Conservative
Type 40
Methodology We take current state-by-state polls and turn the results into probabilities. Each simulation picks a winner in each individual state, based on the probabilities for that state. For example, if Romney has a 55% chance of winning Nevada, he will, in the long run, win Nevada in 55% of the simulations conducted. Some uncontested states (e.g., Utah) will always yield the same result. The simulator does not consider the possibility of split electoral votes in Maine or Nebraska.
The simulator is not a predictor of the election. It provides a range of electoral outcomes that are plausible if the election were today, if the state polls were accurate and if each state were a fully independent event. The election is still months away; state polling is still infrequent and may be stale. While each state is, in theory, a separate election, the reality is that there are usually correlations. For example, if Obama were to win Georgia, he wouldnt lose Virginia. However, something like that is possible in a given simulation. As a result, this model underplays the likelihood of toss-up states breaking heavily for one candidate.
You will get different results based on polls and methods mentioned above. In every case Obama is shown to win. I suppose if you had nothing better to do than run it dozens of times, it might show Romney winning occasionally. I ran it 28 times and Obama won every time from about 289 to 332 electoral votes.
You do realize that that is a left wing site....right?
the point he is missing is that it's a random generator, with no basis in reality. Yesterday, I ran it dozens of times and got Romney to win about 3-4... today I ran it 10 times and got Romney 6 wins.
Course is lying when he says he ran it hundreds of times and Obama wins every time.