Desktop version

Home arrow Sociology arrow Diverging Destinies: The Japanese Case



Like cohabitation, the information on divorce available in the NFS is not ideal. Because exposure to the risk of divorce commences at marriage, analyses of divorce require information on both the date of marriage and the date of divorce (if applicable). However, prior to the 13th round, the only information collected on

Fig. 3.5 Predicted probability of divorce experience, by education and birth cohort

divorce in the NFS was whether or not a woman had ever experienced divorce.[1] Our analyses of longer term trends are therefore limited to whether or not women reported ever experiencing divorce.

Figure 3.5 shows results from logistic regression models of divorce experience as a function of the covariates described above. Here, the analytical sample is restricted to women who had ever married and the results presented are the predicted probability that a 30-34-year-old women reported that she had divorced. The most notable feature of the figure is the rapid rise in divorce among the least-educated women, consistent with predictions of diverging destinies for this group. Very high levels of divorce among women who did not finish high school have been documented by Raymo et al. (2013), among others. The level of divorce in Fig. 3.5 is very low (probabilities are below 0.05 for most combinations of educational attainment and birth cohort), reflecting the fact that the predicted probabilities are for 30-34-year-old women, many of whom have only been at the risk of divorce for a short period of time. It is also possible that low levels of divorce reflect underreporting of divorce experience in the NFS, as noted by Raymo (2008) and Raymo et al. (2013). There is some evidence of growing differences between high school graduates and more highly educated women in recent birth cohorts. This is particularly true of the difference between high school graduates and university graduates born in the 1970s (the solid black circle for this group indicates that the difference between university and high school graduates is significantly larger than in both the first (1940s) cohort and the most recent (1960s) cohort). While this evidence of family bifurcation is much less pronounced than that documented in the U.S. (Martin 2006), it is nonetheless suggestive of growing differences in marital stability at either end of the educational distribution.

The results in Fig. 3.5 describe trends in a limited measure of divorce experience over a long period of time, but are not ideal in that they also reflect changes in marriage timing. The trend toward later age at marriage means that 30-34-year-old

Fig. 3.6 Predicted probability of divorce within 10 years of marriage, by education and marriage cohort

women in more recent birth cohorts will, on average, have experienced shorter durations of exposure to the risk of divorce than their counterparts in earlier birth cohorts. We address this problem by using information on the timing of first marriage and divorce that is available in the 13th and 14th rounds of the NFS. Using these data, we describe educational differences in the probability of experiencing divorce within 10 years of marriage for two marriage cohorts—those marrying in the 1980s and those marrying in the 1990s. These analyses are limited to respondents whose first marriage occurred at least 10 years prior to the survey date. The results, presented in Fig. 3.6, indicate that the probability of divorce has increased somewhat for women in all educational groups and that women with a high school or vocational school education have higher divorce rates than junior college and university graduates. Again, the level of divorce and the pace of increase are most pronounced among women who did not complete high school. There is no statistical evidence, however, that differences across groups have grown over time. This pattern of findings is generally consistent with that presented by Raymo et al. (2013) who used these same data to examine trends in educational differences in divorce.

  • [1] Questionnaires for the 8th and 14th round of NFS did not distinguish divorce from widowhood,but we do not view this as particularly problematic given the very low levels of mortality prior toage 50 (the age of the oldest women in the NFS).
Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >

Related topics