The tricky art of polling

Political scientist says polls are but ‘single frames in an epic film.’


(Edmonton) Recent polls by ThinkHQ Public Affairs and Mainstreet Technologies have produced some surprising results regarding the upcoming provincial election. In both polls the numbers suggest that most Albertans would vote for the Wildrose party, and the bulk of the remaining votes are split evenly between the Progressive Conservatives and the New Democratic Party.

Could Alberta blue be fading?

“What has been revealed in these surveys is that the NDP and Wildrose have a singular shared virtue—they are not the government,” says political science professor Jim Lightbody.

In the April 9 ThinkHQ poll, the vote is nearly evenly split three ways, with the Wildrose in the lead at 31 per cent and the NDP and PC parties statistically tied at 26 and 25 per cent respectively. Meanwhile, the Mainstreet Technologies poll had a similar result with the Wildrose at 24 per cent, PCs and NDP both at 20 per cent provincially and the NDP at 40 per cent in Edmonton.

Whether this news is good or bad depends upon which side of the political spectrum you fall, and how much faith one puts in polls, says Lightbody, as they come with a cautionary tale.

With less than a week of campaigning left before the 2012 Alberta general election, polls showed the Wildrose Party had a wide, if not insurmountable, lead over Alison Redford’s Progressive Conservatives.

The night of the 2012 election, on the eve of what most Albertans thought would be a Wildrose take-over, Lightbody was on the anchor-desk panel at Global Calgary.

“The polls that were released to the public were suggesting the end of the PC era, but polling for the media had stopped Thursday before the Monday vote to meet publication deadlines,” he said. “We had lots of time to chat between segments, and the PC insiders in that booth weren’t concerned. They hadn’t stopped their own polling and they knew that they were going to be just fine.”

By the time the dust had settled that night, the Conservatives had reversed their fortunes and walked away with a comfortable 44 per cent of the vote and yet another majority government.

“Polls are accurate if they are well done,” says Lightbody, adding the more pertinent question may be how long are polls accurate over the course of a 28-day campaign? “And most polls are going to be well done, but they are just snapshots in a moving picture—single frames in an epic film.”

They are also quite nuanced. Lightbody argues, for instance, that while people generally want to tell the truth to pollsters, they also want to be liked and so may say they intend to vote when in fact they simply don’t want to be seen as shirking their civic duty.

“When only 44 per cent of the population comes out to vote, it means that despite what they said in the poll, more than half would rather go curling or watch darts on the BBC than vote.”

Lightbody says Psephologists (a sub-set of political scientists who study elections) call this the Bradley Effect or the Shy Tory Factor, where survey respondents provide false answers because they feel pressured to provide an answer that is 'politically correct.'

He says it’s hard to put much stock in early polls but adds the longer a single government is in power, the more grievances will accumulate.

“It seems that after 44 years of shopping at Safeway, Albertans have noticed that there are other grocery stores in town.”