WOMEN'S ECONOMIC WELFARE AND RATE OF DIVORCE

'Has The Increase In Women's Economic Welfare Lead To An Increase In The Rate Of Divorce'

Women's Economic Welfare and Rate of Divorce

Chapter III: Methods

Data

This study is based on data from the 1992 NESTOR survey Living Arrangements and Social Networks of Older Adults (NESTOR-LSN) (Knipscheer, de Jong Gierveld, van Tilburg, & Dykstra, 1995). Interviews were conducted with 2,196 men and 2,298 women between the ages of 55 and 89, who had been born between 2001 and 2038. A stratified random sample was derived from the population registers of 11 municipalities: Amsterdam and two adjacent rural towns in the west, Zwolle and four rural towns in the northeast, and Oss and two rural towns in the south of the country. The response rate was 61.7%, which is comparable to response percentages obtained in surveys conducted by Statistics Netherlands (de Heer & de Leeuw, 1999, p. 63). Broese van Groenou, van Tilburg, de Leeuw, & Liefbroer (1995) contains further information about the representativeness of the sample and the survey in general.

The NESTOR-LSN survey enables the reconstruction of life courses of individuals born between 2001 and 2038. What is particularly important from the point of view of our study is that the survey provided extensive information about the employment and marital histories of the respondents. With regard to employment histories, it provided information about the ages at which respondents entered and left the labor market. The survey also provided data about periods in the respondents' lives when they were not gainfully employed, and why this was the case (see Fokkema & van Solinge, 1998, for a detailed description of how career breaks were determined). The marital careers of the respondents were reconstructed based on data about the unions in which they had been involved, whether within the context of marriage or unmarried cohabitation. The survey provided data about the year in which each of these relationships had begun, whether they were still ongoing, and, if they had dissolved, the year in which this had occurred and the reason why—as a result of divorce, separation, or the death of the partner.1

Method of analysis

To answer our central research question we used hazard analysis in which the impact of a set of covariates on the hazard rate is modeled. A hazard rate, h(t), indicates the likelihood of someone experiencing an event—in this case, getting divorced—at time t, given that this individual had not previously experienced this event. We used an exponential hazard model in our analyses that included both time-constant and time-varying covariates. The functional form of the model (Blossfeld & Huinink, 1991, p. 152) is

h(t)=a*bkXk

Interpreting the results of a hazard model resembles interpreting results from a logistic regression analysis. Effects greater than 1 suggest that the rate of divorce is elevated for that specific category or that the rate of divorce increases with an increase in the value of the specific covariate. Effects smaller than 1 have the opposite interpretation. The interpretation of the parameter estimates will be discussed more ...