Polls and Predictions

At the time of writing it appears that Obama has 364 Electoral College Votes (EVs)and McCain 173. One EV in Nebraska is still in doubt. Nebraska and Maine distribute their EVs according to the popular vote in each of their Congressional Districts (CDs). Nebraska's CD 2 is still undecided though it is likely to go to Obama.

There was an immense amount of polling during the election. There are 8 National Polling Companies, many of whom published national polls on a daily basis. Some of them published more that one poll result per day as voluntary voting means that pollsters not only have to determine their respondent's choice but also their likelihood of voting. All pollsters have different methods of dealing with these uncertainties, and some publish multiple poll results using different assumptions. Most of the national polls are tracking polls, producing an average of the last few days, usually three days.

The election is actually a series of 50 simultaneous elections, one for each state. Consequently there were polls for all of the US states. For most of the states the result was a foregone conclusion. The major interest was in the Battleground States that could go either way. All of these were states that were won by Bush in 2004. There were many polls taken in these Battleground States where the election was won and lost.


Given these complexities analysis of the polls was necessary, and many Internet sites were dedicated to this analysis. These included, FiveThirtyEight , Pollster , Princeton Election Consortium , and Electoral Vote.
All of the official predictions for the sites slightly understated Obama's EV result - Pollster, Princeton and Electoral vote with an Obama Ev of 352 and FiveThirtyEight with Obama at 348.6 . The problem state for the predictions was Indiana which has not yet been declared but where Obama is in front and probably will win. The polls had McCain just in front in that state. Incidentally, Sam from the Princeton site gave his own prediction which was virtually on the button - 364 Evs for Obama.

I made a modest prediction in this post on 23rd October, that Obama would have a win "possibly by a huge margin". I suggested that it could be as high as 375 EVs, ie 105 Evs more that required. In fact, my "prediction" was 10 Evs too high.


The map below shows the final result, from FiveThrityEight.com. Note that blue is for Obama and red is for McCain. (The Americans seem to have these colors the opposite way around to the rest of the world). A cursory glance at the map gives the impression that the result was close, as there is a lot of red on the map. Many of the red states, though, have a small population and large area.


An alternative method of displaying the result is shown below - from the Princeton site. The shape and size of each state is distorted to reflect its number of EVs. This makes the size of Obama's win much more clear.