Monday, February 13, 2012

Dewar poll, the flaws I see

Paul Dewar's campaign has done a large poll of NDP members to see what the voting intention are of the people.   I think this poll is very, very flawed, here is why:

  • The final party membership is not yet set, new sign ups are still coming in.   A smart campaign holds back a lot of their supporters till the last minute so that other campaigns can not phone them as much
  • The poll can not be sure that the sample is a representative sample of the membership and since it can not know that, there is no way to correct the results to reflect the actual party membership.  They weighted based on provinces but nothing else that I can see
  • I see nothing on the Dewar website indicating who did the polling
  • The poll does not state how many respondents did not know who they would vote for - I tired to work backwards from their margin of error calculation to see what their decided vote response rate was, but when I did so, their margin of error numbers do not work out for their stated numbers
  • The poll lists a margin of error, but that only works if the sample is actually representative of the NDP membership as a whole.   They are using a large number of responses as a way to add legitimacy to their results.
  • This is a poll of the most politically savvy people in the country.  People that are members of political parties are maybe 1% of Canadians.   These are the people most engaged in the political process and the ones most likely to lie because they understand that a horse race makes it easier to get your supporters to act.   In my case, I could have answered the poll because my home was called.   I am not a member of the NDP and the poll had no way to screen me out.
  • NDP members are being flooded with calls.  My 14 year old Ben is a member and therefore our house phone is getting between five and ten calls a week.   We are not answering most of them.   So how does the pollster account for all the refusals of the calls?
  • Finally, this poll was released by one of the campaigns.  Any poll released by a candidate normally is an indication that they results are 'wonky' but show the sort a favourable result for the candidate.  Though the poll has had the result Dewar's campaign wanted, coverage of the campaign.

So here are the headline results:  6,373 calls to 56,522 NDP members

  • Candidate     1st choice  2nd choice  1st + 2nd
  • Thomas Mulcair 25.5        16.7        42.2
  • Peggy Nash     16.8        19.4        36.2
  • Paul Dewar     15.1        21.2        36.3
  • Nathan Cullen  12.8        14.4        27.2
  • Brian Topp     12.7        12.4        25.1
  • Nicki Ashton    9.5        10.7        20.2
  • Martin Singh    4.1         1.8         5.9
  • Romeo Saganash  3.6         3.6         7.2

Where do I start with these results?

I am not sure of the polling dates and they may have happened after Romeo Saganash left the race.  I do not know how to count his results.

The second choice results are suspect because there should be a significant "no second choice" as part of the poll.  Actually the second choice results are meaningless because what is needed is to know how the second choices of supporters of any candidate break out.

The Martin Singh results - they are simply and completely beyond any possible credible belief.  Martin Singh is going to be lucky to break 1%, he is not at 4.1%.

So out of nowhere this poll has Paul Dewar in third, where is that coming from?   BC, the province with the most members, I see no evidence of any sort of strength on the ground.   I do get the sense that Brian Topp is doing well in BC.  I do not see the party heavy weights coming out to back Dewar either.   Could he be signing up tens of thousands of members and winning through that?  I doubt it because I do not see any names among his core campaign team that indicates they can do the mass sign ups needed.

Nathan Cullen ahead of Brian Topp is simply not what is going on out there.   I like Nathan Cullen but there is no team behind him and clearly does not have the financial resources to be doing well in the race.   I will admit it was odd getting a robo-call on his behalf from my friend Corky Evans.

My dead reckoning of the race, and I admit I do not have enough data points to show how I get to this.

  1. Brian Topp - 28%
  2. Thomas Muclair - 28%
  3. Peggy Nash - 20%
  4. Paul Dewar - 13%
  5. Nicki Ashton - 7%
  6. Nathan Cullen - 3%
  7. Martin Singh - 1%

I do not see Peggy Nash getting enough support from the four below her to be able to be on the final ballot.  

This is a closed race and only party members get to vote.  Polling is more or less useless in actually predicting the race because the only thing that really matters is getting new members signed up and then getting them to vote.  

The poll was a very cynical attempt by the Paul Dewar campaign to make up for their lack of support in volunteers and money.   The worst part is that the media has covered it as much as it has.
Post a Comment