A Newsweek poll just released purports showing that the Bush lead has 'evaporated' after the first debate. 49-43 vs. 45-47 (Bush/Kerry). Back in the day, as the Monk likes to say, he and I had a great teacher who memorably declaimed that "[T]here are lies, damned lies, and a statistic." Charles Johnson at Little Green Footballs posts some quick analysis on why the poll bases might be dodgy:
Last poll (Bush 49-43)
391 Republicans (plus or minus 6) 34.58%
300 Democrats (plus or minus 7) 26.53
270 Independents (plus or minus 7) 23.88
Post debate poll (Kerry 47-45)
345 Republicans (plus or minus 6) 29.71%
364 Democrats (plus or minus 6) 31.35
278 Independents (plus or minus 7) 23.94
As someone not particularly well versed in statistical arcana, I don't know what kind of adjustments have been used to correct for any bias in the sample but if you look at the bold % figures (my addition) you can see how those changes alone could effect a big swing towards Kerry if you assume that the prefs break along party lines. This also means that the first poll was likely skewed towards Bush but the "swing" due to the debate is probably quite limited.
UPDATE (by The Monk): See here for more stats and facts about the Newsweek poll. Note also that the Newsweek poll was of REGISTERED voters, the less accurate of samples viz. a sample of LIKELY voters. This Democrat poll is based on likely voters (95% of the folks polled voted in 2000) and found a two-point bump for Kerry and a two-point lead for Bush, 50-48.
No comments:
Post a Comment