Wednesday, September 14, 2011

Two Polls, Two Stories

Let's start with the good news: a Forum Research poll shows the PCs plummeting to John Tory lows in Toronto, sitting at just 24% of decided voters. McGuinty's Liberals sit with 39%, and Andrea Horwath's NDP sit with 30%.

Overall, that would lead to a 17-6 split in the city between the Liberals and NDP, with Davenport and York South-Weston being turned orange. All other 21 ridings remain red, though certain ridings like York West, Scarborough-Guildwood, and Toronto Centre, feature races which are getting mighty close.

The bad news is that Abacus data shows the PCs leading against in Ontario with 41%, compared to 32% for the Liberals, 20% for the NDP, and 6% for the Greens.

These are plausible numbers, to be sure. It would give the PCs a comfortable majority of 62 seats to 27 seats for the Liberals and 18 seats for the NDP. So, at first glance, this poll doesn't seem that crazy.

Then you have to get into the details. Abacus' methodology is very different. Instead of asking the traditional "which party will you vote for: a; b; c; d;", they're instead asking voters to rate on a scale of 1 to 10 their likelihood to vote for a party.

To make this clearer: 0-4 means you're unlikely to vote for a party, 5-7 means you could be swayed, and 7-10 means you're likely to vote for a party.

Now, this is a very interesting way to do a poll, because it gives you the likelihood of getting certain voters. For example, it says that 50% of voters won't consider voting for the Tories. 53% won't consider voting for the Liberals, 55% won't consider voting NDP, while a stunning 74% won't consider voting Green (!).

Again, this is all useful information to gauge which parties can expect to grow, where their votes may come from, and so on, as it includes this data and other data about undecideds and etc. But where in the heck do they find out their voting intentions??

Think of it this way: if I said to Abacus that, on the Ontario PCs, I was, let's say, a 7 in my likelihood to vote for them. Does this mean I'm going to vote for them? Sure, it may mean I'm leaning, but it doesn't actually ask me what my current voting intention is. Maybe I'd like to vote PC, but I have to vote Liberal because of whatever issue (strategic vote!).

So, how does Abacus write me down as? Am I now a PC voter, or am I just a leaner? On that chart, I'm "likely" - so what does it mean? It's such an ambiguous scale that I can't really figure out how they get their voting intentions from.

So, unlike Eric over at, I'm not including this poll in my update. Not until I see more polls of their kind, or unless I see it following a trend (so far, it's bucking it). Maybe this will be a fantastic new way to poll, or maybe it'll be a dud. Right now, it's too early to say.


  1. The Abacus poll was conducted for the Toronto Sun, which might as well be the propaganda arm of the Conservatives. (Not complaining, same with the Star and the Libs). I'll believe it when another pollster confirms the result.

  2. True Dan, true.

    Or should I say "anonymous," from Eric's site!