Horizontal Segregation by Social Origin in Swedish Tertiary Education
It is a well established fact that social origin affects the level of schooling individuals attain. Recently, scholars of social inequality have directed attention to social origin effects on the orientation of educational attainment. In the light of the great expansion of educational systems in the last decades in many European countries, horizontal educational choices have potentially become more important for class reproduction. It has been observed that different fields of study generate diverse outcomes within each level of education.
Some researchers have proposed that when attendance at a given level of schooling reaches saturation, privileged groups look for qualitative differences within each level to secure advantage. There is some evidence that children of manual workers enrol in the less prestigious and less selective institutions. Hence, educational expansion as a means of reducing social inequality has been questioned. Uncovering the rationale of horizontal educational choices is thus highly relevant for the sociology of inequality.
This paper aims at analysing whether tertiary education choice behaviour will accentuate class reproduction. Specifically, I will test three hypotheses on social origin differences in educational decision making, namely (1) that high origin individuals strive for more prestigious degrees, whereas low origin individuals unfamiliar with education pile up in the low end of the prestige scale; (2) that the advantage of high origin individuals is due to differences in the trade-off between risks and rewards, i.e., that low origin individuals behave risk-aversely and make safer choices at the cost of lower pay-offs; and (3) that individuals choose an education that matches either of their parents’ educational credentials (for instance, due to parental role models and/or inherited comparative advantage).
Apart from providing free tertiary education and having an educational system virtually without dead ends, Sweden is an interesting sociological experiment site because of an unusual richness of population level register data. Virtually all university applications are handled by a government authority and stored in a register. The register facilitates an analysis of authentically stated preference embedded in social practice (contrary to realised choices, i.e., life-course events, which is to a larger extent dependent on influence beyond an individual’s control). Since the register contains almost all programmes offered, the researcher has a very good view on the decision process.
The central idea of the paper is to estimate a decision equation, where the risk and reward characteristics of each programme influence the utility gained from choosing that programme. Contrary to many previous studies, educational programmes are not merged into broader fields of study, which will potentially put new light on social origin differences. Each programme will be linked with the labour market conditions for those who have previously graduated from the programme. Building on this, expected risk and reward characteristics are constructed as simple aggregate measures from population level registers using indicators such as high/low wage, skilled/unskilled occupation and unemployment.
The decision equation is estimated by means of a rank order logistic regression model of the ranking and inclusion of alternatives in the choice set, with risk and reward characteristics and their interaction with social origin and GPA as independent variables. The hypotheses are assessed by analyzing the first-time programme choice of graduates of the natural science track in upper-secondary school. Regardless of social origin, individuals attending the natural science track are positively selected on ability and aspiration, and focusing on this group minimizes unobserved differences and thus provides a conservative test of the hypotheses. The choice is occurring late in the educational career, and one could expect this to minimise social differences in decision making.
The estimated decision equation shows that individuals from high social origin do strive for higher rewards. However, this is not due to a risk and rewards trade off. Hence, a difference in risk aversion across classes is unlikely to explain the social differential in horizontal choices. This might in part de due to a negative correlation between risk and reward characteristics; programmes leading to great rewards also have lower risks. Individuals also choose education that resembles their parents’ education, but this effect is very weak in relation to other factors.
In order to provide a more comprehensible picture of the horizontal choice effects, differences in the expected risk and reward programme characteristics across social origin are analysed within a traditional regression framework. Conditional on an extensive control variable set (gender, GPA, number of SATs taken, age and length of studies), I find that individuals from manual origins have 8 percent lower expected chances of obtaining a high wage and 4 percent higher expected risk of unemployment. All due to differences in decision making.
One conclusion is that tertiary level horizontal choices still are characterised by class inequality, even with a conservative test case.
Hallsten, M.