The effects of selected variables on the cognitive achievement levels of students enrolled in Texas adult basic education programs

Abstract

A multivariate case analysis was developed to examine the relationships between specified variables and student cognitive achievement in three Texas adult education programs. Data were collected from an existing statewide reporting system, thereby creating a cost effective mode of analysis. Further, the utilization of a computer stored dataset represented a practical extension of a system for decision-making at both the state and local levels. Variables were organized into five predictor domains as supported by the professional literature in adult education. These included the entering achievement level for each student, a set of five background characteristics for each student, an attendance index for each student, a set of four program variables on each student's class, and a set of five characteristics on each student's teacher. The effects of each predictor domain on student achievement were determined by decomposing variance into partitions using the technique of commonality analysis. Results of the study indicated that only two single variables, entry level achievement and attendance, were directly and substantially associated with learning outcomes in every case. The confounded effects of program variables, background factors, and teacher characteristics were shown to have associations with learning outcomes; however, no other individual predictor variables produced clearly distinguishable trends across all cases studied. In-depth confirmatory analysis is recommended to investigate the areas left unresolved by this study. Of particular interest are the issues of program and teacher effects on the learning process. With additional refinement and development, this particular research technique offers great promise for learning analysis in the field of adult education.