FamilyScape 3.0 microsimulation model

FamilyScape 3.0 is an innovative policy simulation model of pregnancy and family formation. The model, originally developed by the Brookings Institution, is a collaborative effort of the Brookings Institution and Child Trends.

The FamilyScape 3.0 microsimulation tool allows users to model the impacts of behavioral changes (e.g., an increase in the use of long acting reversible contraception) on critical family-formation outcomes. It simulates the key antecedents of pregnancy (sexual activity, contraceptive use, and female fecundity) and many of its most important outcomes (e.g., pregnancy and childbearing within and outside of marriage, children’s chances of being born into poverty, and abortion), and was designed to accurately replicate these dynamics. FamilyScape has been extensively tested and benchmarked to ensure that, in its baseline state, its outputs closely approximate real-world rates of contraceptive failure, pregnancy, childbearing, and abortion. The model can be used to understand the potential impacts of policies and programs, answer “what if” questions, and compare strategies and interventions.  

Child Trends staff have conducted a range of simulations to examine how changes in contraceptive use are linked to reductions in early and unplanned pregnancies, births, and abortions, as well as improvements in maternal and newborn health outcomes. Additional analyses have examined how historical changes in contraceptive use were linked to reductions in teen pregnancy.

Publications

How does FamilyScape work? 

FamilyScape 3.0 is designed to reproduce real-world fertility-related behaviors and outcomes (e.g., pregnancy and childbearing within and outside of marriage, children’s chances of being born into poverty, and abortion) as observed between 2006 and 2010 in the United States. Behaviors and outcomes are simulated at the individual level and are then aggregated to produce population-wide estimates for various outcomes of interest.

The model’s simulation population is nationally representative of women who are of childbearing age in terms of marital status, age, race, educational attainment, and socioeconomic status, and simulated behaviors and outcomes are allowed to vary across these demographic dimensions. Each of FamilyScape’s inputs (sexual activity, contraceptive behavior, etc.) is simulated so as to ensure that aggregate measures of the resulting behaviors are consistent with real-world benchmarks. The model is then validated by comparing its outputs (rates of pregnancy among women who rely on various types of contraception, the incidence of childbearing among teens and adults, abortion rates within and outside of marriage, etc.) to their equivalent real-world benchmarks. FamilyScape’s baseline outputs meet real-world targets, meaning the model can be used to perform simulations in which researchers change the key inputs and model the subsequent effects on the outputs. However, simulations are most accurate for broadly defined groups of women for whom the model can more accurately predict behaviors and outcomes. Additionally, simulated outputs, such as pregnancy rates, are estimates based on available information; they are not intended to substitute for actual pregnancy data.

The figure below delineates the various stages of the simulation. First, the model is populated (using data from the National Survey of Family Growth) with a nationally representative group of women who are assigned a set of behavioral attributes as a function of their demographic characteristics. Next, sexual activity (or a lack thereof) is simulated, and contraceptive use (or a lack thereof) is modeled among women who have sex. In the next stage, some sexually active women become pregnant as a function of their birth control method choice and underlying fertility.  Each pregnancy eventually results in a birth, an abortion, or a fetal loss (often called a “miscarriage”) and after the appropriate post-pregnancy infertility period, a woman is again eligible for pregnancy and the cycle begins again. For pregnancies that result in a live birth, the model assigns a poverty status to each newborn child using Census data. Additionally, child and maternal health outcomes are assigned during this final stage using Vital Statistics data. Specifically, maternal hypertension, maternal smoking, maternal diabetes, infant birth weight, when prenatal care was initiated and the number of visits, and gestation length are determined based on the characteristics of the mother.

For more information on FamilyScape 3.0 please see the technical guide, available at: http://www.brookings.edu/~/media/research/files/papers/2015/05/familyscape3/familyscape-thomas-karpilow.pdf

Newsletters