103

According to an OpEd in the NY Post by John Lott and Michael Weisser that cites a report by the Crime Prevention Research Center, America doesn't lead the world in mass shootings.

"Of the 86 countries where we have identified mass public shootings, the US ranks 56th per capita in its rate of attacks and 61st in mass public shooting murder rate. Norway, Finland, Switzerland and Russia all have at least 45 percent higher rates of murder from mass public shootings than the United States." (bolding mine)

I don't think it would be a particularly controversial claim to say that the US has lower gun violence rates (including lower mass shooting rates) than impoverished countries. However, the claim that Norway, Finland and Switzerland have higher mass shooting rates than the US leaves me skeptical due to the large disparity in overall gun violence.

Does the US not have a higher mass shooting rate than Norway, Finland or Switzerland? Do the above numbers, as presented, accurately depict the reality of mass shootings?

James G.
  • 1,942
  • 2
  • 13
  • 23
  • 117
    I took a look at link [provided with research](https://crimeresearch.org/wp-content/uploads/2018/08/Appendix-1_20180823.pdf). Most amount of cases (I read so far) regarding Russia are episodes of [Chechen War](https://en.wikipedia.org/wiki/Second_Chechen_War). I don't think armed clashes between federal military and local armed separatists during (de-facto) civil war are what people usually call "mass shootings". – Mikhail Gerasimov Sep 27 '18 at 01:24
  • The fundamental confusion/mislead seems to be whether you include "war related" events. – Fattie Sep 30 '18 at 17:22
  • 8
    One should also note that it's a common fallacy to compare large and small entities in relative statements without thorough statistical treatment. Small entities usually have larger relative fluctuations and are much more likely to be the worst or best in many disciplines just because of statistical effects. One maybe should restrict the comparison to countries of similar size. Norway, Finland, Switzerland are all much smaller than the US. – NoDataDumpNoContribution Oct 01 '18 at 09:02
  • 17
    Norway has had 1 mass shooting. I suspect these numbers are doctored by finding the smallest populations having a single mass shooting and then presenting aggregated measures. What a fine way to produce the numbers you want. – Stian Oct 02 '18 at 11:45
  • @MikhailGerasimov it should count. I mean, right to bear arms may cause mass shooting. But it reduces probability of civil war. It provides incentives to "work" things out rather than protesting. –  Oct 02 '18 at 17:22
  • 6
    John Lott has a [dismal reputation](https://thinkprogress.org/debunking-john-lott-5456e83cf326) for dishonesty (or is it an excellent reputation for dishonesty). Even other gun advocates dislike him. In one case, he appears to have confected a survey out of whole cloth. – Andrew Lazarus Oct 02 '18 at 23:47
  • Consider the source. The NYPost is owned by Rupert Murdoch, same person who owns Fox News. It's an incredibly right-wing paper. – Raydot Oct 03 '18 at 20:15
  • 1
    Ad hominem attacks do not constitute acceptable evidence on this site. – Ben Voigt Oct 05 '18 at 03:38
  • According to CDC published figures, the odds of dying in a mass shooting incident are roughly three times the odds of dying from falling out of bed in the USA. Mass shootings are good headlines for the news, bad luck for the victims, but rare enough that you can't draw any conclusions from the numbers alone except that they are rare . – pojo-guy Oct 05 '18 at 04:15
  • 1
    This link highlights the problem with the term "mass shootings": [On average in the US, is there a mass shooting 9 out of every 10 days?](https://skeptics.stackexchange.com/questions/42063/on-average-in-the-us-is-there-a-mass-shooting-9-out-of-every-10-days) – pojo-guy Oct 05 '18 at 04:18
  • Regarding mass shootings many statistics are open to interpretation, but a lot of the data may only be good as the country that collects it (some, like China, may not) or independent data collectors that can’t always be trusted. The definition of a mass shooting is inconsistent — I think many multi-shooting in ghettos and the like may not be counted — in addition to the confusion of definition, there a lot of local and general politics that influence it. Then there the question of cartels/corrupt govt not being held accountable because of terror. – vol7ron Oct 05 '18 at 04:22
  • Nevertheless, the US is often better than what mainstream media reports (they’re often pushed by a liberal anti-gun agenda to reduce NRA influence, because it feeds lots of money into getting conservatives elected). The reality is often somewhere in the middle. This is not an answer; I do not claim that Lott or Weisser are purporting faulty (or accurate) analysis, but hesitate to pass judgment based on who they are or who they write for. – vol7ron Oct 05 '18 at 04:27
  • 1
    @OP The only way to get the answer "no the US doesn't lead the world" (per unit people) would be if you didn't consider atrocities in "areas of conflict". If you don't consider it per unit population you're in a situation where like 4/5 accidents (might be 3/4 but there's 5% diff) are done by NOT drunk drivers, by pure numbers them not drunk guys are causing most of the accidents! Bastards! So this'd mean you'd be safer driving drunk - if you're an idiot. To compare safety you'd have to compare per population, here "crashes per X many not-drunk drivers" and "crashes per X many drunk drivers"-> – Alec Teal Oct 06 '18 at 12:20
  • @OP using text, you want to compare P[crashing|drunk] vs P[crashing|sober] - read P[X|Y] as "probability of X **given** Y", which is P[X & Y] / P[Y], with the 4/5 car accidents thing we have 0.8 (80% = 4/5) = P[sober|you crashed] and the negative, P[not sober|crashed]=0.2 (20%, 1/5). In order to get P[crashing|drunk] = P[Crash & drunk ] / P[drive drunk] - you can only say P[A&B]=P[A]*P[B] (thus getting P[A|B]=(P[A]*P[B])/P[B]=P[A])if they're "independent", obviously this is wrong for crashing and drunk! – Alec Teal Oct 06 '18 at 12:28
  • @OP (I've deleted a premature enter and a failed test) - that's conditional probability, following that through you'll be able to make your comparisons - and apply this to other situations. PS, notice how we need to know who *drives drunk* (including those who don't crash) for the above - that's one of the difficulties with this sort of thing. – Alec Teal Oct 06 '18 at 12:29
  • 2
    [See this NPR article that covers all the aspects](https://www.npr.org/sections/goatsandsoda/2018/11/09/666209430/deaths-from-gun-violence-how-the-u-s-compares-with-the-rest-of-the-world). Short answer, no, but all the countries beating the usa have significant social issues that help explain it. The usa looks out of place listed with those nations. –  Nov 11 '18 at 01:59
  • 2
    The article says "Lankford’s study reported that over the 47 years there were 90 public mass shooters in the United States and 202 in the rest of world. **Lankford hasn’t released his list of shootings or even the number of cases by country or year.**" I know the article is over 3 years old but we had [11 mass shootings in USA just this memorial weekend](https://www.washingtonpost.com/nation/2022/05/30/mass-shootings-memorial-day-weekend-taft-chattanooga-uvalde/). Lankford is likely selective in what he had chosen as a "mass shooting" whether in USA or in Norway, Finland, Switzerland or Russia – alec May 31 '22 at 00:37

6 Answers6

241

As discussed in this answer and this answer when you use a term like "mass shooting", your results become particularly sensitive to your definition. Additionally, results will be sensitive to search strategy, since cases are assembled from searches for news reports. Because the results are sensitive to definition and search strategy, they are susceptible to bias.

The authors of the OpEd are Michael Weisser, aka, "Mike the Gun Guy", and John Lott, the president of the Crime Prevention Research Center the organization that publishes "The War On Guns: Arming Yourself Against Gun Control Lies".

The study the OpEd references was conducted by the same Crime Prevention Research Center, was not peer reviewed, and does not have a coherent methods section with a well defined search strategy or case definition. To me, it reads like an undergraduate student's report or a blog post, not an academic study. Most of the section on definitions and search strategy is a discussion of perceived problems with a peer reviewed study by Adam Lankford that is beyond the scope of this claim. They start with events listed in the University of Maryland Global Terrorism Database, and then add events found using a defined nexis search and an undefined "web search", which included wikipedia.

While it is possible the claim is true, the evidence used to support it is not peer reviewed or thoroughly described, and it is conducted by a biased source.

De Novo
  • 3,176
  • 2
  • 15
  • 27
  • 3
    This is a very good answer - though I feel it could be even better if actual numbers were provided for gun violence in different countries of the world. That being said - I recognize why that might be difficult, as you've already stated that 'mass shooting' and 'gun violence' are loaded terms that run into bias very easily. But references to existing peer-reviewed articles could help that, and rebuff the claim linked in the question (or possibly even prove it true). – Zibbobz Sep 27 '18 at 13:16
  • 7
    @zibbobz That would be a great answer. I have written those answers, and they are a ton of work. Often the numbers come up inconclusive or I cannot find enough data to make a real answer. – BobTheAverage Sep 27 '18 at 15:41
  • 4
    @BobTheAverage yes, they are a ton of work. I tend to avoid writing those kinds of answers unless they're in my field and I'm already familiar with the literature. – De Novo Sep 27 '18 at 16:18
  • 7
    This is a very bad answer, verging on the ad hominem. The data is what it is (and one could validly challenge methodology &c), but the fact that it has been collected by people in favor of gun rights is no more relevant than an anti-gun group trying to conceal or obfuscate it. – jamesqf Sep 27 '18 at 17:31
  • 65
    @jamesqf I believe I did challenge the study directly: there is no coherent methods section, and the web search strategy is undefined. These are objective problems that produce a high risk of bias. The conflict of interest is especially relevant when there is such a high risk of bias. Yes, it would be just as much of an issue in a study produced by advocates of gun control. I'm not sure why that fact would makes it less relevant here. Notice that i didn't say that the risk of bias and conflict of interest made the claim false, it just doesn't provide good support for the claim. – De Novo Sep 27 '18 at 17:41
  • 48
    @jamesqf as far as my earlier comment about this not being my field -- gun violence is not my field. Research methodology is. This is objectively a poorly conducted study with a high risk of bias (see above). – De Novo Sep 27 '18 at 17:44
  • 3
    @De Novo: At least as I read the answer, it seemed that you were reflexively rejecting the conclusions because of the source, then looking for reasons to justify the knee-jerk reaction. Of course your intent and what I read could be different... – jamesqf Sep 27 '18 at 19:28
  • 30
    @jamesqf First you said I didn't challenge the methodology, now you say I did, but was just looking for reasons to justify a knee jerk reaction. The first things you should look for when evaluating any population level rate are the case definition and the methods for identifying cases. There is no knee-jerk reaction or confirmation bias here. The study fails immediately, when investigating what you should look at first. – De Novo Sep 27 '18 at 19:57
  • 45
    @jamesqf Its not an ad hominem at all. When something hasn't been peer reviewed, and doesn't even properly explain its methods, an enormous conflict of interest is sufficient reason to completely discount a source until you find real evidence. – mbrig Sep 27 '18 at 22:12
  • 17
    @jamesqf Do you have any example where the author here has condoned a non-peer reviewed paper with missing methodology section from anti-gun people or is that just an ad hominem? – Voo Sep 28 '18 at 08:30
  • 5
    "This is objectively a poorly conducted study with a high risk of bias" - in that regard, Lott has been very consistent over the years. – PoloHoleSet Sep 28 '18 at 15:35
  • "More Guns, Less Crime". Why is this title bad? Are you familiar with vaccinations and herd immunity? Are you familiar with the disease model of crime? Why is it hard to believe that an armed populace makes a population especially resistant to crime? – Chloe Sep 29 '18 at 02:26
  • 8
    @Chloe those who are familiar with me, my work and post history on SE.Biology might find that question amusing. I used the titles as a proxy for the political leaning of Lott and the CPRC. If you want to ask an on topic question about that claim, I'm sure it would get some answers. – De Novo Sep 29 '18 at 02:39
  • 1
    @mbrig: Is this supposed to be a scientific research paper, published in a journal? If not, then why the complaints about lack of peer review, which is not expected? – jamesqf Sep 29 '18 at 04:16
  • 5
    @Chloe The answer doesn’t say anything about whether such a title is good or bad. It simply uses it to illustrate that the authors of the study are biased towards one of the three possible outcomes the study could possibly have had (more guns = more crime; more guns = less crime; more guns = no difference in crime rate). – Janus Bahs Jacquet Sep 29 '18 at 10:05
  • @JanusBahsJacquet Why does it show bias? What if the title just summarizes the conclusion of the essay/study? If the title was "More guns, more crime", would that make the author biased too? – Chloe Sep 29 '18 at 16:48
  • 9
    @Chloe In an absolute vacuum, being an NRA lobbyist and arms dealer or the president of a pro-gun organisation does not in itself say anything about whether you are pro-guns or not, true. Reality is not a vacuum, though, and in reality, there aren’t many better ways to show a personal bias. Unless you’re arguing that there is no such thing as personal bias or no way to show it? Naturally, if the authors had been anti-gun lobbyists and their previous study called _More guns, more crime_, that would have indicated personal bias just as clearly, just at the other end of the spectrum. – Janus Bahs Jacquet Sep 29 '18 at 16:57
  • 1
    @JanusBahsJacquet That sounds reasonable, but the answer didn't state the person was an NRA lobbyist, arms dealer, or president of a pro-gun organisation. It simply listed two titles of articles/studies, one of which is completely neutral. – Chloe Sep 29 '18 at 17:25
  • 4
    @Chloe Which of the 2 titles is completely neutral? "Arming Yourself Against Gun Control Lies" seem immediately implies a pro-gun publication and that gun-control arguments are lies, and confrontational and "More Guns, Less Crime", has not been proven, unlike "Less Guns, Less Crime", as shown by the UK and most other western countries. – Steve Ives Oct 04 '18 at 14:48
  • 4
    @Chloe you may be right that the words in the title *More guns, Less crime* aren't by themselves an indicator of bias. I was using the word "title" to refer to the actual book. These comments are probably not a good place to argue about the content. Regardless, Lott is not a neutral figure. He doesn't work for the NRA, but he is a gun advocate and a partisan political commentator. This doesn't make him right or wrong, but it does indicate a conflict of interest. I would be just as skeptical of a poorly conducted study from a partisan political commentator on any side of the political spectrum. – De Novo Oct 04 '18 at 16:23
109

In a comment, @MikhailGerasimov gave a link to the list of incidents.

By browsing this, it is clear that the claims include acts of terrorism and war. (But being curiously selective about it, I found no reference to attacks by US or allied forces in Afghanistan, only the Taliban and their allies.)

There is no war going on in the US. There are very few terrorist attacks there. This means that, yes, it is safer in the US than many other places. Mass shootings in the US gain massive headlines, but aren't all that common, really, compared to war zones.

I jumped a bit at seeing Norway mentioned, since we are a peaceful country. Looking at the list, I see one single shooting. It turns out that in a small country with only 5 million people, a single shooting is enough to put us above the US. Technically correct, but hardly statistically significant.

It is an interesting fact that there are very few terrorist attacks in the US. Given that they are "the great enemy" according to many terrorist organizations, one might expect there to be more.

tripleee
  • 127
  • 1
  • 7
Stig Hemmer
  • 1,526
  • 1
  • 12
  • 6
  • 57
    To put this into perspective, [Wikipedia](https://en.wikipedia.org/wiki/List_of_countries_by_intentional_homicide_rate) puts homicide rates of Norway and the US at 0,51 and 5,35 per 100k and year. – JollyJoker Sep 27 '18 at 08:58
  • 14
    From the official statistics, https://www.politiet.no/globalassets/04-aktuelt-tall-og-fakta/drap/drapsoversikt-kripos-2017.pdf , you can see (in 2.2) that the total number of people murdered in Norway in 2009 and 2010 was 60, which is less than the 69 killed in the one mass shooting in 2011. If you look at only gun related murders (in 2.4) the total number of people murdered with a gun in Norway from 2008 to 2017 is 36. – Jan Obrestad Sep 27 '18 at 10:11
  • 64
    The magnitude of that singular 2011 event combined with Norway's small population (less than that of 22 US states) will keep Norway on top of the charts in terms of mass shooting deaths per million people per year from 2011 to the present for the next decade or so. Singular outliers combined with small populations combined with cherry picking oftentimes make the mean a misleading statistic. – David Hammen Sep 27 '18 at 12:33
  • 1
    This doesn't seem to be an answer - more a comment. – Oddthinking Sep 27 '18 at 12:40
  • 11
    We have a really big moat. – T.E.D. Sep 27 '18 at 14:23
  • 6
    Doesn't sound like your definition of terrorism includes homegrown – MicroMachine Sep 27 '18 at 17:12
  • 8
    Imagine the statistic, if something happened in the Vatican. – Chieron Sep 28 '18 at 12:33
  • 24
    @Chieron Quite so: the Vatican has (by a colossal margin) the highest per-capita rate of petty theft in the world, simply because millions of tourists visit every year, and many are pick-pocketed, in a country whose official population is given as exactly 1,000 individuals. – KRyan Sep 28 '18 at 14:52
  • 58
    @KRyan It's like the old joke about Bill Gates walking into a bar, suddenly the average person in the bar is a multi-millionaire. – Barmar Sep 28 '18 at 15:39
  • 1
    There's lies, damned lies and statistics. – RobIII Oct 02 '18 at 13:44
  • @DavidHammen While you are correct about the Norway data being contaminated by an outlier that can't explain the US being 3/4 of the way down the list--you can't have most places being outliers. – Loren Pechtel Oct 03 '18 at 04:29
  • 1
    @LorenPechtel Cees-Timmermans answer indicates reasons why this might be so. Firstly this list appears to be a list of countries which have had mass shootings in the time period. If the author had included countries with no mass shootings (and why wouldn't they) then the US would be near the top. In addition it seems that converting from using the mean to the median (which generally tends to limit the influence of outliers) the US goes back to the top of the list. – Eric Nolan Oct 03 '18 at 11:46
  • @EricNolan No--if we add all the zeroes we would still be 56th or 61st depending on your yardstick. That puts us about 1/3 of the way down the list, not at the top. – Loren Pechtel Oct 03 '18 at 14:20
  • 3
    I said "near the top" but I guess whether you consider being in the top third near the top or not is a matter of opinion. It is certainly not the same as 3/4s of the way down the list. – Eric Nolan Oct 03 '18 at 14:33
87

Snopes explains it as cherry picking:

The first thing to note about the rankings is that Lott has compared the mass shooting death rate in the United States with that of other countries where there was a mass shooting between 2009 and 2015. This might seem obvious, but it’s important to point out that very many countries did not see a single mass shooting as defined by Lott during this period.

The second striking thing about the list of mass shootings in Europe is that it is dominated by outliers. Where the United States saw at least twelve mass shooting deaths every year between 2009 and 2015, some of the other countries on Lott’s list experienced one or two rare but very high-casualty shootings. When you average out the death rates, this creates a highly misleading impression about the consistency and lethality of mass shootings outside the United States.

enter image description here

To further illustrate the point, even counting terrorism, Norway only has one significant event in the past 50 years, whereas it appears that not a day goes by in the USA without somebody shooting multiple people.

According to this 2012 article on gun homicides, the average per 100k people gun homicide rate was 4.9 globally, 3.0 for the US, and 0.4 for Europe (not counting Kosovo, Montenegro, and Russia due to missing data).

Community
  • 1
Cees Timmerman
  • 4,673
  • 3
  • 31
  • 49
  • 38
    The point about leaving out zeroes is a very good one. You get a skewed view of how much rain there is if you leave out days with no rain. – JollyJoker Sep 28 '18 at 08:40
  • 1
    Russia seems very safe. – ANeves Oct 01 '18 at 10:52
  • 3
    I think average is what counts more than median. The probability is small. It doesn't change the fact that you are more likely to be victim of mass shooting in Norway than in US. It's not misleading at all. It just shows that mass shooting in US is overrated –  Oct 02 '18 at 17:51
  • 1
    @JollyJoker How is that point relevant? The OP's question was about the ranking of the US as higher or lower relative to other countries with mass shootings (specifically Norway, Finland and Switzerland). Inclusion of countries with zero mass shootings would not change this ordering. The question was not about how much rain there is, but about which days had more rain than others. – LarsH Oct 02 '18 at 17:58
  • 13
    There's a closely related effect to the one discussed here, well known to statisticians, which is that small counties/states/countries have larger variability in rates than large ones (a kind of 'small county' effect). Consequently any list of rates like this will almost *always* be topped by small countries -- but with something like the rates under discussion, it will typically be *different* small countries in different years (it might be Norway for one period and Serbia for the next period, as one or two events bumps countries up the list and 0 events drops them back out again). – Glen_b Oct 03 '18 at 05:22
  • 3
    @LarsH There are few deaths from mass shootings per year over large areas and long times. The shortish time window and small countries included means those countries that happened to have a mass shooting have a misleadingly high average, since including areas and times when no shootings happened would lower the average. – JollyJoker Oct 03 '18 at 07:53
  • 12
    In light of this it would be interesting to see what happens if were to either include all of Europe together (presumably the metric would drop by a lot) or you broke the US down by state (presumably resulting in a finding that X, Y or Z state are mass shooting hot spots). – Eric Nolan Oct 03 '18 at 11:51
  • Why does Norway have such a high (1.99) value in "Average (mean) annual mass shootings"? –  Oct 03 '18 at 18:00
  • @JollyJoker There are two separate but related issues here... (1) the report omitting countries where no mass shootings occurred in the data set, and (2) the report omitting time periods where no mass shootings occurred in the countries that were mentioned in the report. Cees quoted Snopes about (1), but (1) is irrelevant to the OP's question because it doesn't affect the order of countries. I thought you were making a point about Cees' answer (1), but apparently you meant (2), which this answer doesn't directly mention. – LarsH Oct 03 '18 at 18:41
  • 2
    @J.Chang It does not say that. The average Norwegian in 2011 was more likely to be the victim of a mass shooting. The probability of a Norwegian alive now becoming a victim of a mass shooting is arguably less; at the very least, the small sample size tells us little about the future. – prosfilaes Oct 04 '18 at 06:49
  • 3
    As stated in the article, answer, and comments, "The second striking thing about the list of mass shootings in Europe is that it is dominated by outliers." - [In Norway's case, the terrorist acts of a single Nazi](https://en.wikipedia.org/wiki/2011_Norway_attacks). – Cees Timmerman Oct 04 '18 at 16:16
  • 1
    **BREAKING NEWS: UK is second on the median list behind only the US!! Balkan states [that we hate because we're a racist/nationalist/DailyMail newspaper] included in this list!** \\ not exactly the point, but related: "headline data", even if technically correct, isn't sufficient by itself; just saying "US doesn't lead mass shootings" isn't enough to make any conclusions, without looking at what the data is (are), and what definitions they're using – Sam OT Oct 04 '18 at 21:31
  • There's a valid point about excluding countries with no mass shootings, although the small sample size argument works both ways; of course most small countries won't have mass shootings. Relatedly, it's kind of silly to take the median of mass shootings in smaller countries on a year by year basis; if a particular country had 3 mass shootings in 2015, 1 in 2013, and 2 in 2010, that would be a rather large number of shootings for a country of 5 million, yet the median would still be 0. Trading one dubious statistical analysis for another doesn't help get to the truth. – D M Oct 07 '18 at 14:31
  • 4
    Nice analysis. As The Onion put it: "‘No Way To Prevent This,’ Says Only Nation Where This Regularly Happens". – Oleg Lobachev Oct 08 '18 at 13:26
  • Snopes doesn't use the term "cherry picking". "Cherry picking" is when you choose only the data that supports your position, not when you choose a statistic that supports your position. And the idea that the median is a better measure is absurd. Choosing the median to measure a skewed-right distribution is going to make smaller countries look better. When you add up all countries in Europe, and *then* take the median, it's much less favorable to Europe. – Acccumulation Oct 08 '18 at 18:48
  • The problem with small and big countries could be solved by publishing the same statistics for all 50 US states individually. – gnasher729 Nov 28 '22 at 13:50
56

TL;DR? The USA has so many mass shootings that the average deaths-per-shooting is lower than other places.

Let's break down what he's calculating: The rates of murder from mass public shootings per million people.

This means that, when there is a mass shooting(1) in public(2), that if we account for population(3), more people are murdered(4) .

By using a baseline of one mass shooting = one mass shooting, he's already removed the point of contention: that the USA has many more mass shootings than anywhere else in the world. So already, Norway with one mass shooting is being equalized with the USA, with hundreds.

By specifying in public, he skews things even more in his favor. The United States does not currently have major conflicts or wars on its own soil, which he is counting towards this metric. Anecdotally, more people die in public than in private in a war, though I can't find any statistics to back that up.

By making it per million people, the fact that the USA has 360-odd million people is used as an advantage against smaller European countries.

More people are murdered: this is the thing he's measuring. IF there's a mass shooting, and IF it's in public, how many people die? Norway has had one mass shooting. In it, 69 people died. That's 69 murders (69/1) per mass shooting. Whereas in the US, we've had 155 mass shootings and 1107 killed. That's 7.14 (1107/155) murders per mass shooting.

This has generated a bit of confusion, so here's a simple example:

Let's use apples. I'm a farmer, and the local kids have been stealing my apples. They've stolen 100 apples over 10 incidents, for 10 per incident. However, the rats once got in and ate 15 apples. Does that make the rats a bigger threat, or the kids? I'd argue the kids (the US). They'd argue the rats (Norway).

Carduus
  • 2,096
  • 10
  • 15
  • 17
    Barring small sample size (as in Norway), isn't per capita, or per million people if you prefer, precisely the right way of measuring mass shootings? – Obie 2.0 Sep 27 '18 at 22:49
  • 9
    @Obie2.0 yes, that would be the case when there is enough data to average. Sample size really matters for the viability of averaging, given extreme outliers. If the data had been averaged over the entirety of Europe, comparing the values would be justifiable. – Chieron Sep 28 '18 at 12:40
  • 14
    @Obie2.0 Yes, as a statistician, I'd make sure it was per-capita. But when you're already equalizing it via the per-shooting metric, the per-capita thing is double-dipping. – Carduus Sep 28 '18 at 13:35
  • The only thing I got from this answer is that comparing Norway to the USA is skewing things, and I am inclined to agree with that, but not exactly why that is so. Please end my confusion here. – GwenKillerby Sep 28 '18 at 14:01
  • @GwenKillerby Mass media has been decrying the total number of mass shootings in the USA for years. Even accounting for population size, it's much bigger than any other country. This claims to debunk those critics, when it in fact measures something very different: the number of casualties PER INCIDENT. Let's use apples. I'm a farmer, and the local kids have been stealing my apples. They've stolen 100 apples over 10 incidents, for 10 per incident. However, the rats once got in and ate 15 apples. Does that make the rats a bigger threat, or the kids? I'd argue the kids. They'd argue the rats. – Carduus Sep 28 '18 at 14:19
  • 6
    Now I know the other thing what is wrong with the Op-ed: Time frame. Modern mass shootings started with the Tower Sniper Charles Whitman who killed 16 people. The op-ed cherry picking starts with them JUST looking at 2009-2015, so as to include Breiviks 2011 ideological massacre. If you do a comparison of year by year, it would become immediately clear that 2011 is an outlier for Norway, but presented as regular occurence by Lott and co. – GwenKillerby Sep 28 '18 at 14:42
  • 2
    To extend your metaphor a little -- the rats _could_ be seen as a bigger threat if they had very recently gotten in and eaten the 15 apples. You only have a single datapoint, but if rats got in once, they'll get in again, and so it may make more sense to defend against the rats to avoid losing another 15 apples two days after the first 15, even though the second 15 hasn't happened yet. That doesn't apply here (2018 - 2011 > 2 days) – Nic Sep 28 '18 at 18:31
  • @Carduus - Assuming the measured time period is 7 (whatever units you want), per the original article then 100/7 = 14.28 apples lost to kids per time unit. 15/7 = 2.1457 apples lost to rats. Thus, your claiming that the people doing the study would argue the rats are a bigger threat is clearly incorrect. The kids are nearly 7 times more of a threat. If you do a per capita (where I'd assume # rats is greater than # kids) then kids are even more of a threat. – Dunk Oct 03 '18 at 20:48
  • Your post is hard to follow. The fourth paragraph, in particular, doesn't make any sense. What are you trying to say here? What is your point about making it per million people? What is this stuff about murders per shooting? – Acccumulation Oct 08 '18 at 06:19
23

There is another issue with this study. It states: "Of the 86 countries where we have identified mass public shootings, the US ranks 56th per capita in its rate of attacks and 61st in mass public shooting murder rate". However there are about 196 countries in the world today. It is not clear whether the other 110 countries were excluded on the grounds of missing data or whether they were excluded on the grounds that they had zero mass shootings. If it is the latter, then they have been mistakenly missed from the accounting procedure and the US should be ranked 56th out of 196 not 56th out of 86.

Another issue is that the FBI definition used by CPRC specifically excludes gang and drug related violence but not terrorism leading to an additional source of bias. The FBI report the CPRC reference, specifically states on page 5 that "This is not a study of mass killings or mass shootings, but rather a study of a specific type of shooting situation law enforcement and the public may face". This may decrease or increase the USA's relative ranking.

However in some ways a more important point is that the total death rate from mass killings is low everywhere in the world outside of warzones. One wonders whether a greater benefit would be achieved if the attention directed to reducing mass shootings was directed instead at continuing to reduce the number of deaths due to armed conflicts in the world, addressing domestic violence or even reducing people's sugar intake.

Justin
  • 331
  • 1
  • 2
  • 1
    The first answer already pointed out that the NYP op-ed doesn't even mention the other 110 countries. However, you make this point more clearly and with numbers, so that's better. Of those 110, there must be some at least where there's no mass-shooting whatsoever. Luxemburg, Andorra, French Guiana, Vanatu Island come to mind. So they are unfairly excluded, to make the USA look better. – GwenKillerby Sep 28 '18 at 14:27
12

Note the subtle but important discrepancy between the headline and the detail:

"America doesn’t actually lead the world in mass shootings"

...

A new report from the Crime Prevention Research Center, which one of us heads, has just finished collecting cases using the same definition of mass public shootings used by Lankford.

John Lott/CPRC tend to focus specifically on public mass shootings, which actually excludes the majority of mass shootings. The exact numbers depend on which time periods and which definitions you use, but even this 2014 CPRC report acknowledges that the scope makes a huge difference. Condensing from pp. 4 & 5:

The 2014 CPRC report counts only public mass shootings and excludes "gang fights and shootings which occur in connection with some other crime, such as robbery". They contrast their results to Everytown For Gun Safety, which does include shootings that happen in private and those which are connected to other crimes. Over the period between January 2009 and July 2014, the CPRC criteria find 25 mass public shootings with a total of 180 deaths; in the same period, Everytown finds 110 mass shootings with 560 deaths.

In other words, according to CPRC's own publication, the "public mass shootings" included in their reports only cover about 32% of all mass-shooting deaths and 23% of mass-shooting incidents in the USA.

The headline of John Lott's NY Post article makes a claim about "mass shootings" in general, but the only evidence it offers for that claim is research on a specific subset that covers less than a third of mass shootings. That should be enough to mark it as highly questionable, even before we get into the cherry-picking issues that have been raised in other answers.

GB supports the mod strike
  • 1,214
  • 1
  • 9
  • 11
  • Excluding gang fights makes sense if you're looking at the risk to people. You're rarely going to be hit in a gang mass shooting unless you're a gangster yourself. That part of his analysis is fine, it's downfall is that he's actually finding noise, not a pattern. – Loren Pechtel Nov 25 '22 at 06:08
  • @LorenPechtel Your wording seems to imply that "gangsters" (a term that tends to be defined very subjectively, often by people who have an interest in inflating the number of "gangsters") aren't "people". I would disagree with that, and I'd need some evidence to accept the assertion that gang mass shootings rarely hit non-members (how is membership defined?) The exclusion becomes particularly questionable when comparing across countries, as Lott/CPRC are doing, because it's very unlikely that everybody is defining "gang" the same way, or that gangs work the same way in every country. – GB supports the mod strike Nov 26 '22 at 08:27
  • People are looking at the risk to themselves. Gangland fights pose little risk to the law abiding. – Loren Pechtel Nov 26 '22 at 15:27
  • 2
    @LorenPechtel Are you seriously claiming that there are no law abiding people living in gang-controlled neighborhoods? Are the infants and young children and elderly people wounded and killed in gang violence all law-breakers who deserve what they get? Really not seeing any way to interpret your claim that doesn't make you look pretty terrible. – barbecue Nov 27 '22 at 03:37
  • @barbecue The risk is not zero but such people are rarely hit in gangland fights. Just pay attention to the news--gang fights don't get on the news, bystanders being hit do. You don't see the bystander reports very often. – Loren Pechtel Nov 27 '22 at 03:40
  • 1
    @LorenPechtel I rarely see ANY reports of poor minorities being hurt or killed under any circumstances. So I safely conclude that such incidents are rare I guess. – barbecue Nov 27 '22 at 03:42
  • @barbecue Maybe the reporting varies, but around here when a kid gets caught in the crossfire it definitely makes the news. Things that look like criminal vs criminal never do. – Loren Pechtel Nov 27 '22 at 03:59
  • 3
    My point was that how often you personally hear about something in the news is a TERRIBLE way to determine how often it happens. – barbecue Nov 27 '22 at 04:01
  • @barbecue Not to mention how much potential there is for bias in what gets reported as "innocent bystander" vs. "gangster". – GB supports the mod strike Dec 09 '22 at 04:41