« Final Pre-election Poll Analysis | Main | National Tracking Poll Roundup »

Bush's Disappearing Lead in the CBS News Poll

Yesterday, two days before the election, CBS News reported a 3 point Bush lead among LVs. That survey covered October 28-30. Today, they released survey data covering October 28-31 which show Bush and Kerry now tied among RVs (46-46) and Bush up by only a point (47-46) among LVs. Since these data only differ from the data released yesterday in the inclusion of interviews from Sunday, that suggests Sunday was a strong day for Kerry in this poll--and, perhaps, in others.)

It's interesting to note that, in 2000, Bush was also leading (by 5 points) in the CBS News poll released 2 days before the election. But that lead also pulled a disappearing act between then and election eve. History may be repeating itself.


Very interesting data from Harris Poll about cellphone only voters:

People with cell phones and no landlines
One subject of much discussion in this election is that a growing number
of people have no traditional "landline" phones but have cell phones and
cannot be reached by traditional telephone surveys.
Data from our online survey suggest that six percent (6%) of all likely
voters now fall into this category and that they are voting strongly for John
Kerry by 56 to 38 percent over President Bush.
These results suggest that telephone surveys which miss almost everyone
without landlines will, even if they are accurate, slightly overestimate
President Bush's vote and slightly underestimate Senator Kerry's vote. This
also explains a small part of the difference between our online and telephone

This and other polls seem to indicate the 'break' to Kerry. Is that youe take? Will there be more released tomorrow to calm my nerves?

Bush's LV support on Sunday in the CBS poll was approximately 42.4%. Wow!

This is based on the calculation that about 404 respondents on Thursday through Saturday supported bush (49% of 824 respondents) and 561 respondents did so on Thursday through Sunday (47% of 1194 respondents). The number of respondents on Sunday was 370 (1194-824), so Bush's support was about (561-404)/370=.424, or 42.4%. The calculation is not exact, since it is based on the likely voter model, and could be biased if there was substantial change in the proportion of likely voters on Sunday. I couldn't find the RV numbers from the earlier period, so had to do the calculation for LVs.

In any case, the main point is clear: on Sunday, Bush's support in the CBS poll collapsed.

Did my comment about the CBS poll implying 42.4 percent support for Bush on Sunday go through? My browser flipped out just as I was posting.

P.S. the 95% confidence interval on those Sunday results is +/- 5%.

Do any of the polling organizations feel embarassed?

Following the critical analysis of Gallup on this site, and elsewhere, I read their "final" numbers today with simple astonishment. Apparently, at the last moment, rather than have the election prove their consistent bias, they have fudged, to give the impression, that their poll is predicting a deadheat!

The Corporate Right-Wing Media should and must be destroyed.