After I published my post about where and when the vote was in Harris County, I got an email from HCDP Chair Gerry Birnberg regarding the voter registration numbers. He writes:
One thing to watch out for as you proceed with your analysis: the voter registration numbers upon which the county clerk's "% of registered voters" numbers are based are materially wrong. Kaufman's "Cumulative Report" uses a figure for voter registration in Harris County of 1,892,656. The actual number of persons who were registered and eligible to vote on November 4 was at least 1,956,000 (probably up to 10,000 higher than that, because of the additional voters who were added to the list after early voting closed, but were eligible to vote in this election; I don't know that number, but believe it to be as high as 10,000). The reason for this difference of over 64,000 between Beverly's 1,892,656 figure and the known figure of 1,956,000 is that Beverly's figures are based on people who were actually on the rolls as of September 19 only, and were never updated after that date. This is significant because we believe that the great bulk of applications received and approved after 9/19 came from Democratic areas and voters, because we had an active voter registration effort underway while the Republicans did not, to my knowledge.
Why this matter in the post-election analysis is two-fold. First, the conclusion that voter registration was down (compared to 2004 registration numbers) in some Democratic areas is probably incorrect. If the 64,000 post September 19 registrations were added to the totals, I suspect virtually every (if not every) Democratic area will have experienced an increase in voter registration, rather than a decline in some areas, despite the TEAM results which reduced voter registrations across the board as the voter rolls were "cleaned up" for the first time at the centralized, state level. Second, and on the negative side, however, is the fact that if correct voter registration numbers are used, there is an even smaller percentage of registered voters turnout than Ms. Kaufman's Cumulative Report indicates. (If the denominator -- registered voters -- is increased, as it should be because the 1,892,656 is incorrectly low -- the percentage resulting from dividing the number of voters who voted by the number of registered voters in any area would decrease.) In other words, we voted even a smaller percentage of registered voters in Harris County than Ms. Kaufman's Cumulative Report indicates.
But look, even if you want to quibble about the numbers, the basic point remains the same. Go ahead and assume that the County Clerk's number is too low by 64,000 or so. Assign all of those voters to the Strong D districts if you want. That gets them from minus 25,000 to plus 39,000, which still lags the growth in the Strong R districts. That's using a silly assumption against data that has a similar undercount from four years ago, and it still yields the same conclusion: You cannot account for the increased Democratic share of the vote - which you may recall is on the order of 88,000 more votes than 2004 - without concluding that there must be more people voting Democratic in strong R districts. That doesn't mean we didn't have better performance in the other districts - we did, and I will demonstrate it in the coming days - it just simply means we shouldn't be myopic about where "the base" is. Our voters are everywhere, and we need to have a turnout strategy that reflects that.
Posted by Charles Kuffner on November 16, 2008 to Election 2008