I was curious about how the GOP selection rules would effect the nomination. My sense was the GOP rules make the odds bigger than people think that either Carson or Trump would win the nomination.
Here are the results:
What these show is that if the current numbers hold, and I am sure they won't, the GOP may find it very hard to stop either Carson or Trump. With over 3 months before Iowa, there is time. But the more time passes without the two of them fading, the more desperate the GOP establishment is going to become. As the table below shows, by April 1st over half the delegates will have been selected. Because GOP delegate selection rules favor the top 2 candidates, it is unlikely that a stop Trump or Carson effort would be successful. Moreover, because the Super Tuesday states are going first, the early electorates are made to order for an anti-establishment far right wing candidate.
I would say pass the popcorn - but one of these guys could win.
So the projection tool is here:
https://docs.google.com/...
Here is what is behind the prediction:
1. State polling is used if recent
2. If there is no state polling but the state is in the south, use an average of polling from the south
3. If outside of south, use nation polling average
4. No attempt is made to project super-delegate votes, not votes in the territories other then Puerto Rico
5. Delegate Selection Rules are from the green papers
Because the rules require calculations for individual CD's and have different thresholds, I had to build an estimate in many states for each CD. In some states, the winner of each CD gets 2 delegates, and the second place candidate gets one. In other states there are threshold rules requiring a candidate get at least 15% or 20% of the vote to get a delegate. This math was complicated.
There will be enhancements:
1. The effect of Iowa and New Hampshire on downstream events will be modeled
2. A dispersion formula will be applied. If a candidate gets 20% of the vote, he or she will not get 20% in each CD. As a result, I am going to build a model that randomizes the effect across CD's.
Comments are closed on this story.