The Trib has had a few stories lately about legislative horse races. There’s one point in this piece about House Democratic freshmen trying to win re-election that stood out to me and deserves some fleshing out.
As in any close election, both sides agree voter turnout will be key factor this fall. That built-in Republican advantage is substantial: A generic Republican running for the Legislature registered 15 percentage points higher than a generic Democrat in the September University of Texas/Texas Tribune poll.
The poll cited is this one, and if you click on the toplines data, you see that in the generic state House race question, “Republican” led “Democrat” by a 47-32 margin, or 59-41 if you extrapolate it out. That sounds pretty daunting, but let’s put that in some context. Here are the vote totals for each party in all State House races from the last two cycles, with percentages:
Rep 2008 = 3,828,626
Dem 2008 = 2,962,310Rep Pct = 56.38
Dem Pct = 43.62Rep 2006 = 2,253,543
Dem 2006 = 1,584,064Rep Pct = 58.72
Dem Pct = 41.28
I just simply added the votes each R and D candidate received in all 150 races – blogging can be so sexy sometimes – and voila. You may recall that in 2006, the Democrats netted six seats without losing a single seat they held, to win a total of 70 overall. That’s 46.7% of the available seats, despite the fact that their vote total would suggest they should have won only 61 races. In 2008, of course, Democrats netted three more seats for a total of 74 (one Republican victor from 2006, Kirk England, switched parties prior to the 2008 election), or 49.3% of the total despite receiving only 65 seats’ worth of votes. If there were generic legislative polls those years, I’m not aware of them, but I’d bet they would have skewed pretty Republican as well.
The point of this is to say that such generic polls tell you basically nothing. The reason for this is simple: The votes are not uniformly distributed across House districts. Look at individual results and you clearly see that Republican-held seats tend to have higher turnout than Democratic ones, in no small part because Democratic seats have on average a much higher proportion of Latinos, who are less likely to be citizens, to be registered, or to vote, in them. The 37,625 votes Frank Corte got in 2006 and the 5,201 votes Scott Hochberg got in 2006 were each worth one seat; the same is true for the 73,450 votes Ken Paxton got and the 11,881 votes Ana Hernandez got in 2008. I’m sure there is some correlation between a generic legislative ballot and the number of House seats each party is likely to win, but we don’t have nearly enough data to take a guess about what the one from 2010 means.
As much as I like to dismiss the value of all polling, I especially appreciate your pointing out the inherent fallacy of the generic variety. The conclusion from the Tribune writer you cite is as a flawed as yesterday’s “Keller Cleared” headline.
They need to start doing a much better job over there of interpreting and translating the news.