Thursday, May 08, 2008

 

Fivethirtyeight.com

http://www.nationaljournal.com/njonline/mp_20080507_8254.php


A great article by Mark Blumenthal aka "Mystery Pollster" on Poblano's near perfect prediction on Tuesday night.

For those that don't know, Poblano (he has a background in statistics) is a member of dailykos.com who also put together some mathematical models to predict the primaries on Tuesday night.

The site is amazing, and I visit it often.

But Blumenthal misses one point... Poblano got somewhat lucky in North Carolina. His model is based off of prior results from other primaries. Obama won a large margin on Tuesday night... but it was due more to his strong African-American vote than his high white vote. Obama lost on the white vote and gained on the Black vote as compared to other states. The two acted as a canceler upon one another.

My problem wasn't so much that I didn't see what was going to happen... I just took it to an extreme. I gave Obama less white vote and more African-American vote, but it turned out that I gave Clinton way too much white vote... it was more 60-40 than 65-35.. further Obama got 92% or so of the Black vote vs 90%. Those two in tandem doomed me.

Poblano got it right because his model is based off of prior performance in the primary... and his correct assumption that the Wright controversy wouldn't make such a big difference in the primary voting.

In Indiana, the Black vote showed up for Obama... especially in Indy...

Obama kept his white vote lost to a minimum..... its something that will serve him well come the general... which is where he is heading.

Comments: Post a Comment

Subscribe to Post Comments [Atom]





<< Home

This page is powered by Blogger. Isn't yours?

Subscribe to Posts [Atom]