Friday, September 24, 2010

Polling and Reporting -- The Toronto Mayoral Race 2010

In politics, perception is reality ... except when it isn't.

This week two polls were released on the mayoral contest in Toronto.

At the beginning of the week (September 19), the Globe and Mail. reported that Councillor "Rob Ford has taken a commanding 24-point lead" over his nearest rival, former provincial health minister George Smitherman.

This poll, conducted by Nanos for the Globe and Mail, CTV, and CP24 during the dates from September 14 to 16 ranked the candidates as follows:

Rob Ford ... 45.8
George Smitherman ... 21.3
Joe Pantelone ... 16.8
Rocco Rossi ... 9.7
Sarah Thompson ... 6.4

So for so good, right?

At the end of the week (September 24), and Angus Reid poll conducted for the Toronto Star showed Councillor Ford still in the lead, but by a somewhat truncated margin.  The Angus Reid poll was conducted between September 14-15.  This poll ranked the candidates as follows:

Rob Ford ... 39
George Smitherman ... 26
Joe Pantalone ... 13
Sarah Thompson ... 11
Rocco Rossi ... 8

A somewhat different picture, leading the Star's Urban Affairs Bureau Chief to surmise that "the high number of undecideds suggests Ford’s 13-point lead is surmountable."

This afternoon, CFRB host Jim Richards characterized this Angus Reid poll as indicating that over the course of this week Ford's support has "slipped" by 6 points.

Except (pay attention to the bolds above), while these two polls were released five days apart, they were actually conducted on the exact same days last week. If anything , all the differential between these two polls indicates is the statistical margin of error, which tends to be much wider in municipal polling than, say, provincial or federal polling.

Always be suspicious of horse-race reporting during election campaigns.

The Jack of Hearts does the fact checking that people paid to do it don't.

You're welcome.

No comments: