Monday, May 31, 2010

Angus Reid: Tories Lead by 8

The most recent Angus Reid poll confirms the roughly 8-point Conservative lead shown by EKOS and Harris-Decima. However, in Angus Reid's case, this doesn't constitute movement toward the Tories, as their last poll from 4 weeks ago had the Conservatives up by 7. The Liberals can be slightly reassured by their 8-point lead in Atlantic Canada, and by the effective tie in Ontario. The bad spot for them in this poll is a dismal 16% in BC. As usual, the NDP fares well in this Angus Reid poll.

Adding this poll gives:

CON - 135
LIB - 80
BQ - 53
NDP - 40

The average Tory national lead is 8.0%.

Thursday, May 27, 2010

Harris-Decima: Conservatives Lead by 9

In its most recent poll, Harris-Decima has confirmed the increase in the Tory-Grit gap observed by EKOS. However, Harris-Decima numbers suggest that this is due to a Conservative rise, rather than a Liberal drop as EKOS showed.

This poll is all doom and gloom for the Liberals, showing its electoral prospects deteriorating in all parts of the country, except for Québec, where it only trails the Bloc by 9. The Tories will be pleased with their 7-point lead in Atlantic Canada, 5-point lead in Ontario and 12-point lead over the NDP in BC.

Adding this poll to the mix finally creates some significant movement:

CON - 137
LIB - 79
BQ - 53
NDP - 39

This is the first time since March, when I suggested that things would remain flat until May, that the Tories have been outside a very narrow 130-to-134-seat range. The pattern of having 2-3 months between each significant poll movement continues! This projection is the best for the Tories since the first one of 2010, and the worst for the Liberals since the last one of 2009.

The average Tory national lead is now a very comfortable 7.9%.

EKOS: Conservative Lead at 8.2

This week's EKOS poll yielded results that broadly resembled last week's, save for the Liberal lead in the Atlantic this time. Thus, the 7-to-8-point Tory lead in Ontario might be real, which is very worrisome for the Grits.

The aggregate projection is back where it was for most of the month:

CON - 133
LIB - 82
BQ - 53
NDP - 40

The Conservative average national lead is still 6.6%.

Thursday, May 20, 2010

EKOS: Tories Lead by 9.3

Due to its large sample sizes, EKOS usually gives us numbers that closely track moving averages. This week's poll doesn't, and shows a large national Conservative lead, not so much due to a rise in Tory support as a large drop in Grit (and NDP) numbers: time for the Liberals to worry?

The most notable result is in Ontario, where the Tories lead by 7.1%, which is 6% more than last week and also more than during the 2008 election. The Tories also have a nearly 5% lead in the Atlantic. With respect to last week's EKOS poll, Conservative fortunes have also risen in AB and BC; they have dropped slightly in MB/SK. The dark spot for the Tories is a sub-15% performance in Québec. All in all, this poll would place the Tories near where they are now in the seat count, despite a national lead 2% lower than in 2008.

Meanwhile, Liberal and NDP numbers dropped in every region relative to last week (except the Liberals in AB). Other than the Grits potentially becoming competitive in Winnipeg due to the Tory and NDP drops in MB/SK, there is nothing positive for these two parties in this poll.

Weirdly enough, adding this poll to the mix increases the Liberal seat count:

CON - 133
LIB - 83
BQ - 53
NDP - 39

However, the Liberals are within 1% from losing a seat in each of 5 provinces to the Tories (NL, PE, NB, ON, MB) and also a seat to the Bloc, while they are more than 2% away from gaining any seat. Thus, 83 probably somewhat overstates their expected seat count, and 133 likely understates the Tory one.

The Conservative average national lead is now 6.6%.

Tuesday, May 18, 2010

Harris-Decima: Tories Lead by 4

Harris-Decima has released its latest poll detailing a 4-point Conservative lead, up 2 from 3 weeks ago. The regional breakdown is rather unremarkable: the Bloc and the NDP are a bit high in Québec and BC respectively (well, I guess the NDP is always "high" in BC), but nothing out of the ordinary. The aggregate projection thus remains unchanged for the third update in a row:

CON - 133
LIB - 82
BQ - 53
NDP - 40

The Conservative national average advantage over the Liberals has decreased slightly to 6.0%.

Thursday, May 13, 2010

EKOS: Tories Lead by 6.5

This week's EKOS poll is completely unremarkable in terms of the national and all regional results for the four main parties. It does not change the aggregate projection, which is still:

CON - 133
LIB - 82
BQ - 53
NDP - 40

The average Conservative national lead also stays flat at 6.2%.

Tuesday, May 11, 2010

Ipsos: Tories Lead by 6; Léger Québec Poll

Here are links to two polls conducted last week: Ipsos (via ThreeHundredEight.com) and Léger. Not much surprising in either of these; adding them to the polling average does not change the aggregate projection:

CON - 133
LIB - 82
BQ - 53
NDP - 40

The Tory average national lead is still 6.2%.

Friday, May 7, 2010

UK Election Results

With the results declared in 649 constituencies, here's a summary of the UK election results. First, let's look at the likely number of seats for each party (the remaining one will probably go Tory in a delayed election), with their likely final GB vote share in parentheses:

CON - 307 (37%)
LAB - 258 (29.7%)
LD - 57 (23.5%)
Other - 28 (9.8%)

Of the four polling averages that I calculated just before Election Day, the most restrictive one, counting only last-minute polls by established firms, performed best. Compared to that average, the Tories got +1.2%, Labour got +1.7%, while the Lib Dems got -3.5%. That 5-point Lib Dem loss relative to the other parties explains their woefully disappointing result - they actually lost 5 seats relative to 2005! This election can also be considered a failure for pollsters, especially Angus Reid, which completely screwed up their entry into the UK election polling scene by lowballing Labour's share at 23-24% throughout the campaign.

As it turns out, the uniform national swing (UNS) model would have indeed overstated Labour's seat count and understated the Tory one, though not quite by as much as I thought, and certainly be less than than FiveThirtyEight thought. However, because the Lib Dem vote share barely changed in the last election, UNS, like just about any sensible model, would have pinned down their seat number quite well.

The outcome of this election is right in the middle of the chaos range, where neither Labour+Lib Dem (315) nor Conservatives+Unionists (317) have a majority out of the 645 non-Sinn Fein seats. If the Tories don't get Lib Dem support, they will need some combination of SNP (6), Plaid Cymru (3), SDLP (3) and Green (1) backing. This will be very interesting, as all four of these parties are left-leaning. The excitement isn't over yet!

Of the 14 projections I noted, 10 accurately predicted a lack of working majority for any party, with the Tories as largest party. LSE was inaccurate in having Labour as the largest party, while The Times, Angus Reid and BC Iconoclast were wrong in projecting a Conservative working majority (with unionist support in the case of the Times).

Here are the 3 closest projections for each of the major parties:

CON: 1 (tie). YouGov and UK Polling Report (305); 3. Canadian Election Watch (310)
LAB: 1. Election Prediction Project (245); 2 (tie). YouGov and Electoral Calculus (235)
LD: 1. YouGov (80); 2. LSE (81); 3. The Times (82).

Adding up the absolute deviations for the 3 big parties gives the following ranking of projections:

1. YouGov - 48
2. Electoral Calculus - 62
3. ComRes - 63
3. UK Polling Report - 63
5. Riding By Riding - 67
6. Election Prediction Project - 68
7. Trendlines - 74
8. Canadian Election Watch - 84
9. The Times - 85
10. FiveThirtyEight - 105
11. LSE - 108
12. TNS - 126
13. Angus Reid - 161.5
14. BC Iconoclast - 212

Essentially, due to the Lib Dems' unexpected weakness and Labour's unexpected strength, projections relying more on UNS did better than those taking into account geometric vote transfers. With a middle-of-the-pack ranking, I can't say that I'm very happy, but at least I wasn't laughably bad...

The silver lining - and I think a rather important one - is that the most important number in these projections is the Tory seat count, since that dictates how easy/hard of a time David Cameron will have in forming a government. On that account, I ranked 3rd out of 14, just 1 off YouGov and UK Polling Report!

Thursday, May 6, 2010

EKOS: Conservatives Lead by 7

This week's EKOS poll is available, and east of Alberta, it is good news for the Tories, who score a 4-point lead in both Ontario and the Atlantic as well as a whopping 27-point lead over the NDP in MB/SK. However, they had under 50% in Alberta and under 30% in BC, where there is a tight three-way race. Conversely, the only ray of hope in this poll for the Liberals is BC: it's pretty much doom and gloom everywhere else.

The NDP fell back to 16% nationally, and this was not a good poll for them. Just like several polls last week suggested an NDP surge, we now have two polls today suggesting that it has already abated. Did anything happen between last week and this week, or is this just a coincidence?

Incorporating both this poll and the Nanos poll into the aggregate projection produces little change:

CON - 133
LIB - 82
BQ - 53
NDP - 40

The average Tory national lead is now 6.2%.

Wednesday, May 5, 2010

Nanos: Conservatives Lead by 4

Nanos' latest poll is out, and as usual, it is a pretty good one for the Liberals. While they trail the Tories by 4% nationally, they have politically (though not statistically) significant leads in Ontario and Atlantic Canada and respectable scores in BC (32.2%) and Québec (26.7%). This is also a decent poll for the Tories and the Bloc. It's the NDP and the Greens that get disappointing numbers right across the country. In the NDP's case, it's surprising since they have fared abnormally well in the latest polls by other firms. In the Greens' case, it's also surprising: while they usually get lower numbers with Nanos due to methodological issues (Nanos does not prompt with party name), their 3.8% this time is just dismal.

I will update the aggregate projection once today's EKOS poll is available.

UK Election Tomorrow

Ten polling firms have published polls following the third and final debate last week. Here are the averages of the most recent poll by each firm, and changes from last week:

CON - 34.8% (+1%)
LD - 28.2% (-1%)
LAB - 26.7% (+0.7%)

Three of these firms are newcomers to UK political polling. Excluding them gives:

CON - 35.6% (+1.4%)
LAB - 27.9% (+1.1%)
LD - 27.4% (-1.1%)

If we instead exclude the three firms that did not conduct a poll in the past 3 days (so their most recent poll was either last weekend or conducted over the entire week), we get:

CON - 35.9% (+1.9%)
LAB - 27.4% (+1.1%)
LD - 27.3% (-1.3%)

Only including the six "old" firms that conducted a poll in the past few days gives:

CON - 35.8% (+1.7%)
LAB - 28% (+1.2%)
LD - 27% (-1.3%)

So, what's clear is that the Lib Dems have lost momentum in the past week, while the Tories and Labour have gained some. Also, the Tories are obviously in front, while Labour and the Lib Dems are very close.

I expect the Tories winning by about 8 points, while Labour and the Lib Dems should be roughly tied for second. I expect the final seat count to be roughly:

CON - 310 (including Speaker and the delayed election in Thirsk and Malton)
LAB - 214
LD - 94
Other - 32

Around the web, here are some projections, including both the Speaker and T&M in the Conservative column (Con-Lab-LD-Oth; midpoint if only range is specified):

UK Websites
YouGov: 305-235-80-30
ComRes: 299-233-87-31
LSE: 251-286-81-32
The Times (based on Ladbrokes): 322-213-82-33
Electoral Calculus: 297-235-86-32
UK Polling Report: 305-225-85

International Websites
FiveThirtyEight: 312-204-103-31
Election Prediction Project: 284-245-90-31 (updated)

Canadian Websites
Riding By Riding: 300-230-90-31 (sums to 651)
TrendLines: 299-227-93-31
BC Iconoclast: 337-151-132-30 (updated)

Other Websites (via UK Polling Report)
TNS: 292-204-114
Angus Reid: 330-175-112.5

So the ranges are:
CON - 251 (LSE) to 337 (BC Iconoclast)
LAB - 151 (BC Iconoclast) to 286 (LSE)
LD - 80 (YouGov) to 132 (BC Iconoclast)

Excluding LSE and BC Iconoclast:
CON - 279+ (Election Prediction) to 330 (Angus Reid)
LAB - 175 (Angus Reid) to 242+ (Election Prediction)
LD - 80 (YouGov) to 114 (TNS)

Let's see how these 14 projections, some based on formal models and others (like mine) based on hunches, perform!

Monday, May 3, 2010

Angus Reid: Conservatives Lead by 7 (in Canada)

Angus Reid has a new poll out that broadly confirms what other recent polls have been telling us: the Conservative lead has grown, but just a little, and the NDP is doing relatively well. The only surprising regional results are the strong 31% for the NDP in MB/SK (not that surprising given small sample size), and the whopping 51% for the Tories in BC - especially given that last week's EKOS and Harris-Decima polls put them second there.

Adding this poll in makes the Tories only just regain 1 seat in BC from the NDP:

CON - 132
LIB - 82
BQ - 54
NDP - 40

The Conservative national average lead is 6.0%.

Saturday, May 1, 2010

UK Election: 5 Days before Polls Close

Four polling firms have already completed surveys after Thursday's election debate in the UK. Here are the averages, with the change from the average of these same 4 firms' last results before the debate in parentheses:

CON - 35.5% (+1.75%)
LD - 28% (-0.5%)
LAB - 27% (-0.25%)

Note that, based on their last polls before the last debate, these firms had Labour about 1 point higher on average than all 11 firms (that I know of) combined.

According to the BBC's Election Seat Calculator, which assumes a uniform national swing, we would get CON 286, LAB 251, LD 84. Adjusting for problems with a uniform swing in an election with large changes, I expect something like CON 310, LAB 205, LD 105 if current numbers hold.

In the above scenario, the 30 seats belonging to small parties become very important, since the Tories might be able to govern with the support of small parties rather than the Liberal Democrats. A majority is nominally 326 seats this time, but there are complications:

- In the UK, not only the Speaker, but also three Deputy Speakers do not vote; in the previous Parliament, the Speaker and two Deputies were Conservative (the other Deputy was from Labour).

- Sinn Fein does not take its seats; in the 2005 election, they had won 5.

Thus, if the Northern Irish electoral results are replicated, there would be 641 voting members in the House, meaning that 321 votes would effectively form a majority. If the Tories still provide the Speaker and two Deputies, they would need 324 seats (or 323 if you count the Speaker separately) for a voting majority. If they gain the support of the unionist Northern Irish parties, which are right-wing, and if these parties win 10 seats like they did last time, then the Conservatives would need 314 seats (313 not counting the Speaker). All other small parties lean left, and are thus less natural partners for the Conservatives. Finally, the Tories may decide to provide fewer than 3 of the 4 non-voting members of the House, though they will almost certainly still provide the same Speaker, and probably at least one Deputy Speaker.

So this is how the situation for the Tories looks to me, counting the Speaker (and assuming Northern Ireland results remain unchanged from 2005):
324+: majority
322-323: majority through giving up on one or both Deputy Speaker positions
314-321: majority through alliance with Unionist parties
312-313: majority through both of the above
More than 298 or so: need support of at least some left-leaning small parties to govern without Lib Dem or Labour support
Less than 297 or so: require support of Lib Dems or Labour

Currently, the Tories may well be smack in the middle of that 298-323 ambiguous area, although my guess is that they will look for Lib Dem support before trying to woo over multiple small left-leaning parties. The last 4 days of campaigning will be crucial!

Léger Marketing: Conservatives Lead by 11

A national Léger poll (regional breakdown via ThreeHundredEight.com) shows a bit of a "back to the future" situation, with results close to those of the 2008 election: the Liberals and Tories are down slightly from October 2008, while the NDP is up slightly. The biggest news in this poll is the Conservatives' 9-point lead in Ontario. That's grossly out-of-line with other pollsters, who recently averaged a 2-point Liberal lead there.

Adding this poll to the mix moves one seat from the Liberals to the Conservatives:

CON - 131
LIB - 82
BQ - 54
NDP - 41

Conservative average national lead is now 5.7%.