The Narrative Equation

A quick look at the Furey-era polls

We are headed for an election this coming spring with the provincial and federal Liberals likely tagging one after the other, ideally provincial first.

Last week, Narrative - ex-CRA - dropped its quarterly polls across Atlantic Canada a week early and it had what appears to be some good news for the Liberals in Newfoundland and Labrador.

But you really have to look at the trending over time to see what’s really going on and what could happen next spring.

First of all, let’s understand that political parties and other s can influence these poll numbers. For 20 years, political parties in power in Newfoundland and Labrador have flooded the news cycles with hospital openings, new hospital announcements, sports complexes, fire trucks, cash give-aways and everything else they could find that’s happy and upbeat. It’s called poll goosing. Inevitably, local media report the goosed CRA/Narrative poll numbers without question and that can create an illusion or distortion of what public opinion actually is or would be without the deliberate manipulation.

Second, CRA and Narrative are notorious for discarding undecideds for the party choice question even though they are actually a legit choice for local voters. That also skews perceptions. Notice that *this* time, for the first time, Narrative reported both the raw numbers and the “decideds” only. You can see tossing out the undecideds inflates the Liberal vote by 15 points and the other party choice numbers by comparable amounts. It looks like every second person supports the Liberals when actually about one in three do.

Third, Narrative has disclosed their weighted and unweighted sample sizes this time. There are also results by geographic region, age and sex. This is interesting data if only because it can often show why a poll result is off or why you need to be careful of the result. For example, pollsters in BC weighted samples during one election to the population rather than according to likelihood of voting. The result was they missed who would actually win. Be careful, though. Each of these subsamples has a bigger margin of error than the raw one (and that is hideously large anyway) so don’t rely on them for any fine insights.

Now on with the show.

Understand the Numbers. Be a subscriber.