Marital Quality, Selection, and Causation in the Divorce–Mental Health Link in Japan

• In Event: Family Section Roundtables
  In Refereed Roundtable (60 minute): Table 7: Families and Health

Abstract

A large body of research has established a link between divorce and mental health, but the underlying mechanisms remain debated. Two competing explanations have dominated the literature. Social selection holds that people with pre-existing mental health problems are more likely to divorce. Social causation holds that divorce itself leads to a decline in mental health. Most existing evidence comes from Western contexts, and both processes have rarely been examined simultaneously using longitudinal data outside of Europe and North America. Moreover, both explanations share a common gap: the lack of data on marital quality. Several influential studies have acknowledged that the absence of marital satisfaction data was a key limitation of their work. Using 15 years of Japanese panel data (N = 58,520 person-wave observations; 226 divorce events), this study tests both processes while incorporating marital satisfaction as a key conditioning variable. This study advances two hypotheses: (H1) the association between poor mental health and later divorce is explained by marital satisfaction; and (H2) the effect of divorce on mental health is moderated by prior marital satisfaction.The results support both hypotheses. Regarding H1, the mental health coefficient predicting divorce drops to near zero and becomes non-significant once marital satisfaction is controlled, suggesting that psychological distress leads to divorce by reducing relationship quality rather than through an independent pathway. Regarding H2, divorce is associated with elevated odds of psychological distress onset only among those who were highly satisfied with their marriage before dissolution (OR = 3.732, p < 0.01), while the association is not significant among those who were already dissatisfied.These findings indicate that marital quality is an important missing variable in both selection and causation analyses, and that these mechanisms may generalize beyond Western contexts.

Author

• Yusuke Tsukada, Aoyama Gakuin University