Tennessee, North Carolina, and Wisconsin have found that reducing class size to 15 in the early grades produced substantial long-term gains in student achievement (Achilles et al., 1995). The largest of these studies found that reducing class size from 22–26 to 13–17 children in kindergarten through grade 3 increased student achievement in reading, math, and science through middle school, decreased grade repetition, and increased high school graduation rates (Boyd-Zaharias and Pate-Bain, 2000; Krueger, 1999; Finn et al., 1999).
The distinction between ratio and class size is important, although it is difficult to disentangle class size and ratio in most studies. Undoubtedly, both are important, but the studies that are able to disentangle them (e.g., Mosteller, 1995; Boyd-Zaharias and Pate-Bain, 2000) indicate that improving the ratio without reducing class size does not yield the same effects. Randomized trials in kindergarten through grade 3 compared the effects of adding an aide to the classroom to reductions in class size that produced the same adult-child ratio and found that the added aide did not produce the substantial, persistent gains in achievement obtained from reducing class size (Mosteller, 1995). Thus, a class size of 22 with a teacher and two aides is not an adequate substitute for a class size of 15 with a teacher and one aide.
The existing research is not sufficient to suggest the optimal class size for children at each age. However, it does indicate that smaller class sizes and better ratios than are now commonly required would benefit children, especially children from low-income families. Ratios in the experimental literature rarely exceeded one teacher for every 7 children, which is better than prevailing practices in many education and care programs today (U.S. General Accounting Office, 1995). Only one state requires child care centers to maintain a ratio of one teacher for 7 or 8, 3-and 4-year old children, and regulations in other states range from 10 to 20 children per teacher. Class sizes in early care and education rarely are as small as 15 and frequently exceed 20 (Gormley, 1995). Moreover, unless they are in a Head Start program (where standards are relatively high), low-SES children, who would benefit most from small class sizes, tend to have the largest class sizes in early care and education as well as in kindergarten programs.