Objective: An age-period-cohort (APC) analysis was performed to provide information about age-, period-, and cohort-specific effects on marriage trends in Japanese women. In addition, the relationships of the trends of age-, period-, and cohort-specific effects between marriage and birth were analyzed.
Methods: We obtained data regarding marriages of Japanese women aged between 19 and 38 years for the period of 1985 to 2005 from the National Vital Statistics. Population data used were for an estimated population, obtained from the Population Estimates Annual Reports. Standard cohort tables comprising marriage and population data were analyzed using a Bayesian APC model to identify age-, period-, and cohort-specific effects on marriage rate trends. Previously obtained data for a similar APC-analysis of birth trends were used to compare the trends in the effects of age, period, and cohort on marriage and birth patterns. For this purpose, the estimated values for each effect were normalized.
Results: With regard to the marriage trends in Japanese women, the effect of age was the greatest, peaking at the age of 25 years. The period effect increased after 1997; however, its effect was relatively limited as compared to the other effects. The cohort effect, which was greater than the period effect and less than the age effect, on marriage trends showed a decreasing slope for birth cohorts born after 1966 and subsequent increase after 1982. Comparison of age, period and cohort effects between the trends in marriage and birth rates showed that the age effect distinctly peaked at 25 and 28 years for marriage and births, respectively. The period effect on marriage and birth showed a decreasing trend until 1991 and subsequent increased in 1992 and 1997 for births and marriage, respectively. With regard to the cohort effect on birth rates, a decreasing trend was observed for the birth cohorts after 1961, with increase after 1977. However, with regard to the cohort effect on marriage rates, the decreasing trend observed for birth cohorts after 1966 showed an increase after 1982.
Conclusion: Among age, period, and birth cohort, age is the most influential factor affecting marriage rates. Period effects appear relatively small, but they increased after 1997. Cohort effects reduced for birth cohorts born after 1966 and subsequently increased after 1982. Results of the comparison study showed that changing patterns of age, period and cohort effects had very similar influences on the trends for marriage and birth rates. However, a 3-year difference was observed between the peaks of the age effect on the two rates. A time lag of 5 years was observed between the turning point in the trend of period effects for marriage and birth rates. The changing patterns of cohort effects on marriage and birth rates were similar, but the turning point for the marriage pattern occurred in a 5-year younger cohort compared with the birth pattern.