Thursday, February 24, 2011

My Issues with the Leaked Clark Poll

The polls was leaked to the Vancouver Sun and purports to show Clark in the lead.  I have issues with the poll.

1) It was a poll looking for about ten people from each riding - this means this it is in fact 85 different polls with 10 respondents each.   The pollster also has no way to know if they have a representative sample or now.   The statistical margin of error on each of the 85 parts of the poll is so large as to make the numbers meaningless.  

2) There is nothing indicating how many refusals they got in the poll.  There is also no indication of how they dealt with the 6000 members that were not real members.   There is also no indication of how many people are going to vote and how many not.  My estimate is that they could not reach a very large portion of the membership.

3) They found 27% had not made up their mind.  That is suspiciously high for this poll.    Of the 55000 new members, at least 85% are clearly committed to one candidate or another.  To get to 27% you would have to have a huge number of existing party members not have an opinion, that highly unlikely given the variety people have to choose from.

4) The pollster started calling before all the memberships were processed.   If the pollster was working from an old list and then used a different list later on, this would utterly skew the results.  

5) For Christy Clark to reach 43% of decided voters, she has to be pulling a large amount of existing members, but this does not make sense since she has very little public support I can find of members from before the leadership race.   42 ridings are behind one candidate or another, only one of those is backing Clark.   This would mean 41 of the MLAs would have to have almost no pull with their local riding association.    I have sat on various ridings associations, and a sitting MLA or MP holds a lot of weight with the members.   These are serious political types that sign up people to parties, that work closely with the riding association in the elections and between elections.   These MLAs are a very good barometer of the party members.   Clark has very limited support among members from before the race.

The results of this poll simply do not jive with all the other data that is out there in this race.
Post a Comment