Another look at some recent (outlier?) Presidential preference polls

By gkalyanaram

John McCain’s pollsters have argued that the recent LA Times/Bloomberg is not reliable because the party identification numbers in the sample of this poll are clearly out of the known historical and current data.  I think that they are right.  I have already made this argument in my earlier posting. But here is another look at the numbers.

The historical data for party identification among the voters is 37 percent and 37 percent as Democrats and republicans respectively in 2004, and 39 and 35 in 2000, 40 and 35 in 1996, and 38 and 35 in 1992.  So the historical average over the last two decades is about 3 percent in favor of Democrats.

For the current Presidential election season, the most useful benchmark may be Pew Research Center’s preference measure in the first two months of 2008.  In the height of exciting Democratic primary, the party identification numbers were 36 percent Democrats and 27 percent Republicans.  Given that the sample size was over 5000, this number is quite credible.

So LA Times/Bloomberg poll’s sample of +15 percent Democrats is clearly high, as also +19 in the Newsweek sample.

If you adjust the party identification advantage to be 8-9 percent in favor of Democrats, Obama would lead by about 7-9 points.  There can be no quarrel with that.

Tags: , , , , , ,

2 Responses to “Another look at some recent (outlier?) Presidential preference polls”

  1. Ben K Says:

    I dont doubt Obama is up, but you cant do a serious poll with a sample of 22% Republicans.

  2. gkalyanaram Says:

    Ben –

    I agree with you. That’s my point too.

    However, there is possibility — somewhat distant — that the party identification gap may be unusually high this year because of tough times for the Republican party. May be the party identification is going to be as high 15 points in favor of democrats. Unlikely.

    Another point is that quite a few independents are beginning to identify themselves as Democrats, I think, and that may be increasing the gap.

    GK

Leave a Reply