PDF pending restoration
ERIC Number: ED177189
Record Type: RIE
Publication Date: 1977-Jun
Reference Count: 0
An Alternative Model for the Evaluation of Change. Technical Report.
Corder-Bolz, Charles R.
Previous research has indicated that most mathematical models used to evaluate change due to experimental treatment are misleading because the procedures artificially reduced one of the estimates of error variance. Two modified models, based upon the expected values of the variance of scores and difference scores, were developed from a new theoretical perspective. A Monte Carlo study was made involving one-way analysis of variance using the two modified and two basic models. A total of 1,080 experiments were simulated on three treatment levels, six error levels, and three pretest difference levels. The second modified model estimated treatment effects, without being subject to possible distortions due to initial between-group differences, and yielded valid estimates of treatment effects, regardless of the amount of error variance. While the statistic generated by this model has a theoretical weakness, results of this study indicate that it may be a functional solution to the measurement of change controversy. (Author/CP)
Publication Type: Reports - Research
Education Level: N/A
Sponsor: Office of Child Development (DHEW), Washington, DC.
Authoring Institution: Southwest Educational Development Lab., Austin, TX.