Thursday, March 15, 2012

NDP Leadership Power Rankings

Last month I looked at how NDP leadership candidates stacked up on various metrics, in a mostly futile attempt to handicap the race. With the Dippers' vote a week away, here are the updated Power Rankings (click to view):

MPs: Despite losing an MP to Brian Topp this week, Mulcair continues to dominate with 43 MP endorsements - 10 more than the rest of the field combined.

Endorsements: These are the 308 endorsement numbers, which tell a different tale than the MP endorsements, due to Topp's establishment support. Mulcair still leads at 27.9%, but Topp (26.9%) and Nash (23.9%) are on his tail, with Dewar (13.2%) and Cullen (5.2%) lagging behind.

Donations, Donors, Direct Donors: These numbers come via Pundits Guide and show Mulcair leading with 242k in donations, followed by Topp at 215k, then Dewar, Nash, and Cullen bunched together about 50k behind. The "direct donors" numbers is likely more relevant than the "donors" indicator, since the later includes "pass the hat" fundraisers and, let's be frank, any indicator that shows Martin Singh with six times Mulcair's support is bogus.

Poll: Numbers from the Dewar and Mulcair internal polls released last month. Given the source and the time lag, these numbers should be looked at with caution. Still, in past leadership races, polls of members have tended to be the best predictor of the outcome.

Media Mentions: The number of articles mentioning each candidate. Despite his fall to the second tier, Brian Topp is still being talked about by the media nearly as much as Mulcair.

Social Media: Cullen and Nash have the strongest Facebook presences, while Paul Dewar leads on Twitter followers. The "Twitter Mentions" column comes from a MediaMiser study released today, showing that Cullen and Mulcair have been the most talked about candidates on Twitter. Who knows what that means, but it's interesting to see Cullen generating more buzz on Twitter than with the mainstream media.

Average Share: This is, simply enough, the average share of the candidates across these 10 indicators:

Mulcair 23%, Topp 16%, Nash 16%, Cullen 15%, Dewar 15%, Singh 8%, Ashton 5%

It is by no means a prediction of their first ballot support. Still, it matches the narrative we're hearing, of Mulcair well in front of a tightly bunched pack of four candidates.

Momentum: Shows how each candidate's share has shifted since last month, confirming Cullen's surge:

Cullen +1.9%, Dewar +0.3%, Mulcair -0.1%, Nash -0.4%, Ashton -0.4%, Singh -0.5%, Topp -0.7%

Paul Dewar has quietly generated some positive momentum as well, with Brian Topp losing ground. Despite, or perhaps because of, his new frontrunner status, Mulcair has held steady across these indicators.



  • These seem like good variables for estimating first ballot support, but I'm not sure that is what is going to drive this race. Given the small size of Mulcair's lead (compare to say, Ignatieff in 2006), and the use of a preferential ballot it might make more sense to look for measures that can get at people's second and third choices.

    For instance, lets consider two different pictures one would have gotten from the 2006 race. We could look at who had the most donors - a good measure of depth of support:

    % of donors
    Ignatieff: 24.4%
    Dion: 19.8%
    Rae: 17.8%
    Kennedy: 11.7%

    But if we were more interested in breadth of support (which I believe to be the more important factor in this kind of race), a better indicator would be to look at the distribution of multidonors, ie. how got the most donations from people that also donated to other candidates. That picture suggests a more competitive Dion (compared to say, his poll numbers):

    Dion: 18%
    Ignatieff: 18%
    Rae: 16.5%
    Kennedy: 11.6%

    This can also tell us something about how important endorsements might be. For instance, looking at multi-candidate donors in '06, it appears that Kennedy supporters only had a weak preference for Dion:

    Bennett: 6.9%
    Bevilacqua: 1.2%
    Brison: 10.1%
    Dion: 23.1%
    Dryden: 6.4%
    Findlay: 10.1%
    Fry: .8%
    Ignatieff: 20.2%
    Rae: 18.2%
    Volpe: 2.8%

    That might suggest that Kennedy's endorsement was extremely successful, given that Dion picked up almost all of Kennedy's delegates.

    Of course it is a lot easier to get this kind of data after the fact. But there may be some ways to analyze breadth. Twitter and facebook, for instance, come to mind. Are supporters of Ashton likely to be friends with supporters of Topp? Do they follow Nash?

    The online preferential ballot is another such indicator. Yes it probably misrepresents the relative support of candidates. But I see no reason to expect second choice preferences in the poll to be biased in a systematic way, unless somebody gamed it.

    At any rate, the poll would suggest Nash and Dewar are stronger contenders than they have been given credit for. I'm inclined to find that argument plausible.

    By Anonymous hosertohoosier, at 4:47 a.m.  

  • Yeah, as hard as it is to predict first ballot support, predicting later ballot support is even more daunting.

    The multiple donor data is interesting, and I really like the idea of doing a Facebook study to see how the candidates cluster. Like you said, if Ashton supporters tend to like another candidate, that would suggest they might be more likely jump there.

    I'll be curious to see how the online preferential ballot stacks up to the actual results in terms of 2nd choices, and will definitely do a post on it after the fact. We also have Dewar's poll which asked for 2nd choice. Beyond that, there's been little effort to try and project where the vote will after the first ballot.

    By Blogger calgarygrit, at 10:23 a.m.  

  • Federal Liberal leadership race, Alberta PC leadership race (the Stelmach one) and Calgary mayoral race all fit a pattern where neither of the two front-runners had enough support to win by themselves, which allowed a third person to collect all the anti-X votes, and win.

    Given how flat the support numbers are, evenly divided five ways, this could easily happen again.

    Whoever comes out as the top two (probably Mulcair and somebody) wouldn't be favoured to win, but instead the 3rd placer.

    That also happened in H2H's recent simulation, the 3rd placer being Nash.

    By Blogger Robert Vollman, at 1:52 p.m.  

  • thanks for such a informative post, the statstic really helped me alot.

    By Anonymous dropshipper, at 10:17 a.m.  

Post a Comment

<< Home