Association for Public Policy Analysis & Management

The short-term effects of high-quality pre-K on school readiness are now well-established (Duncan & Magnuson, 2013; Gormley, Gayer, Phillips, & Dawson, 2005; Weiland & Yoshikawa, 2013). At the same time, questions have been raised as to whether these effects persist over time and whether the quality of the pre-k experience can enhance long-term outcomes. Reputable studies have documented both persistence and fade-out for early childhood education programs (Deming, 2009; Duncan & Magnuson, 2013; Garces, Thomas, & Currie, 2000; Reynolds, Temple, Ou, Arteaga, & White, 2011). In addition, stimulating interactions and high instructional quality in pre-k have demonstrated associations with diverse child outcomes in kindergarten (Burchinal et al., 2008). Employing a similar logic, we ask whether pre-k quality is important for the persistence of pre-k effects. Put another way, do children who experienced high quality pre-k, as exemplified by sensitive interactions with teachers and high quality instruction, benefit more and more long-term than children who did not experience pre-k or experienced lower quality pre-k?

To answer our question, we employed data from the Tulsa pre-k project, a longitudinal investigation of about 2000 children who started kindergarten in 2006. About half of these students attended pre-k the year prior. We employed school administrative (from pre-k, K, third, and eighth grades) and a parent survey, collected in fall 2006, for outcome, predictor, and covariate variables. The CLASS (Pianta et al., 2008) also was conducted in the winter of 2005-06 (pre-k year) to assess the quality of the pre-k programs that children experienced.

As a first step, a latent profile analysis was conducted to determine the common characteristics of pre-k quality that children experienced (McCutcheon, 1987). We found evidence of three classes, generally defined as high-quality (13%), moderate-quality (56%), and low-quality (31%). Next, we predicted the long-term (8th grade) outcomes of children who experienced the three different characteristics of care, all as compared with no pre-k, using multiple regression analyses. We also included child, family, and neighborhood characteristics as covariates.

Results suggest differential long-term effects depending on the quality of the pre-k program the children attended. Surprisingly, students in moderate quality pre-k appeared to benefit most from their pre-k experience. Specifically, they demonstrated higher reading scores and GPA than students who were not in pre-k. In addition, they were less likely to be designated as special education, were less likely to repeat a grade, and were more likely to be in the gifted and talented program, all as compared with children not in pre-k. There was less evidence that the highest quality pre-k was associated with benefits beyond those of moderate quality pre-k, but the predominance of moderate quality may have contributed to this result. Future analyses will incorporate multiple imputation to generate 20 imputed datasets, to replicate initial regression results. We also will employ propensity score weighting to balance pre-k and non pre-k groups on observable characteristics.