Seat Predictions Ontario 2018 Election

Ciabatta2 wrote:

Sean in Ottawa wrote:

General comment on efficiency -- all parties will necessarily have various bands of support that are efficient depending on their level of concentration. So a party that is highly concentrated in strongholds would be remarkably efficient at a low level

...

I do not see last minute guesses as cheating. I see earlier guesses without data as baloney.

Great post.  The bit at the start was what I was trying to get at - that I believe that in low levels of support the Liberlas have an efficiency band due to their concentration in Ott/TO.  Yes you can't be efficient at every level, I thought that was just intuitive, but to be fair my posts never state that so fair point in bringing that up.  If I wasn't in my own head, I would have read them the exact same as you did.

I think the last bit there is the only thing we really disagree on (aside from the seat numbers and projections, of course.)  If there is no point looking at current data and postulating into the future, then there is no point in having threads and websites and commentators open for predictions until a day or two before the election, really - it's just a never ending "if the election was tomorrow..." feedback loop.  What matters isn't a fake tomorrow - it's what happens between now and the end date.  But that's just a bug bear of mine, I guess and fair game to anyone who disagrees, since I seem to be in the minority on that.

Definitely, though, its getting late for any Liberal rise.  The post debate hasn't resulted in a bump.  So I'm probablywelllikely going to be wrong!

Thanks for the great chats on this.

Thanks -- my initial point though was the I don't think the Liberals are very highly concentrated and that is why they would be expected to get fewer seats at a low level than the NDP which is more concentrated. The Liberals are a little more spread out so they get a band of efficiency at a higher level than the NDP. This is a problem for the NDP because there is a mid range inefficient place for them but these recent polls suggest they are above that.

So While the NDP is efficient at 20% they are not very at 30% but they are back to efficient by high thirties. The Liberals are inefficient at 20 but become efficient into the low 30s where the NDP is not efficient -- etc.etc.

Now I also have said I think that the Liberals party are probably, while not very efficient at 20% a bit more than what polling models suggest and this is why I would expect them to get a few rather than 1-2. But if they test the bottom of that range only 15% then they could get nothing when the NDP in the same range is likely to get 6-8 seats.

So at any given support level some parties will be more or less efficient. But the argument being made by pollsters is that the NDP are less efficient at all levels which is nonsensical as it is mathematically impossible to be one or the other at all support levels.

On the predictions -- I think some poeple may want to predict these margins in a volatile races but there is no data to be as precise as people want to be so these are just guesses. When you get to the last few days these guesses may have some data attached. But for now the only science would be in saying if the support level is really x the seats are likely to be y. When both are variables it is all pretend and hoping for luck.

My guess, from Edmonton:

PC - 68, 38%

NDP, 49, 36%

Liberal, 6, 18%

Green, 1, 5%

Others, 0, 3%

NorthReport wrote:

http://www.tooclosetocall.ca/2018/06/last-day-of-campaign.html?m=1

If you are gloing to cite "TooCloseToCall" then perhaps you have answered your own question in another post as to why would a Liberal supporter (like me) vote Liberal today (or in my case, yesterday, by Special Ballot).  It is because these are all riding-by-riding fights and the province-wide numbers decide nothing.

There are, according to the TCTC model, 21 ridings in which the Liberals either lead or are closer, and in some cases far closer, to the PCs support totals than are the NDP.  Many are lost causes, with inevitable PC wins but it doesn't make much sense for Liberals to switch their votes to NDP candidates who are even further back in these 21 ridings.  If there were some massive and unpredictable last-minute surge underway for the NDP, perhaps - but not otherwise.

Then there are a further 37 or so ridings that are certain victories for the NDP given the margins by which the NDP leads, so Liberals can vote their party affiliation without risk of electing a PC MPP.

There are a further 12, mostly impending PC wins, in which it is not remotely clear cut which is the party most likely to challenge the PCs (i.e., where Liberal and NDP support are separated by less than 2% points).  In some of them, the Liberals are incumbents, with decent ground campaigns.  In many cases, the Liberals and NDP are so far back that it is almost irrelevant, so given the per-voter subsidy, it makes sense for Liberals to vote Liberal.  

Then that are 16 that are so Blue that there is little point in even thinking about them, as the combined Liberal and NDP vote falls short of PC support. And again, there's the per-voter subsidy.

That accounts for 86 ridings.  In the other 38, it makes sense for Liberals to vote NDP.  If enough Liberals did so, there would be an NDP majority.

NorthReport wrote:

When the results come in tonite just remember to thank yourself for elected Doug Ford Premier.

It is patronizing to say that "Math is Hard" and I won't, because basic arithmetic is actually pretty easy.

Con majority...Then the clown car show begins....

PC - 71 38%

NDP 47 35%

Liberal 7 19%

Green, 1 4%

I hope I am wrong

josh wrote:

The hell with it:

NDP 63 41%

Cons  60 38%

Liberals 1  17%

I'd like to believe that....

Sean in Ottawa wrote:

PC - 71 38%

NDP 47 35%

Liberal 7 19%

Green, 1 4%

I hope I am wrong

I hoped so too. But you were eerily close. And your Liberal prediction was spot on. How did you do that??