Shock: Polls Find That Democrats Don't Like Bush!

I didn't want to comment on this, because, to me, the basics of the argument have been done to death. It's the old correct-for-party-affiliation versus let-the-numbers-stand argument.

Some pollsters correct for party affiliation; some people don't, on the theory that party affiliation is not set in stone and some people call themselves "democrats" when they like the democrats and "republicans" when they like the republicans and "independents" when they don't like either.

I think the latter phenomenon is real but fairly small and should be partially but not wholly corrected for, but... some pollsters disagree.

So I was kinda bored to see that the latest polls, showing bad news for Bush, had a lot more self-idenitifed Democratic respondants than Republicans.

But... the latest samplings do seem egregiously tilted to the left.

For example, the CBS News poll, which shows that Bush's approval rating is at 35%, reports that an unweighted sample shows that 34.8% of its respondents self-identified as Democrats, while 27.6% said they were Republicans. While the unweighted sample yielded a seven-point differential favoring the Dems, a weighted sample had the spread at 11% points in favor of the Dems. This represents at least a 10-11 point swing in the electorate since the 2004 election (and perhaps as much as 14 points), when Bush won by about three points and the Repubs won the aggregate House vote by about four points.

Wait a minute-- I'm not sure if I'm understanding this right, but it seems The New Editor is saying that the CBS poll's survey already had a +7 point Democratic oversampling (the parties are about at parity, so that number should be closer to +0) and then employed weighting techniques not to reduce that partisan differential but to increase it to +11?

What the hell?

In the Ipsos/Reed poll...

Only 80% of the respondents in this poll were registered voters, while 13% of the respondents reported that they were unemployed (the current unemployment rate is about 5%-6%)...

Actually, it's 5.0%, as of today.

And the WashPost/ABC poll had a 52% sampling of Democrats and Dem-leaners, an absolute majority (versus 41% Republican/Rep-leaners). Does anyone believe that there is such an absolute majority of Democratic voters? If so, why can't any Democrat garner more than 49% of the vote?

The WashPost/ABCNews guys think this country is majority Dem, 7% independent, and 41% Republican/Republican-leaner?

The party affiliation of the nation has changed that much since the elections a year ago? Really?

One fudge-factor that I usually applied to polls was that Republicans vote in greater numbers than Democrats, so a poll of citizens -- not actual voters -- would tend to skew towards the Democrats, at least as compared to voting records.

But I'm not sure if that fudge-factor should apply anymore. For years we all assumed that there were greater numbers of non-voting Democrats than non-voting Republicans. When both parties made unprecedented get-out-the-vote efforts in 2004, however, the GOP churned out more formerly-non-voting Republicans than the Dems churned out formerly-non-voting Dems.

Without doubt, Bush's popularity is down. But this far down? Well, sure, if your poll is 52% Democrats.

Why don't they just poll 100% committed liberal Democratic partisans and see i they can get Bush's approval down to where they think it should be-- at zero percent?

On the other hand, all the polls show a higher number of Democrats than we might expect. That doesn't quite prove that the nation has, within a short year, moved 10-14 points in favor of the Democrats, but it is I guess a cause for worry.

posted by Ace at 01:11 PM

| Access Comments