Cameras, not coal, decided the election

How much influence did the Washington Conservation Voters have on the recent elections in Bellingham?

Daniel Pike, Seth Fleetwood and Cathy Lehman were the WCV slate in the Bellingham elections. There is some evidence in the precinct returns that about 10,000 - 12,000 votes were a block that voted the WCV ticket. That’s a big chunk of votes, but it only appears to have been decisive in Cathy Lehman’s defeat of incumbent Barry Buchanan. Larry Farr was not much of a threat to Seth Fleetwood and that race was not considered to be much of a contest. The end result of the WCV attempt to influence the election was: Fleetwood reelected as expected; a surprisingly strong win by Cathy Lehman; and a narrow loss by Pike. Why did Lehman and Fleetwood outpoll Pike? If the coal train issue was as influential as some have claimed, then why did Pike get the boot?

Here’s the conundrum: the mayoral race got a total of about 25,000 votes. However, they were split right down the middle with Pike getting about 12,300 and Linville getting a few more. But at slightly less than 15,000 votes each, Lehman and Fleetwood got more votes than Pike. If the coal trains and WCV were big influences, why should Pike, the banner-carrier for that issue, end up with about 2,400 votes less than two candidates on the same side of that issue?

The 2,400 vote spread between Pike and Lehman/Fleetwood argues against a strong coat-tails effect for Pike’s big money campaign. You can’t have coat-tails when you pull fewer votes. So the question becomes why was there a strong linkage between Lehman and Fleetwood that didn’t apply to Pike.

The big Bellingham vote-getting race was the mayoral contest, not the traffic cameras initiative above it on the ballot. It's reasonable to infer that the huge expenditures in the mayor's race brought up the turnout some. Pike’s expenditures did him little good. The mayoral race drew more voters, but the majority went to Linville, not Pike. The falloff means there were people who voted in the mayoral race who didn’t vote in the other city races -- and they voted against Pike.

Looking down the ballot, the unopposed race set the lower limit for falloff with Terry Borneman barely edging out Jack Weiss with both races turning out about just less than 17,000 each. The mayor's race turned out about 25,000, so there's roughly 8,000 votes up for grabs due to falloff. The two contested council seats both turned out around 22,000, so there were about 3,000 votes uncast between the mayoral and contested council seats. The traffic camera initiative turned out about 22,600 votes, so there’s only a little falloff between the initiative and the two council races.

Interestingly, both Fleetwood and Lehman pulled in almost exactly the same number of votes and so had the same falloff. Both of the council winners outpolled Pike and Linville by about 2,400 votes. The nearly identical city-wide totals for Fleetwood and Lehman are a statistical coincidence. Comparing the precinct numbers shows a sizable overlap, probably not more than 85% at the very most, between Lehman and Fleetwood voters. So most of the votes for Lehman were also votes for Fleetwood, but not all of them.

At most, the people who supported both Lehman and Fleetwood would be about 12,500 votes as a block. Lo and behold, Pike’s total was about 12,300. This sets an absolute upper limit on the size of the WCV voting block. It can’t be bigger than that and it is probably less. It will require careful comparison of the precinct numbers to make certain this isn’t just one of those coincidences. But keep the relationship between Pike and the council races in mind, because I’ll return to it later.

Now let’s look at the possibility of linkage between the traffic camera initiative and the council races. The Buchanan/Lehman race was an upset for an incumbent. That’s unusual, so we should be looking for why it happened. Fleetwood beating Farr is much more the normal sort of incumbent romping home that is the norm in stand-pat city.

The traffic camera initiative is a good predictor of precinct votes for Lehman and Fleetwood. Both the council races overlapped the traffic cameras issue with Buchanan having supported the cameras in council and Fleetwood having been the sole dissenting vote. The statistical correlation shows that the traffic cameras influenced the council elections.

But what about the mayor’s race? In this case, the votes that weren’t part of the WCV voting block appear to have also been yes votes on the traffic camera initiative. Pike owned the traffic cameras, as did all of the city council other than Fleetwood. No amount of posturing about coal trains was going to divert the hostility Pike earned himself over the cameras. Those votes for the initiative were the difference between the big margins Lehman and Fleetwood had in their elections and the narrow loss by Pike.

Here’s how the critical votes went down. The voter comes to the first Bellingham item on the ballot, the initiative to ban the traffic cameras. The voter votes for the initiative and their blood pressure goes up a bit. Next comes the mayors race. Pike, as mayor, signed the contract with ATS. That’s a vote for Linville. Here comes Cathy Lehman and Barry Buchanan. Barry supported the cameras. That’s a vote for Lehman. Now we are at Terry Borneman, unopposed camera proponent. No way to vote against him, so no vote in that race. Along comes Seth Fleetwood and Larry Farr. Seth, the sole dissenting vote against the cameras, collects a vote.

So the furore over the coal trains was less important to the election than the traffic cameras. If the election had been higher turnout, I’d lean toward explanations that people were voting for rather than voting against. If there was a higher turnout, I might think people were drawn to the election by the winning candidates. But the low turnout and ballot falloff makes people voting against the losers the simpler and more likely explanation.

Postscript for math wonks and statistics nerds

The statistical test to see if the traffic cameras cost Pike the election boils down to this question: are the precincts with a stronger vote for the traffic camera initiative also the precincts where the erosion of the WCV voting block support for Pike was strongest? Those of you who have the precinct results in computer readable form and some statistics mojo are invited to pitch in and do a little of the heavy lifting.

The statistical test I’m proposing is the positive correlation - Pearson’s r - between the difference in each precinct’s vote from the citywide average of the traffic camera initiative yes vote, and Pike’s falloff from the Lehman/Fleetwood votes. If the precincts that voted more strongly in favor of the traffic camera initiative also had a stronger falloff from the council to the mayor’s race, then the hypothesis is supported. If the correlation is negative or weak, then the hypothesis is disproven. But the numbers at a glance make a negative correlation nearly impossible, so the question really comes down to how strong the positive correlation is.

-------------------------------

Update - the statistics show there is a small to moderate correlation between the traffic camera initiative and people voting for Lehman and not voting for Pike.  The scatter plot has the strength of support for the traffic camera initiative along the horizontal axis and the falloff (as a percentage of Lehman's total) from Lehman to Pike.  If the correlation was perfect, the dots would form a straight line at 45 degrees.  With those two outliers in the data set, the correlation is about 0.12.  If you remove the two outliers (tiny precincts 210 and 237), the correlation jumps up to 0.26.

That's it.  Case closed.  The traffic cameras were the decisive factor in the mayoral election.