08 Jun Angus Reid’s regional breakdowns

On Friday, contrarian spoke with Jaideep Mukerji, vice president of public affairs for Angus Reid Strategies, about the unpublished regional breakdowns in the Nova Scotia election poll Reid conducted last week for CTV-Atlantic. I’ve been reluctant to post the results for two reasons:

• The regional breakdowns involve smaller subsets of the overall 800-person poll, so the sample sizes are quite small. This means the statistical sampling error, the amount by which the samples can be expected to vary from the populations being polled, is quite large. Sampling error, usually expressed as, “plus or minus X percentage points 19 times out of 20,” is strictly a function[*] of sample size.
• Reid conducts its polls online, a method that seems to invite bias by eliminating non-Internet users. The organization vigorously defends its methodology, claiming to have produced the most accurate prediction of last fall’s federal election. Reid doesn’t simply invite people to respond to a website; it assembles a large pool of interested would-be participants, and then uses demographic information to construct a representative sample within that pool.

The table above shows two regional breakdowns with sample sizes big enough to produce results accurate within six percentage points.[**] The largest is the most striking, encompassing all rural mainland ridings except Guysborough. It shows the NDP with a lead over each of the other two parties of somewhere between 14 and 24 percent.

If true, it shows that Darrell Dexter’s persistent efforts to cultivate support in rural Nova Scotia have paid off. A ~19 point lead over the other two parties in that region is the sort of result that could propel the New Democrats to a majority victory, which Reid predicts.

Underscoring this conclusion, Mukerji says, is Dexter’s personal popularity.

Two-thirds of voters thinks Dexter understands the needs of Nova Scotians. This number usually tracks pretty close to party support, but in this case, only 47 percent plan to vote NDP. So you’ve got people who aren’t going to vote for him who still think he understands their needs. He’s punching above his weight.

Contrarian may have qualms about Dexter’s leadership, but Nova Scotians appear not to. Significant majorities believe he is trustworthy (56%), has a vision for Nova Scotia (55%) and would make a strong and decisive leader (50%).

Find Reid’s full report, with tables and methodology but minus the regional breakdowns shown above, here.

[*] Margin of error = plus or minus 98 divided by the square root of the sample size, 19 times out of 20.

[**] At just 113 voters, Reid’s sample size for Rural Cape Breton and Guysborough was simply too small to produce meaningful results.