A Junkie's Favorite ... New Presidential Vote Analysis!

On Friday, Polidata (the political data analysis firm with possibly the worst website I've ever seen for an organization trying to pass itself off as reputable) released its numbers on November's presidential race by congressional district. Why has it taken nearly five months for this study, you ask? Well, at least in most states, presidential election returns are reported by county, and since most congressional districts make up more than one county, tabulating the c.d. results are slightly tricky and require more detailed number-crunching than the usual media folks are up for (once there's a concession speech, the little white-boards and cool light-up maps that we see every four years get shoved back into the closet until the next time around).

So the Polidata folks do the real math, and then turn it over to the talking heads (and the bloggers too this time around) to yammer about for a while. The Washington Post assigned Dan Balz to the story, and his result appeared on Tuesday, headlined "Partisan Polarization Intensified in 2004 Election." Charlie Cook devoted his weekly column to the new data as well. The finding that most intrigued Balz and Cook is that 2004 saw fewer congressional districts splitting their votes (going for a presidential candidate of one party and a House candidate from the other). Only 59 of 435 congressional districts (13.6%) fit this description, with 41 of those going to Bush and 18 going to Kerry.

This finding is notable because of the significant drop in "split-ticket districts". Four years ago, 86 districts went in opposite directions (40 for Gore, 46 for Bush). In 1996 and 1992 more than a hundred districts fit the bill each time, as Balz notes and Polidata's analysis bears out. Balz attributes the trend away from ticket-splitting to a) redistricting trends, which make congressional districts safer for a given party and therefore less likely to vote for the opposing presidential candidate, and b) increasingly polarized and regionally homogeneous political parties, in which the vast majority of those registered as Republicans will vote Republican across the board and ditto for Democrats. [I'm definitely an exception to this one: since I've been able to vote, I've never voted straight party-line for federal candidates ... and probably never will at the rate the GOP is going at picking presidential nominees.]

Reason b) above is borne out by Charlie Cook's analysis of the congressional districts that went most heavily for Bush but also for Democratic House candidates, and those that went most heavily for Kerry but also for Republican House candidates: of the ten Democratic reps in the heaviest Bush districts, seven are in the South; of the ten Republican reps in the heaviest Kerry districts, five are in the Northeast, four are in the Midwest.

Neither Balz nor Cook include the Senate in their analyses: by my quick calculations, of the 34 races in November, only six (AR, CO, IN, NH, NV, PA), translating to 17.6%, featured split tickets. That was slightly more evenly split, with two races going to Kerry and a Republican Senate candidate (NH, PA) and the other four to Bush and a Democratic Senate candidate. Redistricting certainly plays no role here, but population shifts may.

The tradition of ticket-splitting, most common of late in New England states which have tended to elect Republican governors, is a loss that should be much lamented by moderates of all stripes. The decrease in competitive House district is also bad news for moderates, who tend to fare better in such districts than when put up against more ideological members of their own party in a contest where the primary is the de facto general election. Continued trends in this direction may in fact spell the end for moderates in both parties, as Southern Democrats and Northeastern/Midwestern Republicans grow even more endangered and out of step with their national parties.