Friday, November 11, 2011

NDP and BQ leadership endorsement rankings

With the provincial elections behind us, the leadership races for the New Democrats and the Bloc Québécois now stand as the two most important political events scheduled for the next five months. With this in mind, ThreeHundredEight now turns to these leadership races.

A few weeks ago, I read an interesting piece on Nate Silver's FiveThirtyEight that talked about Republican presidential nominee endorsements, and how they are a good indicator of who is likely to perform well in the primaries. It got me thinking how this could be applied to leadership races in Canada.

That leads me to present ThreeHundredEight's endorsement rankings, which will be maintained and updated throughout the race for both the NDP and the Bloc, and perhaps the Liberals once that campaign gets started. You'll also see charts tracking leadership polls in the right-hand column for the NDP race.

A few caveats: this is, in no way, a scientific method that is meant to have any precise predicting capability. It is just one way to interpret the endorsements and what they might mean in the respective leadership races. And they are just to make the race a little more fun.

So how does it work?

Points are assigned for each endorsement received by a candidate. The amount of points assigned, much like Nate's system, is somewhat arbitrary. But, I did use the 2006 Liberal and 2003 NDP leadership races to calibrate the points system, so they are based on something real.

The chart on the left spells out the points system. Having the endorsement of a former party leader is most important. For every election the former leader led the party, 20 points are assigned.

The endorsements of Members of Parliament are also very important. For every election they have won, 2.5 points are assigned to the leadership candidate. This means an MP who has been around for a long time is worth more than one who was just recently elected.

Former MPs give a candidate two points, while senators (though I don't expect any to come forward in these two races) are worth one point.

Having the support of a current or former provincial premier or party leader is also quite important, but that importance is based on what province the leader comes from. For the NDP race, Quebec, Ontario, and British Columbia are considered "first tier" provinces. Alberta, Saskatchewan, Nova Scotia, and Manitoba make up the second tier, New Brunswick and Newfoundland & Labrador make up the third tier, and PEI is in the fourth tier. The territories are in a fifth tier of their own. These tiers were determined by the number of votes received in the 2011 election, as well as the strength of the party brand in each province and whether the party is in government.

Obviously, these are somewhat arbitrarily chosen. And this is why I emphasise that this is not a scientific exercise, it is just a spot of fun. But the points were not chosen completely out of the air. Let's look at how the system performs for the 2006 Liberal leadership race.
As you can see, aside from Scott Brison earning more points than Ken Dryden, the points sytem does a good job of estimating first ballot support. Of course, this should be expected. The Liberals used a delegate voting method where endorsements were the name of the game. But there were still some delegates undecided going into the first ballot.

After the first ballot, Joe Volpe and Scott Brison went over to Bob Rae while Martha Hall Findlay went over to Stéphane Dion.
Assigning their endorsement points accordingly gives us a good result, with the order of the candidates correct and their share of the points closely lining up with the results of the second ballot. After that ballot, Dryden went over to Rae and Gerard Kennedy went over to Dion.
After assigning their endorsement points, Dion moves to the top ahead of Michael Ignatieff and Rae drops to third. Again, their share of the endorsement points matches up well with the third ballot result.

After the third ballot, Rae dropped out but didn't lend his support to any other candidate. Dryden and Volpe went over to Dion while Brison went over to Ignatieff.
And after assigning their endorsement points, Dion wins more than half of them and his share of the points matches well with the fourth ballot result.

After using the Liberal leadership race to calibrate the points system (which did not need too much tweaking after my own estimates to start the process), I moved to the 2003 NDP leadership race. It used a bit of a mixed system of delegates and one-member, one-vote (OMOV). So, it was a different kettle of fish.

The big difference in this race was that Bill Blaikie had most of the individual endorsements, but Jack Layton had the support of Ed Broadbent. That was crucial, and so I was able to calibrate the points system by taking into account Broadbent's influence.
As you can see, aside from the vote share of the three bottom candidates, the points system matches up relatively well with the results of the first, and only, ballot.

I understand that the points system is fitted for these past leadership races, so they should line up well. But that it can be used across two different kinds of leadership races with generally good results indicates that the endorsement ranking I will be using for the next five months is not meaningless. Will it predict the outcome closely? Probably not, but that is not its purpose. It is simply an interpretation of the endorsements that tells us something about the race. Having endorsements generally means a better and/or more motivated organization, and a better and/or more motivated organization will more successfully deliver votes on the metaphorical convention floor.

So, now that the explanation is out of the way, let's look at how the endorsements stack-up in the Bloc Québécois leadership race.
Daniel Paillé, who is probably best known considering his background as a cabinet minister in Quebec, leads the endorsement ranking with 30 points, or 50% of all available endorsement points. Maria Mourani has 19.5 points to 32.5%, while Jean-François Fortin has 10.5 points ot 17.5%.

This endorsement ranking actually lines up pretty well with the only poll we have, taken in early September. It put Paillé at 11% among BQ supporters, ahead of Mourani (6%) and Fortin (3%). 

As you can see in the chart to the left, this race is not very dynamic. There are only four sitting MPs and two of them are in the race. Their own "endorsements" for themselves are the most important factors. All the other endorsements come from former Bloc MPs.

Perhaps if a few MNAs step-in or one of the former PQ leaders stands behind one of the candidates things could get a bit more interesting. But as it stands, Paillé has the backing of more former MPs than any other candidate.

Now, onto the New Democratic leadership race. This is a much more dynamic one, as endorsements have come in from provincial party leaders, sitting premiers, former federal leaders, and provincial legislators from across the country.
Not surprisingly as he is seen as the "establishment" candidate, Brian Topp has the most endorsement points with 166, or 43.3% of all endorsement points currently on the table. If this result was repeated on voting day, Topp would not win on the first ballot.

Thomas Mulcair comes in second with 104 endorsement points, or 27.2%. He has the most individual endorsements, but as most of them are from first time MPs from Quebec they are not worth as much as Topp's. This, I think, will reflect the organizational weakness the NDP has in the province, as well as its low number of members (though that number is going to increase as the campaign drags on).

(Right-click and open the chart to the left in a new window to be able to magnify it.)

Coming in third is Peggy Nash, thanks in large part to the support of former leader Alexa McDonough. She comes in at 56 endorsement points, or 14.6%.

If we can consider Topp and Mulcair to be first tier candidates and Nash a second tier candidate, we then get into the third tier. Now, any one of these could move up as the campaign goes on. Some of them have just jumped into the race recently. And the fact that someone like Nathan Cullen or Paul Dewar does not have a lot of endorsements might not mean anything in a OMOV system if they have good grassroots support, but the rules of the game being what they are, we have Robert Chisholm, Cullen, and Dewar as what I would consider the third tier candidates.

Chisholm is narrowly on top of the others due to his support from Atlantic Canada (i.e., Darrell Dexter), while Cullen and Dewar so far are relying on endorsements from provincial MLAs in British Columbia and Manitoba, respectively.

Niki Ashton and Roméo Saganash round out the main candidates, in what I'd call a fourth tier. Both of them have the support of two sitting MPs, but Ashton's longer time in the House of Commons puts her ahead of Saganash. Martin Singh has no endorsements at this point.

There we have it. The endorsement rankings put Brian Topp and Daniel Paillé ahead of the others for their respective parties. But I hope the killjoys out there won't take it too seriously. Does it have limitations? Absolutely. But that doesn't mean we can't have some fun with it. And since this isn't set in stone, I'm happy to hear your suggestions on how it can be improved.

The Bloc race is going to be wrapped up in a month, but the NDP race is just getting started. It will be interesting to see how the rankings twist and turn over the next five months.

27 comments:

  1. Very Interesting, I noticed that one of the endorsements you have for Brain Topp is Dawn Black. You have only given her a weight of 1.00, but she is also a former member of parliament for New Westminster-Coquitlam, is it because she is a current MLA you went with the position she has now instead of what her position was?
    Thank you for the clarification

    ReplyDelete
  2. I gave her the weight of a BC MLA because that is what she currently is, but you bring up a good point.

    I suppose that, as I did it with Chisholm and McDonough, I should combine the points for Black, giving her a total of three.

    Thanks for the comment, this change will be reflected in the next update.

    ReplyDelete
  3. This seems like an interesting way and effective to calculate candidate support. It seemed to work well with your example of the 2006 Liberal leadership convention.

    What is interesting is that removing the Ed Broadbent endorsement from Brian Topp's column, would give Thomas Mulcair a comfortable lead.

    Personally, I'd like Paul Dewar to win the leadership, but it looks like it would not happen. His lack of party support is clearly showing with no MP support. Being fluently bilingual is also a must in the post-2011 NDP, and Dewar is weak in French.

    I wonder if other high-ranking NDPers would endorse candidates in the upcoming months. Endorsements from people such as Stephen Lewis, Gary Doer, David Miller, Greg Selinger, Andrea Horwath or Adrian Dix can certainly help.

    - Maple

    ReplyDelete
  4. I think David Miller is most likely going to stay out of this, unless an issue that directly affects Toronto comes up. He's made a habit in the past few years of not voicing any political opinions unless they affect the city.

    This is an interesting way to look at things, Eric. Perhaps arbitrary, but still useful as a quick way to see who might be ahead.

    ReplyDelete
  5. I think Dewar actually has some high profile endorsements from MPs and other luminaries in his back pocket. I think its his strategy to publicize them at a later date.

    ReplyDelete
  6. This is really, really clever, Eric. I was thinking of doing something similar but I couldn't figure out how to put it down to numbers. Good job, hope you don't mind if I copy all this onto my blog (credit of course to you).

    ReplyDelete
  7. What if Olivia Chow changed her mind and entered the race - will she get a boost being Jack Layton's widow.

    What if she decides to endorse someone instead. Will that be worth more than an MP endorsement

    asdf

    ReplyDelete
  8. Two points:

    1) Just noting Broadbent alone represents ~21% of all available endorsement points and he was leader over two decades ago.

    What happens if a weighting is applied to all former leaders, MPs, MPPs etc. that is related to the length of time since they last held office?

    2) Certain individuals like Libby Davies, Cheri DiNovo, Darrel Dexter, Ed Broadbent etc. are exceedingly popular with different sections of the Party. Is is possible to incorporate this "star power"?

    ReplyDelete
  9. Maybe new Quebec MPs who have never run for anything before should have less weight. I doubt many of those MPs can deliver more support than a Manitoba or BC MLA or an Ontario MPP.

    ReplyDelete
  10. asdf,

    That is a good question, but we'll cross that bridge when and if we have to.

    Anonymous 10:46,

    Yes, Broadbent is a huge factor, but his support for Jack Layton was a huge factor. Broadbent is probably the most important NDP figure alive today.

    I'm not sure if length of time since holding office is very important. For example, though Broadbent was leader before McDonough, I'd say he has more influence than McDonough does.

    Aside from DiNovo, that "star power" is already reflected. Along with Godin, Davies is the most valuable MP and Dexter is worth 10 points.

    Jim R.,

    You may have a point, but we'll have to see how many new members they sign up in Quebec. They are working from a virtual blank slate, so there is more growth potential, and having all of those Quebec MPs working to sign up new members should give Quebec a greater share of the membership total when voting day rolls around.

    ReplyDelete
  11. I suspect you're underrating Roy Romanov and Saskatchewan as a whole. The NDP has a solid base here even if they've fallen on hard times lately. On the other hand, I'd drop the Liberals to the third tier in Saskatchewan as they barely exist here anymore. Of course, Bob Rae might declare himself leader for life and there'll be no need for a prediction.

    Another Jack

    ReplyDelete
  12. Hi Eric,

    Very cool. I was wondering whether you have given any consideration to endorsements from non-politicians. I would be inclined to say that the support of, e.g., Duncan Cameron and Murray Dobbin for Peggy Nash should definitely count for something. At least as much as the support of some backbench MP or provincial MLA. And the support of, say, Ken Georgetti or Sid Ryan, should they give it, similarly.

    I would expect this would apply to the Bloc, indeed to any party, as well. There are obvious figures for the Conservatives (think Ezra Levant or Link Byfield) and the Liberals (just think Tom Axworthy ...).

    I could see it being a challenge coming up with something like a comprehensive list of "party influentials", beyond their parliamentary and legislative caucuses. But there is no question that such influentials are there and play a role.

    Best,
    Chris

    ReplyDelete
  13. Something to keep an eye on as the campaign unfolds!

    ReplyDelete
  14. Are municipal politicians that are not included elsewhere given any weight? Having a big city mayor, I believe, would get some power included. I would add a category:

    Major city mayor (i.e. greater than 500,000) - 1.50
    Major city councillor - 0.25
    Large city mayor (i.e. 200,000 to 499,999) - 0.50
    Medium city mayor (i.e. 50,000 to 199,999 or provincial/territorial capital under 50,000) - 0.20

    (Below 50,000 they would have very little influence)

    ReplyDelete
  15. I'd just like to point out that if you applied a similar methodology to the recent BC Liberals' leadership race, as Christy Clark had next to no endorsements.

    ReplyDelete
  16. I assume you meant to finish that thought with the fact that this method would not have predicted Clark's win. You are absolutely correct. I think I've inserted enough caveats in the above post to make it clear this isn't infallible!

    I've actually run it through 13 leadership races in total, and overall it gets more right than it does wrong.

    I'll go through the results in more detail soon, but Clark is one that this method would miss. Alison Redford as well. It would also not work in several other leadership races, but Clark and Redford are the big standouts.

    ReplyDelete
  17. No I totally agree re: the caveats. I wonder though if the nature of the leadership races have something to do with the difference - BC and Alberta both hate votes of the membership with pretty healthy party recruitment numbers IIRC, whereas the federal Liberal race was a vote of delegates. It makes sense that each of these endorsements points would matter that much for a vote of delegates, as each MLA, MP, Senator etc. has a local organization to coordinate with for delegate selection meetings. Whether or not that holds as true for a vote of members... I don't know.

    For the Bloc and the NDP though have you though of including union endorsements? That seems relevant here. Just a private sector endorsements were relevant for the BC Liberal race.

    ReplyDelete
  18. As opposed to the BC NDP which did not have a large amount of membership recruitment...

    ReplyDelete
  19. Both Nash and Topp have endorsements of unions or union members. Could that be worth endorsement points?

    Can certain endorsement be worth negative points (ie. Bob Rae or certain unions that people are angry at like ATUs 113 and 1587)?

    asdf

    ReplyDelete
  20. I'll look at the NDP's race in Ontario and see what union support meant in that race. My system correctly called Andrea Horwath as the eventual winner, but she trails until the third ballot (instead of leading on every ballot).

    ReplyDelete
  21. I agree with Anonymous up there: Roy Romanow nets 5 points? But an endorsement from Glen Clark would give a candidate 10 points?

    ~a diff anon

    ReplyDelete
  22. Small mistakes on one Topp endorsement, Joy McPhail is a former MLA but was also a former party leader in BC - she should be worth 5 points. Carole James is a former leader, but also a sitting MLA that is planning on running for re-election

    Should MLAs also not be worth their points x their elections?

    Also why is Quebec 1st tier NDP and Manitoba not? What is that based on?

    Final note, the BC MLAs endorsing Brian Topp are many of the same that ones that endorsed John Horgan. One 4 backing Nathan Cullen are those on the outs with the party structure in BC

    ReplyDelete
  23. Joy MacPhail was an interim leader, which I'm not counting as making a figure equal to a former leader chosen via a convention (or acclamation).

    That is a good point about James, though, so I'll give her an extra point.

    I've decided not to give MLAs a worth based on their number of elections, since this is a federal leadership race, and I have not done any tests for MLA terms in federal races.

    Quebec is a 1st tier because of the number of MPs from the province and the vote share the NDP got in the province. I realize their membership is small, but it is growing and I imagine it will continue to grow at a high rate. It is also a province that will be kept in consideration by other members because of its importance to the NDP's recent success.

    In any case, it doesn't matter because there are no NDP MNAs or provincial leaders.

    Manitoba is second tier because it is a province with a small population.

    ReplyDelete
  24. Joy MacPhail was called the interim leader, but she was more like a opposition leader than Carole James was. Joy MacPhail is a power in BC NDP politics whereas James never has been. Given that BC has more than a 1/3 of the NDP members at the moment, someone like Joy MacPhail carries a lot of weight.

    That said, the BC NDP membership rose in the run up the the BC NDP leadership race and those memberships lapse in January (I think)

    The one reason I would see Manitoba as 1st tier is because the party has a long term strong history on the provincial level. It also has the 3rd most NDP members

    ReplyDelete
  25. If you're right about MacPhail vs. James, then Topp having both should amount to the same thing no matter who is assigned the points.

    ReplyDelete
  26. You may need to rethink your ranking system in light of the May 11 election. Assigning 2.5 points per election to Members of Parliament does not allow for the fact that many of the first time MP's were elected in areas with no party organization and very little membership. These MP's have far less opportunity to influence voters than a long term provincial MLA with a strong constituency executive.

    ReplyDelete
  27. The system is more abstract than that. The points each endorsement brings to a candidate is not just the direct support an individual can deliver. Endorsements bring indirect support and also reflect the support a candidate already has or has the potential to earn.

    Indirectly, having good caucus support can be an attractive thing and an indication to a voter that a particular candidate would be a good leader of the party.

    In terms of how endorsements act as a reflection of a candidate's support, there are two factors. Most people lending their endorsements want to side with a winner, so they are unlikely to endorse someone they do not believe has a chance of winning.

    But also the innate qualities an endorser sees in a candidate should be visible to others. In other words, if a lot of people endorse Candidate X, it is because they see them as having good leadership potential. If these endorsers see that potential, other people within the party should be expected to as well.

    So, there are many factors at play - it isn't just about how many votes an individual can directly deliver.

    ReplyDelete

COMMENT MODERATION POLICY - Please be respectful when commenting. If choosing to remain anonymous, please sign your comment with some sort of pseudonym to avoid confusion. Please do not use any derogatory terms for fellow commenters, parties, or politicians. Inflammatory and overly partisan comments will not be posted. PLEASE KEEP DISCUSSION ON TOPIC.