Among the usual suspects emerging in the renewed gun control debate is the veteran pro-gun propagandist, professor John Lott, author of the 1998 book, “More Guns, Less Crime.” According to Lott, gun-control laws are not merely futile; they are harmful. They do nothing to keep guns out of criminals’ hands; they only prevent potential victims from getting guns to deter predators.
In a Wall Street Journal article, allegedly substantiating his more-guns-less-crime thesis, Lott presents statistical “evidence” containing an incredible implication: a 33 percent increase in gun ownership would eradicate violent crime (i.e. reduce this crime rate to 0).
This conclusion springs from Lott’s contention that a 3 percent increase in violent crime accompanies each 1 percent reduction in gun ownership in states that enact gun-control legislation.
The critical question, ignored by Lott, concerns the correlation’s limits. Is its validity limited to 1 percent, 10 percent or 100 percent variation in gun ownership? Lott’s more-guns-less-crime thesis sets no limits to the correlation’s application. His empirical surveys (obviously linear regression analyses) can’t possibly permit the extrapolation represented in the extreme conclusion. Moreover, his book, “More Guns, Less Crime,” claims a greater reduction in crime, 4.1 percent instead of 3 percent, for each percent increase in gun ownership. But again, he cites no limits to the correlation, no correlation coefficient to indicate how well the data fit the regression line, no confident level calculation for judging the formula’s dependability.
Lott’s approach is heads-I-win-tails-you-lose. There’s a colossal inconsistency between the rigorous standards he set for opponents and the leniency he permits himself. Illustrating this bias, Lott claims that the Brady Law is associated with increases in aggravated assault. But in the same table listing a 3.7 percent increase in assaults attributable to the Brady Law, he records a numerically larger percentage, 3.9 percent, decrease in robberies. The 3.7 percent assault increase is “statistically significant,” whereas the 3.9 percent robbery decrease in “statistically insignificant.”
His convoluted explanation of this anomaly is unconvincing.
It’s easy to explain how Lott gets away with his statistical flimflam. Mathematically illiterate reporters don’t dare challenge him. Here’s reporter Matt Bai: “There’s no easy answer to that debate partly because it takes an advanced degree in statistics to truly grasp Lott’s methods (we won’t even try).
• C.W. (Bill) Griffin is a retired consulting engineer. He has lived in Ahwatukee Foothills for more than 22 years.
afnanalog posted at 11:44 am on Mon, Jan 14, 2013.
CW is our resident wacko[wink]