Don't Blame the Haystack for the Needle

We need to do what we can to keep horrors like Sandy Hook—and Columbine and Virginia Tech and too many others—from happening again. Some commentators have wrongly suggested that identifying the potentially violent among the seriously mentally ill or even returning to the era of long-term institutionalization would be good policy. As a violence researcher and a psychologist, I am worried that the most politically palatable solution will not be common sense regulation of incredibly dangerous items but treating thousands of innocent people like would-be murderers. Although it sounds appealing, predicting horrific violence among the seriously mentally ill is statistically virtually impossible with the degree of accuracy needed. Worse, like any profiling, because we are talking about a large class of people—the vast majority of whom is entirely innocent and presents no danger—we risk wrongly stealing the civil liberties of hundreds of thousands of individuals.

Like earthquakes and giant meteors, we know the risk of another shooting is out there, but don't believe anyone who can tell you exactly who or when. The science of predicting rare human behavior is little better than our ability to say on which day or even which year the next big earthquake will strike California.

Why is this so? To understand this requires separating the human tragedy from the statistical problem. Even one mass shooting is obviously too many. It is hard to even imagine the suffering of the families in Newtown and my heart aches every time I think about those surviving children and families trying to put their lives back together.

However, that is not the math problem. The math problem is taking a bunch of very common personal characteristics and trying to predict behavior that is rare in relation to the more than 300 million people living in the United States. The class of potential perpetrators is large and most of their personal characteristics are common in the population. The mass shooters in the United States have so far all been male—but there are over 150 million in the country. They have so far all been adolescents or young adults. Statistically, this is a smaller pool, but this is still tens of millions of young men. Racial profiling is of course a sensitive issue, but if you wanted to profile mass shooters by race, then most have been young men of European descent. The public is opposed to these kinds of profiling with good reason—because almost everyone in these groups of people is innocent and will never commit even one murder, much less many. The same is true for the seriously mentally ill. There are hundreds of thousands of people who have or will have a serious mental illness and it is wrong to treat them all like potential shooters.

The other problem with profiling by psychological diagnosis is that, unlike the sex and age of these shooters, there appears to be a lot of variability regarding their psychological condition. As a licensed psychologist, I feel obliged to mention that I have not personally evaluated these people and many of them died on the day they came to notoriety with little formal treatment history. Nonetheless, while acknowledging the limits of available information, it still appears that some mass killers are not psychotic. (This is not to say that they are psychologically well adjusted.) For example, the Columbine shooters, Eric Harris and Dylan Klebold, are usually said to have been motivated by their experiences of being bullied.

With our current knowledge, violence prediction cannot be done without making many more errors than correct predictions. There are two kinds of errors. One is "false positives." These are the errors that lead to treating people more harshly than they deserve because they are said to belong to the dangerous group when they really do not. False positives can lead to forced institutionalizations and forced medications for people who are not really a risk to others. The other is false negatives—missing the people who everyone thought were so quiet and mild-mannered until the day they were not (but let's not lock up all the quiet and mild-mannered types).

Most people have experience with these diagnostic challenges. Many adults know someone who screened "positive" on a mammogram or diabetes screen or other medical test who turned out not to have cancer or diabetes. These screening tests are designed to cast a wide net and they identify a lot of lumps and blood sugar spikes that are not really serious medical problems. Those are the false positives. In medicine, the next step is usually to get a more precise test, such as a biopsy, that will provide a more definitive answer.

Those precise tests are what are missing in the prediction of homicide. We still have not identified specific, definitive signs of future danger—something more specific than being male or young or even seriously mentally ill. Often, these school shootings are the first act of violence—or at least the first act of serious violence—perpetrated by these young, disturbed men, so we do not even have a criminal history to make our predictions more precise. Even more problematically, these acts of mass murder are rare even within the criminal population. Most people who have a history of assault will never become a murderer, much less a mass murderer.

There are a lot of people who try to sell certainty. I would like to be able to identify people before they become killers as much as anybody. But we do not live in the universe of Minority Report. There is no Pre-Crime Unit that can identify murderers before they strike. (John Anderton, the character played by Tom Cruise in the movie version, is an example of a false positive). We need other solutions, solutions that apply the same rules fairly and equally to everyone. Mental health care can be improved and access made easier without going back to the days of never-ending institutionalizations. Mental health initiatives should not be limited to the seriously mentally ill but also focus on interrupting the intergenerational transmission of violence and teaching kids alternatives to bullying. Mental health initiatives alone will not be sufficient. Tougher policy challenges need to be addressed, including serious discussions about the many proposals for gun safety and restricting children's access to the most violent media. It is up to us not to let our fear lead to the systematic oppression of hundreds of thousands of innocent people.