A recent opinion survey (April 26-30) showed the NDP outpolling the Tories 33%-30%, yet it was reported widely as a tie.

Canadian Press and Harris Decima did the poll and spun the results.

Doug Anderson, Harris Decima VP for Public Affairs, said in a telephone interview that he saw no significance in not headlining their news release as “NDP leads by 3”.

“You could report it like that and the data would support you,” he said.

The pretext for this spin was that the margin of error was 3.1%, so the NDP could have been anywhere from 30 to 36 per cent support, and the Tories could have 27 to 33 per cent.

Anderson had no explanation for why they reported it like that when the Conservatives won an earlier poll by 3%.

Harris Decima never mentioned “statistical ties” in its news releases on recent polls when the Tories were ahead: December 2011 (“Conservatives Lead by Six”), January 2012 (“Nationally, the “Conservatives lead by 3”), March 2012 when results were 31-28 for the Tories(“Conservatives hold slight lead on NDP”).

It seems more important to get the results “right” rather than to be consistent in how you report them.

Also not mentioned is the number of undecided in the poll.  They report that 1,006 persons voted, but don’t say how many had no preference. 

“We never report the undecided,“ Anderson said.  When pressed, he said it was 17% (more persons than chose the Liberals).

Pollsters and the media were embarassed after the Alberta provincial election. The “experts” ignored and barely mentioned the large undecided response in opinion polls and predicted a Wild Rose majority victory.  The result -- 61 Conservatives seats and 17 for the Wild Rose.