d16-11e3-a558-00144feabdc0.html#axzz3FUU2GkMQ

19. Viktor Yanukovych. “Can the great survivor of Ukraine politics hang on?” December 2, 2013. http://www.theguardian.com/world/2013/dec/02/viktor-yanukovych-ukraine-protests-orange

Additional information

Funding

The author thanks the Harriman Institute and a grant from Stephan and Tamara Tymkiw for supporting this research.