Meaningless polls tell us little about Sask. Party race outcome
Candidates releasing snippets of polls that cast their chances in favourable light
Those trying to determine who is going to win the Saskatchewan Party leadership race through what’s comically been referred to as polling are finding out it’s like being a thirsty man being cast adrift in an ocean of salt water.
“Water, water everywhere,/ Nor any drop to drink,” says the Rime of the Ancient Mariner.
One can’t completely blame the pollsters — an industry under siege for years now. The combination of a public overly suspicious of pollster/unrecognized numbers on call display that go unanswered and the massive reduction of land lines (especially among younger demographics) has taken its toll on this industry.
However, polling for leadership races has never been especially accurate because pollsters can’t easily get at who is voting (at least when it comes to public polls and not candidate-paid polls where they can dive deeper into membership lists).
In the case of the Sask. Party race, we won’t know who those voting party members are until the wee hours of Saturday morning when party officials (aided by accountants at MNP) verify the last-minute mail-in ballots. Soliciting opinions from mostly nonparty-member masses is nice, but (at best) it’s a guess and (at worst) can be highly misleading.
Still, some media outlets that should know better continue to broadcast such polls as if they are the gospel brought down from on high. As one observer noted on social media Monday, some of the polls are starting to carry the scientific accuracy of betting on one specific floating bathtub toy in the Lumsden Duck Derby.
Alas, a thirsty man can do little to help himself. So as we turn the corner into the final week of the Sask. Party leadership race, we are hearing a lot about potentially specious polling results without nearly enough context that tells why we should not take any of these polls too seriously.
One such poll boldly predicted a battle between the Sask. Party’s Ken Cheveldayoff and Ryan Meili in the next provincial election, based on what amounts to a scientifically insignificant sampling of self-identified party supporters.
In this case, NDP combatant Trent Wotherspoon found it necessary to rebut with a news release hammering the methodology of the survey of 452 federal New Democrats who donated more than $250 to the separate federal wing. (The Wotherspoon news release then went on to release its own internal numbers, painting an inverted picture of him way ahead of Meili. Sigh.)
Another poll Monday commissioned by a local radio station predicted a dead-heat race between Cheveldayoff and Alanna Koch coming down to the wire (which, at least, does seem to be the conventional wisdom). The problem, however, is the drill-down of this online survey only got to 104 actual Sask. Party members (again, who may or may not have voted), producing a result with a margin of error so large it really doesn’t tell us much of anything.
What might have been telling, however, is 48 per cent of the population in this survey who were really unsure who among the five candidates they wanted to replace Brad Wall. Replacing the high-profile leader will be all that much more complicated.
Add to the mix the candidates’ willingness to share selective snippets of their own polling, while refusing to release any polling in its entirety.
For example, Koch’s camp last October issued a news release declaring: “Alanna is positioned in a three-way cluster at the top with Ken Cheveldayoff and Scott Moe within the statistical margin of error.” The poll was said to be a sampling of 500 Sask. Party members with a reasonable margin of error. But with her unwillingness to release actual results, it amounts to nothing more than spin and political gossip.
Finally, for pure entertainment value, we see the completely unscientific Manning Centre survey, which again showed Koch and Cheveldayoff in a dead heat and the top four in a tighter-than expected cluster.
It’s all fun, but it doesn’t quench anyone’s thirst.