Some of you may recall that I used to teach Statistic at the college level. I could not resist running some fairly simple regression analysis based on polling data to date to see if I could come up with a statistically sound range of possible outcomes.
According to Gallop, today in the generic vote for congress, there are 51% Republicans, 41% Democrats and 8% independents. The same poll shows that the enthusiasm level for Republicans is 54% while the enthusiasm level for Democrats is 30%. Here is what that means. Let’s assume that the enthusiasm level is identical to expected turnout. That means that 54% of Republicans would probably vote, 30% of Democrats would probably vote and 50% of Independents would probably vote.
If we calculate assuming nothing changes between now and November 2, this would be what to expect:
GOP 62% DEM 28% IND 10%
Now let’s assume Democrats make a major effort to generate enthusiasm. They are unlikely to change the generic percentages this late in the game, but they may succeed in changing the enthusiasm. Let’s assume they hit it out of the park and increase enthusiasm to 60%, while GOP enthusiasm plateaus at 57 percent. This is what that election would look like:
GOP 48% DEM 45% IND 7%
Now let’s assume that the Republicans surge even more, up to the enthusiasm levels they had in 1994. The Republican enthusiasm could soar to as high as 67% and we will assume that Democrats also increase, up to around 40%. In that case the results would look like this:
GOP 68% DEM 28% IND 8%
Now, lets do a standard statistic analysis and provide even weighting to each outcome. The average result is:
GOP 58% DEM 32% IND 10%
I know these numbers appear to be too good to be true, but recent results are very consistent with them. For example in the recent ballot on Proposition C, the vote to repeal Obama care a poll was conducted of an equal number of Republicans and Democrats just before the election. 61% of Republicans were for the Proposition, 48% of Democrats were against the proposition. In Missouri most of the elected officials are Democrats, but it is also somewhat conservative and John McCain carried Missouri by a very tiny margin in 2008. Normally one would expect more Democrats than Republicans. So most polling firms evaluating a poll of 300 Republicans and 300 Democrats would have adjusted the final estimate in favor of Democrats. Scott Rasmussen, who typically rates Republicans higher than other polling firms estimated that Proposition C would pass 58% to 41%. But the actual results was that Proposition C was passed with more than 70% of the vote.
Even if one assumes that the generic party affiliation in Missouri has shifted from 50/50 to 51/41, it would not explain this result. That would have indicated a final vote of about 62% to 38%. Clearly, at least in Missouri with regard to Proposition C, the difference in Republican enthusiasm was equally important as the difference in the generic ballot. In fact when I used my analysis to factor in the expected voter turnout to be roughly proportionate to enthusiasm, based on the results of that last minute poll, I come up with an expected result of 73% to 27%, not that much higher than the actual 71% to 29% actual result.
My methods are obviously crude, because I do not have access to sufficient data and because I have only spent a small amount of time on this. However when existing data is this skewed, the projected outcomes, baring major unforeseen intervening events, are at least possible. I am not predicting results, just commenting on the probability of some outcomes based on the Gallop Poll today.
I strongly suspect that all of the pollsters are seeing similar results but they are too shocked to report them. I don’t blame them, because these kinds of results are unprecedented. It is really hard to imagine Republicans coming close to getting 60% of the vote under any circumstances and doing better than that is pie in the sky range. It would result in a political landslide of proportions never seen before in this country. It is like predicting that the laughing stock of the National League for the period from 1962 to 1968, the New York Mets, would win the World Series in 1969. Oops. That happened!
TDM