Crime and IQ: A Review

There is a moderate to strong relationship between criminality and intelligence. The evidence for this will be documented below, though explanations for why this is will be saved for another day:


In 1914, the role of intelligence in crime was brought to attention by H.H. Goddard (1914) who found that the majority of people in prison were mentally deficient. After this point though, the relationship between intelligence and criminality was not only ignored, but unfairly rejected (see Hirschi and Hindelang, 1977). This was because of a paper by Edwin Sutherland (1931) titled “Mental Deficiency and Crime” where Sutherland argued that the cause of the association was because of poor testing conditions. He showed that as testing procedures became better, the correlation began to diminish itself. But, he wrongly assumed that over time, it would completely disappear.

Hirschi and Hindelang (1977) re-initiated the crime-IQ debate with a paper titled, “Intelligence and Delinquency: A Revisionist Review”. This paper fought against the sociological bias against the role of IQ in criminal behavior and adult delinquency and pulled from multiple studies available at the time to prove it is unjustified. Some of the most compelling data is explained in the following paragraph (all studies cited in the following paragraph can be found in Hirschi and Hindelang [1977]):

First, they re-analyze data from Hirschi (1969) which had a sample of 3,600 boys from Contra Costa County, California. They find a gamma correlation of -0.31. Wolfgang et al. (1972) uses criminal data from 8,700 boys and splits them into groups by IQ. There was a clear association with a gamma correlation of -0.31 for whites and -0.16 for blacks. West (1973) uses data from over 400 boys from London. They find that one quarter of people with an IQ of 110 or higher had a police record whereas one half of people with an IQ of 90 or less had a police record. He concludes that IQ was a significant predictor of delinquency and that it survived as a predictor after controlling for variables such as family income and family culture. Toby and Toby (1961) showed “intellectual status” was a significant predictor of delinquency/non-delinquency, regardless of socioeconomic status. Finally, Hirschi and Hindelang report on data which shows that even self-reported criminal behavior correlates with criminality.

Spellacy (1977) looks at 40 violent and non-violent adolescent males and tests them on neuropsychological tests and on the MMPI scale. They tested the group on verbal IQ, performance IQ, and full-scale IQ (FSIQ). On FSIQ, there was a 12.4 point difference between violent and non-violent adolescents. The results were consistent across other tests of mental ability. Similar differences are analyzed by Holland, Beckett and Levi (1981) and Holland and Holt (1975).

In the popular book, The Bell Curve, Herrnstein and Murray (1994) argue that intelligence plays a substantial role in criminality. To avoid the intermingling of race and gender in criminality (as men commit far more crime than women and African-Americans have higher crime rates than whites and Asians), they limited their sample to white males in the NLSY. They show multiple tables; the following table has been adapted as it is likely the most significant:

The Odds of Getting Involved with the Police and Courts for Young White Males (adapted from Herrnstein and Murray, 1994)
Percentage Who in 1980 Reported Ever Having Been:
Cognitive ClassStopped by the PoliceBooked for an OffenseConvicted of an OffenseSentenced to Incarceration
I Very Bright18530
II Bright271271
III Normal3720153
IV Dull4621277
V Very Dull3317147
Overall341893

Herrnstein and Murray also report on regression analyses for the data. Before I present those, I would like to show how the data provides a far different correlation when presented as groups, rather than looking at individuals (such as Herrnstein and Murray’s regression analysis does). If we input the data for “Cognitive Class” and “Sentenced to Incarceration” we can find a very strong correlation of r^2=0.926. I only mention this for the purposes of later posts which will be the investigation of how the unit size effects the size of the correlation for IQ and outcomes (e.i. the correlation of IQ and crime for a nation is higher than the correlation of IQ and crime at the individual level).

Regardless, Herrnstein and Murray’s regression analyses show a fairly small correlation between intelligence and “being in the top decile on an index of self reported crime” of r^2=0.0152. (I would point out that they should have broadened the measure. They also find that IQ explains 9.6 percent of the variance in having been interviewed in jail. Despite this, they argue the relationship is obvious, a claim which has been criticized by Cullen et al. (1997). Cullen et al. re-analyze the data but look at more sections of self report data as well as having been interviewed in jail and penetration of the criminal justice system. They find very low correlations; for all of the self report data, less than one percent of the variation is explained by IQ. For having been interviewed in jail, they find only 2.5-2.6 percent of the variation is explained by IQ, and for penetration of the criminal justice system, they find only 3-3.4 percent of the variation is explained by IQ.

Cullen et al. go on to analyze the data on IQ and being in the top decile of self-reported criminality, but they take more variables into account. They find this makes the correlation statistically insignificant. It should be noted that statistical significance is more a matter of sampling rather than an actual interpretation of results. For example, with GWA studies, an increase in the sample increases the predictive power of the model. When a correlation is found statistically insignificant, it should not be understood that no relation exists, but rather that a reliable association could not be found with the given data, but still could be found under different circumstances (hence why many psychologists refuse to publish studies which only find statistical insignificance; it really isn’t useful). On the other hand, they find a statistically significant, strong, negative correlation between IQ and being interviewed in a corrections facility, but they simply point out that other sociological variables are important there as well. Whatever the reliability of Cullen et al.’s results, it does not fall in line with the majority of the data.

Mears and Cochran (2013) followed the results of both Herrnstein and Murray’s (1994) and Cullen et al.’s (1997) studies to design an analysis of intelligence and crime. Like both previous analyses, they used the NLSY data of white men and their AFQT scores and create an index of different forms of delinquency and how much the participants committed those forms of delinquency. They controlled for additional measures, like Cullen et al. in order to refine the correlation to AFQT scores as much as possible. But, instead of purely relying on regression analysis (as they explain it has issues for curvilinear data), they used GPS analysis. First, they provide the bivariate analysis results which show that lower IQ people to tend to commit more crimes, but once you are looking at people in the IQ range of 77-88, the propensity for crime drops off. Essentially, this implies an inverse U-shaped model. Then, when testing the relationship through GPS analysis, they do find an association where people in the 90’s IQ range commit the most crimes, but people below that and above that commit less crimes.

June Andrew (1982) uses digit span tests and verbal IQ scores for a young sample of delinquents. She finds substantial differences in both across non-violent delinquents and violent delinquents. Crocker and Hodgins (1997) follow a Swedish cohort of over 15,000 participants to age 30. The mentally retarded male participants were significantly more likely to have committed at least one violent offense, theft, traffic offense, or “other” offense. Similar results are found for women. A study by Oleson and Chappell (2012) actually looked at a sample of people of very high intelligence, or geniuses (mean IQ of the sample was 154.6). They found that even among geniuses, a statistically significant, negative correlation exists between IQ and use of violence, having killed another human (excluding warfare), and having kidnapped someone.

Diamond, Morris, and Barnes (2012) look at both individual IQ and prison-unit-level (different units of the prisons; groups) IQ and see if it relates to the amount of individual inmate violence. First of all, they find that the average IQ of prisoners is about 2/3rds of a standard deviation below that of the American population. This is line with other research that the IQ difference between criminals and non-criminals is about 8-10 points (Hirschi and Hindelang, 1977). Second of all, they find individual IQ negatively correlates with individual inmate violence and that differences in prison-unit-level IQ negatively predicts the amount of individual-level violence. This may be somewhat in line with the theory I presented earlier that group IQ differences matter more to variation in given outcomes than individual level IQ differences.

Lynam, Moffit, and Stouthamer-Loeber (1993) argue low IQ is a solid, causal predictor of delinquency. This is done through specific procedural measures such as using younger boys, so as to avoid the effect of prison lifestyle on intelligence, as well as controlling for test motivation. The latter procedure is done to combat the hypothesis that the delinquency-IQ correlation can be mediated by the fact that delinquents do not seek to do well in life and will not care about their results on a test. Additionally, multiple studies have shown that the IQ of delinquents was low before said individuals became delinquent (Denno, 1990; Moffit et al., 1981; West and Farrington, 1973). The present study used self reports to measure delinquency in the boys and controlled for social status, race, and test motivation to ensure the correlation remained regardless of these variables. A correlation of r=-0.22 is found for FSIQ and delinquency. Impulsivity mediated relatively little of the relationship; school achievement did not have any effect on the relationship for whites whereas it mediated the association for blacks.

Hodges and Plow (1990) and Ward and Tittle (1994) also control for both SES and race and find that low-IQ remains a significant predictor for delinquency. Wolfgang, Figlio, and Sellin (1972) compare one-time offenders and chronic offenders in intelligence. They control for SES and race and still find an 8.1 IQ point difference for whites and a 10.6 point difference for blacks. The latter difference is particularly interesting because most studies find a smaller association between IQ and crime for blacks (Hirschi and Hindelang, 1977; Lynam, Moffit, and Stouthamer-Loeber, 1993).

McDaniel (2006) used NAEP data to estimate the average IQ of different states. He finds a correlation of r=-0.58 for state IQ and violent crime rate. Bartels, Ryan, Urban, and Glass (2010) use state IQ estimates to create estimates on the relationship of IQ to criminal behavior. They sought to replicate and extend upon McDaniel (2006) by looking at various types of crimes at the state level. Bartels et al. find, like McDaniel, a correlation of r=-0.58 for state IQ and violent crime, despite the years tested being different. They extend with the following, significant correlations of -0.57 (murder), -0.29 (robbery), -0.41 (assault), -0.45 (property), -0.57 (burglary), and -0.29 (theft). McDaniel (2006) was also replicated by Pesta, McDaniel, and Bertsch (2010) who found states with lower average IQs had higher aggregate crime rates. They found a correlation of r=-0.76 between overall crime rate and state IQ estimates.

Templer and Rushton (2011) analyze data from the fifty United States on IQ and criminal behavior. IQ was correlated with murder at r=-0.64, robbery at r=-0.46, and assault at r=-0.47. Beaver and Wright (2011) look at over 200 counties and their IQ estimates. The correlation matrix can seen below in Table 1: primarily the violent crime rate was correlated with IQ at r=-0.58, the property crime rates was correlated at r=-0.40, the aggravated assault rate was correlated at r=-0.52, and the composite crime rate was correlated at r=-0.53.

At the national level, Rushton and Templer (2009) conduct principal components analysis on the effect national IQ has on national crime rates. National IQ correlated with homicide at r=-0.25, rape at r=-0.29, and serious assault at r=-0.21.


Gendreau, Little, and Goggin conducted a meta analysis of 131 studies (1,141 correlations) on the relationship between specific factors and adult recidivism. “Intellectual functioning” had a mean correlation of r=0.07. This relationship was stronger than the SES-Crime relationship, but worse than most others. Kandel et al. (1988) finds that IQ is a protective factor against criminogenic, environmental differences and, in effect, reduces risk of criminality. Many others have agreed with the hypothesis that IQ indirectly affects criminality through causing criminogenic factors such as Herrnstein and Murray (1994), Magdol, Moffit, Caspi, and Silvia (1998), and Ward and Tittle (1994).

The 2009 edition of the Handbook of Crime Correlates finds a large number of studies on the correlation between IQ and criminality (Beaver, Ellis and Wright, 2009). They find that the supermajority find a statistically significant, negative association. Some fail to find a statistically significant association, and very few find a positive association. These results were the same for official offending, self-reported offending, and various forms of psychopathy related to criminal behavior. Additionally, the type of IQ matters; performance IQ has a stronger association than does verbal IQ, but both negatively predict criminality.

Ellis and Walsh (2003) review the international data on IQ and crime; of 68 studies on IQ and delinquency, 60 found statistically significant, negative relationships. The other eight only reported statistically insignificant relationships. Of 19 studies on adult offending and IQ, 15 found statistically significant, negative relationships. Of the 17 studies on self reported and IQ offending, 14 found a statistically significant, negative relationship. Of the (19) studies on the effect of IQ on antisocial personality disorder, all found a statistically significant, negative relationship. Additionally, the international meta analysis provided by the newer edition of the Handbook of Crime Correlates (Ellis, Farrington, and Hoskin, 2019) finds that far and wide, the majority of the studies show a statistically significant, negative relationship between official offending and IQ. Some studies were statistically insignificant and very few showed a positive relationship.


The association between IQ and criminality can be shown both across criminals and non-criminals, and across different types of criminals. Moffit and Silva (1988), Murchison (1926), and Stark (1975) all argue that this relationship is not limited to delinquents vs. non-delinquents. Even within criminals, more intelligent ones are less likely to be arrested. Many studies have commented that the seriousness of a criminal career can be predicted by intelligence (Cymbalisty, Schuck, and Dubeck, 1975; Gabrielli and Mednick, 1980; Lynam, Moffit, and Stouthamer-Loeber, 1993; Moffit et al. 1981).

Multiple studies have argued intelligence plays a role in sexual offending. Okami and Goldberg (1992) meta-analyze the effect of intelligence on pedophilia. The results were mostly non-significant, though some studies show small differences (where pedophiles were less intelligent). Beggs and Grace (2008) test whether intelligence mediates the association between psychopathy and recidivism in child molesters. The study used a sample of 216 males from a prison-based treatment program. They generally could not find significant correlations. But, they find a correlation of r=-0.14 for violent recidivism and a correlation of r=-0.20 for psychopathy. The groups with the highest psychopathy and lowest intelligence were more than four times more likely to be incest and extrafamilial offenders and two times more likely to be repeat offenders.

Across the board, there is a consistent relationship between crime and intelligence.

Bibliography

ANDREW, J. M. (1982). Memory and Violent Crime among Delinquents. Criminal Justice and Behavior, 9(3), 364–371. doi:10.1177/0093854882009003007 

Bartels, J. M., Ryan, J. J., Urban, L. S., & Glass, L. A. (2010). Correlations between estimates of state IQ and FBI crime statistics. Personality and Individual Differences, 48(5), 579–583. doi:10.1016/j.paid.2009.12.010 

Beaver, K. M., & Wright, J. P. (2011). The association between county-level IQ and county-level crime rates. Intelligence, 39(1), 22–26. doi:10.1016/j.intell.2010.12.002 

Beggs, S. M., & Grace, R. C. (2008). Psychopathy, intelligence, and recidivism in child molesters: Evidence of an interaction effect. Criminal Justice and Behavior, 35, 683–695

CROCKER, A. G., & HODGINS, S. (1997). The Criminality of Noninstitutionalized Mentally Retarded Persons. Criminal Justice and Behavior, 24(4), 432–454. doi:10.1177/0093854897024004003

Cullen, Francis T., et al. “Crime and the Bell Curve: Lessons from Intelligent Criminology.” Crime & Delinquency, vol. 43, no. 4, 1997, pp. 387–411., doi:10.1177/0011128797043004001.

Cymbalisty, B. Y., Schuck, S. Z., & Dubeck, J. A. (1975). Achievement level, institutional adjustment and recidivism among juvenile delinquents. Journal of Community Psychology, 3, 289 – 294.

Denno, D. W (1990). Biology and violence: From birth to adulthood.
Cambridge, England: Cambridge University Press.

Diamond, B., Morris, R. G., & Barnes, J. C. (2012). Individual and group IQ predict inmate violence. Intelligence, 40(2), 115–122. doi:10.1016/j.intell.2012.01.010

Ellis, Lee, et al. (2019) Handbook of Crime Correlates. Academic Press, an Imprint of Elsevier.

Ellis, Lee et al. (2009) Handbook of Crime Correlates. Academic Press, an Imprint of Elsevier.

Ellis, L., & Walsh, A. (2003). Crime, Delinquency and Intelligence. The Scientific Study of General Intelligence, 343–365. doi:10.1016/b978-008043793-4/50054-4

Gabrielli, W. F., & Mednick, S. A. (1980). Sinistrality and delinquency. Journal of Abnormal Psychology, 89, 654 – 661.

GENDREAU, P., LITTLE, T., & GOGGIN, C. (1996). A META-ANALYSIS OF THE PREDICTORS OF ADULT OFFENDER RECIDIVISM: WHAT WORKS!*. Criminology, 34(4), 575–608. doi:10.1111/j.1745-9125.1996.tb01220.x

Goddard, Henry Herbert. (1914) Feeble-Mindedness: Its Causes and Consequences. Nabu Press

Herrnstein, R. J., & Murray, C. (1994). The bell curve: Intelligence and class structure in American life. New York: Free Press.

Hodges, K., & Plow, J. (1990). Intellectual ability and achievement in psychiatrically hospitalized children with conduct, anxiety, and affective disorders. Journal of Clinical Psychology, 38, 830 – 837.

Holland, T. R., Beckett, G. E., & Levi, M. (1981). Intelligence, personality, and criminal violence: A multivariate analysis. Journal of Consulting and Clinical Psychology, 49, 106 – 111.

Holland, T. R., & Holt, N. (1975). Prisoner intellectual and personality correlates of offense severity and recidivism. Journal of
Clinical Psychology
, 31, 667 – 672.

Kandel, E., Mednick, S. A., Kirkegaard-Sorensen, L., Hutchings, B., Knop, J., Rosenberg, R., & Schulsinger, F. (1988). IQ as a protective factor for subjects at high risk for antisocial behavior. Journal of Consulting and Clinical Psychology, 56(2), 224–226. doi:10.1037/0022-006x.56.2.224

Lynam, D., Moffitt, T., & Stouthamer-Loeber, M. (1993). Explaining the relation between IQ and delinquency: Class, race, test motivation, school failure, or self-control? Journal of Abnormal Psychology, 102(2), 187–196. doi:10.1037/0021-843x.102.2.187 

Magdol, L., Moffitt, T. E., Caspi, A., & Silva, P. A. (1998). Developmental antecedents of partner abuse: A prospective-longitudinal study. Journal of Abnormal Psychology, 107, 375 – 389.

McDaniel, M. A. (2006). Estimating state IQ: Measurement challenges and
preliminary correlates. Intelligence, 34, 607–619.

Mears, Daniel P., and Joshua C. Cochran. “What Is the Effect of IQ on Offending?” Criminal Justice and Behavior, vol. 40, no. 11, 2013, pp. 1280–1300., doi:10.1177/0093854813485736.

Moffitt, T. E., Gabrielli, W F, Mednick, S. A., & Schulsinger, F. (1981).
Socioeconomic status, IQ, and delinquency. Journal of Abnormal
Psychology
, 90, 152-156

Moffitt, T. E., & Silva, P. A. (1988). IQ and delinquency: A direct test of the differential detection hypothesis. Journal of Abnormal Psychology, 97, 330 – 333.

Murchison, C. (1926). Criminal intelligence. Wisconsin, Wisconsin University Press

Okami, P., & Goldberg, A. (1992). Personality correlates of pedophilia: Are they reliable indicators? Journal of Sex Research, 29(3), 297–328. doi:10.1080/00224499209551651

Oleson, J. C., & Chappell, R. (2012). Self-Reported Violent Offending Among Subjects with Genius-Level IQ Scores. Journal of Family Violence, 27(8), 715–730. doi:10.1007/s10896-012-9468-7

Pesta, B. J., McDaniel, M. A., & Bertsch, S. (2010). Toward an index of wellbeing for the fifty U.S. states. Intelligence, 38, 160−168.

Rushton, J. P., & Templer, D. I. (2009). National differences in intelligence, crime, income, and skin color. Intelligence, 37(4), 341–346. doi:10.1016/j.intell.2009.04.003

Spellacy, F. (1977). Neuropsychological differences between violent and nonviolent adolescents. Journal of Clinical Psychology, 33(4), 966–969. doi:10.1002/1097-4679(197710)33:4<966::aid-jclp2270330409>3.0.co;2-4

Stark, R. (1975). Social problems. New York, Academy Press.

E.H. Sutherland. ‘Mental Deficiency and Crime” In K. Young (ed) Social Attitudes. New York: Henry Holt (1931): 357-375.

Templer, D. I., & Rushton, J. P. (2011). IQ, skin color, crime, HIV/AIDS, and income in 50 U.S. states. Intelligence, 39(6), 437–442. doi:10.1016/j.intell.2011.08.001

Ward, D. A., & Tittle, C. R. (1994). IQ and delinquency: A test of two competing explanations. Journal of Quantitative Criminology, 10, 189 – 212.

West, D. 1, & Farrington, D. P. (1973). Who becomes delinquent? London: Heinemann.

Wolfgang, M. E., Figlio, R. M., & Sellin, T. (1972). Delinquency in birth cohort. Chicago: University of Chicago Press.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s