Staring at the West through Kremlin-tinted glasses: Russian mass and elite divergence in attitudes toward the United States, European Union, and Ukraine before and after Crimea

ABSTRACT

In this paper, we investigatethe divergence in Russian public opinion between the masses and elites in terms of attitudes toward foreign countries in the post-Crimea era. To do so, we combine elite surveys conducted in Russia from 2000 to 2016 with an extensive database of Levada Center mass public opinion polling to test two competing models for explaining the observed divergence in mass and elite opinion: a demographic-driven Common Determinants model and a novel Kremlin Cueing model. More specifically, we assess the extent to which a set of demographic variables trained on a model of mass attitudes is able to predict elite attitudes. Our empirical evidence is more consistent with the predictions of the Kremlin Cueing model, indicating that, in some cases, elite opinion reacts very differently to shocks such as the Crimea crisis due to “where they sit” rather than who they are as individuals.

KEYWORDS:

• Elites
• public opinion
• Russia
• survey research

Notes

1. This last finding may reflect the tendency of a less-informed public to lump the EU in with a perceived “battle” of the West versus Russia, while elites understood better that the EU was taking a much more conciliatory stance with regard to Ukraine and Russia than others (such as the US). Unfortunately, data limitations prevent us from gaining a finer-grained view in time or in quality of opinion.

2. Due to the modest sample sizes involved and geographical sampling approaches employed by public opinion survey firms, we think it very unlikely that there is any more than miniscule overlap between our empirical mass and elite samples.

3. To reiterate, although for simplicity’s sake we describe this as “pressure” to adopt the Kremlin’s position, similar arguments could be made for simply social desirability bias to provide the right answer to a pollster or for a genuine desire to mimic the Kremlin’s position, just so long as this desire is stronger among certain subsets of the elite or during particular periods of time. Ultimately, the methodological approach we have taken here is not able to distinguish between these different mechanisms, but simply to differentiate between the Common Determinants and Kremlin Cueing explanations.

4. Since Levada Center survey data is only available beginning in 2000, we use only the four most recent waves of the elite survey, from 2004 through 2016.

5. For the analyses in the body of the paper, we run OLS models with this four-category dependent variable, whereas in the online appendix (https://s18798.pcdn.co/fas-joshuatucker/wp-content/uploads/sites/1941/2019/08/Buckley-Tucker-Staring-at-the-West-Online-Appendix.pdf) we dichotomize the dependent variable and run logistic regressions.

6. The goal of this indicator is to capture cohort effects – basically related to whether or not one lived as an adult under Soviet rule – above and beyond age effects. All elites in the four waves of the SRE elite survey we use have a higher education.

7. We do so because all elites in the SRE survey are based in Moscow; restricting the analysis to mass respondents in Moscow therefore ensures that our results are not somehow simply picking up a Moscow vs. non-Moscow effect.

8. The way these experiments work is to randomly ask half the sample their opinion on an issue, and then to randomly ask the other half of the sample their opinion on the issue but include a cue as to the opinion of the party in question, which in this case would be the Kremlin, Putin, United Russia, or something along those lines. See Brader and Tucker (2009) for an example of such an experiment in Russia, and Brader and Tucker (2012b) for a more general introduction to partisan cueing experiments.

9. The results from these training models are available in the online appendix in Table A3 (https://s18798.pcdn.co/fas-joshuatucker/wp-content/uploads/sites/1941/2019/08/Buckley-Tucker-Staring-at-the-West-Online-Appendix.pdf).

10. This prediction procedure will involve some error. But of course, if we predicted mass opinion in this way it would also produce some error. As primarily a function of what is included in the model, the error itself is not inherently that meaningful.

11. To recap the estimation strategy described in the previous section, we evaluate whether a common set of demographic determinants are driving attitudes toward Ukraine, the US, and the EU by first pooling the elite and mass survey data. Estimating an OLS model that includes a full set of interaction terms between an indicator for being an elite respondent and each demographic predictor allows us to examine any differences in personal features that are associated with elites’ views toward these countries and, separately, the mass public’s views toward these countries.

12. The primary difference is that for attitudes toward Ukraine, internet usage is a strongly significant predictor of differences between the masses and non-core elites while being male is not; the rest of the results are similar.

13. Although not the primary focus of our research, it is interesting to note that some differences between assessments of Ukraine, the US, and the EU emerge. While in many cases the differences are small, three variables – age, education, and internet usage – have much larger correlations with attitudes toward the US and EU than they do with attitudes toward Ukraine. We suspect this is related to Soviet-era legacies of distrust for the West but not Ukraine; this would be an interesting subject for future research.

14. The primary additional insight we gain is that non-core elites’ attitudes toward Ukraine were too positively estimated after Crimea, at the same time as their attitudes toward the EU were too negatively estimated. Core elites’ views toward Ukraine were even more positively overestimated. These facts lend additional support to the Kremlin Cueing model of elite attitude formation. Elites (especially core elites) receive the strongest cues from the Kremlin on sensitive, politicized, “hot” topics – their opinions of Ukraine were much worse than our demographic model predicted. On more neutral topics such as the EU, where Kremlin cueing is minimal or absent, elites are actually more positive about the entity than our mass-based demographic model would predict. We can only speculate as to why non-core elites are much more pro-EU than are similar individuals in the mass public. Perhaps they are better informed about the true geopolitical positions that the EU is taking, so in the absence of strong Kremlin cueing they end up taking a much more pro-Western position than expected.