11 Oct 2013

The Chronicle Herald/CRA Poll Nailed It.

In the aftermath of non-forecasted victories in recent BC and Alberta elections, pollsters have been examining methodologies and voter turnout to try to understand if polling can still be counted on today as a true reflection of voting intention. Polling for the Nova Scotia election confirmed that pollsters can still get it right.

On behalf of the Chronicle Herald newspaper, CRA conducted rolling polling during the bulk of the recent Nova Scotia election campaign. Throughout, there was strong and sustained support shown for the Liberal Party. However, towards the end of the campaign, Liberal support began to wane somewhat, with orange and blue support rising.

Results from the final few days of polling, published on the weekend prior to the October 8th election, provided results that were within the margin of error of actual votes cast in the election, marking a quarter century of CRA accurately forecasting the election outcomes (see accompanying graph). Further, predictions of voter turnout were also accurate, with the estimate of the Chronicle Herald poll suggesting that there would be just under 60% expected turnout, while the actual turnout was 59%. In addition, our seat projection estimate was that the Liberals would win around 30 seats (they won 33), PCs would win just under 10 (they won 11), and NDP would win around a dozen (they won 7). The collapse of the NDP in Metro was not predicted by anyone, including the loss of Dexter’s seat. In addition, there were some extremely tight races between the PCs and the NDP which could have easily changed the outcome for both of those parties.

So what have we learned?
Telephone research when done properly can still accurately reflect public opinion. That said, there should be ongoing discussion about the difference between polling the public and polling voters. Those who turn out to vote may not necessarily reflect the inclinations of the overall population. We saw differences in voting intention, for example, among younger Nova Scotians, as well as among those in cellphone-only households, who may not have turned out to vote in the same proportions.

Why did the Chronicle Herald commission a rolling poll?
Rather than having a single snapshot of voting intentions, rolling polling evens out daily variations in such intentions and allows examination of trends. In essence, it gives an ongoing pulse of public opinion throughout the course of the campaign. 

Are published media polls important in election campaigns?
Election polls for media outlets have become the norm over the past half-century. Just as political parties often commission polls to guide and gauge their campaigns, so too are published media polls important information for the public about unfolding electoral events. Survey results inform the public and the parties on current public opinion. As such, polls are an important piece of electoral information, akin to editorials or other news coverage. There’s always a question about whether polls influence voting. There has been no research to demonstrate that it does. If it did, election results would be significantly different than polls published prior to an election.

For further comment, please contact CRA at 1-888-414-1336.

To view full commentary with graphic presentation, please click here.

CRA Corporate Research Associates