Observations

• The graph above includes events that may have influenced the candidates' probability of winning. We have tried to find events that were turning points in the election campaigns. If there are other events that you think caused some of the turning points in the election, please let us know via the contact page.
• We can see from the graph that Palin's announcement as VP candidate did little to change McCain's probability of winning, but Palin's RNC speech seems to have made a big impact. From around the time of Palin's RNC speech, Obama started falling precipitously, falling below 50% around September 12 2008 and reaching a low of 36% around September 16.

• It also seems like the financial collapse of 2008, which was "ushered in" by Lehman Bros' collapse on September 15, 2008, along with McCain's handling of the collapse (suspending his campaign to go to D.C. to "help fix" the economy), helped reverse Obama's fall and McCain's rise. One week later, Obama was back at 65-70% and McCain down to 30-35%.
• The final nail in the coffin for McCain's campaign seems to have been Palin's interviews with Couric in late September, especially the third interview on September 30th. Shortly after that, Obama jumped to 90% and never looked back. His probability of winning kept increasing after that until election day 2008, when his probability of winning was estimated to be around 99%.

States with most potential for Obama upside/McCain downside

In adition to an estimate of the probability of winning the election, it's also instructive to see how sensitive this probability is to changes in state poll results. That is, which states impact the result the most if their polls change in favor of one candidate or another?

In the figure on the left, we see the resulting probability of an Obama win, given changes in state polls from the five states with the most potential upside for Obama or downside for McCain.

To keep the figure easy to read, it only includes the resulting probability of an Obama win. The resulting probability of a McCain win was simply 100 minus the probability of an Obama win.

As of November 3rd 2008, just before the election, Ohio, New Jersey, Florida, Virginia, and Pennsylvania had the most potential upside for Obama. The upside was limited because Obama was already around 98.8%. Nevertheless, if any of the state votes changed by 10% in favor of Obama (i.e. 5% switched from McCain to Obama in any state), Obama's probability of winning would have increased to 99.6%-99.8% and McCain's probability of winning would have dropped to 0.2%-0.4%

States with most potential for Obama downside/McCain upside

This figure is similar to the previous one, with the only difference being that it corresponds to changes in state votes from the five states with the most potential downside for Obama or upside for McCain.

Again, to keep the figure easy to read, it only includes the resulting probability of an Obama win. The resulting probability of a McCain win is simply 100 minus the probability of an Obama win.

As of November 3rd 2008, just before the election, California, Pennsylvania, Michigan, Massachusetts, and Illinois had the most potential downside for Obama. The downside was limited because Obama had such a big lead and it would have been diffcult to move him from there. Nevertheless, if any of the state votes changed by 10% in favor of McCain (i.e. 5% switched from Obama to McCain in any state), Obama's probability of winning would have decreased to 98%-99% and McCain's probability of winning would have jumped to 1%-2%

Probability of achieving different electoral vote results

In addition to looking at probabilities of Obama or McCain winning, we can also look at the probabilities of achieving different electoral vote results. That is, what is the probability that Obama or McCain would have collected say, 270, 285, or 300 electoral votes? (Note that the probability of either of them winning the election was the probability that they would collect 270 or more electoral votes).

The figure on the left shows the probabilities of each of the candidates achieving given electoral vote totals, which vary from zero to 538. The points where the curves are the highest correspond to the most likely electoral votes for each candidate, as of November 3rd 2008, just before the election.