Thread by @ProfPaulPoast

@neiltyson is challenging us to look at the data on gun violence.

Well, that's exactly what we do in my Quantitative Security course.

So, "what say the data?"

[THREAD]

https://twitter.com/neiltyson/status/1158074774297468928

To start, a bit of historical context. We should consider Louis Richardson's seminal JASA (@AmstatNews) article from 1949

He looked at "Deadly Quarrels".

These include everything from wars to murders, including mass murders

What he found was that most deaths were caused by a few events (namely massive wars). This led to his famous power law graph

As for murders themselves, he relied on criminal statistics. The quality of these data in 1949 are probably questionable.

Jump forward to the present.

The @UNODC collects data on homicides by country. The @UNODC is also aware of questions about data quality. This is why their GLOBAL STUDY ON HOMICIDE 2019 report contains a whole section on "Data Quality"

unodc.org/unodc/en/data-…

This map (page 15 of booklet 2) shows the data quality by country. You can see that the US is in the "good" category and most of North America, South America, and Europe are in the "good" or "fair" categories.

There are 5 criteria for determining data quality

So the US data are pretty good, but that's not the case for a number of countries.

Working with the data we have, what do we know?

Let's turn to the data collected by @IHME_UW

healthdata.org

More precisely, let's consider the data from their Global Mortality From Firearms, 1990-2016 report, available on the @JAMA_current site:

jamanetwork.com/journals/jama/…

Figure 11 from the report is quite telling.

The US has one of the highest gun-death by suicide rates in the world (2nd only to Greenland...so, the US is actually the highest).

While the US has a high gun-death by homicide rate, there are nations with higher rates.

But the countries with higher rates include (as best as I can tell) no advanced industrialized countries (here is a zoom-in on just the countries with higher homicide rates)

So that's not real encouraging if you live in the USA.

We shouldn't stop there. We could look at gun-homicide data differently.

@UofAlabama's Criminology Prof Adam Lankford has studied data on "mass shootings".

adamlankford.com

He published a 2016 paper in Violence and Victims that presented a cross-national study of mass shootings

In the paper, he defined "public mass shootings" as follows

This is actually the official @NYPDnews definition (this is because his data build off of the NYPD report)

nccpsafety.org/assets/files/l…

What does he find? The conclusion is pretty blunt

Subsequent scholars questioned some of his data and methods (as scholars are apt to do). This led him to write a 2019 paper with a pretty straightforward title:

The whole paper can be accessed via @EconJWatch :

econjwatch.org/File+download/…

This table from the paper is striking (look at the last column of the table).

BTW: The "Lott and Moody" is in reference to this recent paper by @JohnRLottJr & Carlisle Moody

So what the data show are pretty clear: the US has an excessive number of gun-deaths by homicide.

So while @neiltyson's comment was probably well intentioned, he was pointing to the wrong data for understanding why people are upset about gun-violence in the USA.

[END]