Thursday, February 9, 2012

Unpacking Colorado and Minnesota Results

On Tuesday night, the Republican parties of Colorado and Minnesota held precinct level caucuses to select delegates for their district and county conventions. These conventions will then meet to select the delegates that the state party will send to the national GOP convention. The make-up of these delegations will help to determine the Republican nominee in 2012. At the same time, every caucusgoer indicated their preferred national presidential candidate by casting a ballot. These balloted results were reported to the media, while the actual delegate results were not.

The problem is that there is no reliable information about the make-up of these intermediate county and district conventions. A few media organizations have decided to allocate "soft" approximations of delegates based on proportionally splitting the delegates at stake along the preference poll. This is a lazy way to estimate the actual delegate total, and one that's fairly misleading to boot.

It should not be terribly difficult to estimate the actual delegate totals, but we need to know a few things first.
  1. How many delegates to county and district conventions are assigned at each precinct
  2. How many delegates go from the county convention to the state convention
  3. The outcome of the preference poll at each precinct caucus
These pieces of information would make it very easy to create a good-enough prediction about the actual delegate outcome from the state conventions.

I'm having a bit of trouble finding this information directly from either Colorado or Minnesota. The Colorado GOP seems to have lumped together the preference poll at each of the caucus locations, which are each comprised of several distinct precincts. The Minnesota GOP seems to have reported each separately, but I'm having trouble locating the raw data. Neither are making the number of delegates allocated to each precinct readily available.

These models are going to be more valuable than simple proportional splitting because they'll take into account the geographical intensity of support for each candidate, and they'll apply the iterative filtering of support that typically drains delegates from poorly organized campaigns. If you have the data that I'm looking for or anything you think would help, please contact me on twitter at @sillyseasonblog.

No comments:

Post a Comment