CBC confused poll watchers. Finally clarifies its mess.
It took some effort but we finally got CBC to clarify its election poll.
We used the CBC poll tracker for our current federal party polls standing. However, their poll was confusing.
The current standing, 8-26-19, showed the party percentages as Cons: 33.9, Libs: 32.7, NDP: 13.7 but the seats projection projected a Liberal victory. This didn’t make sense. The Cons were ahead in the polls but lose in the seat projections.
Either we were misinterpreting the information or the pollsters were confused.
After numerous attempts at contacting the CBC, poll analysis has been updated and the confusion is clarified below.
The original analysis by CBC POLL TRACKER was very confusing as noted above. However, after determined pursuit of a clarification from the reporter, Eric Grenier, we finally got an explanation from the CBC’s Paul Hambleton, Director, Journalistic Standards and Practises, CBC News.
Mr. Hambleton wrote:
To say that the parties are tied in the polls, but one is likely to win more seats does seem counter-intuitive, I appreciate that. But there is some sense to it. Let me explain what I mean.
The results of individual public opinion surveys may vary somewhat because of the sample size, the time they were done and the methodology they used. CBC’s Poll Tracker combines the results of all the publicly available polls to produce a single average value. As I write this on September 9, the updated values prominent on the page show the Conservatives at 33.8%, the Liberals at 33.4%, the NDP at 13.3%, and so on. (You can find a partial list of the individual polls used toward the bottom of the page.)
The headline numbers reflect voter intentions (at the time the survey was done) and are based on answers to a question usually along the lines: If there was an election tomorrow, what party would you vote for? They reflect the intentions of a carefully selected cross-section of Canadians of different ages, locations, incomes, education, and so on. Although they don’t indicate regional preferences, the results can give you a pretty good idea of the overall popularity of the parties and the political mood of the country.
But that’s just one measurement. It might imply which party would win the election, especially if one party is significantly ahead of the others, but it can’t predict the number of seats a party might win, even though that might be a more pertinent number.
Estimating the seat count means looking at the voter intentions results on a province-by-province basis, comparing them with the number of seats in the province and with past voting patterns and estimating the effect those intentions will have in each riding. Of course, some regions have fewer seats than others. So, while one party might have overwhelming popular support in a particular region, thus pushing its overall voter preference percentage higher, with fewer seats in that region, the preferences may not result in a higher seat total.
The specific example often cited is the western provinces where the Conservatives are overwhelmingly popular, but where there are relatively fewer seats, compared to Southern Ontario where the Liberals are more popular and could win more of the much larger number of seats available.
I hope that clarifies things somewhat. Thank you again for your email.