I use Monte Carlo Simulation and the Normal Distribution function applied to state polling to determine the probabiltiy of Kerry getting over 270 EV's via 5000 simulated election trial runs.
I also use the N.D. function to the KERRY MEAN in 2 national poll groups (the 9 Independents and the combined 18 Indies and Cororates) to determine the probablity of Kerry getting over 50% of the popular vote..
The corporates have Kerry way ahead. The Indies have it dead even.
If its dead even, Kerry wins, since he gets the undecided/other vote.
Its that simple.
This should clear up some of the questions (and incorrect assumptions) regarding the model, that I have seen posted here:
Election Model Methodology: A Primer
For all those non-mathematicians and non-statisticians who may be interested, this is a simple guide to the methodology used in the Election Model:
You can find the latest Election Analysis here:
http://www.geocities.com/electionmodel/There are three basic methods which are used to track elections and predict a winner. Each method utilizes different sources of data as a basis in forecasting a winner.
The first method analyzes economic factors: growth, jobs, inflation, etc. Economic and political forecasters have had success using this approach (after all, this is what they do for a living) by employing an econometric-like model using multiple regression and/or factor analysis. The derived formula weights these variables in order to predict the popular vote. How some forecast a 58% popular vote for Bush, considering the economic pain inflicted on the nation since Jan. 2001, is a mystery to me. But I’m not an econometrician.
A second method tracks national polls and projects the potential movement of undecided or third party voters to predict a popular vote. There are about 15 major national pollsters. Predicting a majority vote does not mean the winner will gain 270 electoral votes, but for all practical purposes this can only occur in extremely tight elections, when the margin is less than 0.5%. In fact, in a 51-49 popular vote split, there is virtually a zero probability that a popular vote winner would lose in the Electoral College. It didn’t happen in 2000, because we know Gore won Florida, but the Supreme Court voted 5-4 for Bush..
A third method tracks individual state polls in order to predict the number of electoral votes. The focus is mostly on the hard-fought twenty battleground states.
In the Election Model, I use a combination of methods two and three. I believe that polls are pretty good indicators, provided that they are fresh and non-biased. They have been pretty accurate historically.
Like others, I use the polls as a starting point and make adjustments from there. The adjustments are simple: I allocate the precentage of undecided/other voters to Kerry and Bush. The undecided voters usually split 2-1 for the challenger, so as a conservative base case, I assume 60% will vote for Kerry. For example, if a poll is tied at 45-45, than Kerry is actually “leading” by 51-49%, since he can expect to receive 6% of the undecided 10%, based on the 60% allocation.
One advantage of national polling is its simplicity. To determine who is “ahead” in the horse race, we only need a single number: the point “spread” between the candidates. If the spread exceeds the polling margin of error (MoE ), which is typically +/-3% for nationwide polls of 1000 sample size, then it is a statistical fact that the leader has at least a 95% chance of winning the election. But that is just the probability for ONE poll only.
Let’s assume three (3) polls (or equivalently a single 3000 sample-size poll) are considered. In this case the MOE tightens to +/-1.80%. If we assume the average of the three polls is 52-48%, then the probability is 95% that the leader will receive somewhere between 50.2-53.8% (a 3.60% range) of the vote. Now add 2.5% for the possibility that he will exceed 53.8%, and he now has a 97.5%+ cumulative probability of exceeding 50.2% of the vote.
For fifteen polls, the MOE is an even tighter +/-0.80%. This means that for the same 52-48 % average spread, the probability is 95% that the leader will receive between 51.2% and 52.8% of the vote. In this case, the probability of EXCEEDING 50% is 99.99+%.
Assume that Kerry has an average 52%-48% lead in 15 polls the day before the election. Then if he loses, it will be most likely due to something other than mere chance (Diebold, maybe).
The 95% confidence interval around the mean comes right out of Statistics 101. The margin of error is 1.96 times the standard deviation, a statistical measure of the spread of polling results around the mean. The standard deviation is used in the normal distribution (the bell-shaped curve) to determine the confidence interval around the sample MEAN (average) poll result.
Of course, the winner must get at least 270 electoral votes. How do we calculate the probability of this result using the latest state polls? The same way as for national polls, except that now we calculate the probability of winning each state based on the latest polls. To determine the probability of an electoral vote win, we utilize these individual state probabilities by running a Monte Carlo simulation. State polls typically sample 500-600, so the MOE is wider (+/-4%) for each state.
The probability of winning a state is based on the latest state poll, adjusted as before for the allocation of undecided votes. In a 50-50 poll split, with a 50% undecided voter allocation, each candidate has a 50% probability of winning the state. If the split is 60-40, the probability of the current leader winning the state is 99.99%. When the polls are close (51-49) the leader has a 69% chance of winning the election. For a 52-48 split, it’s about 83%. For a 53-47 split, it’s 97%. So 53% is a pretty solid number and outside the MOE..
Now, let’s get back to the probability of winning at least 270 votes. In the Monte Carlo simulation: we run 5000 simulated elections to determine Kerry’s probability of winning 270 Electoral Votes. The probability is just the number of election trials he wins divided by 5000.
In each election trial, we generate a random (RND) number between 0 and 1 for each state and compare the RND to the probability of Kerry winning the state.
For example, if the RND generated for FL is .55 and Kerry has a .60 probability of winning the state, he wins since .55 is less than .60. On the other hand, if the RND is .75, then FL goes to Bush. The model generates random numbers for all the states and assigns each state’s electoral votes to the winner, and derives the total number of electoral votes won in each trial. The model runs 5000 of these simulated election trials.
Say Kerry wins 4900 trials. Then he has a 98% chance of winning. The model also calculates Kerry’s average number of electoral votes for the the 5000 trials.
One advantage of the simulation approach is to minimize poll “whiplash”; that is, slight changes in state polls where leaders change hands daily do not affect the results as much as they would if we just assigned the electoral votes to the leader. If the probability is 60% that Kerry wins a given state, then he will win about 3000 of the 5000 simulation trials for the state. And his average EV will not change dramatically based on the one state.
Using both national and state models has the advantage of providing a mathematical confirmation between the two methods. If they do not agree, that could mean that state polls are more current than the nationals, or vice-versa.
The main point is that we need to analyze as many polls as possible, and recognize that this REDUCES the overall margin of error. We will have more confidence in the results.
A final word, perhaps one that cannot be over-emphasized: The analysis produces a PROBABILITY of a Kerry win, assuming the election were held today. It does not PREDICT a Kerry win. There IS a difference. The model cannot determine the probability that fraud will occur, or that there will be an external event which will upset the natural order of things.
Hope that things are clearer now.