Using Data from “Moving to Opportunity” to Investigate Early Influences on Educational Outcomes

The goal of this pilot was to analyze how exposures in early childhood affect educational and labor market outcomes. We compared experimental estimates using data from the Moving to Opportunity demonstration program (MTO) with best-practice nonexperimental estimates generated both within MTO and with data from the Panel Study of Income Dynamics (PSID). Our MTO experimental estimate of the effects of lower neighborhood poverty on reading scores was close to zero and statistically non-significant. Our estimate using the PSID data suggests a small positive association between reading scores and lower poverty that is marginally significant. Unfortunately, our estimates are too imprecisely estimated to distinguish between the experimental and nonexperimental estimates. In the second part of our analysis, we trim and reweight the MTO and PSID samples (using a machine learning algorithm) to make them more comparable to each other on basic demographics such as age, race, parental work status, and public assistance. After reweighting, both the MTO and PSID estimates are somewhat larger. The preliminary work supported by the pilot award helped us obtain a two-year follow-up grant for a project titled “Concentrated Poverty and Social Mobility.”