Friday, May 10, 2013

Gap continues to close, but time running out in B.C.

The newest projection for Tuesday's B.C. election shows the gap between the B.C. New Democrats and B.C. Liberals narrowing further, to 6.3 points. And for the first time, the high seat projection range for the Liberals puts them just over the bar of a majority government (43 seats). But with only a few days remaining, Christy Clark is still just a 10-to-1 shot of being re-elected.

The New Democrats are projected to take 43% of the vote based on current polling, or between 40.5% and 45.5%. The Liberals have increased to 36.8% (or between 34.4% and 39.2%), but there is still enough volatility in the polls to give the NDP as little as 37% by election day or as much as 49%. Suffice to say, at those extremes the Liberals are either decimated or re-elected. But they are extremes.

These numbers give the New Democrats 52 seats to 32 for the Liberals, but the projected high and low ranges now overlap: 40-65 for the NDP and 16-43 for the Liberals. Those are rather wide as the amount of polling data has thinned out (it should beef up again between now and Monday night) and the race is rather close in the Interior and North. But that 43 is the highest the Liberals have been in the projection ranges. The forecast ranges have tightened up a little as some older polls have dropped out of consideration, but they nevertheless still envision anything from an NDP landslide to a Liberal majority. We shall see if they will tighten up further, but with the changes that have occurred in the last two weeks it is perhaps wise to keep our expectations to a minimum.

However, the NDP is still the heavy favourite. A 20-seat edge in the projection will be right 91.7% of the time, so the odds that Clark's Liberals will prove the polls (and thus my projection) wrong are not very high. The odds that they can overcome the 6.3-point margin in six days (four remain before the vote, but the last bit of polling was out of the field on E-6) are slightly larger, at 9.5%, but we're still talking about an exceptional case. The NDP remains the easy favourite to win, but we will have to see what Angus-Reid and Ipsos-Reid have to say in the coming hours and days.
I have some mixed feelings about the two polls that were added to the projection, as they are both from firms dipping their toes into the B.C. campaign for the first time (publicly, at least).

We have already heard from Oraclepoll as they were commissioned to do riding surveys for local newspapers in Kamloops and Prince George. But they haven't put out a province-wide survey since November 2011.

They show a closer race than everyone but Forum has indicated, but we should consider that Oraclepoll is the only firm using live-callers to have put out a provincial poll out in this campaign. Perhaps that methodological difference means something.

If we look at their last poll from 2011, we see that it pegged the NDP at 44%, the Liberals at 25%, and the Greens at 16%. By comparison, the polls taken in the two months before and after that Oraclepoll averaged 41% for the NDP, 30% for the Liberals, and 10% for the Greens. If anything, Oracelpoll was showing lower numbers for the Liberals and higher ones for the Greens than other surveys at the time. That does not give us much reason to suggest that Oraclepoll's methodology is conditioned to over-estimate Liberal support, and thus show a narrower gap, but their methods may have changed in the last two years.

The amount of available information about this poll is disappointing, as we don't even have regional breakdowns. But their provincial results aren't out of step with other surveys to any significant degree.

The survey from Hill and Knowlton came out of the woodwork even more. The Victoria Times-Colonist commissioned the poll from Oraclepoll, but this one from H&K was put out in a press release. It was done with H&K's online panel, and did not show any odd results. Aside from a closer race in (small sample) Vancouver Island, their regional breakdowns fell well within the norm.

Though I don't consider Hill and Knowlton to be a polling firm (they do lobbying and public relations), it should come as no surprise that this important company would have its own public opinion research wing to help it advise clients. And considering they are putting their reputation at stake to some degree by releasing these numbers, it seems safe to conclude that they are reasonably confident in them. They have no particular reason not to be.

But as both Hill and Knowlton and Oraclepoll are publishing for the first time in this campaign, it is impossible to really say what these numbers represent in terms of a trend. Is the gap narrowing? Or if these two firms had been in the field earlier in the campaign, would they have shown a similar gap? We don't know, but other polls have been showing the same sort of trends. It should make Tuesday night interesting.