Izvorni znanstveni članak
https://doi.org/10.11567/met.34.2.2
High Education and Divorce in a Cross-national Comparison: How Do Country-specific Population Characteristics Influence a Micro-level Relationship?
Petra Međimurec
orcid.org/0000-0003-2616-2457
; Katedra za demografiju, Ekonomski fakultet, Sveučilište u Zagrebu
Sažetak
Europe saw a substantial rise in divorce over the last decades (Eurostat, 2018a). However, this general increase was very uneven: divorce occurs more frequently not only in some countries, but also in some socioeconomic categories. This study examines the relationship between high education and divorce from a multinational perspective. More specifically, the following research questions are addressed: a) How does high education affect the odds of divorce in Europe?; b) Does the relationship between high education and divorce vary across countries?; c) Do nation-specific demographic factors mediate this relationship, and how does Croatia compare to other European countries?
Different theories provide different arguments about how education relates to marital (in)stability. For example, some suggest that better-educated spouses have more liberal attitudes that make it easier to opt for a divorce (Levinger, 1976). However, most economic and psychosocial theories see high education as a factor that tends to stabilize marriages among men (Jalovaara, 2003). Becker’s family model (Becker, 1974, 1993; Becker, Landes and Michael, 1977) views the mutual dependence of spouses as a key source of gains to marriage. Because men have typically held a comparative advantage in the labour market and women in domestic work, Becker argues that gains to marriage are maximised when men who have a high economic potential partner with women having low economic potential. For men, high education raises the efficiency of role specialisation, thus reducing the odds of divorce. In contrast, for women, high education decreases the benefits of a traditional gendered division of labour, thus elevating the odds of divorce. A different line of reasoning (Oppenheimer, 1997) asserts that the wife’s high education lessens marital strain (and therefore also lessens the odds of divorce) as it improves the family’s socioeconomic status and living standard. Moreover, education can boost couples’ gains to marriage by bettering communication skills and the ability to solve conflicts (Amato, 1996; Blossfeld, 2014).
Empirical studies report mixed evidence too: divorce appears to be more common among the highly-educated in some countries and in some cohorts, but less common among the highly-educated in other countries and in other cohorts (Härkönen and Dronkers, 2006; Dronkers, 2015). Well-known theories of family change discuss possible reasons.
According to Goode (1962, 1993), one can anticipate a negative or a less positive relationship between high education and divorce in the countries with relatively little barriers to divorce. If barriers to divorce are high, divorce is more affordable to the upper strata. As barriers to divorce decline, lower strata – who generally experience more marital strain (Goode, 1951, 1962) – in turn become increasingly likely to divorce. Other accounts lead to similar predictions. The second demographic transition view (van de Kaa, 1987; Lesthaeghe, 1995) posits that new and less traditional family practices are first adopted by highly-educated innovators, before spreading among the lower strata through imitation (Lesthaeghe and Surkyn, 1988). In this vein, Blossfeld et al. (1995) argue that high education increases the odds of divorce in countries where divorce is still rare, because in such countries divorce violates the established social norms more severely. This association weakens in countries where divorce is more common, and conventions surrounding family life are less strict. In Goode’s words, this makes barriers to divorce lower, and thus reversing the educational gradient.
To test these explanations, this study delivers a comparative multilevel analysis of the educational gradient in divorce across 25 European countries. Individual-level data are drawn from the European Social Survey Round 5 integrated database (ESS, 2010). The sample includes ever married respondents, who were 35 to 54 years old at the time of the interview. Respondents with a missing value on the dependent variable or any of the predictor variables are excluded (around 1,6 %) from the sample. The sample also excludes Israel (the focus is on European countries), and Finland (because of a mistake in the question concerning marital status).
The outcome variable are the odds of being divorced. High education is the main micro-level predictor variable. It was coded as a dummy variable, indicating whether a man or a woman has successfully completed tertiary education (ISCED 2011 level 6 or higher). The following control variables are used: age, immigrant status, and religiosity.
Three measures are used to depict the country-specific demographic context: the percentage share of ever divorced men and women, the percentage share of men and women who have ever cohabitated (these two items are derived from the full European Social Survey Round 5 sample), and the percentage share of extramarital births (Eurostat, 2018b). The three measures form a good scale at the macro-level (Cronbach’s alpha > 0,80), and they are combined into a single variable to proxy the family life environment, i.e. the innovativeness of divorce, its costs, and the rigidity of social norms.
Multilevel modelling for binary response variables is used since the outcome variable is dichotomous. The models are estimated separately for men and women. Five models, starting with the unconditional (null) model to the model with cross-level interactions, are fitted.
The first result reveals that high education lowers the odds of divorce among men and women in Europe. This is in line with the theories asserting that high education improves the economic, cultural, and social resources that benefit marital stability.
The second result pertains to cross-national differences in the educational effect. For men, no random slope in high education is found. However, for women, the educational gradient was found to vary significantly across European countries.
In the next step, the macro-level variable is entered into the models. The logit coefficient estimates confirm that the odds of divorce are higher in countries where less conventional family behaviours are more prevalent. With regard to cross-level interactions, the logit coefficient estimates are found to be significant for both men and women, indicating that the effect of high education on the odds of divorce differs across European countries. This result outweighs the non-significant random slope in high education among men (Snijders and Bosker, 1999: 75, 96). In the countries where barriers to divorce are lower, and family practices are less conventional, high education decreases the odds of divorce. In Croatia, where barriers to divorce are still comparatively high, the social context is found to stimulate the relationship between education and divorce to be less negative (more positive).
My findings are in line with the conclusions from prior studies (Härkönen and Dronkers, 2006; Matysiak, Styrc and Vignoli, 2014; Dronkers, 2015). Although the main goal of the current study was to gain a better understanding of cross-country differences, the findings are consistent with trends reported over time in the relationship between education and divorce. Innovation in family practices associated with the second demographic transition is once again confirmed to produce an increasingly negative educational gradient of divorce, and some authors see this as a mechanism in the reproduction of social inequalities (McLanahan, 2004; McLanahan and Percheski, 2008; McLanahan and Jacobsen, 2015). Such a hypothesis calls for more empirical research, with available evidence suggesting that contextual variables might play an important role in this instance too (Bernardi and Boertien, 2017).
Ključne riječi
divorce; education; cross-national comparison; Europe; multi-level analysis
Hrčak ID:
214257
URI
Datum izdavanja:
29.8.2018.
Posjeta: 2.353 *