Sunday, January 16, 2011

IQ and delinquency

To HBD folks, the idea that low IQ explains criminality is attractive since intelligence is important in many areas of life, and racial rankings on IQ match those of crime involvement. 

Using Add Health data, I calculated the Pearson correlation coefficients between a measure of IQ (a vocabulary test) and frequency of 15 different delinquent behaviors. Here are the results (sample size = 4,082, whites only):


Graffiti -.01
Vandalism .06
Lied to parents .09
Shoplift .04
Serious fight -.11
Seriously injured someone -.06
Ran away from home -.03
Stole a car -.01
Stole something > $50 -.01 
Burglarized a building -.01
Used or threatened with weapon -.03
Sold drugs .04
Stole something < $50 .07
Group fight -.10
Rowdy in public .08

These correlations offer little support for the view that low intelligence is a major cause of crime.  Most of the measures of delinquency are very skewed (which is a problem I will try to look at later) but the correlations aren't consistently negative, and even the less skewed behaviors like being rowdy in public are not negatively related to IQ in a consistent way.   


Statsquatch said...

Interesting, what if you add them up? The measure may be less skewed. I assume the Spearman's corr is the same as Pearsons.

Chuck said...

Refer too. Bartelsa, Ryana, Urbana, and Glass, 2010. Correlations between estimates of state IQ and FBI crime statistics

OneSTDV said...

Do we really need all this fancy-shmancy data when we have our eyes?

Statsquatch said...


Data is fun.

JL said...

Most people with low IQs don't commit crimes, while some with high IQs do, so it's not surprising that the individual-level correlations are low. For this reason, the correct approach is to study aggregates like they did in the study above.

JL said...

Ron, I think your spam filter ate my first post. This is the study I referred to:

Ron Guhname said...

Here's a new study in Intelligence by Beaver and Wright showing higher crime rates in low-IQ counties: