Gender and Perceived Control in the Russian Federation

Abstract

The consequences of the socioeconomic transition for male physical health in Russia are widely documented, but much less research examines psychological well-being and mental outlook, or how these outcomes differ for men and women in Russia. Exploring gender differences in personal control, an important aspect of psychological well-being, we find that, controlling for social and economic factors, Russian women report lower levels of control than men. Gender differences in the social determinants of perceived control highlight the importance of gender roles for understanding psychological stress and inform how social expectations differentially influence men's and women's paths to well-being in Russia.

Notes

Women and men differ in their likelihood of expressing emotions in surveys (Sigmon et al. 2005) and in types of emotional responses to stress (Aneshensel et al. 1991; Rosenfield 1999), but previous research demonstrates that these effects do not explain the gender gap in psychological distress (Mirowsky & Ross 2003).

University of North Carolina-Chapel Hill, Carolina Population Center, ‘Russia Longitudinal Monitoring Survey (RLMS): Project Description’, 31 May 2005, available at: http://www.cpc.unc.edu/projects/rlms, accessed 21 September 2008. We thank the Russia Longitudinal Monitoring Survey Phase 2 of the Carolina Population Center and Russian Institute of Sociology, funded by the United States Agency for International Development (USAID) and National Institutes of Health (NIH) (R01-HD38700), Higher School of Economics and Pension Fund of Russia for making these data available.

Because these data were collected in 2003, when the post-1998 economic upsurge was well underway and the widely unpopular social benefit reforms had not yet been introduced, we are able to examine male and female differences in personal control after the effects of the economic crisis had been substantially ameliorated and before in kind social benefits were monetised. Although it would be useful to examine the social determinants of perceived control before and after the 1998 crisis, the control questions were not included in the RLMS until 2002.

Weights for the multi-stage cluster sample were not assigned to individuals who were followed up after leaving the original dwelling site. We are primarily interested in establishing male–female differences within the sample, and therefore elected not to employ the weights in these analyses. This enables us to include data on 2,323 additional individuals with complete information for individuals for whom no sample weight was calculated. Our main substantive conclusions do not change when the data are weighted.

Cronbach's alpha, a measure of internal consistency between the six selected items, is 0.74, indicating that responses to the six statements are interrelated at a fairly high level.

For women, we found correlations of 0.4 or higher between the independent variable older age and (a) current work (−0.52) and (b) widowed status (0.52). Independent variables correlated over 0.4 for men were younger age group and currently married (−0.42) and older age group and current work (−0.42). Results for these variables should be interpreted with some care when both are in the model. Correlations between these independent variables never exceed 0.52 for women or 0.42 for men and variance inflation factor (VIF) tests were within normal range (<3.4 for women and <2.6 for men).

Because of the relative homogeneity of the sample and the complication of local versus national ethnic majorities, we did not include ethnicity in our models.

In this section, we report the significance of differences between male and female models based on a series of tests using the pooled sample of men and women. To conserve space, these ancillary analyses are not shown, but are available from the authors upon request. These results show that the gender differential is substantial in the bivariate case (−1.17, p < 0.001) and remains highly significant after controlling for demographic, economic and social variables (−0.92, p < 0.001).

We also tested the interactions between (a) current employment and the presence of a person 65 or older in the home and (b) work status and gender role attitude, but they were not significant in the female or male sample.

See, for example, Evans (2002) for an investigation of control in England and Germany; Grob et al. (1996) on Eastern Europe, Western Europe and the United States; Oettingen (1995) on Germany, Russia and the United States; Sastry and Ross (1998) on Asia and the United States; and Wardle et al. (2004) for a comparison of East Central and Western Europe.

These detailed ancillary analyses are available from the authors upon request.