Dimensionality of Measured Achievement over Time.
Abstract
Some type of difference or change score is frequently used to quantify the effects of experimental treatments and educational programs on individuals and on groups of individuals. Whether the change measurement involves the use of simple difference scores, their derivatives, or some more complex methodological design, the measurement process itself assumes that the treatment or instruction results in higher levels of the originally measured variable and that the only change that occurs is a quantitative one. If this assumption is not met, then the computation of any type of difference score is inappropriate and the scores themselves are useless for measuring growth or change. Two studies investigated the tenability of the assumption that classroom instruction results in increases in students' achievement levels while the qualitative nature of that achievement remains constant across time. The data utilized were the item responses to tests in basic mathematics and in general biology administered as pretests and after instruction to students enrolled in those courses. Results included that this assumption was not tenable in the biology data set, where increases in mean achievement level were accompanied by corresponding changes in the factor structure underlying the item responses. For the mathematics data, however, there was no such violation of the assumption: As student achievement levels increased the underlying factor structure remained unchanged. The implications of these results for psychology, education, and program evaluation are noted. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1981
- Accession Number
- ADA110955
Entities
People
- David J. Weiss
- Kathleen A. Gialluca
Organizations
- University of Minnesota