**Update on 2/25/08: With the number of candidates that have a realistic chance of being a nominee of the two major parties down to 3, we’re seeing more head-to-head polls come out. We’ve created a map that displays those polls for McCain vs. Obama and these numbers power our “Probability of 270 feature” which is discussed below. As we mention on the aforementioned map — and want to strongly reiterate here — state polls at this early point, while interesting, are of limited predictive value, and likely will be quite volatile until at least the party conventions.**

===

We’ve added an interesting new feature for 2008 that we call the “Probability of Reaching 270″. This feature brings together your interactive map with current polling numbers and is a powerful new tool that we believe is unique to 270toWin.

When 12 or fewer states remain undecided on your 2008 map, these probabilities will dynamically update as you modify the outcome of a state. The idea here is to give you a snapshot, specific to your customized map and based on current polling, of how likely it is that each candidate will win.

These are mathematical calculations and not our predictions for the election. 270toWin is non-partisan and does not exist to make predictions. The purpose of 270toWin is to give you, the user of the site, the tools you need to make your own predictions.

A little background: Although it is rarely presented this way, statistical theory makes it possible to take the results of a scientifically-conducted state poll and calculate the probability that each candidate will actually win that state’s popular vote, and thus collect its Electoral Votes. Do that for each of the 50 states + DC, and sum up the likelihood of every possible combination and you can generate a probability that the candidate will reach the required 270 Electoral Votes.

Example: Let’s say that a scientifically conducted poll for Missouri (11 Electoral Votes) comes back 48% for Candidate A and 44% for Candidate B, with 8% other, undecided or some other response. The question we are trying to answer is: “Given these results, what is the probability that Candidate A will actually get more votes than Candidate B in Missouri?” Assuming a sample size of 600 (see #3 in Important Notes, below), the answer is about 85%. That means there is an 85% chance Candidate A will win the state’s 12 Electoral Votes.

Now let’s say that Missouri is the only undecided (tan) state on your interactive map and neither candidate is yet at 270, but either would reach 270 if they won Missouri. If you look down in the “Probability of Reaching 270″ area below the map, you would see that the party for Candidate A would show an 85% chance of reaching 270…. as there is an 85% chance they win Missouri, and that is the only thing still to be determined.

It gets more complicated pretty quickly as more states remain undecided, as every possible combination of winning/losing must be evaluated. For example, with 12 states remaining, there are over 4,000 possible outcomes…with 20 states, there are over 1,000,000. However, the fundamental approach remains unchanged regardless of how many states are undecided.

Important Notes

1. Most polls provide a result, with a margin of error (e.g. +/- 4%). Essentially what we are doing here is reversing that presentation to say – “Given the margin of error and the polling percentages, how likely is it that the poll’s predicted winner is correct?” Note that all we are interested in is the probability of winning here…a victory by 10 votes is the same as a victory by one million votes… the winner gets the Electoral College bounty for that state.
2. The calculated probabilities are valid even for results that are within the margin of error. As long as the poll is conducted properly, we can calculate how likely it is that the poll’s predicted winner is correct. Obviously, as the results narrow, the probabilities get closer to 50/50.
3. The sample size is important to the calculations as it impacts the margin of error. Larger sample size means a smaller margin of error. A smaller margin of error impacts the probability that the poll’s predicted winner is correct. For example, a Candidate A 48%- Candidate B 44% poll with sample size of 600 yields an 85% probability that A > B, while the same 48-44 result with a sample size of 1,200 yields a 93% probability that A > B. Sample size is often shown as “n =” when poll results are displayed.
4. Ties (269-269) can occur and the probability of a tie will be displayed in your results.