Monday, September 03, 2007

GOP Primary Predictions - Intro

My goal for this post was/is to make some predictions about the GOP '08 primaries and figure out just how much work Thompson has to do to get in front of Giuliani (or, put another way, figure out how inevitable is it that Giuliani will be the '08 GOP candidate?).

I started by spending some time looking over the Republican presidential primary poll results over at I then proceeded to make some predictions for Iowa, New Hampshire, South Carolina, Florida, Michigan, and California (six of the eight states for which they have polling graphs).

For each state, I tried to follow the trends I saw in it's graph to arrive at a January '08 number. Obviously the trends these graphs show will likely change over the next four months, but the trends were generally quite clear and had been fairly consistent over the last four or more months, so I think using the trend-result numbers will be closer to reality than the current snapshot numbers.

I also made the assumption that Gingrich supporters would ultimately turn into Thompson supporters. Newt is tacitly supporting Fred, and it seems to me that Gingrich supporters probably support Gingrich because they aren't thrilled with Rudy McRomney. Perhaps I'm being myopic to the possibility Gingrich supporters would wind up with the Huckabee camp, but I think overall this is a pretty safe assumption.

Here's what my predictions, per the above, end up looking like on the polling graph overlaying the six states:

Multi State GOP '08 Poll Predictions
The big colored dots are where I think the candidates will wind up. There should be six dots for each (one for each state mentioned above).

The graph is nice and all, but prediction-wise, it's more valuable to see how things actually break down for each of the states, and sort of go state-by-state like it will actually happen.

Prompted my desire for that approach, I compiled the '08 Republican primary schedule and having done that I actually decided I didn't like my original mix of states. I don't think it would really be worthwhile to look at all fifty states, or even all of the "early" (up through "Super Mega Extrodinary Tuesday!!!") states. So, I'm going to leverage the "Influence" index I added to my '08 GOP primary schedule and I'm looking at the states with an influence number greater than 10 who vote before March. Ohio and Texas have >10 influence numbers, but things really are generally decided before they will actually vote in March. Thus, the states that meet my criteria are Iowa, South Carolina, New Hampshire, Florida, Illinois, New York, California, and Georgia.

For Iowa, SC, NH, FL, and CA I'm using the numbers I arrived at in my mega-graph above. For New York, I'm also going to take the same approach using New York's graph from For IL and GA, which don't have graphs at, I'm using opinion polling recorded on Wikipedia.

A final note on my "methodology" is regarding turning voting percentages into numbers of delegates from that state. There are actual rules documenting how that's supposed to work, but each state has some delegates that are sort of flexible (not necessarily tied to the state's popular vote) so it gets extra complicated, fast. I tried to look at historical delegate allocations for percentages and use that. Generally, it appears that the top candidate winds up with say 60-80% of the state's delegates and the remainder get spread among the other top vote getters who got more than 10-15% of the vote.

And now with all this background out of the way ... onto the results.

No comments: