The next mass shooting will take place on February 12, 2014, in Spokane, Washington. It will be committed by an emotionally disturbed, 38 year-old white man who will kill seven people and wound six more at a place he used to work using a semi-automatic handgun he purchased legally in the state.
That, at least, is what a look at the data on past such shootings might indicate. We'll say at the outset: Every assertion in the first paragraph is a function of probability, not fact. The next mass shooting — which will happen somewhere, sometime — will almost certainly not be in that place at that time. But a look at the historic data on such killings, compiled and shared by Mother Jones magazine, makes each of those predictions defensible.
If President Obama's second term has been beset by "unpredictable calamities" like the Navy Yard shooting, in the words of the Washington Post's David Nakamura, we thought we'd try and offer a little prediction. Especially since, if such incidents occur at the same pace for the rest of Obama's term as they have since 2009, there could be 14 more before he leaves office.
[Update, Sept. 18: Read our follow-up note to the residents of greater Spokane (and elsewhere).]
Here's how we came up with our predictions.
Looking at the past 30-plus years of spree- and mass-shootings, this is the easiest trend to spot. Nearly two-thirds of the 67 incidents — 65.7 percent — were at the hands of a white person. Only one was committed by a female. It is indisputable: white men are most likely to commit such acts, though not exclusively.
Shooting at work
Most of the shootings in the database occurred at a place of business, either one where the shooter worked or a restaurant or other such facility. Workplace-related shootings are twice as common as school shootings, for example, which has implications for other demographic predictions. The largest category of shootings is "Other," with nearly half of all shootings taking place at some public location. Workplace shootings comprise just under a third of incidents. Of all of our predictions, this one is the most tenuous.
38 years old
By focusing on the workplace as the site of the incident, it allowed us to better refine the age of the shooter. School shootings are, somewhat obviously, undertaken by younger people. Narrowing the possible venues for an incident to workplaces and other venues, we were able to calculate the average age of men committing such crimes: 38. Include school shootings, and that age drops to 35.6.
While mental health issues are often tricky to diagnose, it's clear that there exists a correlation between a pattern of mental illness and involvement in mass shootings. Those asked after the fact are probably more likely to have seen signs of mental illness, of course; killing a dozen people tends to taint past perceptions. The Mother Jones index has documented examples of treatment attempts, for example, lending as much objectivity as possible. And it shows that about 63 percent of those involved in mass shootings had some history of mental health issues.
Legally-purchased handgun purchased in-state
The vast majority of guns used in mass killings were obtained legally — 81.8 percent. Advocates of gun control will note that this bolsters the case for tighter restrictions, particularly given the overlap between those with mental health issues and those able to buy guns. Opponents of gun control will note that nearly a fifth of shootings used illegally obtained guns, suggesting that new controls won't prevent such shootings.
We were curious about where the guns were obtained, envisioning a possibility that those who live in states or locations with tighter gun control laws might go to another state to buy a weapon (as the shooter in the Washington, D.C., Navy Yard attack did). Almost three-quarters of the time, the guns were purchased in the same state as the incident.
As for the type of weapon, the database did most of the work on this, too. Not every shooter used a semi-automatic handgun, but at each of the workplace or "other" events, such a weapon was the most common one present.
This was one of the more complex calculations. In order to calculate it, we first wanted to figure out how frequently shooting incidents happened. So we figured out how many days passed, on average, between them. Over the past 30 years, that figure is about 222 days — seven months or so. But as we noted on Monday, the pace of shootings has increased dramatically recently. Since the first shooting in 2004, incidents have occurred about once every 149 days. (This is actually less frequently than they've occurred since 2009 alone. Since Obama became president, there's been a shooting every three months, as noted above.)
Which to use? Extrapolating from the Washington, D.C., shooting, the first value gave us a date of April 27, 2014. The latter gave us February 12. We were inclined to accept the February figure but, to make sure, calculated the months in which such shootings were most likely to take place. (See chart at right.) There have been six events in February and four in April. The former, then, is more historically likely. So February it was.
We noticed, in looking at the data, that there was some correlation between the number of incidents in a state and its population — in other words, there were more incidents where there were more people. The most incidents occurred in the most populous state, California, which makes sense.
But there were anomalies. Specifically — Florida had the second-most number of shootings despite being fourth on the population list. So, using California's seven incidents as the peg, we calculated how many incidents would be expected in a state if they were tied solely to population. Florida, with a population of about half California's, would be expected to see about half as many shootings by that standard. The graph at left shows the difference between the number of shootings and the expected number of shootings. Washington state has had far more shootings than its population would predict — five incidents where 1.27 might have been expected. Illinois is on the other end of the spectrum, with only one incident compared to the 2.37 that population would suggest. So we focused on those two states as anomalies in the data. Would the next shooting happen where they happened more than expected or less?
To determine the answer to that, we went back to our other demographics. First, we used Census data to figure out what states had the most men aged 18 to 45 (the smallest category the Census offers) as a function of total population. Washington state ranks ninth in the number of people in that category; 18.46 percent of its residents are men in that age group. Illinois came twelfth. (California, we'll note, came third.)
Then there was mental health. The Centers for Disease Control tracks mental health data by region. We looked at annual data for "[s]erious psychological distress in the past 30 days among adults aged 18 and over" by region, as at left. The blue bar indicates the value for the percentage of the population in each region that suffered such distress in 2010-2011. The red bar, however, shows how that changed between 2009-2010 and 2010-2011. As you can see, the "West" region saw a dramatic increase.
We considered other factors, too, such as the recent report linking gun ownership to gun homicides. But those data correlated more strongly to non-mass-shootings. We were therefore willing to pick Washington as the most likely state.
In order to determine the city, we went back to population data. Calculating the average population of each location where these incidents occurred gave us an average population of about 214,000. The population of Spokane is about 209,000. (The city in Illinois closest to that figure, we'll note, is Aurora, which has about 200,000 people.)
Sadly, this was the easiest figure to develop. At the incidents we focused on, an average of 7.6 people were killed and 6.5 injured. The 67 incidents have seen over a thousand people killed or wounded.
Our sincere hope is that every prediction we made is wrong because no mass killings should happen again. The probability of that happening is not statistically significant.
Photo: Richard Farley was 39 in 1988 when he killed seven people at his former employer. (AP)