we're dealing with small movements within low-balled margins of error here, but if i'm correct that the tory numbers are being inflated in the media due to unreported high levels of undecideds, then you would expect to see the conservative numbers fall over the next few weeks - and that would not mean that they're losing ground, or that the other parties are gaining, but just that the undecideds are slowly registering properly in the sample.
so, you should indeed expect the race to narrow; you shouldn't interpret that as the race changing.
i'm more interested in the following observations:
1) everybody is starting to run out of time, here.
2) when you see both parties inch up by about the same amount, it suggests a high potential for vote splitting.
as mentioned: ontario actually tends to split rather nicely. don't get complacent, but don't start freaking out yet, either. it's hard for conservatives to win here....
the conservatives dominate the rural areas, and the ndp & liberals have their own regions of strength, depending mostly on class demographics. there's not a lot of battlegrounds. so, the traditional fear of a split vote is actually rather minimal, here - i would suspect there's actually a larger swing between the ndp and conservatives than between the ndp and liberals in much of upper middle class fortress toronto, meaning that if the liberals lose these seats it's not going to be due to the split but due to rich liberals voting for ford, an unlikely scenario.
the situation is rather that ford's unusually strong appeal to recent immigrants is going to pull from both parties in areas where the conservatives aren't usually considered competitive - like here in windsor - thereby creating the appearance of a split. and, this could indeed no doubt be largely resolved by consensus candidates. but, it's not quite the same thing as we usually talk about.
further, the ndp are actually up relative to where they've been over most of the last twenty years. so, i suspect that a closer analysis of the data is going to mostly suggest a collapse in the multicultural component of the liberal vote in these urban ridings, with stasis or even gains for the ndp. so, you need to ask where the liberal vote is really coming from and really going. and, there are a few spots where this could even counter-intuitively help the ndp.
consider an urban riding where the last election had this result:
liberals: 38
ndp: 33
conservatives: 21
greens - 8
an eight point swing from the liberals to the conservatives in this riding, along with a two point swing from the liberals to the ndp, would have the following outcome:
ndp: 35
conservatives: 29
liberals: 28
greens: 8
(and the greens do in fact do that well in many urban ontario ridings at the provincial level)
an eight point swing in an urban space would be a huge accomplishment for the tories in ontario. but it wouldn't be enough. and, this is what the numbers aren't taking into account: the tories win seats in the sticks by huge margins, and are uncompetitive in the cities, so they need to make up a very large amount of ground in order to actually swing seats.
it doesn't seem likely that ford is going to swing 15-20% in urban ridings, especially not with goofy candidates, and he'd have to do it to win.
i would like to see more targetted data on exactly where these swings are happening; i'm assuming that the swing is mostly happening in urban ridings, and amongst recent immigrants, but this is a deduction based on what we know about the candidates, and not hard empirical data. if the swing is actually happening in more affluent neighbourhoods, then this could be a harder election. but, the liberals are apparently still ahead in toronto.
i'm still predicting a coalition government.
but i'm still pointing out that the data isn't so great.