Association for Public Policy Analysis & Management

More children than ever before are being exposed to pre-kindergarten (pre-k) programs in the U.S. (Barnett et al., 2013). The short-term effects of high-quality pre-K on school readiness are now well-established (Duncan & Magnuson, 2013; Gormley et al., 2005; Weiland & Yoshikawa, 2013).  At the same time, studies have documented both persistence and fade-out for early childhood education programs  (Deming, 2009; Duncan & Magnuson, 2013; Garces, Thomas, & Currie, 2000; Reynolds et al., 2011). It is unclear why fadeout occurs although hypotheses include pre-k students being unsupported by parents or schools, or subsequent contextual challenges. This paper examines two questions regarding fadeout. On the one hand, do subsequent experiences of instability and poverty contribute to fadeout?  On the other, does attending pre-k somehow “inoculate” children to the potential subsequent adverse experiences of school instability or family poverty? School instability and poverty are examined given their demonstrated adverse links with child outcomes (e.g., Evans, 2004) and their policy relevance. 

We employ data from the Tulsa Pre-k Study, a long-term investigation of over 2,000 children who attended kindergarten in the Tulsa Public Schools (TPS) in 2006 and followed through 8^th grade. Approximately half were enrolled in pre-k in the year prior. Data from a parent survey (collected in fall 2006) and school administrative data are employed in analyses. The main dependent variables include test scores, special education status, school attendance, and repeating a grade, as captured by school administrative data in 8^th grade.  Our independent variables of interest were attendance in pre-k in 2005-06, an unexpected change in school between K through 3^rd grade (or elementary school), and chronic poverty (free lunch status from K through 8^thgrade), all as indexed from school administrative data.

Due to missing data common in any long-term study, we employed multiple imputation to create 20 imputed datasets. As a first step, we used multiple regression with covariates (and imputed data) to examine the extent to which school instability and chronic poverty modified the pre-k–academic outcome link.  We use propensity score weighting to balance the pre-k and no pre-k groups.

Descriptive results reveal that approximately one third of participants changed elementary schools at least one time and a little over half of all participants lived in chronic poverty. However, no evidence from multiple regression results suggests a role for school mobility in fadeout. That said, school mobility is consistently associated with lower test scores. However, there may be differential pre-k effects depending on whether or not the child lived in chronic poverty. Those children who lived in chronic poverty but attended pre-k (vs. poverty with no pre-k) were significantly less likely to repeat a grade by 8^th grade (OR = .47, p <.001), but this association was not significant for those who had not lived in chronic poverty and attended pre-k (vs. no pre-k or poverty; OR = 72, ns). Results provide modest support for the inoculation hypothesis. Additional analyses using path analytic techniques will continue to investigate fadeout.