they're distributing again, which is an error. you need to include undecideds.

they're not even measured. fail.

the best thing you have here is this:

undecided - 37%
brown (pc) - 27%
horwath (ndp) - 21%
wynne (lib) - 15%

if you reduce the sample, you end up with:

brown (pc) - 43%
horwath (ndp) - 33%
wynne (lib) - 24%

that is actually roughly consistent with the published results, which have brown at 40% and the other two at 27% and 25%. you'll note that the greens are at the difference, at 6%. but, that's a coincidence.

the way you should read this poll is actually like this:

conservatives - 30%
not conservatives - 70%

that is, brown has his diehard support, and everybody else is disgusted by the spectrum, but knows they don't want conservatives. that 37% is mostly disappointed liberal voters, who could swing anyway except right. and, that's pretty much exactly what we saw in the last federal election around this time. the ndp were able to briefly jump ahead, but in the end lost because they moved right in an attempt to win the centre, which backfired.

this is why it's an error to distribute.

this is a different situation, because the liberals are incumbents. if there's a change vote coming up, the ndp should swing most of it.

yes, we should wait for the second-choice data to be sure. but, what you'll find is the same thing you saw two years ago, because nothing has changed since...

wynne doesn't have a lot of credibility in moving left, right now. but, horwath should come out and start ranting like a raving socialist - because it's what the province wants, and is fed up that it can't find anywhere.

http://poll.forumresearch.com/data/fae42e93-d6a3-47b5-8395-c451571b0b0aFOLJ%20ON%20Horserace%20Aug%2025%20.pdf