As part of the gun control debate, Health and Human Services Secretary Kathleen Sebelius recently spoke on the connection between mental illness and crime. She offered this factoid:
The vast majority of Americans with a mental health condition are not violent. In fact, just 3% to 5% of violent crimes are committed by individuals who suffer from a serious mental illness.
That sweeps aside the obvious link between suicide and mental illness, but let's press on with what has become a new but erroneous progressive talking point - here is the National Journal telling us that the focus on mental health is misguided:
Perhaps most important, although people with serious mental illness have committed a large percentage of high-profile crimes, the mentally ill represent a very small percentage of the perpetrators of violent crime overall. Researchers estimate that if mental illness could be eliminated as a factor in violent crime, the overall rate would be reduced by only 4 percent. That means 96 percent of violent crimes—defined by the FBI as murders, robberies, rapes, and aggravated assaults—are committed by people without any mental-health problems at all.
And - finally - we see an example of this statistic buttressed by a source; here is The American Prospect:
The stereotype that the mentally ill are very violent is simply incorrect. According to the National Institute for Mental Health, people with severe mental illness, like schizophrenia, are up to three times more likely to be violent, but “most people with [severe mental illness] are not violent and most violent acts are not committed by people with [severe mental illness.]” On the whole, those with mental illness are responsible for only 5 percent of violent crimes.
And now we get to the truth. Following the link to the Institute of Medicine ("only 5 percent of violent crimes") and downloading "Improving the Quality of Health Care for Mental and Substance-Use Conditions: Quality Chasm Series, 2005", we find this (my emphasis):
The first large-scale epidemiological data on the prevalence and incidence of violence (assaultive behavior) among individuals with M/SU ill- nesses were produced in the early 1980s as a part of the Epidemiological Catchment Area (ECA) study, which was designed primarily to determine the prevalence of untreated psychiatric illnesses in community populations across the United States.16 A secondary analysis of these data (Swanson, 1994) found that the vast majority of individuals with mental illness who had not qualified for a substance-use or -dependence diagnosis in the past year were not violent.
Even among individuals with major mental illnesses (such as schizophrenia) having no co-occurring substance-use diagnosis, the proportion committing an act of violence was only somewhat higher than that in the population without mental illness. Only about 7 percent of those with a major mental illness (but without a substance-use or dependence diagnosis) had engaged in any assaultive behavior in the preceding year, compared with slightly more than 2 percent of individuals without any major psychiatric diagnosis.
Individuals with less-severe mental illness were at no greater risk of committing an act of violence than those with no mental illness. Because major mental illness is a relatively rare occurrence, individuals with mental illnesses (but without a substance-use or -dependence diagnosis) account for a very small proportion (about 3–5 percent) of the risk of violence in a community.
Substance-use illnesses by themselves and in combination with major mental illnesses were found to be related more strongly to violence. The ECA study found a 1-year violence prevalence rate of 19.7 percent among respondents with a substance-use or -dependence diagnosis without the presence of a major mental illness, and rate of 22 percent among those with dual mental and substance-use or -dependence diagnoses. Individuals with substance-use or -dependence diagnoses alone represented 26–27 percent of the risk of violence in the community, while those with both diagnoses contributed a much smaller share of the risk (5–6 percent) because of their smaller numbers.17
That is heavy going, but my reading is that people with just mental illness account for roughly 3-5 percent of all violent crime; people with a dual diagnosis of mental illness plus substance abuse account for another 5-6 percent of violent crime. Taken together, people with mental illness account for 8-11 percent of violent crime.
Having said that, let's note that the Prof. Swanson of the 1994 paper under discussion is still writing, and says this:
Epidemiological studies in the community have found that the vast majority of people with serious mental illnesses do not commit violent acts toward others, and that the vast majority of violent acts are not attributable to mental illness (Fazel & Grann, 2006; Swanson, 1994). These studies would suggest that even if we completely eliminated mental illness as a violence risk factor, the population prevalence of violent acts towards others would go down by less than 4 percent.
Having paged through his 1994 paper by way of Google Preview, I see that he attributes 3-5 percent of violence in a community to the mentally ill-only, and another 4.8-5.7 percent to the dual diagnosis group(mentally ill plus substance abuse). I don't see any explanation for dropping the dual group from the total but I do see the opposite - he cites an earlier paper in which he argued against counting mentally ill substance abusers as 'false positives' for mental illness. Haven't tracked that down yet.
Mental Illness Policy Org. (founder) claims to provide fair coverage. They have this summary on the topic, which includes a citation ("The Economic Costs of Mental Illness", H. Harwood, A. Ameen, G. Denmead et al) that leads to the Lewin Group report, p. 5-17 (my emphasis):
Research to date strongly indicates that it is the mental disorders involving psychoses, such as schizophrenia, paranoia, and bipolar disorder (termed severe and persistent mental illness in this study), that involve increased risk of violence. The most recent research has found that among the SPMI population the vast majority of violence is among the population that suffers from both SPMI and substance-related disorders.
The SPMI population without substance-related disorders may be responsible for no more than about 3 percent of violent crime, with 3 to five times as much violence accounted for by the dually-diagnosed (SPMI and substance disorders) population. Costs for the dually diagnosed are accounted for in studies of alcohol and drug abuse. The population experiencing other mental illness (not SPMI) present only a modest increase in risk of violence (about twice as great) compared to the population with no current mental illness.
The Mental Illness Policy Org. summarizes that as follows:
Using National Comorbidity Survey data, this report concluded that, for individuals with severe and persistent mental illness (SPMI), "the SPMI population without substance-related disorders may be responsible for no more than about 3 percent of violent crime, with 3 to 5 times as much violence accounted for by the dually diagnosed (SPMI and substance disorders) population" (section on crime, p. 1.5). The study also assumed that the percentages for homicides were the same as for "violent crime" (section on crime-related costs, p. 6.8). Thus, SPMI individuals with no substance abuse disorders were said to be responsible for no more than 3 percent of homicides, but individuals with SPMI and alcohol or drug abuse were responsible for between 9 and 15 percent of homicides.
So they would claim that the 'mentally ill only' group accounts for 3 percent of violent crime and the "mentallly ill plus substance abusing" accounts for 9 to 15 percent of crime, for a total of 12-18 percent of violent crime committed by the mentally ill.
Well. One study says 8-11 percent, another days 12-18 percent. Neither is suggesting that Kathleen Sebelius is near the mark with her estimate of 3-5 percent.
Which is disturbing. Progressives such as Obama will want to poo-pooh the mental health problem and focus on limiting the public's access to guns. Are they merely misleading the public, which would be troubling but not unheard of in our political class, or are they unaware of the truth and busily misleading themselves?
I'd hate to see our President start a war on guns based on phony intelligence.
AND AS THE TIDE RECEDES... These surveys were done decades ago and the homicide rate has fallen by roughly half since then. If (IF!) we have made equal progress on all fronts then the mentally ill will be committing a smaller absolute number of crimes and their proportion of the mix will be unchanged.
But that is an evidence-free assumption. Suppose we have made tremendous progress deterring and/or detaining common criminals and that the decline of the crack wars has reduced crime by substance abusers but we have made no headway on dealing with violence by the mentally ill. In that case, the absolute number of violent crimes committed bvy the mentally ill would be roughly unchanged but it would represent twice as high a proportion of total violent crime since the base is now so much lower.
In which case, even the baseline of 3-5 percent ov violent crime committed by mentally ill non-substance abusers is outdated and would need to be doubled to reflect our new, lower crime rates.
Which is it? It's interesting to see how the Reality-Based Community, lovers of science and evidence all, have such confidence in their current answers.
I SEE RIGHT THROUGH THIS STAT:
Amongst the analysis presented by the Mental Illness Policy group is this study of the Indiana prison population:
In the first large study carried out in the United States, it has been reported that 10 percent of all homicides are committed by individuals with schizophrenia, bipolar disorder, and other psychotic illnesses, most of whom were not being treated. The study was carried out by Jason Matejkowski, Sara Cullen, and Phyllis Solomon, social workers in the School of Policy and Practice at the University of Pennsylvania. It was published in a recent issue of the Journal of the American Academy of Psychiatry and the Law.
The authors identified everyone in the Indiana state prison system who had been convicted of homicide between 1990 and 2002, a total of 1,397 individuals. The records of a random sample of 723 of these were examined, of which 518 had sufficient information to ascertain whether or not they had received a psychiatric diagnosis. Among the 518 individuals convicted of homicide, 53, or 10.2 percent, had been diagnosed with schizophrenia (n=27), other psychotic disorders not associated with drug abuse (n=14), or bipolar disorder (n=12). An additional 42 individuals had been diagnosed with mania or major depressive disorder, for a total of 95 individuals out of the 518 studied, or 18.3 percent, having a psychiatric diagnosis.
They include this caveat:
It should be noted that the study included only those individuals who committed homicides and were sentenced to prison; it did not include individuals with severe psychiatric disorders who were found to be incompetent to stand trial or not guilty by reason of inanity [sic] and therefore committed to a psychiatric facility instead of prison.
A caveat they do not include - Adam Lanza, the Newtown shooter, won't be standing trial or going to death row. Nor will the Clackamas shooter.
The NY Times surveyed their archives back in 2000 and identified 102 killers in 100 "rampage" attacks going back fifty years. Among their conclusions:
They do not try to get away. In the end, half turn their guns on themselves or are shot dead by others. They not only want to kill, they also want to die.
Acute paranoia, delusions, and depression were rampant among them, with at least 36 of the killers committing suicide on or near the scene. (Seven others died in police shootouts they had little hope of surviving, regarded by some experts as "suicide by cop.")
That is roughly a 70% rate of cheating the hangman. Surveying prisons does not provide a complete picture of the violent tendencies of these killers.
SINCE YOU ASK: Clayton Cramer had an eye-opening letter to Sen. Cruz about the link between deinstitutionalization and violent crime. Here is a 2005 article and a speech by Mr. Cramer on the same topic.