An empirical approach toward the prediction of students' science achievement in the United States and Hubei, China

An empirical approach is adopted in this article to explore a possible model for the prediction of students’ science achievement in China and the United States. The construction of the model was based on the ninth-grade data base from Phase 11 of the Second IEA Science Study (SISS) in the United States, and the SISS Extension Study in the Hubei province of China. The common independent variables of the students’ science achievement are classified into five categories: students’ gender, attitude, home background, classroom experience, and personal effort, according to distinction between visible and latent characteristics, and scree plots from principal component analyses. Latent factors are represented by the first principal components in each of the four latent categories: students’ attitudes, home background, classroom experience, and personal effort. Predictors of the model are constructed by polynomials of the visible and latent factors and their interactions in a multivariate Taylor series. Significant predictors at a = .05 were selected through a backward elimination procedure using the Statistical Analysis System. The structure of the four latent factors and the model complexity are compared between the two countries in terms of their educational, political, social, and cultural contexts.


In Collection: