Herman Rubin said on 12/5/02 7:10 AM: >In article <[EMAIL PROTECTED]>, >wuzzy <[EMAIL PROTECTED]> wrote: >>I read that regression was originally developed as a technique >>to develop propensity scores to be used in contingency tables, not to >>develop models that adjust for confounders as it is used now. > >>This is hard to believe by someone like me who uses regression daily >>for the purpose of adjusting for confounders.. > >>I've lost the book I read this in, but it had no references in this >>regard. > >>Can anyone comment on this? Eg., is there a seminal paper in which >>some turning point visible? > >Regression goes back to the 18th century. The term was >introduced by Galton in the 19th century, who observed >that the heights of sons tended to regress toward the >mean from that of the fathers.
IIRC, based on my reading of Stigler, wasn't it LaPlace who developed the method of least squares to solve the orbit of Saturn? Paul . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
