No doubt you’ve heard about the recent study showing that handgun owners are far more likely to commit suicide than non-owners. I was a little dubious about the reports, so I went to the source: the paper, Handgun Ownership and Suicide in California. Right off, I saw that one of the authors is Garen Wintemute, who has a long history of…questionable collection and use of data. This paper is no exception.
What Wintemute and associates claim to have found is . . .
Rates of suicide by any method were higher among handgun owners, with an adjusted hazard ratio of 3.34 for all male owners as compared with male nonowners (95% confidence interval [CI], 3.13 to 3.56) and 7.16 for female owners as compared with female nonowners (95% CI, 6.22 to 8.24). …
Handgun ownership is associated with a greatly elevated and enduring risk of suicide by firearm.
The researchers started by data mining decades of personally identifiable records — voter registration, Dealer Records of Sale (DROS), and mortality records. They traced individuals by location for 31 years, from January 1, 1985 to December 31, 2016. The idea was to match handgun buyers with suicides. And, sure enough, they found a strong correlation in their 2004-2016 study period.
Once they first threw out the inconvenient data.
The final analytic data set was at the person–period level. It excluded cohort members who had acquired one or more handguns before coming under observation during the study period and cohort members with missing census tracts or birth dates (Fig. S1). We also excluded observation time from registrants younger than 21 years of age, the minimum age for lawful handgun acquisition in California.
In other words, they excluded 1,313,028 living people who had already owned handguns for as long as 20 years without committing suicide. They only looked at new buyers, 676,425 people. They tossed nearly two-thirds of all handgun owners out of their data set because they hadn’t offed themselves yet.
Let’s say you have a pile of nickels and pennies, and you want to know what percentage of them were minted in a given year, say 2005. You sort the nickels and pennies into separate piles. You check nickel dates and find that 3% were minted that year.
Then you sort through your pennies, and toss out any that were minted before 2004. Now you check the dates and discover that three times (!) as many pennies as nickels were minted in 2005; 9%. Big surprise.
That, in effect, is what Wintemute and his fellow learned scholars did. To them, a non-suicidal gun owner is just a bad penny.
At most, they may have found a correlation between new or first-time handgun buyers and suicide. Maybe. But they are falsely presenting this as a correlation with handgun ownership.