The Fallacy of Causal Analysis.
A discussion of a logical fallacy in the use of correlation coefficients to distinguish between alternative causal patterns as an inference tool in social science. Justification for using the technique is superficially convincing, and appears to be supportable by rigorous mathematical analysis. The fallacy of the technique stems from the fact that the conclusions reached from it are based on an assumption usually made as a mathematical convenience. This assumption is that statistical independence of residual terms in the equations represents alternative causal patterns. Unless the analyst is willing to explicitly equate independence of residuals with real-world causation, conclusions about causation drawn from causal analysis are specious. The equations used in causal analysis may be given both a causal and a statistical meaning, and there is no necessary relationship between the two. 8 pp.