Horserace - National and State Poll Update
Greetings, all. Whole slew of polls today. Here is how things break down.
Ohio
Bush: 47.3%
Kerry: 45.8%
MOE: +/- 1.8%
(Based upon 4 polls; 3,202 respondents)
Given this result, we can be 87.7% confident that George W. Bush presently has a lead.
Florida
Bush: 48.2%
Kerry: 45.9%
MOE: +/- 1.4%
(Based upon 7 polls; 5,170 respondents)
Given this result, we can be 98.98% confident that George W. Bush presently has a lead.
Iowa
Bush: 47.6%
Kerry: 45.0%
MOE: +/- 2%
(Based upon 3 polls; 2,426 respondents)
Given this result, we can be 97.06% confident that George W. Bush presently has a lead.
Wisconsin
Bush: 48.4%
Kerry: 45.4%
MOE: +/- 1.6%
(Based upon 4 polls; 3,701 respondents)
Given this result, we can be 99.49% confident that George W. Bush presently has a lead.
New Mexico
Bush: 48.2%
Kerry: 44.6%
MOE: +/- 2.0%
(Based upon 3 polls; 2,225 respondents)
Given this result, we can be 99.13% confident that George W. Bush presently has a lead.
Minnesota
Bush: 44.9%
Kerry: 47.5%
MOE: +/- 1.9%
(Based upon 3 polls; 2,610 respondents)
Given this result, we can be 3.14% confident that George W. Bush presently has a lead.
Michigan
Bush: 44.1%
Kerry: 48.0%
MOE: +/- 2.1%
(Based upon 3 polls; 2,210 respondents)
Given this result, we can be .45% confident that George W. Bush presently has a lead.
Nevada
Bush: 47.4%
Kerry: 44.0%
MOE: +/- 2.1%
(Based upon 3 polls; 2,126 respondents)
Given this result, we can be 98.75% confident that George W. Bush presently has a lead.
Colorado
Bush: 48.3%
Kerry: 45.0%
MOE: +/- 1.9%
(Based upon 4 polls; 2,896 respondents)
Given this result, we can be 99.48% confident that George W. Bush presently has a lead.
Pennsylvania
Bush: 45.9%
Kerry: 48.5%
MOE: +/- 1.6%
(Based upon 5 polls; 4,029 respondents)
Given this result, we can be .91% confident that George W. Bush presently has a lead.
National Horserace
Bush: 49.0%
Kerry: 46.1%
MOE: +/- 1.0%
(Based upon 10 polls; 10,301 respondents)
Given this result, we can be 99.997% confident that George W. Bush presently has a lead.
Bush's Most Probable/Minimal Path To Electoral Success: CO, NV, FL, IA, WI. 270-268. Probability: 93.89%.
Note that these averages are now "Zogby-heavy." More and more state polls have, in my estimation, expired. Zogby, meanwhile, has sampled 1,000 people from each of these states. Thus, the poll results skew toward Zogby's methodology. As I think that Zogby is undersampling Republicans nationwide (he is basing his partisan breakdown on the 2000 partisan turnout), many of these results might be a little pro-Kerry.
Analysis: The national horserace numbers are very striking. Bush's average result has settled at about 49% with a very low margin of error. This is trouble for Kerry. At this point, there is only about 3% of the electorate who are undecided, if this polling average is to be taken (again, this presumes that the methodology of the polls are capable of crafting a sample that reflects the ultimate turnout -- I am skeptical that this is the case). Kerry at this point would basically have to sweep the remaining undecided voters. I find this very unlikely. There is a similar effect in play in some of the swing states -- particularly NM, FL and WI. Bush is over 48% in each state -- with very low MOE's. Kerry is stuck around 46% with about 5% undecided. Kerry would need to run the table on the undecideds there as well.
Important reminder: I have two major problems with polls. The first is strictly statistical (i.e. MOE's are simply too high for one poll). My unweighted averages solve this problem to a good degree. My second is methodological. I think that most of these major polls have methodological problems -- in particular, I think they are underestimating GOP turnout in many instances. However, it is impossible to make any certain claims about the skew that methodological problems might produce.
In my overall thinking about this race, I think the above averages are sufficiently sound in their statistical foundation that they can be used as a measure of analysis, not the measure of analysis. For instance, the situation on the ground in OH, FL and MN incline me toward an opinion that these averages are under-reporting Bush and over-reporting Kerry. Just bear that in mind -- in particular, when I say, "we can be xx.xx% confident..." all I mean is that, assuming that the combined polls provide an accurate sample of the ultimate electorate, we can be xx.xx% confident.
posted by Jay @ 6:38 PM |
