The impact of small class size on children's learning has received increased attention in recent years as evidence from experimental studies (Tennessee STAR experiment) has shown significant benefits of smaller class size for elementary school children. Despite the appeal of experimental studies, their scope remains limited in terms of their generalizeability, ability to examine a wider range of outcomes and explore mechanisms through which class size may affect child outcomes. In this study, we propose to examine how class size is related to children's learning behavior, teaching practices, parental involvement and children's cognitive and social-behavioral outcomes. We will use panel data from the Early Childhood Longitudinal Study-Kindergarten Class (ECLS-K), a nationally representative sample of kindergartners in the U.S., which collected detailed information on children's learning behavior and school outcomes, parental involvement, teaching practices and class size at multiple time points (kindergarten, first grade, and third grade). We will focus on both class size and staff-to-child ratio because teachers in elementary school often use aides in class to assist them, which is likely to impact the amount of individual attention that children receive in class. We will estimate child fixed-effect models that will identify class-size effects using over time variation in class size for the same child. This approach will eliminate any bias generated from time invariant individual and family unobservables that are related to class size and the outcomes of interest. The findings of this study will be important because they will help to understand the mechanisms through which smaller or larger class sizes may affect children's learning and will guide policy making in this area. ? ? ?