The logo used by the federal NDP during the 1980s and early 1990s, featuring hand-holding bathroom people.
In my previous post, I pointed out the flaws in the federal NDP leadership race endorsement point system employed by threehundredeight.com. In this post I will offer two alternative means of gauging relative support for the various leadership candidates.
To recap my primary objections with the threehundredeight endorsement point model, I felt that it lent disproportionate weight to labour endorsements and gave a blanket value to endorsements from past federal and provincial leaders and past MPs, with little regard for the performance or length of service of each endorser. Afterall, an endorsement from Ed Broadbent is worth more than one from Audrey McLaughlin, despite both being past federal leaders, because one is associated with the pre-Layton heyday of the party and the other with a disastrous rout. The same goes for provincial leaders and MPs. Short-term past leaders and MPs are going to command less sway than more successful or long-term former provincial leaders and MPs. There were other flaws I addressed in my previous post, though these were the main considerations I took into account in devising my own endorsement point model. Below is the point award system I have developed:
As you can see, this endorsement point model seeks to weigh endorsements from past leaders and MPs based on electoral performance and length of time served. I admit that the federal leader point score metric is a bit convoluted, though it is easy to explain with an example. In Broadbent’s case, for instance, once his points for time served as party leader and an MP are tabulated, we add scores for electoral performance in the four elections in which he led the NDP (1979, 1980, 1984, 1988). Winning 100% of the popular vote would net 10 points, and all shares of the vote between 0% and 100% earned by the leader would net between 0 and 10 points. Broadbent’s four elections saw the NDP win 18%, 20%, 19% and 20% of the vote, respectively. Therefore, Broadbent gets 1.8, 2.0, 1.9 and 2.0 points added to his score. Alexa McDonough led the NDP to 11% and 9% of the vote in the elections of 1997 and 2000, and as such she earns 1.1 and 0.9 points for said performance.
So, how does the NDP leadership race look now with this endorsement model? Here are the results:
As you can see, compared to threehundredeight’s model, labour leaders, Ed Broadbent and Alexa McDonough aren’t quite the endorsements gods they are on threehundredeight. Moreover, MLAs from Nova Scotia and Manitoba, where the NDP forms governments, are worth relatively more than MLAs from British Columbia and Ontario. Endorsers who aren’t sitting or former politicians or labour leaders also come into play, and the relative success and length of service of past provincial leaders is now a factor, along with the length of service of past MPs and MLAs/MPPs. How this translates into relative standings in the NDP race is illustrated in this chart.
This endorsement model still puts Brian Topp in the lead, but puts Thomas Mulcair in second place and Paul Dewar as a strong third. Peggy Nash, currently ranked in second place on threehundredeight, sits at fourth in this model.
There is another method, a much more simple one, that can be used to gauge relative support in the leadership battle. This is simply giving equal weight to every endorsement, looking at which province the endorsement comes from, tallying up the number of endorsements for each candidate in each province, and dividing up the total number of NDP members in each province in proportion to the share of endorsements held by each candidate in said province. For instance, lets say there are four endorsements from the province of Albertoba. Two are for candidate A, and one each for candidate B and C. Albertoba has 1,000 NDP members. Therefore, those 1,000 members are divided by four, thus giving each endorsement a value of 250 members, or rather votes. Candidate A gets 500 votes from Albertoba while candidates B and C each get 250. This process is repeated for all other provinces. I should add that for endorsements from national union heads, I placed the endorsement in the province in which their main operations or office were based (read, Ontario). As the NDP is using a straight One Member-One Vote voting method for the leadership election, I figured this wasn’t a bad experiment for offering another gauge of support for the various candidates.
I used the membership numbers for each province listed in this article. For provinces not mentioned, I used the membership numbers listed here. New Brunswick’s numbers aren’t available, though I believe that a number of 1,000 is a fair estimate, and as such I used that number (I have a lot of experience with the NB NDP, so I figure I can be granted some license in deriving this figure). When the numbers are crunched, we get a notably different picture of the leadership race. The results are illustrated below:
Using this calculation, we see a very tight race for first place between Thomas Mulcair and Brian Topp, while Peggy Nash jumps to third place yet isn’t far ahead of Paul Dewar.
Moreover, there is yet another gauge one can employ that is pointed out by Alice Funke, the brains behind punditsguide.ca. As she illustrates, the share of total campaign funds raised by each candidate in the 2003 NDP leadership race offered the most accurate guide for the results of the first ballot. Using current fundraising figures for each leadership camp in 2012 suggests that Topp and Mulcair and battling it out for first place while Nash, Dewar and Cullen are in a three-way fight for third. The strong fundraising figures from the Cullen campaign are particularly surprising. It will have to wait until the results of the first ballot are revealed to see if the fundraising metric was mere fluke in 2003 that can’t be repeated, or if it actually is a valid guide for future leadership races.
If all of the aforementioned support gauging models have a common thread, it is that Topp, Mulcair, Nash and Dewar are the four front-runners. Beyond that, any concrete prediction may be prone to spectacular failure. Ultimately trying to rely on all these models and formulas to provide an accurate prediction is really a fools’ game. In the absence of hard polling data of the party membership itself, one can only read various numeric tea leaves in trying to determine the shape of this, or indeed any party leadership race. That is perhaps why these entire exercises should be regarded as innocent amusement as opposed to hard psephology.
Pingback: On uncertain measurements | wellingtonstreetpost