NRA-ILA | That’s Not How Research Works

The NRA calling research papers “anti-gun” may sometimes come across as dismissive or reflexive, or both. Some researchers work to keep their papers sterile, no matter their findings. Others grant themselves leeway to be more bombastic, particularly when it comes to developing theories that explain their findings. Researchers from the University of Central Missouri and the University of Alabama at Birmingham posted a paper online that makes their anti-gun slant abundantly clear through both their text and their model specifications.

David B. Johnson and Joshua J. Robinson posted their paper, “Gun Dealer Density and its Effect on Homicide,” online earlier this month. It has not been peer-reviewed or published in a journal. Johnson and Robinson make several emotional, hyperbolic claims about gun ownership and gun dealers that would likely not be accepted if written in a research paper turned in by a student. The variables excluded from their preferred models initially suggest unfamiliarity with crime data or research, but the use of some variables in later models shows the exclusion was by design. The variables that were included in their preferred model suggest problems with specificity and the exaggeration of small changes in rates shows a commitment to their narrative over sincere analysis.

Let’s start at the beginning of this 32-page paper – a count that excludes the abstract, references, and appendices. In the first paragraph, Johnson and Robinson utilize the tragedy at Sandy Hook to categorize the difference in the number of “gun-related deaths” per year between the individual years 2000 and 2019. The authors make no mention of the fact that most “gun-related deaths” are self-inflicted, or that the population of the U.S. grew by about 50 million people between those years. They also forgot to mention that the homicide rate was lower from 2009 through 2014 than it was in 2000 but that’s a detail that doesn’t promote gun grabbing.

As is the state of existing research: “Much of the literature on firearms – particularly concerning its connection to homicide and crime – is full of null and mixed results.” Johnson and Robinson set out to resolve that unacceptable problem, which just must be the result of a misunderstanding of the relationship between gun availability and homicides, by “creating” a new metric: the number of federal firearm licensees per mile. They do not differentiate between types of FFLs or volume of sales, so a small home-based FFL that transfers a few guns a year is treated the same as a large-volume retailer like Cabela’s.

Instead, they correlate the number of FFLs per mile with the number of NICS checks. NICS checks, readers may know, are the background checks dealers are required by law to run before a sale. The authors correlated required background checks with those required to run background checks. Groundbreaking.

To validate their measure of gun dealer density, they compare metrics using heat maps. Theirs is the only metric that shows north-central Colorado with a high density, and it includes several communities that suffered mass murder attacks, so theirs must be the best. These professors actually used a small number of rare, high-profile incidents to validate a measure they’re using in an analysis spanning decades and the entire contiguous United States. That isn’t a validation – it’s a confirmation of bias.

Alaska and Hawaii are excluded for some reason. Oh, and any year prior to 2003 is also excluded. Their key variable is lagged, for some reason. The authors claim that NICS data is available from 2004 on, but the FBI has data from November 1998 through the present readily available. Also readily available are variables known to be associated with crime in generals and homicides specifically – variables like law enforcement resources, arrest rate, crime itself, poverty rate, unemployment rate, and alcohol consumption.

They tout the effect of gun density they supposedly found, but this model found that homicide increases as income increases and decreases as the percent of men in a population increases.

Those are backwards. Think about it. Men commit most homicides, and crime is known to be associated with lower incomes or poverty. It does not make sense for there to be less homicide as there are more men in a community or there to be more homicide as income increases.

Those are the results of the choice the researchers made, and they clearly made those choices to find a connection between firearms and a negative outcome.

They got what they wanted, but it’s a sadly transparent effort.

The authors seem to believe that “secondary markets” in places like Chicago contribute to violent acts, despite all guns sold by an FFL requiring a background check whether the gun is new or used. They also think that the ATF-reported “time to crime” that shows criminals use a firearm, on average, more than 10 years after it was legally sold, is irrelevant because gun dealers “regularly sell used guns.” Dealers are required to run a NICS check on used guns, too, and criminals don’t buy their guns from dealers. Johnson and Robinson are aware of this research, but it doesn’t support an anti-gun narrative so it’s discounted.

Comments on this paper could easily fill more space than we have. We’ll leave readers with this statement from the authors: “The decrease in the percentage of corporate retailers in these communities also may indicate an increase in the percentage of nearby dealers willing to bend or break federal gun laws.”

That’s not a good way for FFLs to stay in business – and nothing invites ATF and FBI scrutiny quite like breaking federal law.

The researchers may be well aware of that fact and willing to ignore it. They’ve certainly demonstrated their willingness to ignore reality throughout their paper.