ERIC Number: ED168131
Record Type: Non-Journal
Publication Date: 1978
Reference Count: N/A
Blocking Predictor Variables in the Companion Study: Approaches to Data Analysis in Project MITT.
Jovick, Thomas D.
As a companion study to the primary analysis of the Management Implications of Team Teaching (MITT) study, the unitized schools in this study were examined alone for predictors of differential success in implementing the innovation. This report briefly describes how the MITT staff handled a problem in the companion study of reducing the number of predictor variables while still retaining as much information as possible about all of them. They did so by using multiple linear regression to combine conceptually related variables into clusters or blocks. Once the researchers had identified a group of variables that logically comprised a cluster, they regressed the most logical dependent variable in the group (if that were possible to determine) into the others. For each school, they multiplied the resulting raw score regression weights times the obtained score on their respective variables to produce a predicted dependent variable score; conceptually, such a predicted score reflects all the information shared among the variables in the block. (Author/JM)
Publication Type: Reports - Research
Education Level: N/A
Sponsor: National Inst. of Education (DHEW), Washington, DC.
Authoring Institution: Oregon Univ., Eugene. Center for Educational Policy and Management.