Content area
Abstract
Researchers have long sought to identify manipulable factors distinct from student socioeconomic status and race-ethnicity that might increase overall student achievement or reduce differences in achievement between white and non-white students who attend the same school. Numerous quantitative research syntheses that researchers mounted in the 1980's generally uncovered few factors. In recent years, however, researchers have applied different synthesis methods to old data. Some of these analyses have produced evidence of several statistically significant relationships between factors such as class size and achievement. Nevertheless, this evidence is unconvincing for numerous reasons, creating a need for additional research with new data. In this study, I investigate the relationship among several school policies, practices and resources, the reading achievement of fourth-grade students, and differences in reading achievement between white and non-white fourth-grade students who attend the same school by applying a series of multilevel statistical models to data from the 1994 National (NAEP) and Trial State Assessments (TSA) of Educational Progress. Analyses of Pennsylvania data indicate that per-pupil spending, school racial composition, and an interaction between class size and teacher experience are associated with reading proficiency. Analyses of New York data indicate that per-pupil spending, school racial composition, and teacher experience are associated with reading proficiency while teacher certification level is associated with differences in reading proficiency between white and non-white students who attend the same school. Finally, analyses of United States data indicate that school racial composition is associated with reading proficiency while teacher certification level and school location are associated with differences in reading proficiency between white and non-white students who attend the same school. The evidence produced here provides new, potentially useful information to policy makers at national, state, district, and school levels. Lessons learned through the application of multilevel statistical modeling to NAEP and TSA data and through the linkage of several disparate data sets also provide new, potentially useful information to those empowered to modify the future design of surveys producing data sets of these types.
