03 February 2009

Top Ten Risk Factors For Crime

The Archives of General Psychiatry has published a ranked list of the top ten factors that predict a person may engage in future violent behavior. They are, in order:

1. Younger age.
2. History of violence.
3. Male gender.
4. History of juvenile detention.
5. Divorce or separation in the past year.
6. History of physical abuse.
7. Parental criminal history.
8. Unemployment in the past year.
9. Severe mental illness with substance abuse.
10. Being a crime victim in the past year.

Mental illness alone is not a meaningful predictor of future violent acts, but is very significant when accompanied by a history of violence and substance abuse.

There were 3,089 people deemed to have severe mental illness—schizophrenia, bipolar disorder and major depression—but no history of either violence or substance abuse. They reported very few violent acts, about 50, between interviews.

But when mental illness was combined with a history of violence and a history of substance abuse, as in about 1,600 people, the risk of future violence increased by a factor of 10.


The methodology has strong points, but also limitations.

[T]he researchers analyzed data from the National Epidemiologic Survey on Alcohol and Related Conditions. The original survey in 2001-2002 involved more than 43,000 face-to-face interviews with a representative sample of American adults. Three years later, many of the same people, more than 34,000, were interviewed again.

Questions about violence in both interviews included:

—"Ever use a weapon like a stick, knife or gun in a fight?"

—"Ever hit someone so hard that you injured them or they had to see a doctor?"

—"Ever start a fire on purpose to destroy someone's property or just to see it burn?"

—"Ever force someone to have sex with you against their will?"

The large national survey, conducted by the National Institute on Alcohol Abuse and Alcoholism, included people living in shelters, hotels and group homes, as well as houses and apartments, but it didn't include people living in hospitals, jails or prisons.


While the sample size was admirably large, the exclusion of people who are institutionalized or incarcerated excludes an extremely important subset of the total sample for the questions the study seeks to address.

Institutionalization in a mental health hospital frequently involves a situation where someone has a severe mental health issue and is a physical threat either to others or themselves, which creates a much greater likelihood that there is a history of violence; mentally ill people who act mildly don't tend to be institutionalized, particularly in the current era where hospitalization is used as an absolute last resort.

The incidence of mental health problems and of substance abuse problems among incarcerated people is extremely high, with an overwhelming majority of inmates in prisons suffering from one, the other, or both. In Colorado:

Moderate to severe substance abuse is a problem for 82.0% of male and 82.4% of female inmates. Moderate to severe mental health problems exist for 27.8% of male and 34.2% of female inmates.


(The gender of those incarcerated for violent crimes in Colorado corroborates the important role of gender in violent crime, by the way. About 96% of inmates incarcerated in Colorado for violent crimes are men.)

An incomplete sample doesn't mean that the study is useless, of course. It simply means that to get a complete sample, the data need to be integrated with studies of institutionalized mental patients and incarcerated persons to provide a comprehensive picture. The institutionalized population at any given point in time is also much smaller than the incarcerated population at any given point in time.

The reliance on self-reporting of violent acts is also potentially an important flaw. It may tend to underreport liars who are often the worst violent criminals, while overreporting exaggerators. Since there is failure good evidence from criminological studies that a large share of people who are prone to engage in violent act eventually end up involved with the criminal justice system with arrest, prosecution and conviction records, it may be possible to make reasonable estimates of the degree to which people in the survey were truthful with researchers.

Finally, one is concerned, in a study like this one, that people who are at risk of future violent acts may be disproportionately represented among the roughly 17% who couldn't be tracked down for a second round of interviews. One can control for that risk a little, by comparing the information one has about the two time reporters with the information one has about the one time reporters who didn't have a second interview, with bias from the missing data points likely to be smaller if the two groups are otherwise similar. But, there is no completely effective way to remove this concern when the target of violence prone people is a very small and behaviorly distinct subset of the entire sample.

Still, every social science study has some inherent flaws, which is why social scientists do a myriad of studies in a myriad of research designs, in an effort to get at the truth. And, this study, for all its flaws, is probably one of the more reliable research designs out there. Many studies have much smaller and less representative survey sample sizes.

One surprise of the new study is that it did not identify education as a critical risk factor. As I noted in a previous post about the Colorado inmate population, education is a very powerful predictor of incarceration rates:

High school dropouts make up 41.8% of male and 42.5% of female inmates (about 10% of adult in Colorado in the general population are high school dropouts).

Just 0.7% of male and 0.9% of female inmates have an associates degree or higher level of education (about 36% of adults in Colorado have bachelor's degrees or higher levels of education, and one would estimate that another 8-12% have earned associate degrees in the state based upon national figures).

Put another way, in Colorado, a high school dropout is 3.2 times as likely to become a prison inmate as someone who finishes high school but doesn't earn a college degree, while someone who earns a college degree is about 50 times less likely to become a prison inmate than someone who finishes high school but doesn't earn a college degree. A high school dropout is roughly 160 times as likely to become a prison inmate as someone who earns a college degree. A male high school dropout is more than 1500 times as likely to become a prison inmate as a female with a college degree.


While having a degree certainly reduces your likelihood of being incarcerated after committing a criminal act, it is hard to believe that this education effect on the criminal justice process is as great as the numbers above indicate, or that education is fully subsumed in the other factors in the top ten list described above. Perhaps, however, education is less of an important factor in the kind of violent acts that are serious enough to be noted by the survey, but do not produce long term incarceration, than in the acts so serious that one ends up in prison for them.

No comments: