The Fallacy of Causal Analysis,
Abstract
The technique of causal analysis--the use of correlation coefficients as a tool to distinguish between alternative causal patterns among variables--is a popular one in the social sciences. The heuristic justification for t e technique is superficially convincing and indeed, can be hown to have some theoretical foundation in a rather straightforward mathematical model. Moreover, the assumptions made in this model often appear to be reasonable first approxima ions to the structure of interest in the real world. The fallacy of the technique, however, stems from the way in which the assumptions drive the conclusions reached. The conclusions are driven not by the linear relationships chosen to approximate the real world relationships of interest, but rather by a specific detail in the model chosen as a mathematical convenience--the assumption of statistical independence of the residual terms in the systems of equations (causal pattern) which makes the residuals appear most nearly independent. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1971
- Accession Number
- AD0731651
Entities
People
- Ralph E. Strauch
Organizations
- RAND Corporation