Association for Public Policy Analysis & Management

The early elementary grades (grades K-2) are critically important to children’s academic development. Contemporary research suggests that students learn at the highest rate prior to 3^rd grade (Cameron et al., 2014) and that their skill levels in kindergarten are predictive of achievement in later years (Duncan et al., 2007).

Given the importance of early skills, it is concerning that non-white and low-income students continue to enter kindergarten with lower skills than their peers (Reardon & Portilla, 2016). Much of the work documenting these broad trends in student achievement between kindergarten and 3^rd grade has relied on nationally representative data from the two cohorts of kindergarteners followed by the Early Childhood Longitudinal Study (ECLS-K). While these data allow researchers to document national trends, the sampling method employed by ECLS-K yields only 23 student observations at each participating school on average, inhibiting robust within-school analyses. Further, the early cognitive assessment data provided by ECLS-K are not linked to policy-relevant measures such as high-stakes standardized tests at the student level. Lastly, children in the most recent cohort of the ECLS-K started kindergarten nearly a decade ago (2010).

We created a unique statewide dataset to assess trends in student literacy development during the early elementary grades. All public school kindergarteners in all but one school division in Virginia are assessed using the Phonological Awareness Literacy Screener (PALS). We link data from the PALS assessment to student’s results on the high-stakes 3^rd grade Standards of Learning (SOL) reading assessment for two recent cohorts of kindergarteners in Virginia. This paper marks the first that we know of to link statewide data from a school readiness assessment to a high-stakes, policy relevant outcome in 3^rd grade. We leverage this unique data to answer two research questions:

To what extent does the relationship between students’ literacy skills at kindergarten entry and their outcomes on a high-stakes 3^rd grade exam vary by subgroup? Are these relationships explained by sorting of students into different schools?

Consistent with prior literature, our findings indicate that non-white and economically disadvantaged students enter kindergarten with lower literacy skills than their white and economically advantaged counterparts. Strikingly, we also find that the link between students’ early literacy skills and their reading outcomes on the high-stakes 3^rd grade exam varies significantly across subgroups, with white and high-income students far more likely to recover from literacy struggles in kindergarten than their non-white and disadvantaged counterparts. For example, white students scoring in the lowest quintile on the kindergarten PALS exam are more than 20 percentage points more likely to recover to reach proficiency standards by third grade than Black students scoring in the same quintile. Notably, these discrepancies do not appear to be explained by student sorting into different schools. Our findings that non-white and economically disadvantaged are substantially less likely to be proficient in reading by 3^rd grade than their white counterparts who enter kindergarten with comparable skills provide policymakers with valuable data that can be used to inform an equity-based approach to early literacy.