ERIC Number: ED184914
Record Type: RIE
Publication Date: 1979-Sep
Reference Count: 0
Regression Weights and Communication among Researchers from Different Disciplines.
Richard, James M., Jr.
This report examines the difference in approaches between sociologists and psychologists when using multiple regression techniques in the analysis of behavioral data. Psychologists and sociologists are often divided in their orientation toward regression techniques, and this division could be a substantial and unfortunate barrier to communication among researchers. Many sociological methodologists argue that unstandardized (raw score) regression weights should be used in comparing groups, whereas psychologists rely almost exclusively on correlations and standard score regression weights. This report attempts to clarify the issue by illustrating how raw and standard score regression weights behave in various circumstances. The report is presented in four sections. Section one explores the reasons why sociologists prefer unstandardized weights. Section two discusses reasons why psychologists prefer standardized weights. Section three examines circumstances that do not fit conveniently under the rubric of comparing sociological and psychological perspectives on regression weights. The concluding section considers the implications of these analyses. It is found that each kind of weight is invariant in some circumstances where the other is not, and each kind can provide scientifically meaningful information that the other cannot. (Author/AV)
Publication Type: Information Analyses
Education Level: N/A
Sponsor: National Inst. of Education (DHEW), Washington, DC.
Authoring Institution: Johns Hopkins Univ., Baltimore, MD. Center for Social Organization of Schools.