A longitudinal study of the relationships between township-level characteristics and teen birth rates in Taiwan

The influence of a place on adolescent health behaviors has received increasing research attention. The argument has been that there are contextual processes operating at the scale of communities or other geographical areas, which are responsible for health inequality. However, few studies have examined the contextual effects on health in a longitudinal data set. This study used township-level data and a pooled ecological design with repeated measures to examine the relationships between contextual characteristics and birth rates to teens aged 15-19. Data were extracted from Taiwan census data and Health and Vital Statistics in the year of 1980, 1985, 1990, 1995, 2000, and 2005 (N=2154). Contextual characteristics included the percentage of residents with a college degree, the percentage of divorced/separated, the percentage of residents older than 65, the percentage of aborigines, residential mobility, and urbanization. We used two-level mixed models in which years are nested within towns and cities. Contemporary and time-lagged covariates were modeled. In the contemporary model, a lower percentage of residents with a college degree, a higher percentage of residents older than 65, a higher percentage of aborigines, and a lower level of urbanization were associated with increased teen birth rates. Results of the 5-year and 10-year time-lagged models both showed that a lower percentage of divorced/separated, a lower percentage of residents with a college degree, a higher percentage of aborigines, and a lower level of urbanization were associated with increased teen birth rates, suggesting that minority composition, community education level, urbanization, and local community norms toward marriage and family determined teen birth rates.