Our Insights

Class Size in Preschool

New findings add to what we know

Class size reduction policy has been long debated in early education as well as the K-12 sphere. This is not surprising given the ambiguity in defining class size and the mixed findings across several research studies. The only true experiment for preschool class size found that children assigned to smaller classes had larger gains on measures of general knowledge, language, and social behavior (Ruopp, Travers, Glantz, & Coelen, 1979). Yet, subsequent research on class size (and ratio) produced inconsistent findings with many studies finding no significant associations between class size and child outcomes (e.g., Burchinal, Zaslow, & Tarullo, 2016; Howes et al., 2008; Mashburn, Pianta, Hamre, Downer, Barbarin, Bryant, 2008. Meta-analyses differ in their conclusions Bowne, Magnuson, Schindler, Duncan, & yoshikawa, 2017; Perlman, M., Fletcher, B., Falenchuk, O., Brunsek, A., McMullen, E., & Shah, P.S., 2017).

The Tennessee Star experiment has received the most attention in the K-12 literature, as an experiment in which children (grades K-3) were randomly assigned to class sizes of 13-17 or 22-25. They found positive effects on reading and math in every grade for small classes beginning in kindergarten (Finn & Achilles, 1999). Since these children are just slightly older than children in preschool, it raises questions about the inconsistent findings in the preschool literature.. Of course, there are many reasons class size reduction can fail to meet its promise when implemented at scale, and not every class size reduction initiative has been successful in K-12  (Blatchford, Chan, Galton, Lai, & Lee, 2016; Schwanzenbach, 2014).

A study we recently published in Early Childhood Research Quarterly adds some interesting information regarding the effects of class size. This study implemented a unique design, in which 22 preschool teachers in the city of Chicago each taught one morning and afternoon session with different class sizes. Sessions were randomly assigned to be either 20 children (regular) or 15 children (reduced) with one lead and one assistant teacher in each classroom. All classroom sessions were publicly funded, staffed by a certified lead teacher with BA, and were 2.5 hours in length, 4 days per week.

The 354 children who participated in the study were administered assessments in early math (WJ-III Applied Problems), literacy (TOPEL), and language (PPVT-III) at the start and end of the school year. Teachers were observed with two measures of classroom quality (CLASS and Snapshot). Mothers of children participating in the study were administered brief questionnaires related to their demographic characteristics (education, income, home language, race).

We found significant literacy gains from smaller class size but no significant advantages of reduced class size for language of mathematics. We also, found no differences in classroom quality scores, though there was evidence that smaller classes led to more one-on-one interactions. Unfortunately, we had no measures on other important domains including executive functions and socio-emotional development. What should we conclude from these modest findings? First, they are one more indication that class size reduction alone can improve outcomes. Implementing different class sizes within each teacher is a strength of this study because it allowed us to control for a host of variables that might have otherwise differed. However, one limitation of this approach is that teachers may have been less likely to adapt their teaching practices to fully take advantage of smaller class size because they continued to teach another session with the larger class size. Further, no professional development was provided to help teachers modify their teaching to maximize gains from class size reduction (Graue, Hatch Rao, & Oen, 2007). Another consideration is that instructional quality was fairly low; more intensive and frequent interactions with a teacher may not be particularly helpful if they are not well-planned, intentional, and rich.

Broadly speaking, in the context of the broader literature, our findings suggest that class size indeed matters in preschool. Some may be tempted to focus on the tradeoffs among program features including class size and such supports as professional development. However, an alternative is to consider what sort of program supports are required to provide classroom experiences that will produce the gains in learning and development programs wish to see for their children. This may require a host of improvements implemented simultaneously, rather than just choosing from among them.


The National Institute for Early Education Research (NIEER) at the Graduate School of Education, Rutgers University, New Brunswick, NJ, conducts and disseminates independent research and analysis to inform early childhood education policy.