Thursday, July 07, 2005

Terror Statistics (or Correlation is not Causation)

Here's a post at BTC News about the number of terror incidents reported in recent years. (Via Atrios)

It is an interesting post about the numbers and how they are arrived at. As with all statistics there are many possible data sources and Terrorist Attack has to be defined in certain terms in order to create an accurate picture. There is some debate about how things are defined and depending on how the statistic is defined we get different pictures of the situation. This isn't really surprising and I will leave this whole debate to people who know more about international terrorism than I do.

I would like to point out the final paragraph though under the graph

"The axis labels on this graph are hard to read, but the terrorism rate decerases steadily through Clinton's second term, bottoming out in the year 2000, and then it rises again dramatically. I think this graphis one of the best ilustrations I've seen yet of the difference between what Karl Rove described as the "liberal" and "conservative" approaches to combatting terrorism."

The graph itself shows a steady decline in terrorist attacks until 2000 and then they rise again quickly.

This final paragraph is one of the best illustrations I've seen yet of how not to use statistics. Though some may disagree with me I do not think that the President of the U.S. has that much of an affect on terrorism and this graph is not convincing evidence.

This graph and the conclusions drawn by the writer of this article is a great example of the thing that I must have explained 5 times in every lab I taught during college. Correlation does not imply Causation. Just because two things LOOK related does not mean that they ARE. There are mediating factors, one big one being the economy. It is my humble opinion that a less thriving economic athmosphere in the world could result in larger numbers of unhappy people who think it is a good idea to blow things up. And besides the economic factors there are any number of political and sociological factors that could have contributed to the sudden increase in terrorism. Some of these things may relate to US Foreign Policy and some of them may not.

For liberals to imply that the sudden rise in terrorism during Bush's term in office is because of conservative policies is simply false. Just like it is incorrect for conservatives to imply that the sudden increase is due to the Liberal policy of placating them until they gained enough power to strike hard and fast.

Just because I go to the bathroom everytime I'm in the bathroom doesn't mean that the bathroom causes me to pee. Get it? There are other factors at work here besides who is President. To discover what they are will require a much more detailed analysis of the terrorist activity levels, events leading up to 2000 and who exactly the terrorists are that are committing these acts. And even then we will still be guessing about what caused it.

I'll tell you whose fault it really is, the damned terrorists.


Left_Wing_Fox said...

That's a strawman I'm afraid. The argument is not so much "Conservative politics cause terrorism" but rather that for a president who makes a "War on Terror" the certerpiece of his policy and politics, those policies and politics don't seem to have any sort of impact on the number of terrorist incidents.

In other words, if you use the toilet every time you go to the bathroom, the bathroom isn't necessarily to blame. But if you take a pill to cure incontenance, and wind up using the toilet more often despirte the claims of the package, wouldn't it be fair to assume that the pill isn't working?

Shinobi said...

Way to stay with the bathroom analogy. I think that is an excellent point. Obviously the Bush Administration has not yet accomplished their mission or anything close to it judging by these statistics.

But I don't think that is the point that was being made. Maybe I'm reading wrong. It seems like they were trying to compare Clinton v. Bush on terror and saying that there were fewer terrorist attacks during the Clinton Administration because he was doing a better job, and I don't think these statistics can show anything of the kind.

Left_Wing_Fox said...

I can see where you're getting that from, but I don't think the goal of this is to necessarily claim that Clinton is better, but rather to combat claims by the Republicans that they are more successful at combating terrorism than liberals, especially the comments from Karl Rove. Remember? "Conservatives saw the savagery of 9/11 in the attacks and prepared for war; liberals saw the savagery of the 9/11 attacks and wanted to prepare indictments and offer therapy and understanding for our attackers."

Far from being repudiated, Rove's remarks have been repeated and reinforced by the political right in America. The meme that only the Republicans can save us from terrorism has been a constant and consistant line from virtually every right-wing politician and pundit since 9/11. If this graph offers little proof that Clinton was necessarily "better" at combatting terrorism, it certainly dosen't support in any way, shape or form that idea that Bush or the Republican controlled government has been a success in his "War on Terror" by even the most basic measure: the number of attacks and fatalities from terrorist actions.

hubris said...

I don't think you were fighting a strawman, Shinobi, I think you were right on.

The other thing that isn't considered in their interpretation: The opposition is not necessarily static in terms of their efforts/attempts.

In other words, if you are going to the bathroom more often but it's because someone is forcing you to drink two-liter after two-liter of coke, it isn't necessarily a sign that your incontinence pill is failing.

Of course, it could be argued that strategy could have an effect on the overall coke supply (i.e. cutting off the causes of terrorism at its roots), but to ignore the variable is a gross oversimplification, I think.