Comments on: It’s Obama! (Or is it?): Forecasting the 2008 election

By: UnderstatementJones

UnderstatementJones — Thu, 16 Oct 2008 20:07:33 +0000

I’m late to the party, but…

So great. Elections equals economy plus voter identification.

Therefore, if you want to win the next election, you should get more people to identify with your party.

Until political scientists can identify and model the factors that underlie voter identification, I’m not sure they’ve produced anything of real use.

By: Okobojicat

Okobojicat — Mon, 22 Sep 2008 15:05:16 +0000

I am so glad they guessed at the national popular vote, considering that does not mean a single thing in an election. They should be guessing (forecasting) and the Electoral College vote and if they need to, explain the three or four critical states in this election: Colorado, Virginia, Ohio and possibly Pennsylvania or Florida. That makes a heck of a lot more sense than just giving % of vote received.

By: Polemarchus

Polemarchus — Sun, 21 Sep 2008 17:23:35 +0000

Before I decide whether to trust these projections, I’d like to see how these models did in another extremely close election: Bush-Gore 2000. Was Bush the safe bet then, according to these forecasts? (Or was that very election your “almost”?)

Next question: what effect does the historical success of these models have on the controversy surrounding rational choice theory as a methodology for political scientists? Do forecasts like these work as well on smaller scales like, say, local or state elections?

Lastly, and somewhat facetiously: if these models have been working accurately for many elections, predicting the winner with the accuracy Professor Dickinson cites, how come Presidential elections in this country are dragged out for so damned long? Doesn’t this make the daily campaign tit-for-tat moot? If the winner is predicted so far in advance, why bother with tracking polls, debates, gaffes, and sleazy campaign ads?

Having asked all this, I wonder how these methods differ from things like the Iowa Electronic Markets…

By: Bhima

Bhima — Sun, 21 Sep 2008 17:21:04 +0000

But Matt, where is YOUR prediction ?

Given that the predictions are a measure of the popular vote, I would imagine that unless the vote is close, the electoral vote would represent the popular vote, hence the margin of error in the predictions is very critical. Can you comment on this ?

It would be interesting to know to what quantitative extent the above sources have been succesful in prediction the vote outcomes of past elections. Ascribing a confidence level based on this (a weightage if you will) would perhaps be useful rather than carrying out a simple arithmetic average..

As you had stressed in the past, this is an election of many firsts. One for which there is, as I understand it, quantitative data is the discrepancy between intent and action in voting for black candidates. Are there other such effects that can be factored in along with this ?