Angus-Reid also posted their most recent poll today, with results very similar to EKOS'. This poll is good news for the Liberals in Ontario, but not so good in BC. The national numbers (34-33 for Grits) are basically the same as in the EKOS poll, and my projection based on this poll alone also has the Tories winning by a very small number of seats. We are really in a dead heat!
Overall, it looks like if an election had occurred this past week, we would have basically had a rerun of the 2006 election outside Québec. In Québec though, the Liberals are a lot stronger, and the Conservatives are weaker, as is the Bloc.
Both polls from today suggest that the dominant opposition party in Manitoba/Saskatchewan has become the Liberals, by over 10%. The NDP beat the Grits by 7.6% in this region in the last election. This doesn't have big seat implications since the Tories are far ahead there, but it suggests that the Liberals are a true national party again.
My aggregate projection still shows the Conservatives ahead by 15 seats:
CON - 123
LIB - 108
BQ - 46
NDP - 31
That's because there are relatively few polls this summer (basically only EKOS and Angus), so I'm temporarily using more than one poll from each company. But if more data confirm today's polls, the numbers will keep moving in the Liberals' favor.
Thursday, July 30, 2009
Weekly EKOS
Latest projection:
CON - 124
LIB - 106
BQ - 46
NDP - 32
Not much change for any of the parties. In this new poll, the Grits have taken a 34.1-32.5 (statistically insignificant) lead in the national popular vote. However, even looking at this poll alone, my seat projection gives the Tories a marginal lead. The Liberals have suspiciously good numbers on the Prairies in this poll, but these don't really help seat-wise. The Grit Ontario lead stays at 4%, which is very low considering their national lead.
This is the 7th EKOS poll in a row where the difference between the top two parties is statistically insignificant (keeping in mind that the uncertainty is higher when evaluating the difference between two parties than evaluating a single party's support).
CON - 124
LIB - 106
BQ - 46
NDP - 32
Not much change for any of the parties. In this new poll, the Grits have taken a 34.1-32.5 (statistically insignificant) lead in the national popular vote. However, even looking at this poll alone, my seat projection gives the Tories a marginal lead. The Liberals have suspiciously good numbers on the Prairies in this poll, but these don't really help seat-wise. The Grit Ontario lead stays at 4%, which is very low considering their national lead.
This is the 7th EKOS poll in a row where the difference between the top two parties is statistically insignificant (keeping in mind that the uncertainty is higher when evaluating the difference between two parties than evaluating a single party's support).
Friday, July 24, 2009
First Mapped Projection
This is not a new projection; I'm just mapping this projection from July 23, based on polling through July 21. Mostly, relative to 2008, we have the Liberals gaining in close suburbs, but still not quite there in the outer ring of most cities. I will probably map another projection when there's significant movement, or when an election gets called. Stay tuned!
2008 Result Maps!
I have decided to spice things up on this blog by adding maps! In this post, you can view the results of the 2008 election, and if there's enough demand, I'll be happy to post maps of the 2004 and 2006 results as well (yes, including the different boundary between Acadie--Bathurst and Miramichi in 2004).
In the future, I will occasionally be mapping my projections. I will not always do so because:
1. While aggregate seat projections are a hazardous exercise, specific projections are even more prone to error: the idea behind the aggregate projections is that if some ridings swing more, and others less, things will roughly balance out in the total.
2. It takes time!
I will, however, certainly map the last projection before the election, and possibly other "significant" ones (e.g. at the start of the campaign) as well.
Enjoy!
In the future, I will occasionally be mapping my projections. I will not always do so because:
1. While aggregate seat projections are a hazardous exercise, specific projections are even more prone to error: the idea behind the aggregate projections is that if some ridings swing more, and others less, things will roughly balance out in the total.
2. It takes time!
I will, however, certainly map the last projection before the election, and possibly other "significant" ones (e.g. at the start of the campaign) as well.
Enjoy!
Where the jobs actually are
The New York Times argues in this editorial that "with low-wage work expected to be the most plentiful in the years to come, raising the minimum wage and growth opportunities should be a priority of the White House." (Quote is not in the article, but in the snippet provided when I shared this article on Facebook.) It states that "according to the Labor Department, 5 of the 10 occupations expected to add the most jobs through 2016 are 'very low paying,' up to a maximum of about $22,000 a year. They include retail sales jobs and home health aides. Another 3 of the 10 are 'low paying,' from roughly $22,000 to $31,000, including customer-service representatives, general office clerks and nurses’ aides."
This seemed fishy to me, so I went ahead and checked out the actual Labor Department data. It turns out that by definition, low-paying job categories tend to be larger. To see why this is a problem, consider the following example. Suppose for simplicity that there are 1M low-paying jobs divided into 10 categories, and 1M high-paying jobs divided into 100 categories. Also suppose that the low-paying categories are projected to grow by 5%, and high-paying ones by 20%. Then each low-paying category would add 5,000 jobs, while each high-paying category would only add 2,000 jobs, so by the NY Times' methodology, we'd reach the false conclusion that low-paying industries are expanding faster.
The actual data is not as stark, but the pattern is definitely there. In fact, “very low” paying occupations are projected to add 3.65M jobs from 2006 to 2016, “low” paying 3.38M, “high” paying 3.34M, and “very high” paying 5.22M. These classifications are quartiles, each containing occupations representing 1/4 of workers in 2006. So as you can see, the data cited by the NY Times actually undermines their point, since "very high" paying jobs are projected to grow the fastest, with all other categories growing about equally fast. Sadly, this statistical sleight of hand forms the basis of much of their argument, so the entire editorial is pretty worthless.
This is not to say that I oppose a higher minimum wage in the U.S. or Canada: I haven't made up my mind on the issue. But what this data suggests is that, in fact, education needs to be a high priority in the U.S., since it is jobs in the top quartile, i.e. those that require the most skills, that are going to be created the fastest over the next few years. And I doubt that the situation would be very different for Canada. In fact, in the long-run, education is the best tool for combating inequality: increasing the supply of high-skill workers and decreasing that of low-skill workers will automatically reduce wage differentials without economically distortionary government interventions.
So, thanks to the NY Times for pointing me to that interesting data, but FAIL for reaching the wrong conclusion. (Again, I'm not saying that higher minimum wages are bad, just that the facts mentioned by the NY Times do not support their case.)
This seemed fishy to me, so I went ahead and checked out the actual Labor Department data. It turns out that by definition, low-paying job categories tend to be larger. To see why this is a problem, consider the following example. Suppose for simplicity that there are 1M low-paying jobs divided into 10 categories, and 1M high-paying jobs divided into 100 categories. Also suppose that the low-paying categories are projected to grow by 5%, and high-paying ones by 20%. Then each low-paying category would add 5,000 jobs, while each high-paying category would only add 2,000 jobs, so by the NY Times' methodology, we'd reach the false conclusion that low-paying industries are expanding faster.
The actual data is not as stark, but the pattern is definitely there. In fact, “very low” paying occupations are projected to add 3.65M jobs from 2006 to 2016, “low” paying 3.38M, “high” paying 3.34M, and “very high” paying 5.22M. These classifications are quartiles, each containing occupations representing 1/4 of workers in 2006. So as you can see, the data cited by the NY Times actually undermines their point, since "very high" paying jobs are projected to grow the fastest, with all other categories growing about equally fast. Sadly, this statistical sleight of hand forms the basis of much of their argument, so the entire editorial is pretty worthless.
This is not to say that I oppose a higher minimum wage in the U.S. or Canada: I haven't made up my mind on the issue. But what this data suggests is that, in fact, education needs to be a high priority in the U.S., since it is jobs in the top quartile, i.e. those that require the most skills, that are going to be created the fastest over the next few years. And I doubt that the situation would be very different for Canada. In fact, in the long-run, education is the best tool for combating inequality: increasing the supply of high-skill workers and decreasing that of low-skill workers will automatically reduce wage differentials without economically distortionary government interventions.
So, thanks to the NY Times for pointing me to that interesting data, but FAIL for reaching the wrong conclusion. (Again, I'm not saying that higher minimum wages are bad, just that the facts mentioned by the NY Times do not support their case.)
Thursday, July 23, 2009
New Projection
EKOS released a new poll today, and its results have been incorporated into the latest projection:
CON - 124
LIB - 104
BQ - 47
NDP - 33
Overall, the Conservatives lost 3 seats, while the Liberals gained 3. This is mainly due to some changes in Ontario, where the Grits are once again approaching 40%. Accordingly, you may have noticed that the background of this page got a bit lighter - indicating that we've moved closer to a tie and farther away from a Conservative majority.
While the latest poll has the Liberals within 0.3% of the Conservatives, my numbers show the Tories ahead by roughly a dozen seats based on this poll alone. This is what I referred to in this earlier post; note that the Ontario gap in this poll is indeed only 4.2%, which is good news for the Tories.
This poll also asked Canadians what kind of government (majority/minority and Liberal/Conservative) they prefer. Canadians are roughly equally divided into 4 groups, desiring:
1. a Liberal majority
2. a Conservative majority
3. a Liberal or Conservative minority
4. none of the above.
Worryingly for the Tories, only roughly 1/3 of Ontarians and 1/4 of Quebecers want to keep the CPC in office, while about 45% in both provinces want the LPC instead. If Ignatieff can start converting that goodwill into votes, the Grits can move up pretty quickly. However, even if the Liberals manage to win Ontario by a dozen point and get back to their 36-seat apex (2000 election) in Quebec, they would still be a dozen seats short of a majority without significant gains in other regions. Perhaps this is why Ignatieff is trying to win support in the West: it may not net seats in this coming election, but without the possibility of sweeping Ontario or Quebec in the foreseeable future, the Grits need a healthy number of seats out West to even hope for a majority.
Finally, an interesting snippet: in the EKOS poll, the Greens are in the lead among voters under 25, with 27% support. The sample size for that group is small though, so this lead is far from statistically significant.
CON - 124
LIB - 104
BQ - 47
NDP - 33
Overall, the Conservatives lost 3 seats, while the Liberals gained 3. This is mainly due to some changes in Ontario, where the Grits are once again approaching 40%. Accordingly, you may have noticed that the background of this page got a bit lighter - indicating that we've moved closer to a tie and farther away from a Conservative majority.
While the latest poll has the Liberals within 0.3% of the Conservatives, my numbers show the Tories ahead by roughly a dozen seats based on this poll alone. This is what I referred to in this earlier post; note that the Ontario gap in this poll is indeed only 4.2%, which is good news for the Tories.
This poll also asked Canadians what kind of government (majority/minority and Liberal/Conservative) they prefer. Canadians are roughly equally divided into 4 groups, desiring:
1. a Liberal majority
2. a Conservative majority
3. a Liberal or Conservative minority
4. none of the above.
Worryingly for the Tories, only roughly 1/3 of Ontarians and 1/4 of Quebecers want to keep the CPC in office, while about 45% in both provinces want the LPC instead. If Ignatieff can start converting that goodwill into votes, the Grits can move up pretty quickly. However, even if the Liberals manage to win Ontario by a dozen point and get back to their 36-seat apex (2000 election) in Quebec, they would still be a dozen seats short of a majority without significant gains in other regions. Perhaps this is why Ignatieff is trying to win support in the West: it may not net seats in this coming election, but without the possibility of sweeping Ontario or Quebec in the foreseeable future, the Grits need a healthy number of seats out West to even hope for a majority.
Finally, an interesting snippet: in the EKOS poll, the Greens are in the lead among voters under 25, with 27% support. The sample size for that group is small though, so this lead is far from statistically significant.
Wednesday, July 22, 2009
If we can't get your vote, we won't fund you
So given this, it looks like the Conservatives have decided to stop even pretending that Ablonczy's demotion had nothing to do with Toronto gay pride funding. The Conservatives are gambling that not funding Divers/Cité is going to do more good appeasing their base (which may be a bit disgruntled about all the bailout money) than harm losing the gay vote and the Montréal vote. Divers/Cité seems like a prime target Tories since clearly not many gays vote conservative, and Montréal is a wasteland in terms of Conservative electoral hopes: the entire metro area (even far-away suburbs, unlike Toronto or Vancouver) wasn't even close to electing a single Conservative in the last election.
This calculation may end up backfiring, though, if it costs Tories votes in the Québec City area, since they may find themselves in several tough fights there to retain their seats. I doubt that this will have an effect outside Québec, in ridings where the Tories have a shot.
This also fits into the pattern of Tories redistributing public money to further their political ambitions. Remember the $2.2 billion fiscal imbalance "resolution" given to Québec in 2007, and how when it was subsequently clawed back through tweaks in Equalization, there was suddenly money for GST harmonization for Ontario?
This calculation may end up backfiring, though, if it costs Tories votes in the Québec City area, since they may find themselves in several tough fights there to retain their seats. I doubt that this will have an effect outside Québec, in ridings where the Tories have a shot.
This also fits into the pattern of Tories redistributing public money to further their political ambitions. Remember the $2.2 billion fiscal imbalance "resolution" given to Québec in 2007, and how when it was subsequently clawed back through tweaks in Equalization, there was suddenly money for GST harmonization for Ontario?
Tuesday, July 21, 2009
Whither Policy Debate?
The first thing that I noticed about the difference between American and Canadian politics when I moved down here is how partisan some U.S. networks are. Say all you want about CTV and CBC's biases; they are nothing compared to Fox and MSNBC's. On the surface, this gave the impression that Canada's political discourse is superior to America's.
But over time, I have noticed that while there are more political media fireworks in the U.S., there is also more (of course, still not very much) substantive policy debate. For example, the merits of the GM bailout were much more debated down here than back home, even if relative to each country's economy, the amount contributed by Ottawa is roughly twice the amount contributed by Washington. Of course, the debate didn't get very deep in the U.S. mainstream media either, but at least some substantive issues were mentioned (underlying causes, beyond the recession, of GM's woes; long-term viability; desirability of instead letting GM fail, etc.). In Canada, beyond the reporting of the basic facts and stereotypical reactions, virtually nada.
Or take the last election campaign in both countries. Other than the short-run state of the economy, the main issue in the U.S. was health care, while it was the Green Shift in Canada. In the U.S., the discussion touched upon whether a mandate is desirable (comparing the Clinton and Obama plans), the long-run impact of the policies on the fiscal situation (costs of the Obama plan and creeping tax increase implied by the McCain plan), and other slightly technical issues that are nevertheless possible to explain to the general public, and important for understanding the different proposals' implications. In Canada, the media basically said, "The Green Shift will raise your taxes, but reduce pollution. Other than that it's too complicated for you to understand, oh stupid Canadians. By the way, the NDP wants cap-and-trade, while the Conservatives want regulation." It went on to repeat thousands of times that the Green Shift is too complicated, while never attempting to report on the differential economic and environmental impacts of the carbon tax, cap-and-trade, and regulation approaches.
There are many reasons why there is less substantive policy debate in Canada, not all of which are bad, and here I consider a few. First, Canada arguably faces fewer big problems than the U.S.: the long-run fiscal outlook is not nearly as dire, it is not mired in Iraq, its healthcare system (despite all its flaws) does provide reasonable care to almost everyone, etc. This, obviously, is good for Canada, but doesn't mean that there isn't anything important to debate.
Another possible reason is that Canada has fewer high-profile independent policy analysts. It helps when Nobel Prize laureates (like Paul Krugman and Gary Becker) have their own blogs. Even though academics are biased like everyone else, the fact that they don't speak on behalf of a party, interest group, company or think tank reduces the likelihood that they face outside pressure to further bias their analysis. What do you find more informative: that Greenpeace supports a carbon tax, or that Greg Mankiw, George W. Bush's former chief economic advisor and now Harvard professor, supports a carbon tax? Part of the problem is that Canada is much smaller than the U.S., and that can't be changed. But our academics becoming less timid and our media giving them more room could go a long way toward changing things.
A third factor is that Conservatives and Liberals have viewpoints that differ less than Republicans and Democrats do. I'm not sure whether this is a good thing, though I'm certainly happy that there isn't a sizable Sarah Palin wing of the CPC.
Also, Canadians spend much more time worrying about the provincial distribution of federal money than Americans. This is mainly because relative to state/provincial governments, the U.S. government spends more money directly than the Canadian government, which instead transfers more money to fund provincial governments. These transfers make regional inequities more transparent: it's easier to get outraged at Quebec and the Maritimes getting so much in equalization, than to get outraged at provinces with higher unemployment getting so much in EI premiums. So as we debate more about inter-regional distribution issues, we think less about "big" national issues. I don't think its bad that Canadian provinces have more spending power than U.S. states, but I do think it's unhealthy for Ottawa to constantly tweak the equalization formula and bail out regional industries that are in a semi-permanent state of crisis.
Finally, there's Canada's iron-tight party discipline. When one argues against another party, it's easy to simply appeal to broad philosophical differences and gut feelings. But when one argues against someone from their own party, one is forces to have a more careful discussion. Surely some Liberals were for cap-and-trade, some Dippers were for a carbon tax, and some Tories were uneasy with regulation. But while in the U.S. and many other countries, these intra-party debates are in good part public, in Canada, they tend to stay mostly private.
In the end, this means that if you're a policy wonk like me, Canadian political discussions leave you wanting (or maybe they don't even whet your appetite). Don Newman retiring won't help. Let's hope Andrew Coyne isn't planning to go anywhere anytime soon. But at least, our electoral system, with its potential for dramatic swings in very short periods, keeps things interesting! (Of course, it also contributes to the stifling party discipline mentioned above...)
But over time, I have noticed that while there are more political media fireworks in the U.S., there is also more (of course, still not very much) substantive policy debate. For example, the merits of the GM bailout were much more debated down here than back home, even if relative to each country's economy, the amount contributed by Ottawa is roughly twice the amount contributed by Washington. Of course, the debate didn't get very deep in the U.S. mainstream media either, but at least some substantive issues were mentioned (underlying causes, beyond the recession, of GM's woes; long-term viability; desirability of instead letting GM fail, etc.). In Canada, beyond the reporting of the basic facts and stereotypical reactions, virtually nada.
Or take the last election campaign in both countries. Other than the short-run state of the economy, the main issue in the U.S. was health care, while it was the Green Shift in Canada. In the U.S., the discussion touched upon whether a mandate is desirable (comparing the Clinton and Obama plans), the long-run impact of the policies on the fiscal situation (costs of the Obama plan and creeping tax increase implied by the McCain plan), and other slightly technical issues that are nevertheless possible to explain to the general public, and important for understanding the different proposals' implications. In Canada, the media basically said, "The Green Shift will raise your taxes, but reduce pollution. Other than that it's too complicated for you to understand, oh stupid Canadians. By the way, the NDP wants cap-and-trade, while the Conservatives want regulation." It went on to repeat thousands of times that the Green Shift is too complicated, while never attempting to report on the differential economic and environmental impacts of the carbon tax, cap-and-trade, and regulation approaches.
There are many reasons why there is less substantive policy debate in Canada, not all of which are bad, and here I consider a few. First, Canada arguably faces fewer big problems than the U.S.: the long-run fiscal outlook is not nearly as dire, it is not mired in Iraq, its healthcare system (despite all its flaws) does provide reasonable care to almost everyone, etc. This, obviously, is good for Canada, but doesn't mean that there isn't anything important to debate.
Another possible reason is that Canada has fewer high-profile independent policy analysts. It helps when Nobel Prize laureates (like Paul Krugman and Gary Becker) have their own blogs. Even though academics are biased like everyone else, the fact that they don't speak on behalf of a party, interest group, company or think tank reduces the likelihood that they face outside pressure to further bias their analysis. What do you find more informative: that Greenpeace supports a carbon tax, or that Greg Mankiw, George W. Bush's former chief economic advisor and now Harvard professor, supports a carbon tax? Part of the problem is that Canada is much smaller than the U.S., and that can't be changed. But our academics becoming less timid and our media giving them more room could go a long way toward changing things.
A third factor is that Conservatives and Liberals have viewpoints that differ less than Republicans and Democrats do. I'm not sure whether this is a good thing, though I'm certainly happy that there isn't a sizable Sarah Palin wing of the CPC.
Also, Canadians spend much more time worrying about the provincial distribution of federal money than Americans. This is mainly because relative to state/provincial governments, the U.S. government spends more money directly than the Canadian government, which instead transfers more money to fund provincial governments. These transfers make regional inequities more transparent: it's easier to get outraged at Quebec and the Maritimes getting so much in equalization, than to get outraged at provinces with higher unemployment getting so much in EI premiums. So as we debate more about inter-regional distribution issues, we think less about "big" national issues. I don't think its bad that Canadian provinces have more spending power than U.S. states, but I do think it's unhealthy for Ottawa to constantly tweak the equalization formula and bail out regional industries that are in a semi-permanent state of crisis.
Finally, there's Canada's iron-tight party discipline. When one argues against another party, it's easy to simply appeal to broad philosophical differences and gut feelings. But when one argues against someone from their own party, one is forces to have a more careful discussion. Surely some Liberals were for cap-and-trade, some Dippers were for a carbon tax, and some Tories were uneasy with regulation. But while in the U.S. and many other countries, these intra-party debates are in good part public, in Canada, they tend to stay mostly private.
In the end, this means that if you're a policy wonk like me, Canadian political discussions leave you wanting (or maybe they don't even whet your appetite). Don Newman retiring won't help. Let's hope Andrew Coyne isn't planning to go anywhere anytime soon. But at least, our electoral system, with its potential for dramatic swings in very short periods, keeps things interesting! (Of course, it also contributes to the stifling party discipline mentioned above...)
Monday, July 20, 2009
First Projection and the "Ontario Gap"
CON - 127
LIB - 101
BQ - 47
NDP - 33
This projection is based on Angus-Reid, EKOS and The Strategic Counsel polls conducted from early July to July 17. The provincial breakdown is available at the top of this page until the projection is updated.
These results are very similar to the ones from 2006, even though the Conservatives are leading by much less than 6%, which was their margin of victory. Why? The Conservative vote seems to have gotten quite a bit more efficient lately: Conservative support in Ontario, a province with a disproportionate share of swing seats, is up relative to the rest of Canada. In 2006, the Tories lost Ontario by 5% while winning the country by 6%, which means that Ontario was 11% less favorable to them than Canada as a whole - call this the "Ontario gap." Now, they're leading the national totals by about 2% and trailing in Ontario by about 2%, so the Ontario gap has shrunk to 4%.
In fact, it appears that while in 2006, a popular vote tie would have resulted in a Liberal win, right now, it would result in a Conservative win. However, because Ontarians seem to vacillate between Grits and Tories more than the rest of Canada, this situation could well change before or during the next campaign.
LIB - 101
BQ - 47
NDP - 33
This projection is based on Angus-Reid, EKOS and The Strategic Counsel polls conducted from early July to July 17. The provincial breakdown is available at the top of this page until the projection is updated.
These results are very similar to the ones from 2006, even though the Conservatives are leading by much less than 6%, which was their margin of victory. Why? The Conservative vote seems to have gotten quite a bit more efficient lately: Conservative support in Ontario, a province with a disproportionate share of swing seats, is up relative to the rest of Canada. In 2006, the Tories lost Ontario by 5% while winning the country by 6%, which means that Ontario was 11% less favorable to them than Canada as a whole - call this the "Ontario gap." Now, they're leading the national totals by about 2% and trailing in Ontario by about 2%, so the Ontario gap has shrunk to 4%.
In fact, it appears that while in 2006, a popular vote tie would have resulted in a Liberal win, right now, it would result in a Conservative win. However, because Ontarians seem to vacillate between Grits and Tories more than the rest of Canada, this situation could well change before or during the next campaign.
"Why Is the Background so Ugly?" and Other Questions
You may wonder:
- Why the background is so ugly: Its color reflects the latest projection! The background is pure blue if I project a Conservative majority, pure red if I project a Liberal one, etc. It is white if the top two parties are tied, and its intensity reflects how close the leading party is to a majority. A bit of an eyesore, but dorky-fun and informative!
- Why arithmetic rather than geometric swing: First and foremost, it's a matter of personal preference, given that neither method is obviously preferable to the other. I haven't done an in depth analysis of what method is better, but intuitively, winning 32-30 doesn't feel 1.5 times safer than winning 47-45. In fact, the former result suggests a three-way race or the presence of many small parties, neither of which usually makes the incumbent safer in the following election. For what it's worth, the Paulitics blog made both an arithmetic and a geometric projection in 2008, and the total absolute error of the arithmetic method was lower by 2.
- Why the use of the Liberal logo is inconsistent: It officially changed after the 2008 election! So I use the old logo in the left column to show the 2008 results, but the new one for my current projections. I plan on doing the same for all parties in the future.
- Why I don't make a popular vote projection: Most of the time, I end up with something that is very close to an arithmetic mean of the most recent polls. As a result, my popular vote estimates aren't very exciting, and I've decided that they're not worth the clutter. I will, however, post the gap between the two top parties according to popular vote (i.e. the Conservatives and the Liberals). Other blogs have various weighting/estimation procedures, and ThreeHundredEight.com lists all recent polls so that you can easily make your own!
- Why the pie charts in the left column do not reflect the seat distributions: They represent the popular vote. Again, in the interest of aesthetics, I did not include the numerical values of the vote shares. But they are embedded in the URL of each pie chart, in decreasing order (which shows up on the charts as clockwise, starting at 3 o'clock).
More questions concerning this blog? Just ask!
Note: Edited on March 3, 2011
- Why the background is so ugly: Its color reflects the latest projection! The background is pure blue if I project a Conservative majority, pure red if I project a Liberal one, etc. It is white if the top two parties are tied, and its intensity reflects how close the leading party is to a majority. A bit of an eyesore, but dorky-fun and informative!
- Why arithmetic rather than geometric swing: First and foremost, it's a matter of personal preference, given that neither method is obviously preferable to the other. I haven't done an in depth analysis of what method is better, but intuitively, winning 32-30 doesn't feel 1.5 times safer than winning 47-45. In fact, the former result suggests a three-way race or the presence of many small parties, neither of which usually makes the incumbent safer in the following election. For what it's worth, the Paulitics blog made both an arithmetic and a geometric projection in 2008, and the total absolute error of the arithmetic method was lower by 2.
- Why the use of the Liberal logo is inconsistent: It officially changed after the 2008 election! So I use the old logo in the left column to show the 2008 results, but the new one for my current projections. I plan on doing the same for all parties in the future.
- Why I don't make a popular vote projection: Most of the time, I end up with something that is very close to an arithmetic mean of the most recent polls. As a result, my popular vote estimates aren't very exciting, and I've decided that they're not worth the clutter. I will, however, post the gap between the two top parties according to popular vote (i.e. the Conservatives and the Liberals). Other blogs have various weighting/estimation procedures, and ThreeHundredEight.com lists all recent polls so that you can easily make your own!
- Why the pie charts in the left column do not reflect the seat distributions: They represent the popular vote. Again, in the interest of aesthetics, I did not include the numerical values of the vote shares. But they are embedded in the URL of each pie chart, in decreasing order (which shows up on the charts as clockwise, starting at 3 o'clock).
More questions concerning this blog? Just ask!
Note: Edited on March 3, 2011
It's a Launch! Projection Methodology
Welcome to Canadian Election Watch, a blog where I will make federal seat projections based on the most recent publicly available polls. Of course, several websites like this exist, so in addition to politically neutral seat projections, I will also occasionally provide my own biased view on political issues in Québec, Canada and the United States.
The backbone of the projections is a uniform arithmetic swing model within each region: if polling average says that the Greens have gained 4% in Ontario relative to the last general election, I assume that they have done so in every riding in Ontario. The six regions are: Atlantic Canada, Québec, Ontario, MB/SK, Alberta and BC.
Of course, while the underlying projection method is simple and objective (though its selection is of course subjective), judgment calls are needed* to deal with the following issues, among others:
1. Poll inclusion: I define a poll's age as the time between its midpoint date and the most recent poll's midpoint date. In normal times, a poll up 7 days old is included at full value (unless it is not the most recent poll by a firm), and a poll up to 27 days old is given some weight. These periods may be lengthened when polling is infrequent (e.g. during summers), and will be shortened during election campaigns. The point is to get enough polls to make the projections consistent, but not so many that they become unresponsive.
2. Poll weighting: National polls of n respondents will receive a weight of sqrt(n). I do not use full proportionality because part of the appeal of using multiple polls is to prevent one firm's methodological biases from driving the results. For the same reason, polls that are not the most recent by a firm are discounted by half (in addition to the time discounting).
3. Intra-provincial breakdowns: I will generally ignore these, since it is often hard to figure out what these regions mean (does "905" mean the Greater Toronto Area outside Toronto, the Toronto Census Metropolitan Area outside Toronto, or the actual area of the 905 area code?). Even when we know what the regions refer to, they may not correspond to riding boundaries, and I do not have the desire to go through results by polling station. However, in special cases where there is evidence that a region might be moving (not just voting) very differently from the rest of the province, I will make use of these breakdowns. For example, if the Conservatives are weak in Québec, numbers specific to the Capitale-Nationale area will be very helpful.
4. Ad hoc intra-provincial/regional patterns: a) Tight Montréal ridings move much less than the rest of Québec. So if the numbers tell me, for example, that the Liberals will gain a Montréal riding from the Bloc by a narrow margin, I will likely override that change. b) New Brunswick normally behaves very much like Ontario. So when projecting New Brunswick ridings, I will give consideration to both Atlantic Canada and Ontario numbers. Have you noticed any other (useful) patterns? Something that can improve projections in British Columbia would be especially helpful...
5. Riding polls: These will be given very high, but less than proportional, weight. The main issue here is that I will likely miss quite a few of these, so if you know of one, please inform me by posting a comment about it!
6. Special circumstances: Northern ridings, scandals, by-elections, open seats (especially when the incumbent was popular), seats held by independents, etc. These will be considered on an ad hoc basis, but adjustments for these factors will be minimal due to my lack of local knowledge about most of the country, and the fact that most electors in most ridings vote for a party rather than a candidate.
*Yes, if I built a sophisticated statistical model, I'd be able to push back where the subjectivity comes in. But it is unlikely to help that much, time is a finite resource, and I don't enjoy programming...
Note: Edited on August 28, 2010 and March 2, 2011
The backbone of the projections is a uniform arithmetic swing model within each region: if polling average says that the Greens have gained 4% in Ontario relative to the last general election, I assume that they have done so in every riding in Ontario. The six regions are: Atlantic Canada, Québec, Ontario, MB/SK, Alberta and BC.
Of course, while the underlying projection method is simple and objective (though its selection is of course subjective), judgment calls are needed* to deal with the following issues, among others:
1. Poll inclusion: I define a poll's age as the time between its midpoint date and the most recent poll's midpoint date. In normal times, a poll up 7 days old is included at full value (unless it is not the most recent poll by a firm), and a poll up to 27 days old is given some weight. These periods may be lengthened when polling is infrequent (e.g. during summers), and will be shortened during election campaigns. The point is to get enough polls to make the projections consistent, but not so many that they become unresponsive.
2. Poll weighting: National polls of n respondents will receive a weight of sqrt(n). I do not use full proportionality because part of the appeal of using multiple polls is to prevent one firm's methodological biases from driving the results. For the same reason, polls that are not the most recent by a firm are discounted by half (in addition to the time discounting).
3. Intra-provincial breakdowns: I will generally ignore these, since it is often hard to figure out what these regions mean (does "905" mean the Greater Toronto Area outside Toronto, the Toronto Census Metropolitan Area outside Toronto, or the actual area of the 905 area code?). Even when we know what the regions refer to, they may not correspond to riding boundaries, and I do not have the desire to go through results by polling station. However, in special cases where there is evidence that a region might be moving (not just voting) very differently from the rest of the province, I will make use of these breakdowns. For example, if the Conservatives are weak in Québec, numbers specific to the Capitale-Nationale area will be very helpful.
4. Ad hoc intra-provincial/regional patterns: a) Tight Montréal ridings move much less than the rest of Québec. So if the numbers tell me, for example, that the Liberals will gain a Montréal riding from the Bloc by a narrow margin, I will likely override that change. b) New Brunswick normally behaves very much like Ontario. So when projecting New Brunswick ridings, I will give consideration to both Atlantic Canada and Ontario numbers. Have you noticed any other (useful) patterns? Something that can improve projections in British Columbia would be especially helpful...
5. Riding polls: These will be given very high, but less than proportional, weight. The main issue here is that I will likely miss quite a few of these, so if you know of one, please inform me by posting a comment about it!
6. Special circumstances: Northern ridings, scandals, by-elections, open seats (especially when the incumbent was popular), seats held by independents, etc. These will be considered on an ad hoc basis, but adjustments for these factors will be minimal due to my lack of local knowledge about most of the country, and the fact that most electors in most ridings vote for a party rather than a candidate.
*Yes, if I built a sophisticated statistical model, I'd be able to push back where the subjectivity comes in. But it is unlikely to help that much, time is a finite resource, and I don't enjoy programming...
Note: Edited on August 28, 2010 and March 2, 2011
Subscribe to:
Posts (Atom)