In the field of teen pregnancy, recent progress has been made in identifying and funding evidence-based prevention programs. Less is known, however, about evidence-based policies in this area.

Despite declines in the teen birth rate in all states, substantial variation in the rate of decline exists across states. This paper assesses the effects of state-specific policies on declines in each state from 1990 to 2007.  We adjust for between-state differences in the teen birth rate, controlling for differences in state population characteristics, such as the percent of people in poverty. In addition, given that policies change over time, an analysis of how changes in policies correspond to changes in the teen birth rate is needed.

We use intensive longitudinal data techniques to capture both the time-varying and time-invariant relationship between state-level characteristics (predictors) and the teen birth rate. Specifically, the study uses Time-Varying Effects Models to examine the effects of health- and education-related policies on changes in the teen birth rate from 1990 to 2007.  The data were lagged by one year (e.g., state level data from 1989 were used to predict 1990 teen birth rate, and so on).  Instead of being constrained to linear assumptions, Time-Varying Effects Models (TVEMs) allow coefficients of time-varying variables to vary across time to capture the changing relationship between predictor variables and outcome over time.  We incorporate TVEM techniques with Multi-Level Modeling to account for time nested within state data. 

Time-varying variables used in the models include demographic, economic, education, health and social policy factors, including the percent black, the male-to-female ratio, percent of freshman who graduate, educational expenditures per pupil, family planning expenditures per woman 15-44 in poverty, percent of population in poverty, percent of births to unmarried women, percent of children aged 6-18 currently insured, violent crime rates, and prevalence of abortion providers.  Time-invariant variables modeled as fixed effects include the state percent fundamentalist, female labor force participation, unemployment rate, and welfare benefits.

Preliminary results limited to time-varying predictors reveal the following selected results:  As the average freshman graduate rate increases, the teen birth rate decreases.  As state expenditures per pupil increase (from 1989-2006), the teen birth rate decreases (1990-2007).  Similarly, as family planning service dollars per woman in poverty increase from 1992-2003, the teen birth rate decreases (1993-2004).  A positive relationship between the percent of insured children ages 6-18 and the teen birth rate exists across all years, though the strength of the relationship changes over time.  From 1995-2007, as the proportion of abortion providers decreases over time, the teen birth rate increases.  As the percent of all births born to unmarried females aged 15-44 increases, its positive relationship with the teen birth rate also increases. This relationship is strongest from 1995 through 2007.    

Next, fixed effect predictors, such as religious fundamentalism, will be added to the model.  Results will be discussed in the context of research on individual- and program-level predictors of teen pregnancy.