22 October 2013

Commentary – After the election #nlpoli nspoli

Following is a commentary by Don Mills of Corporate Research Associates in response to the post “CRA, Abacus, and the 2013 Nova Scotia General Election”. – EGH

Don Mills’ commentary is available two ways:  via Scribd and underneath the Scribd insert, as a post within SRBP.  The Scribd version is Mills’ original submission complete with the tables as originally submitted.


Following on from some recent commentary regarding public opinion polling and election forecasting on a Newfoundland and Labrador blog (http://bondpapers.blogspot.ca/2013/10/cra-abacus-and-2013-nova-scotia-general.html) Corporate Research Associates Inc. (CRA) is pleased to respond to a number of key points. Mr. Hollett, on his blog, raises important areas for discussion, and also points to a number of “problems” with CRA’s performance, methodologies and public disclosure of information. What follows is some more detailed information on the recent Chronicle Herald / CRA poll, as well as on our unblemished 20+ year track record.

The blogger first points to three key initial problems with our recent research results:

Problem number 1: this commentator stated that anyone could have called the Liberal majority win in advance of the actual election. So, could a dead parrot have called the Nova Scotia election? Maybe, but the reason someone could have forecast a Liberal win was because CRA has been publishing the results of its quarterly polling results (from the Atlantic Quarterly Survey), as it has for the last 20+ years, and those results clearly pointed to a swell of support for the Liberal party in the last year. Without that pulse of Nova Scotia public opinion, commentators may have known that Nova Scotians were ready for change, but would have had no idea where support stood, or the degree of backing for the Liberals.

Problem number 2: Despite serious issues seen in polling in advance of recent elections in BC and Alberta, telephone polling of prospective voters can be a useful and accurate way to gauge public opinion and voting intention. As some commentators have noted, not only did the Chronicle Herald/CRA’s poll results accurately show how voters would cast their ballots, Abacus Data survey results released close to the election were on the mark as well. Having multiple firms post results that are accurate strengthens trust in the industry and in the methodology of telephone research.

Problem number 3: Comments have indicated that the Chronicle Herald/CRA did not release enough of its data to show that the polls “could explain what was going on in the campaign”. In fact, there were public releases and discussion of regional differences, and the increase in support for the Liberal party in Metro in particular. Although extremely tight races such as Mr. Dexter’s 21-vote loss would have been difficult to predict based on polling, as shown in the following quote in the Chronicle Herald (September 20), support in Metro was clearly turning red:
“But recent polls from Corporate Research Associates have shown the New Democrats trailing the Liberals in Halifax. The two parties have essentially flipped positions since 2009, said Corporate Research vice-president Margaret Chapman.” http://thechronicleherald.ca/novascotia/1155916-grits-see-metro-breakthrough
Moving on to respond to other comments, the blogger said that the Chronicle Herald / CRA did not release seat predictions. This isn’t the case. On October 7, CRA published general seat predictions on its website: http://cra.ca/nova-scotia-2013-provincial-election-polling-commentary/

Further, the blogger said that CRA “relies on luck” in its research to understand voting intention, and that we “don’t poll for voters”. This is far from the truth. The Chronicle Herald / CRA poll asked a battery of questions of individuals to determine their voting intention as well as their likelihood of turning up on Election Day. Based on these results, the Chronicle Herald / CRA results accurately estimated voter turnout at just under 60%. Further, CRA’s analysis of party support takes into account likelihood of voting. That’s how our polling has been accurate for over 20 years.

He has also claimed that in recent elections, including the most recent in Newfoundland and Labrador, that “CRA just didn’t come close at all.” That’s not true. CRA’s results are shown below and compare actual election results to published CRA polls (both from the Atlantic Quarterly survey, as well as from a commissioned poll by the Telegram), which clearly show that CRA numbers for decided, or decided and leaning voters are within a maximum of 3 percentage points for each party:

Newfoundland and Labrador Election
Oct 11, 2011
NDP Liberal PC
Actual results 25% 19% 56%
AQ (Q3) 24% 22% 54% Aug 15-Aug 31 Decided
Telegram 26% 17% 57% Sept 29-Oct 3 Decided / Leaning

What’s more, CRA polling has been accurate for every election within the last 20+ years. Here’s some backup, showing published results from both CRA’s nearest Atlantic Quarterly survey to each election, as well as commissioned results for various news outlets for each of the last few elections in Atlantic Canada:

Nova Scotia Election
June 9, 2009
NDP Liberal PC
Actual results 45% 27% 25%
AQ (Q2) 37% 31% 28% May 7-May16 Decided/Leaning
CBC 44% 28% 26% May 18-May30 Decided/Leaning

Prince Edward Island Election
October 3, 2011
NDP Liberal PC
Actual results 3% 51% 40%
AQ (Q3) 7% 59% 31% Aug 10-Aug 28 Decided
Guardian 5% 53% 36% Sept 23 - Sep 27 Decided/Leaning
New Brunswick Election
Sept 27, 2010
NDP Liberal PC
Actual results 10% 34% 48%
AQ (Q3) 16% 41% 36% August 4 - Aug 24 Decided/Leaning
Telegraph Journal 11% 36% 46% Sept 13-Sept 19 Decided

These results show decided or decided and leaning voting intentions, and we believe that this information, coupled with predictions for voter turnout, is a better measure of election outcomes than looking at the public in general. It’s voters who determine the outcome of an election, after all.
The commentator claimed that research firms themselves don’t follow the standards set by the MRIA on disclosure of methodology to news media. Not true. CRA’s poll for the Chronicle Herald outlined sample size, margin of error, questions asked as well as polling dates, on a daily basis.

Finally, accusations that CRA only issues information on the ‘horse race’ and nothing else are unfounded. CRA conducts substantial public opinion polling on a wide variety of topical issues, and releases this information every quarter. The recent Chronicle Herald poll delved into most important issues facing Nova Scotians and which leader was best able to handle various crucial issues. These results were published in the Chronicle Herald on September 19
[http://thechronicleherald.ca/novascotia/1154968-herald-poll-economy-tops-voters-agenda] and were widely discussed in various forums.