Holinshed's Mayoral Poll; How it was screwed up

Let me preface this post by saying that I just finished reading the Ottawa Sun's editorial suggesting Clive Doucet should pull out of the mayoral race. The editorial even went so far as taking a moment out of context and quoting his initial reaction to the Holinshed poll that came out Monday morning while he was doing a Ottawa Sun editorial board interview. What a terrible move on the Ottawa Sun's part.  Fortunately for my readers, I have a treat.  Read this article and then we'll see some lovely images on the Holinshed Poll with some thoughts.

I have called the polling company in question three times this week and have been backed up with phone calls from others to request a report on said poll; they refused to give us anything or flat out ignore the requests.  I emailed them two nights ago and asked for a methodology, who requested the poll, who supplied a calling list, what time of day were these calls done at and in which wards were they polling from.  All of these need to be considered when doing a sample poll.  Consider that Ottawa has 23 wards.  On average it would be 17.3 people per ward making up that poll.  That doesn't sound right.  Could you imagine if one ward tipped the average while others had one or two respondents?  I am convinced that is one of the errors, but sadly I do not have the proof.

When Clive Doucet took the microphone on CBC Radio on Tuesday this week, he called into question the poll.  Host Kathleen Petty reigned Doucet in, but the following day she had a segment on polling.  I will get to that one a little later.  However, even CFRA, the ones believed to have requested said poll, also had something to say about it during The Lunch Bunch on Wednesday.  Michael Harris called into question both the sample size and the margin of error.  He also speculated that, "the only person who would benefit showing you dead in the water is Jim Watson."  I am not big on conspiracy theories, but I doubt it was Watson himself.  Could it be someone from his campaign or in his interest perhaps?  Could be.  I doubt we'll know before October 25th.

What about Kathleen Petty's spot on polling?  October 6, 2010 Petty began her 8 minute segment on polling with Doucet calling for more and creditable polls.  Nik Nanos of Nanos Research spoke in the segment to speak on the matter and he immediately said there needs to be more than one pollster to give perspective and let voters focus on issues.  Nanos and Petty also emphasized that polls can polarize issues or candidates and tell people how to vote. One of the most curious things Nanos brought up was the screening process of the respondents. It is important to screen people who were likely to vote and then there needs to be a way to counter the fluctuation and opinion.  Overall, polling is difficult and can cost money, but we have seen three online polls with telling information.

First, the Debate 2.0 Poll that was run during the live televised debate.  It showed Clive Doucet in a commanding lead with 58% of the vote.  Now, my poll is a little bias and I recognize that, but it too has a lead for Doucet with Watson trailing; Haydon and O'Brien are out.  Finally there's the Twtpoll that shows Doucet in a slight lead, Watson trailing and again Haydon and O'Brien are out.  Sure, online polling can be difficult to call accurate, but all three of the online poll flies in the face of the Holinshed poll.  If that doesn't convince someone that O'Brien's pretty much out, his own campaign went on the attack following Debate 2.0 and has been swinging for the fencing and cranking them foul.

So what about the Holinshed poll?  The company did not send me a copy of the report nor did they answer my email on understanding the poll.  However, I did receive a copy of said report with the caveat that my source neither knew who requested the poll nor it's methodology.  We simply have the questions asked and the numbers.

 

That's the first question.  Can you imagine the script if this was the first question asked? "Hi, we're polling about the mayoral race.... (insert that question)".  Nik Nanos said you want to start off by screening and in the report there is no indication of this being done.  Not even on the follow up questions.  Let's consider the question itself though. I am not sure I agree with including support and leaning in the same category.  Someone who is leaning toward a candidate can be won over by another.  They are, for all intents and purposes, undecided.

Generally speaking? Generalizations are also asking people to waiver.  However, the tone of the question is set with "is it time to elect someone as Mayor?"  I am going to assume the calling script had the word 'else' in there, but regardless, the question immediately leads those who are "leaning" to also say it's time for a new mayor.

 The next three questions already swing in Watson's favour.  The wording doesn't hurt listing Watson at the end (albeit, they did it alphabetically, however omitting the names of the candidates would likely be best since they did so for the first question).

The final question was one on the issues, but actually made me laugh.  Third on the list is accountability and honesty, but this poll clearly lacked both.

W.

Edit: Check this cover story on Geo Vote and how Holinshed uses it to ensure their clients get the best results possible: http://www.holinshed.com/TheBusinessPost.aspx#top
Edit: The Ottawa Sun is now trying to host a poll of its own to see if Doucet and Haydon should drop out of the race or not!  Go there, vote and have your opinion weighed.  Media should never be allowed to dictate who is and is not a candidate, let alone do so based off of a poorly conducted poll specifically designed to engineer a favourite candidate... and then try to make it look somehow legit with their own little poll to back their own Op/Ed piece up!