STUMP » Articles » Misleading Gun Death Stats: "Kids'" gun deaths vs. motor vehicle deaths » 31 May 2022, 15:27

For obvious reasons, we’re seeing a bunch of people throwing around gun death statistics.

Unfortunately, some are misleading.

Guns kill more “kids” than motor vehicles do

If you define “kids” as being age 1-19 years old.

I generally don’t.

The original source of this should be like me, that is, from people who know how to get info out of CDC WONDER. That means they are being deliberately misleading in doing this statistic.

Or, maybe they’re being lazy or low-skilled. This other possibility is because they went with the 5-year age groups, which is how you lump in adults with kids. I mean, if they wanted to say “well, 18- and 19-year-olds are adolescents!”, they should have gone whole hog and included the 20-24 age group, where the peak of gun homicides occur. Of course, people may have noticed the game at that point.

I will do a later post where I will focus simply on causes of death for children (and I will be including ages 15-19 in a young adult analysis, not with the children), but for now, let’s look at this purported statistic and see what is misleading about it.

Twitter examples and original graph

Here are a couple example tweets:

Has anyone seen this study from the New England Journal of Medicine on the increase in gun violence in 2020?

In 2020, firearm deaths overtook car accidents as the most likely cause of death for a child 19 and under.https://t.co/vClw6E55Fn

— NYC Angry Mom (@angrybklynmom) May 27, 2022

May 19, 2022 – NEJM

"Firearm-related injuries were second only to motor vehicle crashes as the leading cause of death among children/adolescents. Since 2016, that gap has narrowed. In 2020, firearm-related injuries became the leading cause of death."https://t.co/SQdc9UZwpr pic.twitter.com/pv5wwVbTal

— Aaron Richterman, MD (@AaronRichterman) May 25, 2022

Here is the graph if the tweets didn’t work for you:

To say I’m unimpressed with graphs where they used Excel defaults is an understatement.

Why did they exclude infants?

You may wonder why they started at age 1, and not at birth.

Deaths from age 0-1 are considered infant mortality, and if you want to make a big deal over gun deaths being the top cause of death for children, you can’t include infants… otherwise infant mortality causes will swamp all the older age groups, yes, even the older teens.

More babies die than the whole age 15-19 group in a usual year. Heck, more babies die than those age 20-24. That’s just by number.

If we went by death rate, you’d have to get into middle age before you ran into age levels that were comparable in death rates to infant mortality rates. There is a reason I don’t like talking about infant mortality. It’s pretty unpleasant, even if it is a lot lower than it used to be, even 20 years ago.

Very few babies are killed by either guns or motor vehicle accidents, so for this analysis to work, they have to exclude them. It is mainly physiological reasons that babies die, especially given how the U.S. counts infant mortality compared to other nations.

So, they did know that there were some very different mortality risks in the span of the 0-19 years old age, but there are also some very different mortality risks within 1-19. So let’s get back to that.

Fixing the graphs and splitting out ages

I’m just going to show you how gun deaths break out by the standard “5-year” age groups: (1-4 years, 5-9, 10-14, 15-19).

Also, motor vehicle accidents, which was the prior number one cause of death for age 1-19, if you grouped them all together. These are large causes of death for all age groups.

First, a redo of their ugly graph:

You can see that this is more a story of the improvement in traffic safety over the decades, though yes, you can see gun death rates have been increasing since 2015.

Now to do an age breakout of gun deaths:

And motor vehicle deaths:

In both cases, the primary contributor to deaths are the age 15-19 group. Now, I can actually get statistics by single year ages, and remove ages 18 and 19, but I’m not going to do that for this post.

There is a useful point in the original letter being linked: there is no reason to believe that the drug ODs (which also jumped up), the motor vehicle deaths, or the gun deaths will go down even if COVID-19 deaths decrease.

They call firearm deaths a preventable cause of death, and that might be true, but looking at the trends, it seems to me that the actual decrease in motor vehicle deaths show that was really preventable. The gun deaths look to have had a stubborn floor. Maybe think about that one a little more. (and yes, yes “The only acceptable rate is zero!” is all very well, but it seems to me “firearm violence” is really outside the CDC’s purview.)

Using statistics in argumentation: know the data and stop abusing the data

In some cases, it’s people who don’t know any better who have been flogging the graph, and others who have redone it to make it look better (like Financial Times.)

But many others had to know it was deliberately being misleading. What was the point of folding in the age 1-9 group, for instance? They basically contributed nothing. Age 10-14 did see a step up in both motor vehicle deaths and gun deaths, so maybe yes, keep that there… but maybe not.

But the people who wrote this letter should know better than to combine over such a wide range of ages, especially since so few children (pre-teens) die from anything and it’s mainly the older teens and young adults who are dying from external causes. This was true pre-pandemic, and yes, it got worse during the pandemic.

They had to know that throwing in a low-risk group with a much higher-risk group would be misleading.

SHAME.

via GIPHY

If you’re going to use statistics for your argument, while trying to hook into the emotion of “think of the children!”, my advice is to drop the stats.

Some of us know the relative risks to little children of all sorts of things.

My best advice is if you want to make an emotional argument, just make the emotional argument and eschew the statistics if you don’t understand them well. Statistics really aren’t all that convincing, especially if you have to torture them like that. People will notice.

Pathos, ethos, and logos are a very strong combination, but if you don’t have the logos, either learn it or just drop it.

And if you have to be misleading about the statistics….. don’t.

Related posts:

U.S. Gun Deaths are Mostly Suicides, not Homicides

Drug Overdose Deaths, Part 2: U.S. Age-Related Trends 1999-2020 with Provisional Results in 2021

Motor Vehicle Accident Deaths, Part 2: Age-Related Trends with Provisional Results in 2021

Homicide: Trends, 1968-2020, and Provisional Counts Through June 2021