I wanted to follow up on my previous post where I analyzed how black voter turnout would affect races in the Deep South.  In this post I’ve added some educated guesses on what we should expect Obama’s draw among white voters to be.  I also reduced black voter turnout to 85%, because this appears to be the projected black turnout in Georgia based on early voting.  Other states are projecting even higher.  Here are the predictions:

The worst case is where Obama only gets his proportional share of the Kerry white vote based on Obama’s primary numbers.  The best case is where he matches Kerry’s take.  An explanation of the “prediction” follows.

There are some assumptions baked into how I calculated Obama’s share of the white vote.  I took an average of Obama’s share of the white vote in the Democratic primaries, and then compared each state to the average Obama share.  Whatever the variance was, I used that to shift, up or down, the projected share of Kerry voters that Obama can persuade.

The assumption baked into all of this is that Obama will not outperform Kerry among white voters, and conversely that Obama will likely significantly under-perform Kerry.  In South Carolina, for instance, Kerry secured approximately 34% of white voters, while I’m projecting that Obama to win only 24% of the white vote.  If you believe that Obama will do very poorly among white voters in the South, this model bears that out.

The only exception to this is in VA, where Obama significantly over-performed.  Because of this, the model used a white vote share for Obama similar to Kerry’s numbers for the “prediction”, and a best case where Obama outperforms Kerry among white voters by about 2.75 points.  Obama will need a result like this (or depressed turnout among Republicans) to win North Carolina.  Nonetheless, he is performing significantly better there than Kerry did in 2004, when there was a native son on the ticket.

In states where you don’t expect a huge suppression of Obama’s numbers among white Democrats or if you expect Obama to over-perform Kerry, the model is too conservative to be useful.  This results in a huge spread and depressed numbers.  In Pennsylvania, for instance, we ended up with a 19% spread and an expected result of 45.92% for Obama.  That is really unlikely.

I’ve also provided the table that I used to calculate Obama’s state by state white vote variance, in case anyone was curious:

Support Eyes On Obama!

Back to Blogs