1. I'm not an expert on game theory but I believe the 2000 and 2004 races are significantly different from 2008 and the narrowing we saw in those elections may not occur. Certainly events could drive such a narrowing and I'll make no predictions here but let me say this: the psychology of 2-point pre-election race such as 2000 or 2004 is different from the psychology of a 7-point pre-election race which is where we've been for most of October. When your preferred party/candidate is only 1 or 2 points behind you will be highly motivated to pour money and/or time and effort into closing that gap. Of course, the other side will be working hard too but I propose that things might trend toward equilibrium in such a highly competitive environment. Now contrast this with a 7-point race where all the energy and momentum appears to be on one side. If you've got money to spend (dubious in current economic climate) but your candidate is down a lot and many are saying it is hopeless, are you going to spend that money or exert that time and effort? I think it's quite possible you won't and the tendency toward equilibrium we saw in 2000 and 2004 will never materialize. In fact, if resignation sets in (and there are signs) it could move the other way. 2. The likely voter poll results have tended to favor Republicans. It's one thing to be registered to vote and another to actual show up and vote. You know who always shows up? Older people, especially retirees. As you would imagine McCain is better with this demographic than others. Young people, who favor Obama by large margins, tend to disappoint. They also favored Kerry by a significant amount and may have turned up a bit more for him than they did in 2000 but still at a lower rate than seniors. But for the pollsters who are relying on historic trends to make prediction this year, and some are, that may not work at all. African American turnout is definitely going to exceed any election in recent memory but the exact rate is obviously unknown. And early indications are that A) Obama's lead among young voters is much higher than Kerry's and B) they are registering and early voting in huge numbers. I still think their final numbers may disappoint some but they will make up a higher percentage of voters than typical. The Obama ground game may also impact turnout in ways that confound the likely voter models. Also, the "enthusiasm gap" which has been well documented may also depress turnout for McCain. Republicans who say they are going to vote may be more prone to "forgetting" than Democrats this year. Early voting really changes the game this year as there are many more states offering this option and turnout is reportedly very high in many states. Pollsters should all be asking the following question in their remaining polling: Have you voted already? This question is even stronger than "like voter" measures and would provide very telling data. But regardless, unless pollsters have adopted more sophisticated measures of "likely voters" I still prefer looking at polls based on registered voters. 3. Because of early voting the exit polling will be more useless than ever. And don't be surprised, in states with early voting, if McCain leads in exit polls in some close states. Because we expect, based on recent elections, that exit polls are biased toward Dems, this may cause some panic on election day. But if Democrats are disproportionately voting early, as anecdotal evidence indicates, there simply won't be as many out voting on election day. Exit polls may show a conservative skew this year. |
