ABSTRACT
The authors analyzed the Early Childhood Longitudinal Study–Kindergarten Cohort (ECLS-K) national data set to investigate gender differences in ability group placement in American kindergartens. They found that in kindergarten, within-class ability grouping was widely used for reading instruction, with boys being underrepresented in high-achieving reading groups and overrepresented in low-achieving ones. Gender differences in reading group placement were consistent across classrooms and were explained by student-level characteristics. Boys’ underrepresentation in high reading groups was explained by their lower reading skills at kindergarten entry, as measured by the reading test scores available in the ECLS-K. By contrast, boys’ overrepresentation in low reading groups was only partially explained by their lower test scores. Compared with girls of similar social background and reading test scores, boys continued to have higher chances of placement into a low reading group. This remaining gender difference was explained by the lower teacher evaluations of boys’ reading skills and approaches to learning. Boys’ disadvantages in reading group placement at school entry raise concern over their further academic success.
ACKNOWLEDGMENTS
The authors dedicate this article to the memory of their colleague and friend Dr. Lynn Mulkey, who passed away in February 2010. They honor her scholarly contributions, which have been integral to their research on ability grouping in the early grades. Support for this research was provided by the National Institutes of Health, National Institute of Child Health and Development branch (Grant #5R01HD45614). The project was developed while Sophia Catsambis served as the American Educational Research Association (AERA) Jean Griffith Fellow at the National Center for Education Statistics, Washington, DC. The project benefited from the comments of Jerry West, Project Director of the ECLS-K, and the staffs of NCES and AERA.
Notes
The subsample had a slightly lower mean socioeconomic status than the entire ECLS-K sample. The proportion of White students was lower by 6%, whereas the proportion of African American and Hispanic groups was higher by 3% each when compared to the total sample.
In preliminary analyses we considered alternative models with classroom-level predictors, which produced very similar results. For this reason we present here the simpler, more parsimonious models with Level 1 predictors only.
The multinomial logit link function was ηmij = log (Prob(m)/Prob(M)), where m is equal to 1 or 2 for placement into the low or high group and m is equal to 3 for placement into the average group (reference category).