Econometric Modeling as Junk SciencePDF VersionThis article by sociologist Ted Goertzel of Rutgers University discusses how multiple regression math models have been misused in sociological studies. Most of his discussion centers on the way John Lott of Yale and David Mustard misused such models to "prove" that increased handgun ownership as measured by "Shall Issue" laws leads to reduced crime rates. During the 90’s Lott published a popular book based on this research titled More Guns, Less Crime that was immediately embraced by the NRA and other Far-Right interests as little less than divine revelation. Soon thereafter it was discovered that Lott's models had been poorly designed and were riddled with "clustered" data (i.e. - independent variable data that was inadvertently correlated with other variables and therefore not truly independent). They were even found to have basic computer code errors that prevented them from even calculating many of Lott’s foundational results correctly. Needless to say, none of this made any dent in the book's popularity with the Far-Right or dissuaded them from canonizing Lott. This article first appeared in condensed form in the Spring 2002 edition of the Skeptical Inquirer as Myths of Multiple Regression & Murder: Econometric Modeling as Junk Science. It is reprinted here in full with the author's permission.