Population Research and Policy Review 21: 505–533, 2002.
© 2003 Kluwer Academic Publishers. Printed in the Netherlands.
Good marriages gone bad: Health mismatches as a cause of
later-life marital dissolution
SVEN E. WILSON & SHAWN L. WADDOUPS
Brigham Young University
Abstract. This study explores the impact of health status on marital dissolution for couples
in late mid-life. A key feature of the empirical framework is that it incorporates the interaction
of health between the spouses. This speciﬁcation allows not only a general test of whether
health matters but also a speciﬁc test of an important implication of cost-beneﬁt models
of marriage dissolution. In particular, cost-beneﬁt models imply that marriages exhibiting a
“health mismatch” (where one partner has substantially better health than the other one) are
more likely to get divorced than couples who have similar health (whether good or bad).
Using a Cox proportional hazards model, we test this hypothesis by estimating the impact of
different spousal health combinations on the probability of marital dissolution (as indicated
by separation). Data are taken from four waves of the Health and Retirement Study (1992–
1998) and consist of 4,241 couples where at least one spouse is between the ages of 51 and
61 in 1992. We do ﬁnd evidence for the health mismatch hypothesis, but only among couples
in which both couples report their marriages to be very satisfying. Among other couples, no
effect is found. This suggests that health is of minor consequence for already unhappy couples,
but health mismatches pose a signiﬁcant risk of dissolution to happy couples within this age
cohort, possibly because of the unexpected nature of poor health at a relatively young age.
Keywords: health, marital dissolution, rational models
A robust empirical regularity in demographic research is the strong asso-
ciation between health and marital status. The consistent ﬁnding of studies
addressing the topic has been that married people enjoy higher levels of well-
being, including both physical and mental health (see Joung et al. 1998 for
a recent review).
As an illustration, consider Table 1 below, which doc-
uments the correlation between health and marital status using the 1992
Health and Retirement Study (HRS). This table shows how the prevalence
of “fair” or “poor” health varies across sample respondents for both men and
women. Only the never married women have health similar to their married
counterparts. Particularly striking are the high values for separated and wid-
owed, with rates of poor health two to three times as high as their married