While women have surpassed men in terms of participation in higher education in almost all countries of the world, women and men continue to choose different fields of study. In two important papers Charles and Brady (2002, 2009) explored cross-national variation in gender segregation in higher education. In a cross-country cross sectional perspective they found that some characteristics of modern societies seem to increase levels of gender segregation in fields of study (while they seem to decrease gender differences in educational attainment). More specifically they argue that that structural features of modern labor markets and educational systems "support the cultivation, realization, and display of gender-specific curricular affinities" (Charles and Bradley 2009: 924) and thus increase levels of gender segregation in fields of study.

The first goal of the current paper is thus to replicate und update Charles and Bradley's (2009) recent paper that is based on data from the mid 1990s with the most recent data that is currently available. To this end we compiled a macro-level dataset on 42 countries based on statistics on gender segregation provided by Eurostat, UNESCO, The World Bank and The United Nations Development Programme from the year 2000-2011. While our sample of countries is not identical to Charles and Bradley, there is considerable overlap between the two studies. Furthermore we also calculate field-specific gender segregation (Aj) with the method introduced by Charels and Grusky (1995) as our dependent variable and try to explain variation in gender segregation with economic growth, labour market functions, tertiary system indicators and other non-economic indicators such as societal growth as our independent variables. The second goal of this paper is to go beyond previous work that is based on crossectional cross-country comparative models by exploiting the availability of time series data for twelve years. More specifically we run fixed-effect panel regressions models that enable us to estimate whether within country change in the key macro variables indeed leads to increasing levels of gender segregation. Our bivariate and multivariate analyses demonstrate that gender segregation in tertiary education is still evident and that the gender divide in soft and hard sciences persist. Our fixed-effect models show some interesting results that partly contradict Charles and Bradley's (2009) cross-sectional findings.

For further information about Gendering in Research:
Lea Skewes etnols@cas.au.dk
Hanna Thaler hthaler@cas.au.dk
Emma von Essen vonessen@econ.au.dk