Questionable Poll Reporting from the Globe

I’ve been following the polls pretty carefully in the lead-up to the October election and a story caught my eye yesterday. The Globe headline read “Parties tangled in three-way tie as Duffy drags on Tories: poll”. I’ll forgo the discussion about covering the horse race over the issues (after all, that’s what I’m writing about as well); if the Globe is going to cover the horse race, it would be nice if they were at least precise about it.

The poll in question is the Nanos tracking poll, which, if I understand it correctly, surveys about 250 people each week and is reported as a four week average of a total of 1000 people. It is useful, then, largely as an indicator of longer-term trends, than as a marker of where the electorate is now, because it includes opinions from a month ago.

This is the trend we see if we look at the Nanos Poll alone. But, if we take Eric Grenier’s approach, a very different picture emerges. Poll Tracker includes the results from Nanos, but weights more recent polls more heavily. So, three polls that began surveys after August 14th show the NDP leading with between 34% and 37% of the electorate’s support, the CPC at 29%-30%, and the Liberals with between 24%-28%, and the Greens between 4% and 6%. The general trend of these polls is slightly up for the NDP, which is reflected in Poll Tracker‘s projections and trends.

PollTracker‘s model shows the NDP at 35.4%, the CPC at 29.8%, the Liberals at 24.7%, and the Greens at 4.3%, which seems to reflect the recent polling results and trends. Which brings me to the problem with the Globe’s reporting. In proclaiming a “three-way tie”, they’ve neglected to mention that this is an AVERAGE over the last month. It’s a very selective reading of a single poll as opposed to an analysis of the status of the campaign as a whole.

The author, Chris Hannay, fails to convey the context to the reader, which, as a reporter, is his job. He even fails to mention that the poll was conducted over the span of a month, saying instead that “the most recent interviews took place on August 21st”, implying that the poll is far more current than it actually is. It is shoddy reporting like this that allows voters to find the information they want instead of the information that is accurate. I expect better from you, Globe.