The Multisite Findings of the Impact of Assisted Housing on Educational Well-Being

• In Event: 25.033 - A Multisite Study of Assisted Housing Programs and Educational Well-Being Using Integrated Data Systems

Abstract

The purpose of this multi-site study was to examine the educational outcomes of children participating in assisted housing programs relative to their peers not living in assisted housing. This national policy issue was explored by six sites in the Actionable Intelligence for Social Policy (AISP) network. Each of the study sites integrated, at minimum, individual-level student education records and assisted housing records from the U.S. Department of Housing and Urban Development (HUD). This study employed a quasi-experimental design to capitalize on existing administrating records at the federal, state, and local levels. Each site examined two cohorts of public school students—one third-grade and one seventh-grade group. Sites followed these two cohorts from 2005 to 2012 which allowed for the examination of education records through elementary, middle, and high school.

All sites were able to employ a similar base research design. The base research design (one the mirrors the methodology used in much of the previous research literature) defined treatment and control groups as such: children were categorized as “assisted housing” if they participated in an assisted housing program (low-income public housing, project-based housing, low-income housing tax credit program, and tenant-based vouchers) at any point before the study period began and for which HUD records existed (2000-2005). Four of the six sites were able to produce advanced research designs that improved their ability to isolate the effect of assisted housing programs – these approaches included selecting only new participants in assisted housing programs and matching on either “pretests” for all education outcomes (e.g., achievement, attendance, suspensions) or the rich social service data available in IDSs (e.g., other forms of public assistance, child welfare involvement, prior homelessness). Control group children were selected using propensity score matching so that they closely resembled children in assisted housing on demographic and other relevant characteristics from education records.

All sites then utilized panel regression approaches to compare children in assisted housing programs to their matched peers on outcomes including academic achievement (reading and mathematics), attendance, and on-time graduation. Findings from the base research design across sites showed a common trend of small negative or no effects of assisted housing programs on educational outcomes. This is consistent with limited number of previous studies examining housing impacts on education outcomes. However, findings from the advanced research designs that were better able to isolate the effects of assisted housing found that most negative effects disappeared and, in some cases, small positive effects were found. Without accounting for baseline education variables or other determinants of assisted housing involvement identified in IDSs, these sites would have mistakenly overestimated negative effects and underestimated positive effects of assisted housing participation on educational well-being. These findings demonstrate the value of IDSs for understanding the intersection of housing and education programs and policies, and offer potential for time and cost-efficient evaluations of other social policy areas as they related to children’s education well-being.

Authors

• Whitney A. LeBoeuf, University of Colorado, Boulder

• Katie Barghaus, University of Pennsylvania

• Cassandra Henderson, University of Pennsylvania

• Dennis P. Culhane, University of Pennsylvania