January 18, 2018

Race, IQ, and Crime

This article will analyze the role that racial intelligence differences play in racial crime differences. First, I’ll review the national and individual level data showing that crime has a negative association with IQ which persists after controlling for all home environment variables. Second, I’ll show that the so-called differential detection hypothesis is false and that IQ probably has a causal relationship with criminal activity. Finally, it will be shown that a large proportion of the Black-White crime gap can be accounted for by controlling for IQ.

IQ and Crime: Macro Studies

Lynn and Vanhanen (2012) summarized the data relating national IQ to national crime levels. Studies consistently find a negative correlation between national IQ and crime. As can be seen, the size of this correlation varies from weak to strong depending on the study.

IQ a national crime.jpg

Bartels et al. (2010) looked at the relationship between US state levels of crime, IQ, and racial demographics.When each variable was analyzed on its own, the percent of a state which was Black and its mean IQ score (estimated based on NAEP data) predicted various categories of crime. When entered into a regression at once, however, they sometimes lost statistical significant due to the small sample size (n=50). However, both variables retained significance when predicting murder and % Black was a better predictor than IQ.

IQ and state crime.jpg

Beaver and Wright (2011) examined the relationship between crime, IQ, and other variables across 243 US counties. Beaver and Wright created a measure of “concentrated disadvantage” by performing factor analysis on the following variables: % black, % female headed household, % poor, % of households on public assistance, and % unemployed.

IQ and coutny crime.jpg

Whether predicting property or violent crime, IQ and concentrated disadvantage mostly had significant independents effects with IQs effect size normally being larger.

IQ and Crime: Individual Studies

Ellis and Walsh (2003) meta-analyzed research done on the relationship between IQ and crime, delinquency, and related variables. Out of 68 studies on IQ and delinquency, 60 found a negative relation (88%) and the remaining 8 found no significant relationship. Out of 19 studies on IQ and adult criminal offending, 15 (79%) found a negative correlation. Out of 17 studies on self-reported offending and IQ, 14 (82%) found a negative relationship. Out of 5 studies on IQ and anti-social personality disorder, and 14 studies on childhood conduct disorder, all 19 found a negative relationship. Thus, the vast majority of research establishes IQ as a correlate of crime and related constructs.

On the other hand, only 7 of 19 (36%) of studies on recidivism and IQ found a negative relationship. Ellis and Walsh explain this by positing that the population of people who have already committed a crime have a far lower mean IQ than the general population (and a reduced range) and, among this subset of the population, intelligence differences don’t explain nearly as much variance in crime as it does among the general population.

Schwartz et al. (2015) found a linear negative relationship between IQ and various measures of criminality in a complete finish birth cohort for 1987 (n=60,069).


Frisell et al. (2012) looked at the association between IQ and crime in a total male birth cohort sample from Sweden (n=700,514). IQ was measured at age 18. IQ was negatively associated with crime (-.19) and this only fell to -.18 when adjusting for single motherhood and income. Furthermore, IQ differences predicted crime differences between half-siblings and full siblings raised together. The fact that IQ predicts crime within families shows that home environment variables which vary between families cannot possibly account for the relationship.

Levine et al. (2011) looked at the relationship between IQ and criminality in a sample of 12,686 Americans, 496 of which had been incarcerated. Those who had been to jail had a mean IQ of 89.61 while those who had not who had a mean IQ of 100.6. This is a difference of .77 SD and was statistically significant. The groups also differed in SES (SMD = -.37, p<.001). IQ and SES were put into a regression and shown to have effects independent of one another and had a significant interaction such that the effect of low IQ was worse for low SES people and vice versa.

Yun and Lee (2013) looked at how the kind of neighborhood someone lives in impacts the association between their verbal IQ and their probability of being arrested. In a sample of roughly 2,000 Americans, they found that verbal intelligence had a negative relationship with criminality even after controlling for neighborhood poverty, unemployment, % Black, % female headed household, and % on public assistance, as well as individual age, sex, race, poverty, self-control, and age.

Yun Lee 1.jpg

Yun and Lee also found that the relationship between IQ and criminality was much stronger in well-off areas than it was in disadvantaged areas.

Yun Lee 2.jpg

There is also a significant longitudinal literature linking IQ in childhood to criminal activity in adulthood. For instance, Farrington (1989) found that IQ at age 8 was a significant predictor of being convicted of a violent crime as an adult in a sample of 411 boys. Similarly, Stattin et al. (1993) looked at the longitudinal relationship between IQ and crime in a sample of 122 boys and found that criminals scored lower on measures of intelligence than non-criminals even at the age of 3.

Stalin 1.jpg








Controlling for SES reduced some, but not all, of these associations to statistical insignificance, but this is probably due to Stattin et al’s small sample size.

Stalin 2.jpg

Finally, Fergusson et al. (2005) conducted a 25-year longitudinal study on 1,625 participants. They found that IQ at age 8-9 predicted criminality in adulthood. This relationship was also found to be mediated by childhood conduct problems, which just tells us that IQ begins to have an effect on criminality at an early age.

The Differential Detection Hypothesis

Some people have argued that having a low IQ doesn’t actually cause crime but, rather, simply causes people to perform crimes in such a way that they are more likely to get caught. This line of argument is refuted by the previously referenced self-report data showing that low IQ predicts not only arrest but also self-reported criminal activity.

Moffitt and Silva (1988) examined this question more closely and looked at crime and IQ in a  sample of 654 boys using two different kinds of measures of criminal activity: self-report and police report. They found that delinquent adolescents had lower than average IQs and that their IQs did not differ depending on whether or not they had been arrested. In other words, IQ was not predictive of whether a criminal was caught. This directly contradicts the hypothesis that IQ correlates with crime simply because dumb criminals are more likely to get caught.


Given that this explanation of the relationship between IQ and crime fails, and home environmental variables cannot explain it, nor can related psychological constructs such as self-control, and IQ in toddlerhood predicts criminality in adulthood, it seems safe to say that IQ probably has a causal relationship with crime. With that being said, let’s look at the role that IQ plays in racial crime differences.

Racial Differences in Crime Holding IQ Constant

Two studies have looked at what happens to racial crime differences after IQ is held constant. First, Beaver et al. (2013) looked at the degree to which racial differences in crime disappeared after controlling for self-reported life time violence and verbal IQ. Their sample consisted of  3,029 males.

African American men were 43% more likely to be arrested than White men. However, this dropped to a statistically insignificant 13% after controlling for life time violence and IQ. Before applying the controls, Black men were 56% more likely to have been incarcerated. After applying controls, this figure dropped to a statistically insignificant 18%. Finally, once arrested Black men were 50% more likely to end up incarcerated and, after applying these controls, that value dropped to a statistically insignificant 24%.

Secondly, Herrnstein and Murray (1994) analyzed a large nationally representative data set and found that the Black-White incarceration gap decreased by nearly ¾ after simply controlling for age and IQ.


Thus, racial differences in IQ probably explain a good deal of the Black-White crime gap, though not all of it.

Facebook Comments
  • Shane Willis

    Is there any way to control for both IQ differences and differences in testosterone levels to see the remaining incarceration rates by race?