“What ifs” always raise unpleasant issues; after all, as the French say, “if bubbe had cojones she’d be zayde”, and if the Democrats had cojones they’d be in the White House and we’d be in Darfur instead of Iraq (but Michael Moore would still be complaining).
In the aftermath of this year’s congressional primaries, many “what ifs” have been raised about the race in the 11th District; both supporters of runner-up David Yassky (ruefully, and under their breathe, or after a few beers) and also-ran Chris Owens (by the candidate himself, in his concession speech) have claimed, contrary to the initial conventional wisdom that the presence in the race of additional black candidates helped Yassky, that Owens cost Yassky the primary.
Interestingly, several Owens supporters have used the blogs to raise the converse; that it was Yassky (27% to winner Yvette Clarke’s 31%) who cost the election to Owens (19%). There is an implicit presumption here that Yassky, being white, didn’t really have the right to run, and that, if he’d done the right thing, the proper and progressive victor would have emerged.
Although I’ve happily done my best to puncture many an Owens balloon, my aim here is merely to examine the statistical likelihood of both of these claims. Facially, they both begin with an idiotic presumption that: 27 + 19= 46 > 31. While mathematically correct, this is logically inapt. The illogic is that all the votes from either Yassky or Owens would be transferable to the other. That’s highly unlikely in the real world.
Owens got 9,971 votes. 4,792 of the votes came from Assembly Districts (ADs) where blacks were the majority (at least within their 11th CD portion). Of these, 1,311 came from ADs (40, 55, 56, 58) which, within the 11th CD, had no significant blocks of white voters. It would be idiotic to presume that the Owens voters in these ADs were any more likely than any other black voters to support Yassky in the event that they could not vote for their first choice. Another 1,927 of the Owens votes from black ADs came from areas (41, 42, 43) which, within the 11th CD, contained a significant bloc of white votes which were predominately Orthodox Jewish. These Orthodox voters may have been white, but few, if any, voted for Owens. Compare Owens vote in the two ADs (45, 48) which, within the 11th, were predominately Orthodox Jewish; in those ADs, Owens got 8% of the vote, so it is safe to assume that the votes transferred from Owens in the 2nd category of black ADs were also largely from black voters, who would have no special proclivity to switch to Yassky. There was one black majority AD (57), in which there was a significant white vote of the Brownstone variety, as well as some black voters who might have been enticed to vote for Yassky on the basis of an issue (he, upon Owens’ departure, becoming the most anti-Atlantic Yards candidate left in the race), but otherwise presuming any great switch of votes from Owens to Yassky in the ADs where the vote was predominately black was probably unlikely.
Owens got 5,179 votes within ADs where, within the 11th CD, the overwhelming majority of votes were cast by whites. 136 of the votes came from the predominately Orthodox ADs, although it is probably safe to assume that most of the Owens votes were not cast by Orthodox Jews. 2,817 came from a nearly all Brownstone AD (52), although that AD also contains two NYCHA housing projects. The final white AD (44) contains a section (Windsor Terrace, part of Park Slope, part of Kensington) which overlapped the 20th SD (producing 2,624 votes in the State Senate primary) and had a large Brownstone element. It also contains a portion of the 21st SD (producing 3,017 votes in the State Senate primary), which mixes Orthodox and Russian Jews with a polyglot of Yuppies, South Asians, Blacks and Latinos; in the State Senate primary, it gave Noach Dear 53% against Kevin Parker, which probably reflects its proportions. Another small predominately Orthodox portion of the 44th overlaps the 27th SD.
Were these white votes likely to transfer to Yassky? One assumes they chose Owens over Yassky for one of a number of reasons. Those motivated by Atlantic Yards (in the 52nd , part of the 44th, as well as the 57th) were likely Yassky votes, though unenthusiastic ones. Those motivated by altruisms or white guilt to support a black candidate were unlikely, but not impossible, Yassky votes, since that desire was likely to fade in the face of alternative black choices with machine connections or incomplete matriculations. Those who chose Owens because he was most “progressive” were a strong possibility to move to Yassky, since the other remaining candidates, even if actually more liberal, were generally less able to articulate such a worldview beyond their talking points, and had less resources to sell their message (although many “progressive” voters were likely to prefer, all things being equal, to vote for a black candidate). The answer seems to be that Yassky could pick up most of these votes, but surely not all of them. Those who chose Owens because they were impressed by the service rendered to their community by his father were probably not statistically significant, but were unlikely to have any predisposition to vote for either Una’s daughter or Clarence’s adopted son.
Now let’s look at the converse. Would the Yassky votes switch to Owens? In the nearly all black ADs, where Yassky managed 12% (989 votes), is there any reason to assume that these voters were more likely to vote for Owens than either of his opponents? I think not. The same must be said of the black votes in the black majority ADs with an Orthodox Jewish element (which account for 2,377 Yassky votes). As to the Orthodox votes in those ADs, as well as the more significant Orthodox vote in the predominately Orthodox ADs (1,242 Yassky votes), it is almost mind boggling to contemplate any of these votes transfering to Owens, who had the least institutional support, and the least palatable position on Israel. This is also true of the Orthodox votes Yassky got in the 44th AD (where Yassky got 4,041 votes), where the Orthodox probably accounted for at least a third (and probably more) of Yassky’s total.
Only in the 52nd (4,334 Yassky votes) and the 57th (945 Yassky votes) is the white vote entirely of the Brownstone sort most likely to switch to Owens (although somewhere between 1/3 to 2/3 of Yassky votes in the 44th probably also so qualifies).
So how would one deign to predict where such votes would go in the event a candidate dropped out? Well this is a primary. Unlike a general election, where, if the Republican left the race, it is unlikely his supporters would go to the Democrat, the voters in a primary in any particular election district voting for one candidate, are pretty likely to have a lot in common with the voters in that primary living in the same election district who are voting for a different candidate. In the event a particular candidate drops out, those votes are likely to split pretty much like the votes cast by their remaining neighbors. As such, I’ve attempted to reallocate the voters in our two scenarios in that manner.
Ideally, one would want to do this Election District by Election District (ED). There are two reasons I’ve not done this. First is that ED by ED results from this race cannot be pulled from the web. The second is that, even provided with the results, the time it would take to do this would be tremendous, although anyone willing to pay me to undertake the project is free to contact me at Gatemouthnyc@hotmail.com.
Instead, I’ve done this AD by AD. Thus, in calculating the race minus Yassky, in each AD I took the vote captured by each of his opponents, totaled them, allocated them a percentage and then multiplied that percent by the Yassky vote to reallocate it, dropping the fractions to avoid creating extra votes.
Is this a fair allocation of votes to Owens? In the four nearly all black ADs, I think it is very fair. In the black ADs with an Orthodox Jewish element, I think it overstates his vote, since it is unlikely that Orthodox votes would transfer his way (this is especially true in the 43rd, where a large share of Yassky’s votes were a Hasidic-control vote and likely to have transferred en masse to a candidate other than Owens). For similar reasons, the Owens vote would also be overstated in the all-Orthodox ADs as well.
Where Owens vote was likely to be understated somewhat by this process is Brownstone Brooklyn. But does anyone really believe that, in a three way race with no incumbent, Owens was going to receive a much higher percentage in the 52nd AD than 62%, especially with Yvette Clarke attracting some significant civic support, and Carl Andrews representing a portion of the area in the State Senate? In the 44th, the understating of Owens’ Park Slope vote is probably offset by the overstating of his Orthodox vote, which almost certainly accounts for a disproportionate amount of Yassky’s total in the AD. Only in the 57th, do I think that Owens gets significantly cheated in my formulation, but Yassky had only 945 votes to reallocate there. Anyway, I think that, in the district wide aggregate, the variables in every area wash each other, and to the extent they don’t, this would only matter if the result of the re-allocation were close.
The results are not close. With the reallocation of the Yassky vote, Clarke would get 20,059 (39%), Owens 15,847 (31%) and Andrews 15,379 (30%). Even positing that Owens was disadvantaged by the reallocation, an Owens victory seems highly unlikely.
The Owens-less race presents its own caveats. I think the reallocation to Yassky in the all-black ADs is fair, but in the black ADs with an Orthodox element, Yassky is surely being awarded bonus votes, since the voters who made up his proportion of the total vote were less likely to resemble their neighbors, especially their neighbors who voted for Owens. On the other hand, I think the reallocation of Owens votes (1,554) in the 57th (Yassky would get 22% of them) wildly understates Yassky’s potential support, since as the most anti-Yards of the remaining candidates, he’d be likely to pick up most of the Owens’ white vote (which Owens probably won) and a decent share of Owens’ black votes. I also that it is unlikely that Yassky would get nearly 73% of the vote in the 52nd. Less significantly, I think that Yassky’s allocation of the Owens vote (over 80%) in the Orthodox ADs is lunacy, since these were unlikely to have come from the same type of voters who voted so overwhelming for Yassky in those areas (however, we are only talking about 136 Owens voters in those ADs). The impact of some of these flaws would be minimized if the reallocation was done on an ED by ED basis. Nonetheless, we have a decent rule of thumb for measuring whether the scenario that Yassky would have won the race without Owens has any plausibility.
The scenario is plausible. With the reallocation, Yassky would get 19,036 (36.65%), Clarke 18,843 (36.27%) and Andrews 14,066 (27.08%), a victory of 193 votes (.38%). Since I think the reallocation overstates the Yassky vote, I think the question of who really would have won is up in the air. Those who’d like to see this reallocation redone ED by ED are free to contact me at the address above.
The Numbers:
The Actual Vote:
For further discussion of some of the topics within please see:
http://www.r8ny.com/blog/jerry_skurnik/an_analysis_of_nyc_primary_results.html#comment-61713