During the 2008 election, 270toWin introduced an election simulator. This was an attempt to aggregate state-by-state polling into a range of plausible election outcomes. Visit the 2012 election simulator to learn more about the methodology at work, and to conduct your own 2012 election simulation.
On this page, we take a look back at how the 2008 simulator performed against the actual 2008 election. The graphic below shows the range of results in the final 1,000 Obama vs. McCain simulations, with summary statistics to the right. The simulator correctly foresaw an Obama victory, with 100% of these 1,000 simulations showing that. The median electoral votes were 342 vs. 365 in the actual election.
It was pretty obvious who was going to win by the time Election Day 2008 rolled around, so this wasn't like predicting a Cubs World Series victory in April and having it come true. However, the results do tell us a few things:
Late polling was pretty accurate in 2008, as were our calculated polling averages (these feed the simulator). The simulator is only as good as the inputs. Given the proliferation of cell phones and other factors, the polling environment is probably more difficult in 2012. It will be interesting to see how accurate the pollsters are this time around.
As we discuss in the methodology, the simulator assumes each state is an independent event. As a result, it underplays the possibility of most or all the toss-up states going to a single candidate. We saw that in 2008, as the most common simulator results did not capture the strength of Obama's victory.
The simulator is very sensitive to changes in the polls. As late as September, 2008, the outcomes were closer to 50/50. For 2012, we will create a URL that will track and display each day's simulations, to make it easier to follow the ebb-and-flow of the campaign. Keep in mind that state polls tend to lag behind national ones, so 'race-moving' events (e.g., a convention bump) may take several weeks to fully show up in the simulations.
* How to read: If 1,000 elections were held, we would expect Obama to have between electoral votes 50% of the time.
Median: This is the midpoint. If you rearranged all 1,000 simulations from high to low electoral votes, this would be the middle outcome.
Average: This is a weighted average. If you added up all the electoral votes from the 1,000 simulations and divided by 1,000, this would be the average result.
Most Common Outcome: This is the result that happened most often in 1,000 simulations. While interesting, it may bear no relationship to the other statistics. In fact, it might show a result favoring a different party than the other statistics. (As the number of swing states decreases, the most common result will move closer to the median and average.)
Range: This is the full range of outcomes that occurred over 1,000 simulations. Essentially, it says that – given the current polls – the election outcome will not be outside this range.
Run a Simulated Election The simulator is closed for the 2008 election.
Summary statistics from the last 1,000 election simulations.
Median Electoral Votes1
Average Electoral Votes2
Most Common Electoral Votes3
Range of Electoral Votes4
Content Display Issues
A few people have reported problems viewing certain 270toWin election maps and/or polls. If you have an Ad Blocker in place, please disable it. Separately, you may not be able to view our maps in the new IE10 browser due to some changes Microsoft has made regarding the display of Flash content. This issue will not be fixed prior to the election, so you may want to visit 270toWin using a different web browser. Sorry for any inconvenience.