In this column, Broder writes about going to Michigan to "check the public reaction to the first debate of 2008." He does this by doing "man-on-the-street" interviews with people, "outside a Sam's Club in one shopping center, outside a Target in a second and on a downtown...corner with three fast-food restaurants." This, on the face of it, is really stupid. Anecdotal evidence is almost completely useless when one is trying to gauge public opinion. Standing outside a Sam's Club or some fast food restaurants doesn't change that, despite the implicit suggestion that "real Americans" can be found in these places. These types of interviews can be used to illustrate findings in scientific polls, certainly, but Broder's column is based entirely on them, despite the fact that there has been ample public polling coming out of Michigan in recent days.
Broder does mention these polls at the very top of his column writing:
The Michigan polls I had seen...drew a confusing picture. One had Obama ahead by four points; another by seven, a third by 10 and the last by 13.
A confusing picture? Really? There were four polls, all showed Obama ahead, three of the four showed Obama ahead by statistically significant margins, and two of them showed him with double digit leads. What's confusing? These polls all point strongly towards a substantial Obama lead. For some reason, Broder pretends that four polls showing an average Obama lead of 8.5 points is "confusing."
Then Broder explains that he's going to talk to voters in Oakland County, Michgian because Oakland County, "is a classic bellwether area." It's such a bellwether, writes Broder, that:
In 2000, when Al Gore carried Michigan by 5.1 percentage points, he won Oakland by 1.2 points. In 2004, both were even closer. John Kerry won Michigan by 3.4 percentage points and Oakland by one-half a point.
Ok, this is getting ridiculous. Generally, when we talk about "bellwethers" we're talking about places that reflect the views or preferences of the larger public. A "bellwether" state, for example, is one in which the vote closely aligns with the national popular vote. Similarly, if Oakland County were really a "bellwether" county for Michigan, then its votes in 2000 and 2004 should have been in line with the statewide vote. But it wasn't. In fact, as Broder tells us, in both years, Bush did about 3 percentage points better in Oakland County then he did statewide. Oakland's not a bellwether, it's more Republican leaning than the state as a whole.
Broder seems to be confusing a "bellwether" with someplace that's evenly divided. Think about the 1988 election, for example. In that year, George H.W. Bush won the popular vote by about 8 percentage points. In Pennsylvania, however, he won by only about 2%. Does that make Pennsylvania a "bellwether?" Of course not. If you treated Pennsylvania as a good indication of the election as a whole, you'd have a skewed picture of the outcome. Ohio, on the other hand, went for Bush that year by 9%, a much better "bellwether."
Look, I don't mean to harp on this one point, but a lot of people read David Broder and this column is just full of laziness. It really reads like he wrote the column two weeks ago and didn't bother to make it fit reality. Michigan's not actually all that close, Oakland County is not a good bellwether, anectodal interviews are utterly useless and don't illuminate anything, and finally, the column ends without making any coherent point.
At the risk of being glib, Broder is a good "bellwether" for the chattering classes. They aren't evil or stupid or biased, they're just lazy and suffer terribly from a lack of creative and critical thinking. This column is a perfect example of those weaknesses.
No comments:
Post a Comment