Where - and when - the vote was

I have finally received precinct data from Harris County. It's going to take me a little while to work through it to start getting some answers about how the votes were distributed - Matt Stiles, who got his draft canvass earlier, produced a colored map from it that you should see - but in the meantime, I want to discuss a couple of trends that are fairly easy to spot even with the data that I had at the time. Here is a Google spreadsheet that compares the number of votes, registered voters, and early votes in each State Rep district from 2004 to 2008. There are two points that need to be made, because I think they go a long way towards explaining what happened these past few weeks.

First, let's look at where the voters are. I began by classifying the 25 State Rep districts in Harris County as Strong D, Strong R, and Swing. They are:

Strong D - HDs 131, 137, 139, 140, 141, 142, 143, 145, 146, 147, 148
Strong R - HDs 126, 127, 128, 129, 130, 132, 135, 136, 138, 150
Swing - HDs 133, 134, 144, 149

Note that all four of the swing districts were held by Republicans going into the 2004 election. Dems won HD149 in 2004, 134 in 2006, and 133 this year.

The first thing to notice on the spreadsheet is that there's a whole lot more voters in the strong R districts than there are in the strong Ds:

Dist type   04 voters  04 avg   08 voters  08 avg   Change

==========================================================

Strong D      700,638  63,694     676,691  61,517  -25,563

Strong R      860,816  86,082     907,006  90,701  +46,190

Swing         314,842  78,710     310,575  77,644   -4,267

What had been a 160,000 voter registration gap in favor of the Strong R districts is now 230,000. Of the 14 districts now held by Democratic State Reps, only three - HDs 131 (+838), 141 (+582), and 149 (+2,350) - gained voters. Six of the eleven Republican districts grew - HDs 126 (+1,371), 127 (+4,406), 130 (+20,948), 132 (+18,743), 135 (+1,475), and 150 (+9,393). Whether this is demographics or a good illustration of the Bettencourt Effect is unclear, but the challenge to Democrats running countywide isn't. You can't win as a Democrat in Harris County solely by turning out voters in Democratic districts. There's far too many voters outside of those districts, and you need as many of them as you can get.

It's not just registrations. The percentage of voters who turned out increased in every district, though as Marc Campos noted, in some cases that meant the number of voters declined, thanks to the lower registration rates. But as above, the increase was greater in the Republican districts:

Dist type    04 votes  04 pct    08 votes  08 pct   Change

==========================================================

Strong D      348,135  49.69%     374,217  55.43%    +5.74

Strong R      546,774  63.52%     607,814  67.01%    +3.49

Swing         193,423  61.43%     201,785  64.97%    +3.54

The rate of increase in the Dem districts was highest, but turnout still lagged the R districts by double digits. What's more, the Republican districts were a greater share of the electorate in 2008 than in 2004. In 2004, 50.24% of all votes were cast in the Strong R districts, while the Strong Ds accounted for 31.99% and the Swings the remaining 17.77%. In 2008, the percentages were 51.34% Strong R, 31.61% Strong D, and 17.05% Swing.

Yet with all these factors seemingly working against the Democrats, the results in 2008 wre completely different, Every statewide Dem won at least a plurality of the vote. All appeals court candidates won majorities. Dems won 27 of 34 contested countywide races, and picked up a State House seat to increase their majority of the delegation to 14-11.

How could this be? The answer is obvious: More Democrats live in the Strong R districts than vice versa (and also in the swing districts, three of which are held by Democratic reps), and more of them are there now than were in 2004. This is clear not only from the countywide results, but also from some of the state rep races and from the March primaries. Consider some of the State Rep races. In 2004, the winning margins for Republican candidates over their Democratic challengers in HDs 126, 127, and 138 were 69.3/30.7, 70.4/29.6, and 63.8/36.2, respectively. In 2008, those results were 59.4/38.4, 65.7/32.3, and 59.0/41.0 (there were Libertarian candidates this year in 126 and 127). It's highly likely that other Dems did better in those districts, as well as in the others, than they did in 2004. I'll know for sure when I get the precinct data, but it should be clear to see: The Dems gained a lot of ground in the reddest places. They even turned some of them blue. I'd call that the fruition of the "Run Everywhere" strategy.

I'm discussing all this at such great length because I want to make this point crystal clear, and I haven't yet seen it in any of the analysis I've seen so far about turnout, the Latino vote, etc. That point is that Democrats won up and down the ballot despite having fewer voters, in both an absolute and a relative sense, than they did in 2004. The Democratic base doesn't just exist in the same familiar places any more. It's everywhere in the county, and any strategy for winning the county again in 2010 has to take that into effect.

The other topic to discuss is early voting. As we saw this year, early voting increased by a lot, which caused some projections of the final turnout to be far higher than the actual amount. It also made for a change on Election Day, as the usual dynamic of early Republican leads that got whittled into as Tuesday returns trickled in was reversed. The question is whether there was a late surge of Republican voting, or if it was just the case that the Democrats shot their load early and had little left for Tuesday. Here's how these numbers looked:

Dist type    04 early  04 pct    08 early  08 pct   Change

==========================================================

Strong D      121,938  35.03%     220,763  58.99%  +68.43%

Strong R      212,103  38.79%     341,767  56.23%  +44.95%

Swing          77,726  40.18%     114,161  56.58%  +40.79%

Voters in all districts went to the polls early at a higher rate than they did in 2004, but proportionally the Democrats did more of this. Voters in the Strong D districts, which as we know are more Democratic than the Strong R districts are Republican, made up 32.62% of the early vote this year, as opposed to 29.61% of the early vote in 2004; for the Strong Rs, the numbers were 50.51% in 2008 and 51.51% in 2004. The flip side of this is that these voters were a much smaller share of the Tuesday vote. Strong D voters were 33.43% of the E-Day vote in 2004, and 30.26% of it in 2008, whereas Strong Rs were 47.47% of the E-Day vote in 2004, and 52.46% of it in 2008. I don't have enough evidence to judge whether or not the Republicans had a late rush to the polls, which is to say if they wound up with more voters on Tuesday than they had originally expected, but it seems clear that the Democrats had a greater shift towards early voting, which put the Republicans in the catchup position afterward.

All righty then. I hope this holds us all off till I work my way through the precinct data.

Posted by Charles Kuffner on November 12, 2008 to Election 2008