Yesterday I was using the Global Terrorism Database to check some suprisingly low figures on what percentage of terrorist acts are committed by Muslims. (Short answer: Worldwide since 2000, about 80%, rather than 0.4 - 6% as given in various sources.) But I found some odd patterns in the data for the United States. Look at this chart of terrorist acts in the US which meet GTD criteria I-III and are listed as "unambiguous":



There were over 200 bombings in the US in 1970 alone, by all sorts of political groups (the Puerto Rican Liberation Front, the Jewish Defense League, the Weathermen, the Black Panthers, anti-Castro groups, white supremacists, etc., etc.) There was essentially no religious terrorism; that came in the 80s and 90s. But let's zoom in on 1978 onward, after the crazy period we inaccurately call "the sixties". First, a count of Islamic terrorist acts worldwide:

Islamic terrorist acts worldwide
This is incomplete, because the database contains over 400 Islamic terrorist groups, but only let me select 300 groups at a time. (Al Qaeda is one of the groups not included here.) Also, this doesn't list any acts committed without direct supervision from a recognized terrorist group, nor acts whose perpetrators were not identified (about 77% of the database, estimated from a sample of 100, with the vast majority of those unknowns in Muslim countries). But we can see there's an increase after 2000.

Now let's look at terrorist acts of all kinds in the US:

Terrorist acts in the US, 1970-2013

We see a dramatic drop in terrorist acts in the US after 2000. Sampling them, I found that except for less than a handful of white supremacists, there are only 3 types of terrorists still active in the US: Nutcases, animal liberation activists, and Muslims. If we exclude cases of property damage (which has never terrified me), it's basically just nutcases and Muslims.

Going by body count, it may still be an increase, because even if you exclude 9/11, just a handful of Muslim attacks still accounted for 50% of US fatalities in terrorist attacks from 2000 through 2013. But counting incidents, by 2005 there were about 1/3 as many per year as just before 2000. From 2000 to 2013 there were only 6 violent terrorist attacks in the US by non-Islamic terrorist groups that were not directed solely at property damage, resulting in 2 fatalities over those 14 years. Violent non-Islamic organized terrorism in the US has been effectively eliminated.

Some of this reduction is because we've massively expanded our counter-terrorism agencies. But if that were the explanation, given that homeland security doesn't stop all of the Islamic attacks they're focused on, surely we would see more than 6 attacks by other groups in 14 years.

Much of the reduction might be for non-obvious reasons, like whatever happened around 1980. But I think the most-obvious hypothesis is that Islamic terrorists gave terrorism a bad name. In the sixties, terrorism was almost cool. You could conceivably get laid by blowing up an Army recruiting center. Now, though, there's such a stigma associated with terrorism that even the Ku Klux Klan doesn't want to be associated with it. Islamists made terrorism un-American. In doing so, they reduced the total incidence of terrorism in America. Talk about unintended consequences.



On a completely different note, I couldn't help but notice one other glaring thing in the US data: terrorist acts attributed to "Individual" (a lone terrorist not part of an organization). I checked 200 cases from other countries and did not find one case tagged "Individual". But half of all attributed cases in the US from 2000-2013 are tagged "Individual". The lone gunman thing, where someone flips out and shoots up a Navy base, or bombs a government building because of a conspiracy theory, is distinctively American.

Perhaps Americans really are more enterprising than people of other nations. Perhaps other countries can't do the detective work to attribute acts to individuals. Perhaps their rate of non-lone wolf terrorism is so high that the lone wolf terrorists disappear in the data. Perhaps we're more accepting of "defending our freedom" as an excuse for shooting people. Perhaps psychotic delusions of being oppressed don't thrive well in countries that have plenty of highly-visible oppression. But perhaps Americans really do have a staggeringly-higher rate of mental illness than everyone else in the world. (Yes, suspicious study is suspicious, but... it is possible.)
New Comment
73 comments, sorted by Click to highlight new comments since:
Some comments are truncated due to high volume. (⌘F to expand all)Change truncation settings
[-]gwern240

Much of the reduction might be for non-obvious reasons, like whatever happened around 1980.

Exactly. Much of the reduction is for the same reasons you no longer hear much about cults, or about hijacking planes to Cuba (or about anarchists car-bombing Wall Street or trying to shoot the Queen, for that matter), or about communism. There seems to be something about the shift from a traditional partially-industrialized society to a post-industrial one which triggers this sort of great social upheaval, which manifests in part as new religious movements (labeled cults) and violent action (terrorism or guerrilla), which eventually get discredited and cease to be alternatives. In Japan, you had the 'rush hour of the gods' with many syncretic Buddhist groups and the Red Army (to name the most infamous one) with a last gasp in Aum Shinrikyo; in America, you had those but also Weathermen etc; in South Korea with its later development the process is still ongoing, with the cults take on a Protestant Christian form - the recently deceased cult leader associated with the Sewol Ferry disaster an interesting example - and the violence tends to be associated with North Korea (various assassinations or attempts come to mind).

6JoshuaZ
If this is correct then we should expect a reduction in Islamic terrorism but I'm not sure what the expected timeline for such a reduction would be.
[-]gwern130

we should expect a reduction in Islamic terrorism

On some time-scale, yes. But saying when is a bit of a sucker's game. To use the American example: are we still in the '60s or have we passed into the '70s yet or maybe even '80s?

I'd be more confident in predicting that both cults and violence are on a long-term secular decline in South Korea, which seems to be over the hump.

but I'm not sure what the expected timeline for such a reduction would be.

That's a little difficult to say. I think you could probably use per capita GDP to try to pin down when one would expect 'the troubles' to begin, and extrapolate from there. (This would probably yield predictions like: East Asia to continue to quiet down; Middle East and nearby Islamic regions to remain stable in violence; Africa to increase in cults and movements like Boko Haram even as larger-scale violence and disorder decreases.)

I know you didn't make the graphs, but ...

I don't like line graphs. They're a kind of smoothing, but almost never the right choice. The first two graphs are smooth enough that it hardly matters what you do, but I redid the third as a scatterplot plus a loess curve.

Here's the loess superimposed on the line graph, if you want to compare them. I got the data here, linked from here

A find your hypothesis interesting but let's discuss a few different possible causal mechanisms than your primary suggestion (and it is possible that this drop is some combination of causes):

1) You bring up the possibility that the threat of Islamic terrorism has lead to a substantial increase in security levels, so more terrorist incidents are stopped or never get off the ground. If this is the primary cause then I'd expect most of the drop off to occur after the first bombing of the WTC in the 1990s, and a heavy post-9/11 drop-off but your graph shows a drop-off staring before then but then shows the most clear cut drop-off in the mid 1990s, so this may be part of the effect. The fact that the drop-off rate doesn't get massively faster after 9/11 possibly undermines this. I think your point is also a valid one which suggests this isn't what is happening or at least is only a small factor of what is going.

2) The American political scene may in some ways simply be more civilized than it was in the 1960s and 1970s. True in many ways things at the congressional level feel more partisan, but it may be that the groups which felt like they had no options other than violence don't fe... (read more)

We see a dramatic drop in terrorist acts in the US after 2000.

Do we? I see a sawtooth-decline starting in 1995/6, with 2000/1 not deviating from the trend.

4NancyLebovitz
It's hard to say, but I'm seeing a decline in both the peaks and valleys a little after 2000, with a big drop around 2003.

Your analysis is surprisingly Americo-centric. The 1970s saw very serious terrorism (far worse than America) in the UK, Germany and Italy, all of which are now very peaceful countries. Did 9/11 also make terrorism un-Italian?

Secondly, your timing is all wrong. The fall in terrorism worldwide long predates the rise of specifically Islamic terrorism.

Thirdly, Islamic terrorism is the intellectual and organisational descendant of secular Arab terrorism and is received in much the same way. The only innovation is the suicide bomber. Yet in the period that you c... (read more)

5PhilGoetz
It is not surprisingly Americo-centric for a post titled "How Islamic terrorists reduced terrorism in the US". I acknowledged that there was a bigger (numeric) fall in terrorism in the US in the 70s. The fall in the US after 2000, though, is probably as big or bigger when expressed as a percent drop rather than as an absolute drop. Equal efforts at reducing a variable results in drops that are similar by percentage more than by absolute number. (That means the effort to go from 100 cases per year to 10 is more similar to the effort to go from 10 to 1 than to the effort needed to go from 900 to 800.) There has been no fall in terrorism worldwide; just the opposite. It was at its lowest point in 1971-1975, and is over 10 times as high now (as measured by GTD incidents). For Europe at a whole, it was at its lowest in 1970, then was high from 1976 to 1997. For "(USSR & the Newly Independent States (NIS))", there was no terrorism until 1989 (imagine that!), then a steady rise until 2010. The fall I pointed out for the US after 2000 also happens in a graph for Western Europe, which I would expect, but not for the world as a whole. There is a sudden dramatic fall in Central America bottoming out in 1998, to just a few incidents per year. This might be due to a similar thing happening with the drug wars there, or might just be bad data.
6ailyr
Maybe, the total number of incidents rises just because with the spread of the Internet and other communications technologies, it's easier to get information about terrorists attacks. For example, there definitely were terrorist attacks in USSR before 1989(https://en.wikipedia.org/wiki/Category:Terrorism_in_the_Soviet_Union), but they aren't mentioned in GTD.
0Salemicus
If you see the same phenomenon all over the developed world, then it is very likely to have roughly the same causes throughout that class of countries. It is parochial in the extreme to explain that phenomenon in one country solely in terms of causes specific to that country, rather than to causes that could have affected all the countries in the relevant class. Otherwise you are essentially arguing for some staggering co-incidence. For example, if we are asking why did crime fall in New York in the 1990s, and all your explanations are specific to New York, you are missing key factors. Crime fell across America, and across the developed world. Explanations specific to New York can only explain the difference between New York and the rest of America, and explanations specific to America can only explain the difference between the US and the rest of the developed world, and so on. So yes, you are being Americo-centric.
-1PhilGoetz
I think you should have checked the database, which you obviously didn't, before writing two long replies. You see the decline after 2000 in the US and in Europe, which are the regions affected by Muslim terrorism (outside of Muslim countries, where this dynamic would not apply). You don't see it in Japan, Russia, India, China, or in Muslim countries. You don't see it in the world overall. This is entirely consistent with my hypothesis; your hypothesis that it is coincidence is the one that requires coincidence. And, yes, when I write a post that says it's about America, it's going to be Americo-centric. I live in America. It's okay for me to talk about America.
0Salemicus
Your replies bear little or no resemblance to what I've written. I never made a hypothesis of 'coincidence,' I never said you shouldn't talk about America, and overall I have to conclude that either I am writing unclearly, or you are misrepresenting me. In either case, I am disinclined to continue such a conversation.
4JoshuaZ
I wonder how much of this general pattern then is attributable to the end of the cold war? Both the Germany and Italy examples are due primarily to Marxist groups. And the USSR did support non-Marxist terrorist groups in the West on occasion as well. So the winding down of the Cold War may have meant both a loss in funding and a loss of ideological motivation.
2alienist
I don't think this applies to the German or Italian Examples, but the IRA was largely funded by Irish Americans, and 9/11 made funding a terrorist group seem at lot less like a fun expression of ethnic solidarity.
6Salemicus
That may have been a factor in the failure of the IRA splinter groups to get off the ground, but Irish terrorism was already coming to a halt. The IRA called its final ceasefire in July 1997 and signed up to the Good Friday Agreement in 1998.
6knb
As I understand it, the reduction in terrorism in Ireland was essentially complete by the time of the Good Friday Agreement in 1998.

half of all attributed [terrorist acts] in the US from 2000-2013 are tagged "Individual".

The NSA/FBI might be doing a great job of stopping coordinated terrorist attacks and infiltrating and deterring groups that might launch attacks on America. Plus, I get the sense that a significantly higher percentage of U.S. than European Muslims are willing to cooperate with the police to stop terrorism.

[-][anonymous]70

You're graphs are too low resolution to read.

As for individual actors in other nations, what about Anders Behring Breivik? You might find this list helpful:

https://en.wikipedia.org/wiki/Lone_wolf_%28terrorism%29

8PhilGoetz
The graphs look large and clear to me. They're about 570 pixels wide. Are you using a cell phone? I'm using Chrome. Now that I think about it, even if other countries had a similar base rate of lone-wolf terrorists, I might not see any in 200 samples, because many of them have much higher rates of terrorism than ours (in some cases, a hundred times higher). Also, those with high rates probably can't do the detective work to attribute attacks to "Individual" rather than "Unknown".
0[anonymous]
Latest Firefox, and they're still unreadably blury...
4pragmatist
I use Firefox, and the graphs aren't blurry at all.
2NancyLebovitz
The graphs are nice and clear on Chrome.
1PhilGoetz
Doing a side-by-side comparison, they are a tiny bit blurrier in Firefox, which is bizarre, since they're JPEGs. But they're still large and clear on my screen. I can't make them any bigger; that's the size the GTD website produces them in.
6[anonymous]
Is this what you're seeing? http://imgur.com/94RsYDV

I think you're behind some kind of image-fucking (i.e. recompressing) proxy server. Are you tethering on a cellphone network by any chance? (Or on an airplane?)

2Douglas_Knight
If that's it, Phil could solve the problem by using https. That would requiring switching servers to, say, imgur.
8VincentYu
The crappy resolution is due to the image host (postimage.org), which is downsizing some images served to some IPs (verified this through a few private VPNs).
69eB1
I use Firefox on a mac and they look nothing like that on my machine. http://imgur.com/7l9Xwua

I checked 200 cases from other countries and did not find one case tagged "Individual".

Breivik..?

6PhilGoetz
He was not in the 200 cases I checked, which is not surprising, since there were 48,108 cases. Breivik is an anecdote. A sample of 200 is data.

The lone gunman thing, where someone flips out and shoots up a Navy base, or bombs a government building because of a conspiracy theory, is distinctively American.

A conspiratorial explanation: perhaps the data has been manipulated to downplay terrorist coordination in America. There would be a couple reasons to do that, depending on the audience. For ordinary citizens, a collection of nutcases might be less scary than a dark underworld of coordinated cells. For terrorists and wannabe terrorists, a collection of uncool loners might less glamorous (as you... (read more)

8gjm
It looks to me (admittedly mostly from the outside; I don't live in the US, though I travel there sometimes) much more as if the government has played up the risk of terrorism by organizations like al Qaida than as if it's trying to make it seem less threatening. (It seems like there's less of this under the present government than its predecessor, which cynically would make sense because there's some evidence that people's votes tend to shift "rightward" when they are afraid for their lives.)
4fortyeridania
Yes, you're right, they do seem to play this up. That's strong evidence against the conspiratorial explanation I mentioned. Yes, I think this is true. But I seem to remember (but cannot cite) that people also "rally around the flag" and support the leader more--even if the leader is not "right." Moreover, the present government still has some additional political incentive to play up the risk of terrorism, especially if the administration can blame it on the previous one. "Al-Qaeda is really scary--man, our predecessors sure did let their guard down."
2JoshuaZ
As someone also inside the US, I also get this impression.
[-][anonymous]00

If you stratified the data on terrorist ideology by sect, would Wahhabism by the underlying confounder, rather than Islam?

Islam explicitly advocates for violent Jihad. Countries like Australia should deal with terrorism as a policing and arts/culture issue, than a military problem. In fact, having no military might be in our best interest. There are a number of nations in the world with no militaries, some in particular unstable and criminal parts of the world, like Costa Rica, which don't have any military issues, as far as I know. In fact, it would seem ... (read more)

[This comment is no longer endorsed by its author]Reply