Association for Public Policy Analysis & Management

In recent years, significant policy changes have taken place in publicly funded early education, including increases in elementary charter schools (Keaton, 2012). These educational initiatives aim to reduce inequities in access to high-quality early learning environments as levers for diminishing achievement disparities (U.S. Department of Education, 2007). However, national trends on children’s academic performance in charter versus traditional public schools remains understudied, particularly whether charter school enrollment mitigates or exacerbates early achievement gaps.

The current study investigated whether charter school enrollment significantly moderated income disparities in early math achievement (K - 2^ndgrade) after accounting for parents’ propensities to select charter versus traditional public schools. Data were drawn from the ECLS-K:2011 and included 840 children enrolled in schools districts offering both traditional public and charter schools. Math and reading skills were measured in the fall and spring of kindergarten and spring of first and second grades through direct assessments developed by NCES. Household income was reported in the spring of kindergarten, first grade, and second grade by parents, and three longitudinal categories were created: never low-income, always low-income, or ever low-income.

Missing data were imputed in Stata 14 using the mi impute chained command (m = 10; StataCorp, 2013). Propensity scores were generated to account for the non-random assortment of children into charter schools by regressing school type (charter or regular) on a large set of demographic background characteristics (Hong & Yu, 2008; Lunceford & Davidian, 2004; Rosenbaum & Rubin, 1983). Two weighting specifications were compared: child population weights (W1C0) and the inverse of each child’s propensity score based on their school type (e.g., Miller & Votruba-Drzal, 2013). Three-level growth curve models (i.e., repeated observations within individuals within schools) were estimated predicting achievement over time.

Math and reading achievement increased each month and slowed over time, but neither initial skills, linear growth, nor quadratic growth differed by school type. A significant interaction on the linear slope between family income and school type emerged for math, indicating that always low-income children showed more growth in math skills when enrolled in charter schools than in regular public schools. This differential growth compounded over time; by the end of second grade, always low-income children scored 0.34 SD higher in math when attending charter schools, whereas differences were non-significant or in the opposite direction among children who were never low-income or ever low-income. These results suggest a causal link between charter school enrollment and enhanced math growth for chronically poor children. Examination of numerous structural (e.g., school size, teacher experience) and process (e.g., use of small groups for math, minutes of math instruction per week) characteristics of schools has not yet revealed significant mediators that explain this effect.