Wednesday, September 22, 2004


Battling polls

Bush still appears to be ahead of Kerry in Ohio, but is it a small margine or a large one?

On one hand, we have as Exhibit A, the new Zogby poll done Sept. 13-17 that shows:
Bush 50.1%
Kerry 46.8%
Nader 0.3%
MoE 2.4%

The Zogby trend shows the two campaigns approaching a dead heat again. Zogby has somewhat of an unusual methodology in that it conducts online polls (for the Wall Street Journal) twice a month. This covered 1,718 Ohio respondent and demographic info submitted by respondents is cross checked for accuracy.

On the other hand, Exhibit B is the new Ohio Poll from the Univ. of Cincinnati of 456 Likely Voters conducted Sept. 12-18:
Bush 54%
Kerry 43%
Nader 2%
MoE 4.6%

The previous Ohio Poll (August) had the race as a dead heat. It appears that the Swift Boat ads and the RNC had and effect. Bush’s net favorability rating is 11 percentage points higher while Kerry’s net favorability rating fell from a positive rating (+5%) negative rating (-7%).

The accuracy of the Ohio Poll has been way off in some elections. I also don't think their focus on Likely Voters adds accuracy. I also find it odd that they did it over 7 days. That's an awfully long time in the polling world. Also I find it odd that they only polled 456 since they contacted 812 in August. Regardless of the stated MoE, that small number has the potential for introducing a lot of other error sources. For example, that means that they had roughly 5 respondents per county. The chances of each of those sets of 5 being representative of the county as a whole is not desireably high.

Another major warning about potential methology problems is that the poll shows undecideds at 1%. That, in itself, is absurd.

More polls are due in the next few days.


<< Home

This page is powered by Blogger. Isn't yours?