Most studies of school achievement use free-lunch eligibility or other basic indicators to adjust for differences in students’ socioeconomic backgrounds. This study determines whether these variables are enough to separate the confounding effects of students’ backgrounds from the main variables of interest in education studies. The Early Childhood Longitudinal Study dataset from the kindergarten class cohort of 1998-99 (ECLS-K) provides an unusually vast array of information regarding children’s home resources and experiences. This plethora of parent-reported data raises questions about which variables researchers should include in their analyses, and it provides an extraordinary opportunity to examine this question. Using a split-sample design, stepwise regression, and multi-level modeling, this study systematically examines over 200 ECLS-K student background variables to determine which factors predict reading and mathematics achievement after typical SES controls are employed. The study identifies several variables that are important supplements to traditional SES measures, including the number of children in the household, mother’s age at first birth, and children’s books at home. Results indicate the extent to which “value added” studies can be flawed when using only basic demographic variables. The findings hold implications for data collection and accountability efforts, including NCLB, teacher evaluation plans, and the design of state longitudinal data systems.
social class; socioeconomic status; Early Childhood Longitudinal Study (ECLS-K); large- scale data; multi-level modeling; mathematics; reading; elementary school; United States, 1999–2005.