More on polls

[mmcirvin]

This Terry Neal column in the Washington Post does a good job of explaining the systematics in various opinion polls, specifically on the presidential job approval question. The Post/ABC poll has traditionally been relatively sympathetic to Bush, with some times when it was a real outlier. That can happen for reasons having nothing to do with the bias of the pollsters. In general, the absolute values of these numbers mean much less than the trends.

Neal mentions the odd fact that Nixon, Carter, Reagan, Bush I, and Clinton were all in the same neighborhood of public job approval at this point in their administrations. I think this is more coincidence created by small sample size than anything significant. Including Jimmy Carter is actually a bit of a cheat: he only got up to above 50 percent because of the rally-round-the-president spike associated with the beginning of the Iranian hostage crisis, which dissipated in a big, big hurry. (It's hard to remember now that there even was such a spike.)

If you want a really eerie, probably meaningless coincidence, look at how closely Bush's numbers are tracking Kennedy's. Kennedy's late spike was from the Cuban missile crisis. I think it's amazing how popular he was initially given that he had nothing analogous to the Sept. 11 spike and was elected almost as narrowly as Bush II. There was something very unusual in the air in the early JFK administration.

(Charts, as always, from Pollkatz).

Comments

[ronebofh.livejournal.com]

You know, as we head into late November of Bush's third year, telling me about Bush's number tracking JFK's is a bad thing. YOU ARE BAD!

Plus i just saw "Zoolander".