But is the plural of anecdote data? I can never get that straight.

That being said, there are good ways to do participant observation. Dick Fenno's done some great stuff on Congress that way.

A practice which is carried out in an entirely scientific manner, which accounts for the fact that you will typically get better odds on some horses than on others :-)

---

Heh.

We can actually predict election outcomes with a pretty high degree of consistency. Right now, for example, without having done any specific research, I'll predict that every member of the US House of Representatives that runs for re-election, hasn't been involved in a major scandal, and has raised more money than his opponent will win. I can guarantee these results within a .05 confidence interval. Presidential elections are harder, of course, because we're dealing with a small 'n' problem and wide temporal variance in conditions.

But I suspect my predictions are likely to be as precise as any specific predictions as to the impact of, say, a 1/4% increase in interest rates on the US economy.

See e.g. here and here. Note that Fair is an economist, but Tufte (was) a political scientist (before he went off the reservation); not sure about the others.

However, I would say that the most important thing in the social sciences is explanation rather than prediction. Even in physics, the fact that g=9.8 m/s^2 (32 ft/s^2) is less interesting than the explanation of why g=9.8 m/s^2. That this allows us to normally duck the controversy associated with Lott et al. is merely a beneficial side-effect. :)

Ummm. . . I can be pretty precise about a larger change in interest rates in most situations. Political scientists still seem to me to be calling horse races. I speak as the child of one and the niece of several more, though of course none of them are in elections theory. But my friend in elections theory doesn't seem to be able to predict elections until they're almost upon us either, which rather leads to the conclusion that they're guessing.

The social sciences, into which I'd lump polisci and econ, aren't that good at prediction because there are just too many variables. Economists cop out with "ceteris paribus," when they know damned well that all other things are never held constant; the world just doesn't work that way. How often does the Fed change rates by more than 1/4% at a time? Not that often. Sure, I can make a pretty good guess what's going to happen with, say, a 10% rate hike or a $5 an hour increase in the minimum wage--but how useful is that?

Plus, there is a common fallacy that crops up here-that the purpose of theory is to predict. It isn't--the purpose of theory, whether in physics, biology, economics or political science, is to explain. A good theory may well have predictive qualities, but that is because it well explains the object of study.

A simple example of a political science theory that is strong in its predictive power (and its explanatory capacity):

Duverger's Rule: That single-member district electoral systems in which winners are declared by plurality, will produce two-party systems. This is almost universally true, and the theory can account for the deviations.

Duverger's Hypothesis: That electoral systems which use proportional representation formulae to allocate seats in a legislature will produce a multi-party systems. This is also essentially universally true. Indeed, I cannot think of an exception.

And before anyone jumps on the fact that I qualify the theories, consider how many times in biology, for example, that actual practical outcomes do not fully comport with ideal theory.

I'm really not trying to piss all over poli-sci (sociology, yes, but it's generally held in contempt by most social scientists.) I think what they do is interesting and valuable, and I think it potentially can be fairly scientific. I think that economics is farther down that road than poli-sci, largely because monetary behavior is easier to measure, but I'd never argue that it's all the way there. That part of the post came out of a conversation with a friend who studied poli-sci undergrad & is considering returning for graduate studies. I think the tragedy of political science is that, even more than economics, it got hijacked by the social agendas of the fifties and sixties, a dark ages from which it only now seems to be emerging; the Marxists make interesting observers, but they tend to shy away from rigor, since rigor tends to make them look like idiots.