Sunday, August 02, 2020

Time for Another Comment on Not Trusting Polls

Anybody that has read my posts before knows that I do not trust the public polls.  Having been in the survey business, during my career, I know that designing and conducting surveys is as much art as science.  I also know that there are ways to crank survey results that still meet the accepted standards for doing surveys.

Companies do not use polls to make money they use them as a marketing tool because there is no money in conducting polls.  If a polling company is conducting a poll on behalf of a media company it is usually on a cost-recovery basis and the media company will want estimates that will allow it to support whatever narrative they are pushing.  If a polling company is doing a poll just to do a poll they are spending their own money on it so they will be as economical as they can be, in which case it becomes a situation of "You get what you pay for."

As well, do not get me started on poll aggregates and seat models.  There are no industry standards for them so how they create their models is totally up to the modeler and none of them are without bias.  I remember during the last election a poll came out that was very favourable to the Liberals and when I went to check a certain seat model, that shall remain nameless, that poll was not included on the list of polls used to create the model.  No reason was given of course but the polling company was one of those companies with a generally good reputation so it was a mystery as to why it was excluded. 

Which brings me to the polls we have been seeing for the past few months.  Many Liberals cheered on the Liberals' rise in the polls as the pandemic unfolded.  Now the Liberals are sinking in the public polls and many Liberals are worried, citing the WE issue as being the problem.  I am not so certain that WE is the culprit as it has been in the news for a month and it had no impact on the public polls until week four.  Seems odd to me and a little convenient.

At any rate, good or bad, none of that really matters.  I have always said if you want to see what the true state of politics is in the country watch the political parties.  

The Liberals are just doing their job as the government.  You would not have known the polls, during the spring, were saying they could win more that 200 seats.  On the other hand when the WE issue hit they did not try to change the channel like alot of governments do when they find themselves having to deal with controversy.  Again they have just been doing their job.  

The Opposition Parties on the other hand have been having conniption fits.  Even before the WE controversy they have been trying to create some kind of scandal.  Anybody remember the Harrington Lake issue a few short weeks ago?  Their desperation has been palpable for awhile.

The reason is simple.  The Liberals and the Conservatives and maybe the NDP keep polling and data analytics teams on retainer.  They would have continued to do so to help them fine tune their message and they are not just the minimalist, superficial polls like the public polls.  They do in depth data analysis using the latest data analytics techniques.

What would these data be telling the the parties.  My guess for the Liberals is they did get a boost from their handling of the pandemic and its economic fallout.  More importantly I believe the data is telling them that the favourable opinion of the government is hardening as time goes on.  They expect some fluctuation but it will be from a higher baseline than just a few months ago.  As this situation continues it will only harden more because change in public opinion is glacial.  It does not change quickly.  Further I would bet that their internal data are telling them that the issue most important to Canadians is the pandemic.  Other issues are well down on the list of concerns.  So, all they need to do is stay the course.  It is the best course for winning the next election.

The problem for the Conservatives and maybe the NDP (if the have the money for internal polling) is their data is probably telling them the same thing.  They know the longer this situation goes on the harder it will be to bring that baseline back down to just prevent the Liberals from winning a majority government, let alone unseating them as the government.  Hence, the increasingly strident attacks on the Liberals, about everything and anything they can think of.  They need some issue, besides the pandemic, to break through the consciousness of Canadians.  To make things worse they are dealing with a minority government, which does not give them the luxury of time in which to do it.  Ironically, if we had a majority government there would not be the same urgency as there is now.

So, as usual, I would say ignore the polls.  They are not reliable in really assessing the political situation in the country.  Watch the parties and what they do.  To the discerning eye they are a much better indicator of what is happening.  Just watching them it looks like the Opposition Parties are in much greater trouble than the Liberals.  That could change but for now I would say that is the situation.

2 comments:

Jackie Blue said...

I've been freaking out about polls too. But I also have to keep telling myself that polls outside the writ period are more like temperature gauges than prognostications. There's also push polling, where questions can be skewed in such a way as to influence the answers. Some of the more recent ones that put vote intention at the very end of an issues survey look a lot like they are. You had some time to think about this or that issue and how it affects your judgment of the government. Then answer accordingly. Call and response.

People can lie to pollsters. Especially in the era of Internet panels where you sign up in exchange for an Amazon gift card or some other incentive. They select from the pool they want to send a weekly survey and rotate eligibility to win a prize.

Nevertheless, if the opposition parties really were stupid enough to take advantage of the public polls to make good on their threats of running a campaign in the midst of a pandemic, they'd commit electoral suicide. I dare them to do it.

What I do know is that PMJT wouldn't have gone to that Benghazi committee farce if he didn't see there would be any political benefit to it. He did, after all, skip two of the debates last year. So it could be they have some internal data telling them how the public would respond to this.

Although I did like the comment from one of the panel pundits this past weekend where she said that PMJT denied Skippy Poilievre his desired social media soundbite. That's got to throw the Cons off their data game at the very least.

ottlib said...

"Some of the more recent ones that put vote intention at the very end of an issues survey look a lot like they are."

Like I said cranking polls in a way that still meet the industry standards for conducting a poll.