In his guest post yesterday, Trent Steidley challenged the simplistic use of data from the National Instant Criminal Background Check System (NICS) as a measure of “gun sales.”
Today he critically engages the second of three common narratives emerging from the great gun buying spree of March 2020: that March 2020 saw the most guns sold in a single month in the history of NICS.
By Trent Steidley
NARRATIVE #2: March 2020 saw the most guns sold in a single month
This is true, at least for the numbers from NICS sales. But March 2020 is not a large increase considering the effect of population size and in relation to previous spikes.
When we look at the NICS data (corrected for sales, as noted in my first post) there are three clear spikes in the NICS sales data: December 2012, December 2015, and March 2020.
I’ll let you guess why these were peak months, but for the moment look at how similar they all are in their maximum. They all peak between 215,000 and 230,000. But the US population continues to grow over time. So, there are more potential buyers in 2020 than there were in 2012.
Does the 2020 peak really translate into the highest month ever per person? When we calculate the number of sales for every 1,000 people in the US, the rates are remarkably similar.
In the grand scheme of things, is it accurate to say that 6.95 guns for every 1,000 people is a huge increase compared to the 6.89 per 1,000 in 2012?
To give a sense of the practical effect, imagine a city with the same population of one million people in 2012 and 2020. The spike in 2012 amounts to 6,887 guns sold and the spike in 2020 amounts to 6,953 guns. So, the 2020 spike had 66 more guns sold than the previous spike for the entire million-person city.
It certainly seems that the March 2020 is technically unprecedented, but its effect is not great in magnitude.
About the Author
Trent Steidley is Assistant Professor in the Department of Sociology and Criminology at the University of Denver. His research seeks to understand how social movements, politics, and the criminal justice system interact to affect policy and criminal justice outcomes in the United States, particularly with regard to firearms. His current research specifically focuses on the determinants and consequences of concealed carry weapons laws in the United States, the determinants of police spending on military equipment, and the determinants of firearm demand in the United States.
Recent publications include:
Malone, Chad and Trent Steidley. 2019. “Determinants of Variation in State Concealed Carry Laws, 1970-2016.” Sociological Forum 34(2).
Steidley, Trent. 2019. “The Effect of Concealed Carry Weapons Laws on Firearm Sales.” Social Science Research 78:1–11.
Steidley, Trent. 2018. “Big Guns or Big Talk? How the National Rifle Association Matters for Conceal Carry Weapons Laws.” Mobilization: An International Quarterly 23(1):101–25.
Gun Curious 
Is there an uncertainty on those calculations, i.e., 95% c.l.?
LikeLike
They describe the population not a sample so no confidence intervals?
LikeLike
Parsing numbers is always greatly amusing, but every week in my CCW classes I use this data to inform students on how prevalent guns in the USA really are: specifically the week we broke the 500,000 NICS checks PER WEEK – and how CCW permittees and private sales are NOT accounted for in that stat. We have consistently been selling that many every week for a couple years now and it is a staggering number. I also point out that a while back, an ISLAND nation (Australia) of about 19 million people spent 2 YEARS stealing the guns from the population. In that time they stole about 650,000 guns. When people realize that dealers in the USA sold that many last week and the week before and next week and the week after that, they begin to see that we have a staggering number of firearms in the population. It generates a lot of wonderful conversations about various sub-topics.
LikeLiked by 1 person
David has the idea. You could estimate a confidence interval but I’m not sure what it would achieve. Given the fuzziness of the NICS data to proxy gun sales and the fact this is *the number* of NICS checks not an estimate inferred from a sample its fair to use just the number. If someone wants to do the t-test to compare 2012 to 2020 they could, I suppose.
LikeLike