October 27, 2023

Content warning: This post analyzes suicide and mass shootings in a dispassionate manner.

A few years ago, I was watching the news program, Democracy Now!, which was doing a segment on suicide prevention. They brought on Dr. Kelly Posner Gertenhaber, a psychiatrist from Columbia University, who opened with a seeming whopper of a statement: because 9 out of 10 mass shooters have suicidal issues, it is imperative to assess people for suicide in order to stop the next mass shooting. Keep in mind that the doctor said this directly following a news segment on two Stoneman Douglas High students who died by suicide that week after surviving the 2018 mass shooting.

But that’s not even the main reason why this statement was so egregious. It was the blatant abuse of statistics in service of ableism that I found truly appalling.

First, a little background. Almost 10% of adults experience suicidal ideation or suicidality during their lifetimes, and about 4.3% of the adult population experiences suicidal ideation in a given year. In the United States, these statistics mean that approximately 11 million U.S. adults have recently had or are currently experiencing suicidal thoughts.

Now, it’s also important to understand a little bit about conditional probability, because the doctor made a common error that often, perhaps intentionally, leads to erroneous conclusions. By conditional probability, I mean any statement of the form P(A|B) that gives the probability of some outcome given that another event has occurred. For example, we can use conditional probability to determine the chance, P(A|B), that someone will be a mass shooter (event A) given that they are suicidal (event B), if we know both the probability that someone is a mass shooter and suicidal, P(A&B), and the probability that someone is suicidal, P(B). Knowing P(A|B) would be potentially useful for prevention efforts, but despite the unfortunately increasing frequency of mass shootings in the United States, we don’t have reliable data on P(A&B) to calculate P(A|B). However, that’s not even the main problem with the doctor’s statement.

The main problem with the doctor’s statement is that she reports that “90 percent of shooters have suicidal issues,” which is a conditional probability of the form P(B|A), or the equivalent of the statement that the probability that someone is suicidal given that they are a mass shooter is 90 percent. This is not at all equal to P(A|B), but it’s a common misconception that the two statements mean the same thing, which is known as the inverse fallacy. For this reason, I don’t hold it against the hosts of Democracy Now! for not pushing back against this misinformation, as it is easy to be confused by statistics.

The reason why P(B|A) or that “90 percent of shooters have suicidal issues” is a misleading statement in this context, is because it completely ignores P(B). Let me explain visually.

If I told you that “90 percent of shooters have suicidal issues,” you would probably envision some type of Venn diagram that looks like this:

Where Yellow represents the set B of people who are suicidal, and Blue represents the set A, people who are mass shooters. The overlapping region, Green, represents the subset A & B, which is people who are suicidal and mass shooters. The probability P(B|A) would be calculated by dividing the size of subset A & B (the Green region) by the size of set A (Blue and Green regions). (Notice how set B (Yellow region) is excluded from the equation.) Unfortunately, when the public thinks about these statistics, this is the image that they have in their heads, in no small part because of how the media disingenuously reports on these issues by making salient the fact that many mass shooters appear to have “mental health problems.” And you too would probably think that the statement P(B|A) is very convincing, because the set of people who are mass shooters (Blue, set A) seems almost to completely overlap with the set of people who are suicidal (Yellow, set B). However, the reason why this is incorrect is because the set B is actually exponentially larger than the set A. Here is what the Venn diagram might actually look like:

The above diagram reveals why we don’t need to know the actual value of probability P(A|B) to know that it is a fallacy that people who are suicidal are at risk for committing mass shootings, because we know that the amount of people who are suicidal and don’t commit mass shootings (Yellow region) is exponentially larger than the amount who do (Green region). Unfortunately, the media and law enforcement will continue to scapegoat marginalized people in order to push for more surveillance and reporting tools that don’t work, and which will instead be used to codify violence towards marginalized groups, especially Black and Brown people.

Now, what about the supposed correlation between mental illness and mass shootings? If we look at the statistics, over 20% of adults, or approximately 52 million U.S. adults, are currently experiencing mental illness, and based on two recent studies, it is estimated that between 50% to 80% of the population will experience mental illness in their lifetimes. So the probability that someone will be a mass shooter given that they are experiencing mental illness is even smaller than the already small probability that someone will be a mass shooter given that they are suicidal.

It is time to look beyond mental illness as the reason for America’s violence problem.

**For further reading:**

Stop blaming school shootings on bullying, by Srishti Bungle, New York University, December 8, 2021

Racism isn’t a mental illness and bigotry isn’t a disability, by Moirha Smith, University of Vermont, September 9, 2020

Gun violence rhetoric stigmatizes disability, by Jesser Horowitz, Vassar College, October 4, 2017

Obama-era gun control legislation fuels stigma against disability and disorder, by Thomas Finn, University of California, San Diego, March 6, 2017