Monday, 22 September 2003

Understanding science

David Adesnik apparently has been drinking the Perestroikans’ Kool Aid:

The secret to success in America’s political science departments is to invent statistics. If you can talk about regressions and r-squared and chi-squared and probit and logit, then you can persuade your colleagues that your work is as rigorous as that of a chemist, a physicist, or (at worst) an economist.

Funny, I just came back from spending a month with people who told me that the absolute worst way to get a job in political science is to “invent statistics.” If David means “understand and be able to utilize” by “invent,” that is. If he means something else, I can’t figure it out.

Still, it is absolutely impossible to explain the tactics of Al Qaeda or Hamas without reference to their perverse ideologies.

It is? Actually, it’s pretty easy to explain their tactics—historically, they’ve been quite effective. What’s (slightly) more difficult to explain is why Al Qaeda and Hamas engage in terrorism while the Sierra Club and Libertarian Party don’t.

The real problem is that [Robert] Pape, like so many political scientists, abandons all nuance in deriving policy programs from his work.

Fair enough. But what exactly does that have to do with the fact that Pape uses quantitative methods in his research? Adesnik claims:

As I see it, the cause of this unsubtle approach is political scientists’ obsession with statistics, a pursuit that dulls their sensitivity to the compexity of real-world political events. If numbers are your thing, you’re going to have a hard time explaining why Israelis and Palestinians have spent five decades fighting over narrow tracts of land.

So then, what is to be done? As you might of heard, many political science programs require training in statistics but not foreign languages. That trend has to be sharply reversed.

Great. Now we can have more social scientists who are completely incompetent at quantitative methods, but at least can express that incompetence in multiple languages. Where do I sign on to this initiative?

Look, I’m more than willing to concede that quantitative research doesn’t—and can’t—answer every interesting question in political science. But the rigorous study of politics can, and IMHO should, be scientific: founded on the scientific method, no matter whether the actual methods used are qualitative or quantitative.

And—irony of ironies—the APSR piece that Adesnik vents his wrath at is completely qualitative (at least in terms of its method of inference). Not a p-value, χ², or logit model in sight.

Anyway, you can read the piece yourself courtesy of Dan Drezner, at least until the APSR’s copyright goons come after him.

Expectations management

Why am I getting a weird feeling of dejà vu from reading the New York Times’ alleged sneak preview of Tuesday’s UN speech by George Bush?

According to the officials involved in drafting the speech, for an audience they know will range from the skeptical to the angry, Mr. Bush will acknowledge no mistakes in planning for postwar security and reconstruction in Iraq. ... In the speech, Mr. Bush will repeat his call for nations — including those that opposed the Iraq action — to contribute to rebuilding the country, but he will offer no concessions to French demands that the major authority for running the country be turned over immediately to Iraqis.

Wow. Maybe he’ll also storm out of the room in anger and call people in the audience names.

9/11, Terror, Saddam, ad nauseum

Steven Taylor of PoliBlog notes a Wall Street Journal editorial on Iraq’s al-Qaeda ties and the capture of Abu Abbas on the outskirts of Baghdad.

Meanwhile, my friend Scott Huffmon forwards a collection of quotes from administration officials that juxtapose 9/11 with Iraq (Scott therefore wins the longstanding Signifying Nothing “no-prize” for forwarding evidence of the adminstration linking Saddam and 9/11). Perhaps more interesting is the associated article discussing how the public’s belief in a 9/11-Saddam connection came about. Key graf:

A number of public-opinion experts agreed that the public automatically blamed Iraq, just as they would have blamed Libya if a similar attack had occurred in the 1980s. There is good evidence for this: On Sept. 13, 2001, a Time/CNN poll found that 78 percent suspected Hussein’s involvement—even though the administration had not made a connection. The belief remained consistent even as evidence to the contrary emerged.

Or, as I am fond of saying, when it comes to politics, it’s all heuristics.

Mystery Red Hat upgrade bugs

I’m spending most of this afternoon slowly unravelling whatever went wrong with upgrading one of our boxes from Red Hat 7.3 to Red Hat 9. Main problem: none of Red Hat’s GUI administration tools work—they all die with segmentation faults. Neither did sendmail (which I promptly booted out in favor of postfix.)

In the process of straightening everything out, I installed apt-rpm. We’ll see if that makes the system slightly more administerable (is that a verb noun?).

The mystery deepens. Apparently, somehow the PyGNOME installation is hosed. However, it's only intermittently hosed; most of the Red Hat admin tools segfault, but some don't (they're just oddly buggy, like the Package tool that won't let you select things in the Details view). And Foomatic-GUI runs just fine (once wget is installed—no, don't ask me why). Damn strange.