Ontario 2014 (June 12th)

[Linus Van Pelt]

What are you all actually doing, mathematically, to get these projections?

I use a ratio projection on a regional basis, but altering results in ridings that have had by-elections (taking into account polling averages around the by-elections). Unlike Brendan, I will also be doing some riding level tweaks to get more believable results. Tweaks will be based on math as well though.

But what is the actual formula?

Or are you deliberately keeping the details secret, so that it can't be replicated, like a for-profit pollster?

[Hatman 🍁]

What are you all actually doing, mathematically, to get these projections?

I use a ratio projection on a regional basis, but altering results in ridings that have had by-elections (taking into account polling averages around the by-elections). Unlike Brendan, I will also be doing some riding level tweaks to get more believable results. Tweaks will be based on math as well though.

But what is the actual formula?

Or are you deliberately keeping the details secret, so that it can't be replicated, like a for-profit pollster?

No, I should be transparent for conflict of interest reasons (as I may or may not have access to internal polls that I cannot report on in my model)

In the ratio model, if a party doubles their support in a region, then they double their support in each of the ridings. It definitely has its flaws (which is why the Greens can often be shown to "win" seats), but I have seen it used in the polling industry (not by Forum Research obviously). I first started using it when I was running a Canadian election game in the Individual politics sub forum here.

[Linus Van Pelt]

I see; thanks.

[adma]

The PC's "base" in TJB is Timmins. But it's not much of a base for the party in the least. They'd have more luck trying to get Kenora.

I thought of writing that, but it's not really a PC base. It's less NDP-friendly than the rest of the riding, but not a PC base.

Actually, in TJB last time, the PC "base" happened to be Kapuskasing, because Kap Mayor Al Spacek was their candidate.  (Which what I mean in emphasizing that 2011 was an exceptional circumstance.)

[DL]

I think I have figured out the secret to Forum's so-called seat projection model. They apply a uniform swing across the province to the 2011 election results - then they arbitrarily subtract 10 seats from the NDP and give them all to the Liberals for no other reason than that the principals at Forum are Liberal supporters and that is what they like to believe will happen. Its highly unprofessional and misleading - but if you are them - who cares?

Here is an excellent blog posting that shreds the ridiculous Forum seat projection model

http://tcnorris.blogspot.ca/2014/05/turning-poll-results-into-estimated.html

"Forum would have us believe that almost identical polls taken less than four months apart would produce two quite different outcomes for the Liberals and PCs even though their vote shares are identical. It simply isn't believable. A second and in my mind more serious issue is that even though the NDP in these polls has a vote share less than 1% different from its 2011 vote percentage in both cases, the seat projection suggests the NDP would lose 8 seats from its 2011 total. This would occur despite the fact that most of its competitive races are run against the Liberals, who are down significantly from their 36.5% vote share in 2011 in both Forum surveys. Again it isn't believable. Newspapers use Forum and that dominates secondary and social media coverage, but my recommendation would be to pay attention to the seat forecasts of 308 or LISPOP or any of the others before taking Forum seriously. For what it is worth my own seat projection using Forum's latest numbers is PC 49 seats, Liberal 36 seats, NDP 22 seats (a majority is 54 seats). Almost never would a party trailing by five points finish ahead of its rival in an election. It is true that the Liberal vote is more efficient as Forum says, but there limits to vote efficiency."

[Hatman 🍁]

Yes, even 308 will have better numbers than Forum

[RogueBeaver]

Don't worry, they're now onto noting whether any of Hudak's photo-op locations have ever received a nickel in government subsidies. Total game changer. That said, the PC tour folks are doing it on purpose for some reason. My advice would be avoid any negative shinies, but what do I know about tour planning?

[DL]

How do you like this? Hudak's ad on job creation uses stock footage of people at work - in RUSSIA!

http://blogs.ottawacitizen.com/2014/05/06/hudaks-ontario-job-creation-ad-uses-video-clips-outsourced-from-russian-federation/

[RogueBeaver]

Shiny (and I'd say the same about Wynne spandex or Horwath hard hat) That said, better research so he doesn't get nitpicked to political death.

[RogueBeaver]

Pension FAQ.

Wynne: Only Grits can be trusted w/transit.

[DC Al Fine]

Pension FAQ.

I like the idea behind the Liberals's plan, but the numbers are sketchy. Take the $45k example. How do they think they can fund 60% of a CPP payout on 40% of the CPP contribution rate?

[RogueBeaver]

Horwath: raise the minimum wage to $12 while cutting small business tax.

[Hatman 🍁]

My Week #1 projection (same numbers as I reported on yesterday):

http://canadianelectionatlas.blogspot.ca/2014/05/2014-ontario-election-projection-week-1.html

[DL]

Hatman, i have to take issue with one thing in your projection. you say at the end

"Pumping the regional numbers from both polls into my new Ontario election model, I get a very narrow PC lead, with the Tories at 46 seats, the Liberals at 43 and the NDP at 18. This would result in a minority Progressive Conservative government in the 107 seat legislature."

Its correct to say this would result in  PC "plurality" in the 107 seat legislature...who would actually form a GOVERNMENT if that was the seat distribution is a whole other question....In fact, i suspect that if the PCs were only a couple of seats ahead of the Liberals there is a strong likelihood that the Liberals would form another minority government with a more explicit accord with the NDP.

Anyways, it just irritates me when people project that such and such a party is projected to form a "minority government". - we don't elect a government, we elect a legislature and which party forms a government is not always the party with the most seats. in 1985 Ontario elected 51 PCs, 49 Liberals and 25 NDPers - the results was a Liberal minority government.

[Hatman 🍁]

You're right DL, Wilf made the same comment on my blog. I will call it plurality next time.

[Krago]

Hatman, i have to take issue with one thing in your projection. you say at the end

"Pumping the regional numbers from both polls into my new Ontario election model, I get a very narrow PC lead, with the Tories at 46 seats, the Liberals at 43 and the NDP at 18. This would result in a minority Progressive Conservative government in the 107 seat legislature."

Its correct to say this would result in  PC "plurality" in the 107 seat legislature...who would actually form a GOVERNMENT if that was the seat distribution is a whole other question....In fact, i suspect that if the PCs were only a couple of seats ahead of the Liberals there is a strong likelihood that the Liberals would form another minority government with a more explicit accord with the NDP.

Anyways, it just irritates me when people project that such and such a party is projected to form a "minority government". - we don't elect a government, we elect a legislature and which party forms a government is not always the party with the most seats. in 1985 Ontario elected 51 PCs, 49 Liberals and 25 NDPers - the results was a Liberal minority government.

In 1985 Ontario elected 52 PCs, 48 Liberals and 25 NDPers - the result was a PC minority government that lasted approx. 8 weeks.

[DL]

In 1985 Ontario elected 52 PCs, 48 Liberals and 25 NDPers - the result was a PC minority government that lasted approx. 8 weeks.

True, but keep in mind that in 1985 the PCs were the incumbent party therefore Frank Miller was premier until the Liberal/NDP accord pushed him out - but he never passed a throne speech or established confidence so there never really was a PC minority government - there was just an interregnum that in effect was a continuation of the pre-election PC majority government.

[mileslunn]

My understanding is while the governing party gets first crack at forming, usually its customary if the government wins fewer seats than another party for the leader to visit the lieutenant governor and resign.  For example in both 1979 and 2006 federally, the Liberals in theory would have been given first go at trying to form a government, but in neither case it happened as Trudeau in 1979 and Martin in 2006 submitted their resignations to the GG and thus opening the way for Joe Clark in 1979 and Stephen Harper in 2006 to form government.  Now to be fair there are some differences as in 1979, the PCs were only six seats shy of a majority and there were six Socred MPs who likely wouldn't have supported the Liberals.  In 2006, The BQ likely wouldn't have backed the Liberals, whereas in the next election its probably fair to say the NDP would prefer a Liberal government over PC.  The only barrier is due to how long the Liberals have been in power and the amount of baggage they've taken it might just set up the conditions to ensure the PCs win a majority next time around while if they pass the throne speech and let the PCs introduce a budget which will include some unpopular spending cuts there will be less of a backlash or they could go to another election in which things would be more favourable for them.

[mileslunn]

Pension FAQ.

I like the idea behind the Liberals's plan, but the numbers are sketchy. Take the $45k example. How do they think they can fund 60% of a CPP payout on 40% of the CPP contribution rate?

Definitely could be a vote winner, but could also hurt them.  Most people like the idea of more secure pensions, but if it means having over $1,000 docked from their paycheque it might be less popular.  Add to the fact many small businesses may lay off or close shop and so depending on how much noise small businesses make this could have an impact.  Finally those most likely to support this are over 50 (who generally vote) and thus are close to retirement and concerned about lack of savings.  The problem is they would just pay more with little benefits as payouts are based on contributions so they wouldn't have paid in enough to get the full benefits.  Never mind because its provincial only there is the issue of those who only live part of their working life in Ontario.  The one's who will benefit most from this are those in their 20s and 30s, but that group has a low turnout as well as few people at that age are thinking much about retirement.  So it will be interesting.

As a disclosure, it won't affect me since I work in the banking industry which is under federal regulations as well as I have a workplace pension so I won't have to pay into it.  I was though worried if it impacted my company that there would be layoffs or annual salary increases would be put on hold for a while.

[RogueBeaver]

In practice I'm quite sure whoever wins a seat plurality will form government.

[Linus Van Pelt]

What are you all actually doing, mathematically, to get these projections?

I use a ratio projection on a regional basis, but altering results in ridings that have had by-elections (taking into account polling averages around the by-elections). Unlike Brendan, I will also be doing some riding level tweaks to get more believable results. Tweaks will be based on math as well though.

But what is the actual formula?

Or are you deliberately keeping the details secret, so that it can't be replicated, like a for-profit pollster?

No, I should be transparent for conflict of interest reasons (as I may or may not have access to internal polls that I cannot report on in my model)

In the ratio model, if a party doubles their support in a region, then they double their support in each of the ridings. It definitely has its flaws (which is why the Greens can often be shown to "win" seats), but I have seen it used in the polling industry (not by Forum Research obviously). I first started using it when I was running a Canadian election game in the Individual politics sub forum here.

Wait, sorry: upon further reflection, I actually still don't get it.

Suppose a party had 20% at the last election in a region and 60% in one riding in the region. Then according to the bold, if the party's support doubles to 40% in the region, they double in the riding to 120%? That can't be right.

[Citizen Hats]

you can recount the seat totals, a if you're goal is to project winners, they are likely to win

[Hatman 🍁]

What are you all actually doing, mathematically, to get these projections?

I use a ratio projection on a regional basis, but altering results in ridings that have had by-elections (taking into account polling averages around the by-elections). Unlike Brendan, I will also be doing some riding level tweaks to get more believable results. Tweaks will be based on math as well though.

But what is the actual formula?

Or are you deliberately keeping the details secret, so that it can't be replicated, like a for-profit pollster?

No, I should be transparent for conflict of interest reasons (as I may or may not have access to internal polls that I cannot report on in my model)

In the ratio model, if a party doubles their support in a region, then they double their support in each of the ridings. It definitely has its flaws (which is why the Greens can often be shown to "win" seats), but I have seen it used in the polling industry (not by Forum Research obviously). I first started using it when I was running a Canadian election game in the Individual politics sub forum here.

Wait, sorry: upon further reflection, I actually still don't get it.

Suppose a party had 20% at the last election in a region and 60% in one riding in the region. Then according to the bold, if the party's support doubles to 40% in the region, they double in the riding to 120%? That can't be right.

You're right, it isn't quite like that. Their raw votes would double, so their percentage would be compared to the raw votes for all the parties in that riding. This would heavily skew things, as all of a sudden turnout in that riding would be huge (though, the end result could look realistic).

[RogueBeaver]

From the LG's Twitter feed: what a writ looks like. 214 of them.

[Citizen Hats]

From the LG's Twitter feed: what a writ looks like. 214 of them.

Why 214 of them?