A week before the GPC convention, a poll puts them at 10%. What’s the real story?

An article published September 21st on ipolitics.ca carried the headline More Canadians than ever support Green Party. Reporting on an EKOS poll, the article quoted pollster Frank Graves as saying the party was “doing as well as they ever have”, with support “around 10 points”, and then stating “If there was an election tomorrow, they’d probably win a handful of seats.”

This would appear to be positive news for GPC supporters heading into the party’s national convention next weekend. The reality, however, is less inspiring.

Some matters of accuracy

Unfortunately, the article’s headline, and several of the quotes within, do not stand up to examination. I’ll start with some matters of accuracy, then get to the larger problem (and more troubling implication) in the second section of this post.

First off, Graves’ quote stating that the party is “doing as well as they ever have” is questionable. EKOS themselves polled the GPC at 11.5% support in their polling for May 2010. In fact, based on the graph in the linked EKOS article, they polled the GPC above 10% for over a year straight prior to the 2011 election, exceeding 12% several times.

Polls, of course, have margins of error, so perhaps we should consider “around 10%” and 12% to be statistically equal. (A margin of ±2% is nothing shocking in this type of polling.)

Furthermore, the article’s text implies that Graves might be speaking about a bigger picture that includes not only national support, but also where those votes are concentrated. So, let’s put a friendly asterisk on this first quote, but allow it to stand.

The article’s headline, however, raises a second red flag. Do “more Canadians than ever” support the Greens? Since that headline is based on EKOS’ recent poll, let’s compare with an earlier EKOS poll to answer the question:

• In May 2010, EKOS polled the Greens at 11.5% support. The Canadian population at that time was less than 33,476,000 (the number reported later in the 2011 census), which implies at most 3,849,740 Green voters.
• In September 2018, EKOS has polled the Greens at “around 10%” support from a Canadian population now standing around 37,067,011 (based on Stats Canada’s quarterly population estimates), which would imply 3,706,701 Green voters.

That’s a difference of 4% fewer voters today. We could invoke the margin of error again, but even that would only justify a headline stating that “as many Canadians as ever” support the Green Party. Even when we take into account the margin of error, it doesn’t seem that the article’s actual headline, suggesting more Canadians than ever support the Green Party, can be justified.

This is especially true considering Graves himself is not quoted as saying the party has more support than ever, only that they’re doing “as well as they ever have” [emphasis mine].

The final misleading aspect of this article is another of the quotes from Graves. He states that “If there was an election tomorrow, they’d probably win a handful of seats.”

Here my complaint is not with Graves’ statement itself. He is right, mathematically speaking, that if everyone got up tomorrow without any election campaign taking place, and didn’t put any second thought into their vote intention, then the Greens might win several seats.

The problem is that this is a fantasy scenario that has no relationship to the actual, upcoming 2019 election. It’s misleading because it implies that the Greens might be in a position to win several seats in 2019. As of today, there is no evidence that they are in such a position—and that brings me to the second half of this post.

The historical significance of one-year-out polling for the GPC

As long-time GPC supporters and observers know, the Green Party always polls higher between elections. Quite a bit higher, in fact:

• One year before the 2011 election, GPC vote intention (averaged across major polling firms) was 9.6%. The actual outcome of the election was 3.91% popular vote for the GPC. The year-out polling was 2.45 times higher (or 5.69 percentage points higher) than the election outcome.
• One year before the 2015 election, GPC vote intention (averaged across all polling firms) was 6.4%. The actual outcome of the election was 3.43% popular vote for the GPC. The year-out polling was 1.86 times higher (or 2.97 percentage points higher) than the election outcome.

Based on these numbers, a poll result of 10% today would suggest the party is headed for a 4.1–5.4% election outcome in 2019. This would be an improvement on the last two elections, though not a very large one.

However, there’s one other fact that must be considered: EKOS has historically polled the GPC 2–3% higher than other firms. Each polling firm has its own methodology, and for whatever reason, EKOS’ methodology has generally been the most favorable toward the Greens’ chances. This is not to say that EKOS is doing anything wrong, but simply that their polls represent the high end of the spectrum for GPC poll results.

As of today, the average of all firms’ polling for the GPC is around 6%. (See, for example, the CBC poll tracker.) This is the number we need to plug in to the ratios I gave above, if we want to see where the GPC really stands in comparison with past elections.

An average polling of around 6% today would be consistent with the GPC heading toward a 2.4–3.2% outcome in 2019.

If that support was extremely concentrated in a few ridings, it could still translate to a “handful” of seats. However, based on past riding-by-riding results, a maximum of two is much more likely. Even with a gain of one seat, seeing the popular vote drop for three elections in a row would have to be a concerning development for the party’s supporters.

So, if this poll is news at all, it’s not very good news—and it’s definitely not historically good news, as the article’s headline suggests.

[Edit 2018-Sep-24: Edited to improve clarity.]
[Edit 2018-Jul-18: Edited to add percentage-point differences for past pre-election polling and results, and to improve clarity.]