October 7th, 2004


(no subject)

Over the past few weeks, I have noticed an odd trend. Whenever someone presents cogent evidence to the effect that this is anyone's race, it's almost inevitable that someone will say...

"But what about electoral-vote.com? They have Bush up by ____."

Yes, I say, but that doesn't necessarily mean what I think you think it means...

Recently, electoral-vote.com changed their methodology in a way that is less favorable to Kerry. They are averaging the results of both new and old polling data to get their results. Prior to this change, they could easily be accused of "cherry-picking" the polls they used. By averaging poll results, however, cherry-picking becomes less of a potential issue... however, it also means that their results less accurately predict where we are right now as opposed to where we were when the data was taken.

If you look at their polling data, you'd see that some of their latest data is over a month old. As far as this race is concerned, that's forever. By averaging significantly older data with new data, all you are doing is making the new data less accurate... and when polling itself is such an inexact science, that's asking for trouble.

Out of the 343 state polls used at electoral-vote.com to create their current projections, only 48 of the polls, or approximately 14%, were taken after the first debate... and even those results are being averaged with results that, in some cases, are over a month old.

In short, electoral-vote's methodology sucks because they do not have enough fresh polling data to accurately judge this election by. Using their current methodology, it is unlikely that we will ever see the full impact of even the first debate in their eletoral vote results prior to Election Day-- much less the impact of all subsequent debates and events.

So, where do you get good information on where this election stands? Wherever the data is the most current with the best methodology. That tends to mean large polls, such as Gallup, Zogby, etc. Even then, you have to take the poll results with a grain of salt, as what is being polled is usually "likely voters". Recent Gallup polls balanced their likely voter methodology by using 12% more Republicans in their polled group than Democrats. While Republicans do tend to vote more often than Democrats, they do not vote nearly 30% more often per person, which is what a focus group of 12% more Republicans might indicate.

It should be pointed out that Gore trailed in the polls by 5% amongst likely voters on the eve of the last election, but won the popular vote... and there are more organizations and mechanisms in place to get out the vote this year than in prior elections. These "get out the vote" drives make a real difference in the race, and this year, they are larger and better organized than we have ever seen... but they are also the kind of factor that isn't seen in polls. Why? Because likely voters are generally screened on the basis of their participation in previous elections. College students who have never registered to vote before, by definition, aren't counted in such polls. Likewise, other demographics -- such as minorities and women -- are underrepresented amongst "likely voters".

Which goes back to what I've said in the past. This election will be decided by unlikely voters and by turnout. Whether you look at Zogby's battleground states poll or even at electoral-vote.com, many of the races are simply too close to call. There is a very real possibility that the concentration of the Kerry campaign, advertising, and the numerous "get out the vote" drives on key battleground states could pay off, potentially leading to a race where George Bush receives the most votes -- largely due to significant majorities in uncontested states in the South -- but John Kerry wins the overall race with victories in the swing states. There is every indication that we will see some incredibly tight races and several surprises... again.

The only reasonable thing for any of us to do is assume this race will be another nailbiter and act / vote accordingly. Although in the last election every vote was not counted, there is no denying that, on the whole, every vote counts.