One of the things that was hotly debated in the aftermath of the 2004 elections was whether Martin Frost’s campaign for the 32nd CD had a positive effect on downballot races, or if the resources he expended only benefitted his own totals. I think I have some data that may help answer this question.
What makes this tough to answer is finding an appropriate basis for comparison. Looking at turnout in a given precinct, even in a strongly Democratic precinct, isn’t enough, since you can’t say for certain based on higher than usual numbers who those people were actually voting for. Similarly, you can’t just look at returns in a given precinct, since you don’t know what the baseline performance would have been. What you need is to compare returns in precincts where Frost was on the ballot to those without him, and you’ve got to compare the results for a downballot candidate to a baseline result for each.
I got an inkling for how to do this while doing an analysis of Chris Harris’ SD9, which was requested of me awhile back. What gave me the idea was noticing that within SD9 there were four contested Congressional races, and that there was a wide spread in performance among the four Democratic candidates.
Results in SD9
CD26 - Michael Burgess versus Lico Reyes
Candidate Votes Vote Pct Vote Diff Pct Diff
========================================================
Burgess 38,709 78.90 +833 +4.45
Reyes 10,353 21.10 -2,643 -4.45
Bush 37,876 74.45
Kerry 12,996 25.55
CD06 - Joe Barton versus Morris Meyer
Candidate Votes Vote Pct Vote Diff Pct Diff
========================================================
Barton 22,824 52.54 -893 +0.03
Meyer 20,618 47.46 -780 -0.03
Bush 23,717 52.57
Kerry 21,398 47.43
CD24 - Kenny Marchant versus Gary Page
Candidate Votes Vote Pct Vote Diff Pct Diff
========================================================
Marchant 56,759 61.67 -3,210 -0.35
Page 35,278 38.33 -1,440 +0.35
Bush 59,969 62.02
Kerry 36,718 37.98
CD32 - Pete Sessions versus Martin Frost
Candidate Votes Vote Pct Vote Diff Pct Diff
========================================================
Sessions 13,104 52.87 -2,111 -6.65
Frost 11,680 47.13 +1,333 +6.65
Bush 15,215 59.52
Kerry 10,347 40.48
“Vote diff”, in case you can’t tell, is the difference between that candidate’s total and the corresponding total in the Presidential race. “Pct diff” is the same for the two-party vote percentage. An eleven point range between Frost, who got more votes than John Kerry, and Reyes, whose opponent got more votes than Bush. Clearly, Frost’s campaign made a difference for himself. How can we translate this to other races?
What I decided to do what this: Within SD9, there were two Dallas County races, one of which you may have heard about. Those races were for Judge of the 303rd District Court, between Beth Maultsby (R) and Dennise Garcia (D), and for Sheriff, between Danny Chandler (R) and Lupe Valdez (D). Did Martin Frost help either of those two Democratic candidates get more votes than they otherwise would have?
The method I chose was this: Compare the performance of Garcia and Valdez to that of John Kerry in the precincts where Frost was on the ballot to those where he wasn’t. The theory here is that since there was essentially no campaigning by Kerry in Texas, his performance could be considered a baseline, a minimum of what a Democrats should get (it’s probably a little higher than that since there was some campaigning by Bush, but that’s not really important for these purposes.) Only the Marchant-Page race covered any significant part of Dallas County in SD9 (there were two precincts in the Burgess/Reyes race, totalling about 50 votes), so we’ve got a fairly straightforward basis for comparison. Looking at the less-heralded judge’s race as well helps us to filter out whether any differences are due to Valdez’ well-known effort.
With that in mind, here’s what we get:
Dallas County portion of SD9
Candidate Votes Vote Pct
=================================
Bush 43,097 60.66
Kerry 27,950 39.34
Maultsby 38,867 57.59
Garcia 28,627 42.41
Chandler 39,113 57.29
Valdez 29,159 42.71
Sessions/Frost precincts
Candidate Votes Vote Pct
=================================
Bush 15,215 59.52
Kerry 10,347 40.48
Sessions 13,104 52.87
Frost 11,680 47.13
Maultsby 13,680 56.35
Garcia 10,597 43.65
Chandler 13,865 56.41
Valdez 10,714 43.59
Marchant/Page precincts
Candidate Votes Vote Pct
=================================
Bush 27,846 61.28
Kerry 17,591 38.72
Marchant 25,820 59.75
Page 17,397 40.25
Maultsby 25,149 58.26
Garcia 18,021 41.74
Chandler 25,211 57.77
Valdez 18,432 42.23
Both Garcia and Valdez clearly did better overall in the Frost precincts, but looking at how they did relative to Kerry in each area, there doesn’t appear to be much difference – indeed, if there is anything significant, it’s in Page’s favor.
Differences in Sessions/Frost to Bush/Kerry
Candidate Votes Vote Pct Pct Ratio
============================================
Sessions -2,111 -6.65 0.89
Frost +1,333 +6.65 1.16
Maultsby -1,535 -3.17 0.95
Garcia +250 +3.17 1.08
Chandler -1,350 -3.11 0.95
Valdez +934 +3.11 1.08
Differences in Marchant/Page to Bush/Kerry
Candidate Votes Vote Pct Pct Ratio
============================================
Marchant -2,026 -1.53 0.98
Page -194 +1.53 1.04
Maultsby -2,697 -3.02 0.95
Garcia +430 +3.02 1.08
Chandler -2,635 -3.51 0.94
Valdez +1,035 +3.51 1.09
“Pct ratio” is the ratio of the given candidate’s vote percentage to that of the corresponding Presidential candidate for those precincts. Another way of putting this is that Frost did 16% better than Kerry, while Garcia and Valdez each did 8% better in his districts; Page did 4% better than Kerry while Garcia was again 8% better and Valdez 9% better in Page’s districts.
From this, it would seem fair to conclude that on a performance basis, it didn’t matter to Garcia and Valdez if they were in Page’s precincts or Frost’s – they each did about as well relative to the national ticket in each. Therefore, if there was any Frost effect at all, it would have to come from turnout alone.
I didn’t know what I was going to find going in to this. I do not consider this to be fully conclusive. For one thing, I’ve not looked at all of Dallas County; in particular, I’ve not looked at any of the other Congressional races there, including the precincts in which Eddie Bernice Johnson ran unopposed. More importantly, we only looked at some turf that was pretty hostile overall to Democrats; we’ll need to see how things shake out in the more Democratic areas in these districts. I also don’t have the relative turnout numbers at hand, so I can’t say if there was a real benefit from an absolute vote total perspective in Frost’s district. At the very least, however, this should cast doubt on the notion that Frost’s tide lifted all boats. There’s still more work to be done, but it doesn’t look good at this point.
The full spreadsheet is available here if anyone wants to check my math.