I have no evidence whether this claim that Kerry and Edwards are not among the most liberal Senators is any more accurate than the rival claims that they are. I'm inclined to trust it at least somewhat because it references sources and describes exactly what data is being used. It does bring up some interesting points, though.
Remember, I'm a statistician by profession. I get paid for figuring out how to collect and analyze data. You have been warned.
First and most obvious, statistical claims are only as valid as the data on which they are based. Further, they are only as good as the sampling methods used to select a manageable number of points from that data. For example, say you decided to poll for the current Presidential election and decided that a good place to corner ordinary people would be at gas stations. That could bias the data in several ways: 1. Some men still gas up their wife's car, by virtue of the traditional gender model in which the car is the domain of the male. That could get a higher percentage of male respondants. 2. Possibly poorer people are less likely to drive, either taking the bus or carpooling to save money. Or maybe they buy smaller cars that use less gas. Or maybe they tend to drive older cars that use more gas. Whichever it is, it's a possible source of bias. 3) More people in crowded cities have access to good public transit systems and so drive less. Those city centers tend to skew more liberal, especially the Eastern cities, than rural areas.
The article linked above makes the point that 2003 voting data is not representative and thus not a valid sample.
Next, I'm getting very, very tired of hearing the word "liberal" used as a pejorative. The root is from "liberty" (or its Latin equivalent) and was formerly used to mean free in the sense of generous, as in "a liberal host". These days it's used to describe a certain cluster of political views. The thing is, it's a perfectly valid set of viewpoints. So is the set usually denoted "Conservative". Neither is intrinsically dishonorable.
The even more crucial point is that very few people believe in or even agree on all the viewpoints denoted by either convenient label. So how about we give up on empty convenience and actually argue real issues?
Final point: it's rare for a Senator to be at either extreme except as compared to other Senators. Most of them are more centrist than many of the people who elect them, precisely because they are elected by large numbers of people and because even the most homogeneous areas are not totally unanimous. The exceptions are where the only opposition is clearly unsuitable for one reason or another or where a politician has tried to appear more centrist during the election than he or she turned out to be in actual use. (The Shrub-in-Chief comes to mind here. And there I've displayed my biases, which are another thing an intelligent voter ought to bear in mind when reading anyone's statistical claims.)
Posted by dichroic at August 10, 2004 02:23 PM