Predicting Reading Achievement from Measures Available during Computer-assisted Instruction

A high correlation between on-line rate of prog,ess and student achievement on a standardized test was found for a computer-assisted instruction (CAl) program in initial reading. In most cases, CAl measures of progress were better indications of spring test performance than was the pretest given in the fall. Rates of progress in the parts or strands of the CAl program were highly correlated with each other, but certain strands proved to be better predictors of spring test scores than an overall rate measure. Regression models were developed to relate spring test scores to amount of time spent in CAl; results from these models were in accord with data from an earlier experimental study designed to evaluate the effectiveness of the CAr program. Using a stepwise regression, which included both pretests and rates of progress in the strands, multiple correlations were obtained of .79 for the Cooperative Primary Test and .84 for the Metropolitan Achievement Test.