Sunday, January 31, 2016

Measuring General Antisocial Behavior

I have been wrestling with how to measure efficacy of one gun a month laws as measured by effects on the Northeastern state crime rates in response to passage and repeal of South Carolina and Virginia 1 gun a month laws, and I found myself wondering if change in gun-crime rates (murder, aggravated assault, robbery) relative to non-gun crime rates (like rape, larceny-theft, burglary, motor vehicle theft) might be a fit measure.  So I decided to see if nationally these various antisocial felony rates were strongly correlated or not.
Surprise: the following shows correlation of murder rates, rape rate,s larceny-theft rates, motor vehicle theft rates, and burglary rates to murder rates for 1960-2012:


1 0.53031789 0.734603441 0.80169651 0.934104

murder rates correlation of course is 1.00, rape rate is very poor, probably because of low rates of rape reporting in the 1960s and early 1970s. But larceny-theft is a surprisingly strong correlation, and auto theft and burglary rates are very strong. Murder and burglary are so strong that changes in one predict changes in the other with startling accuracy.

No comments: