The stability of school effects across academic subjects

• Main Author: Liu, Bing-Zheng
• Format: Others
• Language: English
• Published: 2009
• Online Access: http://hdl.handle.net/2429/4016

The effects of schooling on the cognitive development of children has been a central theme in educational research (Heyns 1978). One of the problems in this area of research is the stability of school effects. Studies have examined the stability of performance indicators across random samples of students from the same grade level, across subject areas for students at the same grade level, across grade levels at one point in time, and across years for the same grade level. The purpose of this study was to examine the stability of school effects across the curriculum subjects, mathematics, reading, science, and history. This was accomplished by reanalyzing the NELS:88 data, which is a nationally representative sample of nearly 25,000 eighth-grade students attending over 1000 schools in the United States. A major problem for analysts confronted with educational data and achievement outcomes is choosing the most appropriate strategy for analyzing multilevel data. In addition, the stability of school effects across subjects so far has been measured by correlation coefficients, so another problem is how to estimate the covariances among school effects properly. To meet these two challenges, this study used a three-level Hierarchical Linear Model. By adding a within-student level model using dummy coding, the outcomes of different subjects were combined into one equation. This permitted the estimation of the covariance matrices for the original four curriculum subject outcomes at the higher level analyses. The results of this study indicate that there are significant differences between U.S. secondary schools in both their Type A and Type B school effects. In other words, there is significant variation in schooling outcomes between U.S. secondary schools, after controlling for (1) student background only, and (2) both student background and school contextual effects. The results also show that there is a uniformity of both Type A and Type B school effects across outcome measures, that is, schools with superior levels of performance in one curricular subject tend to have superior levels in other subjects even after controlling for (1) student background only, and (2) both student background and school contextual effects. === Education, Faculty of === Educational and Counselling Psychology, and Special Education (ECPS), Department of === Graduate