Visit our sponsor:    Visit our sponsor: Your Ad Here    Visit our sponsor:

Posts Tagged ‘Strategic Voting’

Three Strategic Lessons and Seven Kinds of Ridings from the Ontario Election

June 16th, 2014 | 14 Comments

[Welcome, National Newswatch readers!]

As always, the real data story is found in the cross-tabs. The Ontario Liberals and Ontario NDP each gained a percentage point in popular support in Thursday's election, but they did not do so evenly, nor in the same places. Unfortunately for the Ontario Progressive Conservatives, they lost nearly everywhere.

The table below is a slightly updated version from the one I published on Twitter Saturday. Like Saturday's version it shows changes in turnout, raw vote counts, and vote shares from the 2011 to 2014 provincial general elections, broken down by sub-region of the province; but it also now shows seat changes at the side, and totals at the bottom, as well.

[Click on image to open full-sized version]

Regional Vote Shifts, 2014 Ontario provincial general election

These vote shifts demonstrate that while the NDP lost votes in well-educated downtown ridings in Toronto and Ottawa, the Liberals tanked in working and middle class southwestern Ontario, including Windsor, Sarnia, Chatham, London and the Niagara region – more or less as could have been predicted from last year's Liberal leadership race that saw Toronto's Kathleen Wynne defeat Sandra Pupatello the doyenne of Windsor.

At the same time, the starkness of the commitments in Tim Hudak's Million Jobs Plan to cut 100,000 public sector jobs appears to have cost his party vote-share across the three 905 regions of Durham, York and Peel, and into Central Ontario – and even cost raw vote-counts in some of them – as primarily Liberal voters stampeded to the polls to stop him. Even the NDP reinforced and grew its claim to parts of Brampton in the 905 West, as the Liberal vote also grew, leaving the Ontario PCs trailing in third place across the northern half of Peel Region.

The Ontario Conservatives also tanked in southwestern Ontario, coming within 5 points of losing Sarnia-Lambton to the NDP, and only hanging on in Chatham-Kent–Essex and Elgin-Middlesex-London thanks to the splits in three-way races where "a vote for the Liberals was a vote for Hudak", contrary to the motto being peddled by the red team in downtown Toronto and Ottawa. A bit further north, the PCs' drop benefitted both the NDP and the Liberals, as the blue team lost ground in Brant, lost their seat in Cambridge, and fell to third place in their former seat of Kitchener-Waterloo.

Downtown pundits and university professors scoffed at the NDP's strategy of appealing to small-town and working class soft conservative voters, but they shouldn't have, as it worked in Oshawa and Brampton, across the southwest of the province from Sarnia to the Niagara, and in northern (but not northwestern) Ontario.

The bad news for the Conservatives is that they can be beaten by Liberals in wealthier areas and by New Democrats in less-well-off areas of the province. Meanwhile the Liberals will be emboldened to fight both the NDP and the Conservatives in the big cities and suburbs.

Three Strategic Lessons

  1. The Conservatives can be beaten. Their strategy of playing to a small but loyal core base vote can be cannot survive forever, especially with an unforgiving hard-right platform. It's possible to win over some Conservative votes to other parties, and to also outnumber them by bringing previous non-voters to the polls. Indeed a hard-right platform can demotivate non-ideological low-information conservative voters and motivate progressive ones.
  2. The assumptions behind the Ontario NDP's strategy – however well or poorly that strategy might have been executed – did prove accurate in the intended parts of the province. We political-science-trained observers may think of politics in a left-right spectrum, but voters don't. Conservatives have spent a lot of time thinking about how to get working class voters to vote against their economic interest, but there is a way to win those voters back with the right message (and ideally a far-better-funded campaign next time!).
  3. The assumptions behind so-called "Strategic Voting" did not prove accurate. "A vote for the NDP is a vote for Hudak" read Liberal leaflets dumped in NDP-Liberal ridings which had no chance of electing a Conservative, but did dump some of the most progressive NDP MPPs in return for a three blue Liberals. But it wasn't NDP-to-Liberal switching that defeated Conservatives, contrary to what some very good pre-election opinion research from Innovative Research suggested might be the case. Indeed in every region where the Ontario PCs lost seats, both the Liberal and NDP vote-counts increased. Perhaps the only seat where an explicit strategic voting campaign helped defeat a PC MPP was in Oshawa, where the Elementary Teachers went to bat for one of their members running for the NDP in concert with other groups, to persuade people that the strategic vote in that case was NOT for the Liberals.

Seven Kinds of Ontario Ridings

Another lesson we can learn from the Ontario election is that the province's electoral map – federally and provincially – is changing. Gone are the days of just blue seats, red seats, and red-blue swing seats. Today different parties are engaged in quite different contests with each other in different parts of the province, and the three traditional categories account for only half the seats in Ontario.

The next table shows the complete federal and provincial electoral history of Ontario's 107 ridings, from the 2004 federal general election through to the 2014 provincial election we just finished (with the 2000 federal transposition added in for context, although with the split between the PCs and Canadian Alliance, the Liberals show up as the winner in all but 2 seats). It then classifies then by historic contest, and whether safe, swing or changing, and shows which seats are now held by different parties at the federal and provincial levels.

[Click on image to open up 2-page PDF]

Ontario's 107 Ridings, by federal and provincial electoral history, 2004-2014<br /><br /><br /><br /><br />

  • There are 14 ridings which have returned a federal or provincial Conservative in every election from 2004 to 2014, four of them with the Liberals always in second place, and 10 with varying competitors over the years
  • There are 11 ridings where a Liberal has always been elected, in six of which the Conservatives always came second, with the other 5 seeing various different competitors
  • There are 3 ridings in which the NDP has always won from 2004 onwards, Toronto-Danforth where the Liberals were always number 2, and Hamilton Centre and Timmins-James Bay where they faced various competitors
  • 17 ridings have mostly featured an NDP-Liberal contest over the years, but in four of which the Conservatives displaced the Liberals federally in 2008 and 2011, and in another seven where they did so in 2011. Eight of the 17 ridings elected a Liberal MPP last Thursday where a New Democrat MP won the last federal election. This will be troubling for the NDP federally.
  • 45 ridings have mostly featured a Conservative-Liberal contest since 2004, but in 13 of them the NDP replaced the Liberals in second place federally in 2011. The Liberals won 6 of those 13 seats provincially last Thursday, and 30/32 of the straight-up blue-red swing seats. This has got to be very worrisome for the federal Conservatives.
  • 1 riding has been a Conservative-NDP contest since 2004: Oshawa.
  • Perhaps the 16 most interesting seats in the province not only defy categorization, but appear to be changing in their allegiances over time — perhaps due to changing demographics, solid candidate search and/or local organization, or some combination of all three. Half of them are held by the NDP provincially, while three-quarters are held by the Conservatives federally.

What's clear is that Conservatives now look beatable in the very parts of Ontario that granted the Prime Minister his majority government in 2011, and even in seats his party won before then. If the Liberals can knock the Conservatives back in the well-educated ridings around Toronto, and the NDP cut further into their seat count in the working class ridings southwest and north, that's the pincer action the Conservatives don't want to face. Not even the new riding boundaries would save them in that situation, so they'll have to run a much more adroit campaign than Tim Hudak did to forestall that possibility in 2015.


For full Ontario provincial election results, you can consult the Ontario Pundits' Guide database, located at Unfortunately, a system "upgrade" at my web-host messed up the charting, but I'm working on a more permanent fix for that.

Rae Attack Ad a Political Red Herring

March 21st, 2012 | 30 Comments

[Welcome, National Newswatch readers!]

The origin of the term "red herring" to describe a diversionary tactic actually originates not with the British dog hunt, but in some political analysis that drew on its imagery.

Pundits have spent the better part of two days now chasing their tails; trying to divine the tactical imperative behind attacking Bob Rae on election day in the Toronto–Danforth by-election that the NDP was heavily favoured to win. Was it intended to muddy the waters on voting day in the by-election, toss a grenade into Liberal ranks just as they were about to lose a by-election in their interim leader's former riding in a campaign that had been in many ways more closely associated with Rae than his candidate, was it a misfire altogether, or do the Conservatives now view the Liberals as more of a threat than the NDP?

Lost in all the combing of entrails about the online ad recalling Mr. Rae's time as the NDP premier of Ontario was the coverage of two simple facts:

  • At 5.3%, the Conservatives' share of the vote in Monday's Toronto–Danforth by-election was amongst the worst ever showings posted by that party since it was formed in 2004, and
  • The gains posted by the Liberals in their vote-share in Danforth Monday came – not out of the NDP share – but at the expense of the Conservatives.

In fact, apart from 17 ridings in 2004, just after the merger of the Canadian Alliance and Progressive Conservative parties, the Conservatives have only obtained a lower vote share on three occasions: Justin Trudeau's Papineau riding in Montréal in 2011, and Gilles Duceppe's old seat of Laurier Ste. Marie in 2008 and 2011.

Meanwhile, though they didn't quite return to 2008 levels, the Liberals posted a roughly 10-point gain in their vote share in Toronto–Danforth on Monday, and in fact raised their raw vote absolutely by just under a thousand votes. But earlier Liberal bleeding to the NDP there was already baked in the cake: the NDP maintained its 2011 vote share of roughly 60%, having gained from the Liberals in 2004, 2008 and 2011.

Not to overread the results of one by-election, but if anything it confirms the least reported aspect of recent national surveys as well: the Conservatives remain in the lead, but have come down significantly from their election high. That support will go different places depending on the province, but strategically the Conservatives cannot afford to have it migrate back to the Liberals in ridings where demographically they are the only alternative to the government, particularly not in the coveted and hard-won 905 area code seats around Toronto which handed them their majority.

Turnout in Toronto–Danforth Monday was actually pretty healthy by recent by-election standards, coming in at 43.4% — the top-end of the recent range of 20-40% over the last few decades. And while that's down from 65% in last May's general election in Danforth, it was higher than Fort McMurray–Athabaska posted last May, with another few ridings coming in only points higher then as well. So, while we ought to apply the prudent proviso that changes in vote-share should not be compared between cases of different turnout, it would be wrong to cast this by-election as well below the norm for turnout.

I don't fully buy the "blue Liberals" explanation about why the final election polls didn't predict the election outcome, but it certainly did account for the key vote shifts in the ridings located just outside Toronto. And if people were picking up concern about Bob Rae at the doorstep back then as a reason for that shift, it makes perfect sense for the Conservatives to raise the spectre of Rae's record again now.

The case at hand and the general pattern it suggests also raise questions about the strategic rationale for the so-called Cullen Plan of joint nominations. (Not that he intended it to apply to ridings not held by the Conservatives such as Toronto–Danforth, but no-one has successfully argued why voter migration patterns would be so different in those ridings than others.)

The question is this: if the presence of a Liberal candidate in a given riding could be counted on to cut into the Conservative vote, why would the NDP want to take them out of the mix? Across the prairies, one of the major reasons the Conservatives have consolidated their hold on a large number of ridings is because outside the few remaining Liberal incumbencies, the party's vote share had dropped to low single digits nearly everywhere. When the NDP last won prairie seats in any number, they had similar vote shares to the 2011 election, but could count on the Liberals taking a chunk out of what is now the Conservative vote there. Same goes for seats in the interior of B.C.

Taking the Liberals out of the equation could make for a higher gap for the NDP to make up in reaching the Conservatives if the government become unpopular again. If the Prime Minister has a continued strategic reason for wishing the demise of Bob Rae's new party, that's what it is.

Meanwhile, Cullen's plan would see the NDP abandon seats such as the 905 to the Liberals, but with no guarantee that the Liberal MPs thus elected would support an NDP minority or coalition government, rather than a Conservative one. For all the power of the "cooperation" mantra (and it's clearly significant at this moment amongst Cullen's very committed supporters), none of the plan's proponents have ever addressed that strategic point. Being from British Columbia, where the NDP wins government provincially only when the more right-leaning parties split the vote, one would think Cullen was acutely aware of the dynamic.

Pros and Cons of the Cullen Plan: a Sceptic’s Guide to Electoral Coalitions in Canada

December 24th, 2011 | 31 Comments

[Welcome, National Newswatch readers!]

Four-term Skeena M.P. Nathan Cullen launched a bit of a hand grenade into the NDP Leadership race not long after he entered it, with his proposal to allow local riding associations to endorse joint nomination meetings in Conservative-held ridings.

Citing the absence of majority support for the Conservative Party's majority government, his campaign's news release quoted him as follows:

Cullen said he does not a support a full merger. But also said that he senses a moment where people from across party lines are open to new approaches to defeat Harper, and asked people who feel that way for their support. He also said his own experiences and those of Parliaments that featured co-operation convince him that working with others achieves more.

“I am convinced that there is a hunger in Canada, far removed from Ottawa, among people who pitch in every day to make things better—and wonder why the political establishment can’t,” he said. “I think we can, and that Quebec’s new openness to progressive federalism makes now the time.”

Cullen asked for a mandate to negotiate an arrangement whereby local riding associations of "progressive federalist parties" (i.e., the Liberals and Greens) could be given the option of deciding to hold joint nomination meetings for the purpose of unseating a local Conservative M.P.

As further explained in this FAQ posted to his website a month later, each riding association would duly nominate a candidate, and then the three parties' candidates would enter a run-off amongst the membership of all three riding associations. The winning candidate would run under the banner and platform of their own party, and the others would stand down.

The idea is based on the assumption that if you could take the last election's results as a starting point, it should be possible by reducing competition between the NDP, Liberals and Greens to scrape together enough votes to defeat the Conservative incumbent. Do it in enough ridings and you could defeat the Conservative government. On several of the occasions where I've seen Cullen discuss it further, he has suggested that the arrangement would only be required for one election, in order to achieve a referendum asking Canadians which form of electoral reform they wanted, after which new arrangements would apply in any event.

Cullen has not received the backing of any other leadership candidate for his proposal, and for many in the NDP it was an instant deal-breaker, although he received the blessing of national press gallery for raising the idea, notably a favourable profile from Maclean's columnist Paul Wells. It also received the endorsement of Project Democracy (why am I not surprised).

Given that I've become the designated hitter against strategic voting, however, I've been asked more than once recently to assess the merits or demerits of the idea. Well, here goes. But, first some background.

How are candidates currently selected?

The provisions governing the selection of candidates are laid out in Part 6 (ss. 66 to 92.6) of the Elections Act. Basically, anyone who is not an electoral officer, prisoner, provincial or territorial elected representative, is not ineligible to vote, and, if they had ever run federally before, did not fall behind on their election filings, can be nominated as a candidate, so long as they file the proper nomination forms, have an official agent, enough signatures from eligible electors in that riding, and can cover the deposit.

Candidates who want to be endorsed by a registered political party, and have that party's name next to theirs on the ballot, also have to present a letter signed by that party's leader or their designate along with their nomination papers (s.67(4)(c) of the Act). And a political party can only endorse one candidate at a time in each riding during an election writ period (s.68).

In order to gain the right to have that endorsement, candidates must follow the rules set out by the party to obtain the endorsement, which typically include a vetting process, perhaps some other eligibility requirements, and then either winning a contested nomination amongst the membership in good standing of that party's riding association who live in the riding, or else being acclaimed by that body, or else receiving an appointment by the party leader.

If there is a nomination meeting, the riding association has to file a report of its results to Elections Canada (the reports are available in a searchable database on their website), and the candidates have to file a financial return for their nomination campaign, and keep their spending under a defined expense ceiling.

Candidates who are duly nominated by their parties, as with independents, all have to be formally nominated under the Elections Act with the local Returning Officer in that riding as well, with all requirements met and the nomination accepted before 2:00 PM local time, on the Monday three weeks before voting day. A candidate can withdraw by 5:00 PM on nomination, but may not be replaced by another candidate from that party after 2:00 o'clock. Any candidate who withdraws after 5:00 PM remains on the ballot, regardless of whether they are campaigning or not.

Issues with the Cullen Plan

a) Party Governance – Cullen stresses that it would be up to the discretion of the local riding associations of the various parties to agree to this proposal. That might be true, but it would also require the consent of the parties' national offices – and more than one party has a convention resolution or even constitutional provision on the books, stating that it will run a full slate of candidates. Also, don't forget that s.366(2)(j) of the Elections Act states that, in order to be eligible for registration as a party under the Act:

… one of the party's fundamental purposes is to participate in public affairs by endorsing one or more of its members as candidates and supporting their election

Suppose a local riding association decides not to field a candidate, or to endorse a candidate from another party. That decision could be overruled by headquarters, by having the leader appoint another candidate there instead. The party could even deregister the riding association (under s.403.2(2), though not during an election writ period), and then set up a new one to replace it.

Here's another problem: we often use the term "riding association" to mean two different things. Sometimes we mean all the party members who live in that riding and/or are members of that riding association (only members actually living in the riding may vote on who that riding's candidate should be). Other times we mean the members of the riding association executive or board of directors.

But which body should have the right to enter into such an agreement with other parties and/or their riding associations? The riding board, or the riding membership? That would be governed by the riding association's constitution and by-laws, and presumably those of the party as well.

What if all the parties and all their governing bodies agreed on the general principle, but disagreed as to whether the tactic would work in a given riding?

b) Due Process During the Run-off Contest – OK, suppose you have the riding association members, executives/boards, and the party headquarters all on board to run a two-step joint nomination process, and suppose everyone trusts everyone else not to game the system, and to gracefully accept the outcome.

Now, how do you ensure it's a fair contest? Let's say we accept that each party's run-off candidate will have already passed that party's vetting and eligibility requirements, a membership eligibility cut-off date for that party's nomination process would have been set, the membership list was agreed upon, the various candidates for the nomination within each party gained access to that list, they conducted a campaign, identified their supporters, and got them out to the first meeting to vote.

Now you have three duly-selected candidates, one from each of the 3 so-called progressive federalist parties, entering into a new race. Who sets the rules for that race? Who governs it? Do each of the riding associations give their membership lists to the other parties' candidates so that they can conduct a proper campaign? (ding, ding, ding: big privacy bells should be going off here) Were their membership eligibility cut-off dates co-ordinated with one another? Or would new membership sales be allowed? If a person was a member of two different parties, could he or she vote twice? Would it be fair if the membership fee in each party was different? Is there a spending limit? What about a requirement to file a meeting report, or candidate financial returns with Elections Canada?

Here's still another problem: given that we will have a redistribution between now and the next election, and that the boundaries won't be known until 12-18 months before the next election, many riding associations will have to be re-organized along the new boundaries, and thus will be undergoing a lot of governance changes themselves during the same period nominations are to take place. This adds an extra layer of uncertainly to the whole biznak.

c) Acceptance of the Outcome of a Joint Nomination – The problems aren't over, because people who say they will accept the outcome, often don't afterwards. Now you have to trust that all the following conditions will hold true:

  • that none of the candidates who had already waged nomination contests to win their own party's entry into the joint nomination race decide after losing the joint nomination to either (i) appeal to the party to appoint them as a candidate regardless, (ii) run for another party that was not part of the joint nomination process (hello Canadian Action Party, Unity Party, Progressive Canadian Party, etc., etc.), or (iii) run as a serious independent
  • that none of the party leaderships decide they can't live with the outcome, and appoint another candidate instead
  • that the party whose candidate won the joint nomination can properly fund that campaign, because under s.404.2(2.2) of the Elections Act only that party's riding association or headquarters could transfer funds into that candidate's campaign, not other parties or riding associations. And no riding association may incur advertising expenses during the writ period, only candidates or national parties. Presumably in the pre-writ period, other participating parties' riding associations could spend some money advertising their support of the outcome.

Now, certain local kamikaze candidates could go rogue, I suppose, by accepting their party's endorsement and getting duly nominated with the Returning Office, but then withdrawing from the ballot after the deadline for the party to replace them, if they really wanted to support a common candidate but couldn't get their own party headquarter's buy-in. I can't imagine that kind of tactic would be well-received by the electorate, but I suppose anything's possible, given the right set of circumstances.

d) Public Acceptance of the Process – The parties would also have to assure themselves that the local electorate would be accepting of such a proposal. Electors have the right to vote, but it's political parties who decide on who the available choices will be. It's a big leap to assume that the public will be accepting of a process run by a tiny proportion of the whole riding population to conspire to eliminate certain choices from the ballot in the hopes of torquing the election outcome. It's another big leap to assume that this would be done in a vacuum, given that the Conservative Party would be following along closely, and reserving all their strategic and tactical options.

Of course, for the NDP and the other parties, there are other strategic and tactical considerations, such as:

  • would it actually even work, or would enough orphaned voters switch to the Conservatives or stay home if their preferred choice was not on the ballot?
  • is it democratic or politically wise to be advocating the elimination of political choice for the public by a small group of political activists
  • in particular could the NDP then run as effectively as one of the two main choices in the "consideration set" of the general election (a term coined by Innovative Research's Greg Lyle to describe the process by which consumer choice is whittled down to two viable choices, and drawing parallels to voting behaviour), if it was at the same time enabling the election of one of its non-"consideration set" competitors.
  • tactically, what criteria would be applied for deciding whether a joint nomination would achieve the intended result or not in a riding with newly formed boundaries
  • what is the cost-benefit analysis of foregoing the opportunity to build in that riding next time, foregoing the room under the national spending ceiling (which is affected by how many candidates a party runs in an election), and foregoing the rebate, organizational opportunities and other team-building in a riding association that result from waging a local campaign, even if it doesn't win, as against the probability of being able to defeat the incumbent in a riding


I believe there is already so much evidence that these tactics don't work. We had four either purposeful or accidental cases of ridings without full slates of candidates in the 2008 election (Central Nova, Cumberland-Colchester-Musquodoboit Valley, Portneuf-Jacques Cartier, and Saanich-Gulf Islands), to no effect whatsoever. Meanwhile Elizabeth May, for example, was finally able to win a seat, in a situation without the massive "Shun Lunn" push for joint nominations in that same riding the time before.

At the best of times, party votes from the previous election are not additive if one party decides not to contest a riding in the next one. But that would be doubly so in the case of the Liberal vote in a riding where one of their longstanding incumbents got defeated in 2011. The experience with incumbents being defeated in 2008 and the later impact on party vote in 2011 was that the vote-share obtained in defeat was often a ceiling for next time, rather than a floor vote.

But anyway I just find the whole thing unnecessarily undemocratic. We already have a process by which each party selects a candidate and then those candidates engage in a run-off. It's called an election. Why would parties want to expend so much energy on an elite-driven process (yes, riding association executives and memberships are a small elite, relative to the entire electorate in that ridinng), when they could be talking about the issues they think are important to take to Ottawa, instead.

To the extent that there are any points in its favour, I can think of one situation in the U.K. where it did work. After the "Cash-for-questions" scandal, one of the implicated Conservative M.P.s (Neil Hamilton) was renominated over the howls of even part of the constituency association, and former BBC foreign correspondent Martin Bell decided to run as an Independent "anti-sleaze" candidate. Tony Blair's legendary aide Alisdair Campbell managed to get the Liberals on board with the Labour Party in a decision not to run against Hamiton and Bell, giving the latter a free run at the race. Hamilton's earlier 16,000 margin was massively turned around on election day, and Bell was elected for a term as M.P. But this was against an unpopular incumbent, and with both other parties agreeing to support an independent candidate, rather than one of their competitors'.

The other possibly good thing about the proposal from the NDP's perspective, is that it puts the party's nearer competitors on the defensive rhetorically, for a change. Now they have to justify why they wouldn't co-operate with the proposal or with other political parties, while the NDP benefits from the fact that they extended the olive branch in the first place.

To me the few cases in which the proposal would work, and the less-than-democratic way that it would have to function, weaken the otherwise conciliatory aspect of this approach. It will be interesting to see if this one proposal continues to be a lightning rod for Mr. Cullen's campaign.



For the latest on the NDP Leadership Race, don't forget to follow the half-hourly news updates, and social media tickers at the Pundits' Guide NDPLdr portal page:

Strategic Failure

May 14th, 2011 | 94 Comments

[Welcome, National Newswatch readers!]

As the designated curmudgeon against Strategic Voting in numerous news stories throughout this election, readers might not be surprised to see me comment on its poor success rate afterwards.

I fully expected to do so, noting the disappointment of the various groups at their incorrect calls and the evident reelection of the Conservative government with a majority mandate.

However, now I'm being told that the groups view their efforts as being a success, and that they're planning to repeat them all over ahead next time. Oh brother. I think we need to be really clear on their record in that case.

The problem with the strategic voting websites is that their electoral analysis was incompetent and utterly wrong in most of the ridings where it could be said to have mattered — leading to incorrect recommendations in many cases where it would have made a difference, and no recommendations in others that were overlooked.

Largely the reviews of the sites' performance have concentrated on whether they made the right calls or not in their "key ridings" (an example from the "Data Journalism" blog expanded that a bit; here's another from Erin Weir at, UPDATE: and one I missed earlier from the Jurist at Accidental Deliberations).

But the issue also needs to be raised about their calls in the ridings that wound up being key, but which they missed identifying; or in the ridings where they declared the seats to be safe that actually turned over.

Project Democracy – Critical Errors

by Alice Funke and Stephen Yardy

Here is the project's self-description of its purpose:

About Project Democracy

Project Democracy is an information resource for progressive Canadian voters looking to topple the Harper Conservatives by harnessing the votes of the majority of Canadians. In Canada we don’t have an electoral system that directly reflects the will of Canadians. With our 'first-past-the-post' electoral system, vote splitting could mean Canadians end up with a majority Harper Government with as little as 35 percent of the popular vote.  This is not the outcome most Canadians want, and avoiding it requires voter knowledge and cooperation. Project Democracy provides that knowledge through careful use of available polling data and analysis of local “on the ground” riding information. If progressive Canadians vote smart, the majority of Canadians will get a government that better reflects their values.

10 Missed Calls –> 8 Missed Opportunities

Because the site's criterion for picking ridings was to look at previous results, ridings which were not previously close were overlooked for inclusion in the project's "Key Contests". 10 previously non-close ridings were called completely wrong, leading to 8 missed opportunities to try and prevent a Conservative candidate from winning. Emphasis is ours.

Riding Expected Winner Recommendation Actual Outcome
 Their Rationale & Discussion
Etobicoke Centre Lib Vote your preference Cons-Lib, by 25 votes
This has been a safe Liberal seat, and so far the Conservatives are significantly behind in second place.  Unless the polls indicate any major change you should be safe to vote your preference.
Elmwood–Transcona NDP Vote your preference Cons-NDP, by 0.9%
Jim Maloway now looks safely on his way to defeat his Conservative rival. Vote your preference.
Mississauga East–Cooksville Lib Vote your preference Cons-Lib, by 1.4%
This is a safe Liberal seat. Vote your preference.
Willowdale Lib Vote your preference Cons-Lib, by 1.7%
This is a safe Liberal seat. Vote your preference.
Don Valley East Lib Vote your preference Cons-Lib, by 2.2%
It looks like Liberal MP Yasmin Ratansi can fend off her 2nd place Conservative challenger. Vote your preference.
Pickering–Scarborough East Lib Vote your preference Cons-Lib, by 2.5%
Liberal MP Dan McTeague has made it his mission to try to keep those oil companies honest on price gauging, and it has made him very popular with his constituents and he looks to be on track to win this riding again. This appears to be a safe Liberal seat. Vote your preference.
Scarborough Centre Lib Vote your preference Cons-Lib, by 3.9%
The Conservatives do not present a real threat. Vote your preference.
Mississauga–Brampton South Lib Vote your preference Cons-Lib, by 9.6%
This appears to be a safe Liberal seat though the Conservatives have been working hard in multicultural Brampton over the last couple of years. It appears safe vote your preference.
Scarborough–Rouge River Lib Vote your preference NDP-Cons, by 10.7%
This has been a safe Liberal seat and remains so despite the departure of Derek Lee. Vote your preference.
Pierrefonds–Dollard Lib Vote your preference NDP-Lib, by 3.7%
This riding is generally a secure Liberal seat, but it is important to note that the Conservatives came in second in 2006 and 2008. Vote your preference, but keep an eye on the Conservatives.

12 Wrong Calls –> 11 Counterproductive Recommendations

In twelve of its "Key Contests", Project Democracy made completely wrong calls.

Note how they insinuated in several cases that their calls were supported by what "the latest polls say". In fact no local polls were conducted in any of these cases.

If enough people followed the Project Democracy recommendations, it would have undermined the party that actually had the better chance of beating the Conservative candidate. Also, take note of some of the overheated rhetoric that proved to be utterly, utterly wrong (esp. in Saint John and Fredericton).

Riding Expected Winner Recommendation Actual Outcome
Their Rationale & Discussion
Montmagny–L'Islet–Kamouraska–Rivière-du-Loup TCTC BQ NDP-Cons, by 5 votes
Former Bloc MP Paul Créte won this riding convincingly in 2008, but when he gave up his seat to run in provincial politics in 2009, the Conservatives won by a little more than 1,000 votes. If you look at the "Best regional polling Apr 28, it would seem like the Bloc is in a better position to take back this riding from the Conservatives.
Bramalea–Gore–Malton TCTC Liberal Party Cons-NDP, by 539 votes
Liberal incumbent Gurbax Malhi is in for a tough fight.  The NDP and Greens are not competitive in this riding, we recommend progressives back Liberal MP Gurbax Mahli.
Scarborough Southwest Lib Liberal Party NDP-Cons, by 3.2%
This has appeared to be a safe Liberal seat but the race appears to be tightening. Check back on Sunday.
Saint John TCTC Liberal Party Cons-NDP, by 19.1%
Three-term MP Paul Zed lost by under 500 votes in 2008, a classic case of vote-splitting electing a Conservative. A new Liberal candidate is running, which will give the incumbent an edge along with vote-splitting. Progressive voters can pull together to foil Stephen Harper in Saint John by supporting Liberal Stephen Chase.
Brant TCTC Liberal Party Cons-NDP, by 20.4%
This is a tight Liberal/Conservative race. Former MP Lloyd St. Amand is attempting to take this riding back from Conservative Phil McColeman. The NDP are a distant third in this riding despite drawing 9,000 votes last election. This is a classic split vote situation. The Liberal candidate has the best chance of taking this seat away from the Conservatives.
West Vancouver–Sunshine Coast–Sea to Sky Country TCTC Liberal Party Cons-NDP, by 21.9%
This is a straight up battle between the Liberals and Conservatives. Even with former Liberal Blair Wilson incumbent running as a Green last election they were not competitive. This election the race will be closer and it is possible for Liberal Dan Veniez to defeat Conservative John Weston. A correction for Blair Wilson's absence as a Green incumbent has been added. Progressive voters should come together and vote Liberal in this riding to unseat the Conservative in a vulnerable swing riding.
Edmonton Centre TCTC Liberal Party Cons-NDP, by 22.7%
Longtime Liberal incumbent Anne McLellan lost this seat by 6 percent to the Conservatives in 2006. The Liberals were uncompetitive in 2008 with Conservative Laurie Hawn garnering almost 50 percent of the vote. If progressives avoid vote-splitting, latest polls show the Liberals have a hope of regaining this seat. We recommend voting for Liberal Mary MacDonald.
Fredericton TCTC Liberal Party Cons-NDP, by 24.5%
The Liberals lost this seat in 2008. The new Liberal candidate is Randy McKeen, a former local radio host. The Liberals have the only shot of beating the Conservatives.  The stakes could not be higher — there is a real opportunity to eliminate a seat from Harper's quest for a majority, and a vote for the Greens or the NDP will not be effective.  We strongly recommend supporting Liberal Randy McKeen. 
Chatham-Kent–Essex Cons No call Cons-NDP, by 27.4%
It looks like Liberal Matt Daudlin has the best chance of ousting Conservative Dave Van Kesteren. The NDP have fallen off in this riding to a more distant third. We will not pick at this point as the riding looks out of reach.
Miramichi TCTC Liberal Party Cons-NDP, by 29.3%
Five-term Liberal Charles Hubbard lost in 2008 by about 1,400 votes, a classic case of vote-splitting electing a Conservative. The Liberals have a new candidate, Keith Vickers, and they appear to have the only shot at defeating the Conservatives and denying Harper his coveted majority. We recommend voting for Keith Vickers.
Huron–Bruce TCTC Liberal Party Cons-NDP, by 29.6%
This was a Conservative pick-up from the Liberals last election with the New Democrats running a distant third. The Conservatives won this riding easily in 2008. It will be tough to win it back, however Liberal Charlie Bagnato is the candidate best positioned to make a run.
Simcoe–Grey TCTC Liberal Party Cons-NDP, by 32%
This would normally be a safe Conservative seat. However Helena Guergis is running as an independent. It is possible that the Conservative vote will split between Guergis and Leitch. The NDP and Green have 11,000 votes between them and no hope of winning. The Liberals got over 12,000 votes last time so Alexander Smardenka could run up the middle and win. With collective voting and a Conservative split vote this is now a good place to defeat the Conservatives.

Underestimating the Conservatives – Overestimating the Liberals (7 cases)

Obviously some of these ridings wound up being closer than others. The fact is that none of them were flagged, either because the Conservatives were underestimated, or the Liberals overestimated. The analytic errors would be more forgiveable, except that they were usually stated with such certainty.

Riding Expected Winner Recommendation Actual Outcome
Their Rationale & Discussion
Vancouver Centre Opp Vote your preference Lib-NDP-Cons, by
5% between 1st & 3rd
The Conservatives don't appear to be a threat in this riding.  Vote your preference. 
Hamilton East–Stoney Creek NDP Vote your preference NDP-Cons, by 9%
Incumbent NDP candidate Wayne Marston won easily. This now looks like a safe NDP seat with the Conservatives too far back to be a factor. Vote your preference.
Hamilton Mountain NDP Vote your preference NDP-Cons, by 14.1%
The Conservatives are far back in this riding which now seems to be a safe NDP riding. Vote your preference.
Algoma–Manitoulin–Kapuskasing Liberal/NDP Vote your preference NDP-Cons, by 21.4%
This is a tight Liberal/NDP race where it doesn't appear that the Tories have a chance of winning. The only question we have is the effect of the gun registry issue. We suspect it will not be enough to make the Conservatives competitive, but we'll keep close watch.
Thunder Bay–Rainy River Liberal/NDP Vote your preference NDP-Cons, by 21.5%
This is an NDP Liberal race. Vote your preference. 
Sudbury Liberal/NDP Vote your preference NDP-Cons, by 21.6%
This is a Liberal-NDP race as always, but was not without a significant Conservative percentage in 2008. Vote your preference.
Vancouver Kingsway Liberal/NDP Vote your preference NDP-Cons, by 22%
Liberal/Conservative David Emerson is not running and the general Liberal upheaval in 2008 has begun a new ball game in this riding. NDP Don Davies won with almost 3,000 votes ahead of the Liberals, but the Conservatives were close behind. The most recent polling indicates this is a NDP-Liberal race, with the Conservatives well behind, so vote your preference. Please sign up to our list and/or check back again as we approach election day.

Overestimating the Conservatives and Bloc – Underestimating the NDP (3 cases)

Well, no-one can fault anyone for not being able to call Quebec, but these examples show how even local polls could be completely wrong even two weeks out from Election Day.

Riding Expected Winner Recommendation Actual Outcome
Their Rationale & Discussion
Chicoutimi–Le Fjord BQ BQ NDP-BQ
In the last election, Bloc candidate won by only 3000 votes.  A Segma Recherche poll for the Courrier du Saguenay released this week-end shows the Bloc in the lead with Conservatives 2nd and NDP 3rd. We recommend voting for the Bloc candidate to ensure this riding doesn't go to Harper.
Québec BQ Vote your preference NDP-BQ
The Bloc won this riding by 11,000 votes in the last election, but it has been close before. Check back in, but unless there is a dramatic change, vote your preference.
Richmond–Arthabaska BQ Vote your preference BQ-NDP
This riding is now a safe Bloc riding. A Cara Telecom poll for the Nouvelle Union released this week-end shows the Bloc in the leading way ahead of the Conservatives. You can be comfortable voting your preference.

Attempted Corrections (6 cases)

In some cases (I believe such as Essex, though I can't confirm through the Twitter feed anymore), the site's methodology managed to eventually get the answer right, even if only at the end of the campaign (though who knows how much confusion the earlier, erroneous information sowed).

In other cases, readers raised complaints about the site's faulty calls, leading to some ridings being selected for riding-level polling by the project's pollster, Oracle Poll (also the pollster for Elizabeth May's Green Party campaign in Saanich-Gulf Islands).

In the 4 Quebec ridings selected, the riding polls confirmed that the NDP candidates were all in first place, or best positioned to challenge for it. In the BC riding of Esquimalt-Juan de Fuca, the poll confirmed that it was a Conservative-NDP race, and that in spite of retiring MP Keith Martin having been a Liberal, the Liberal candidate was far behind the pack.

However, some really ill-informed decisions were made about the other ridings. For one thing, most pollsters will tell you that it's extremely difficult to draw a reliable sample in the northern and remote parts of the country, so that polling in Nunavut, or Desnethe-Missinippi-Churchill River was close to utterly pointless, and money down the drain. The Nunavut poll reported that Conservative MP Leona Aglukkaq had 70% of the vote (she got 50%).

The DMCR poll found that Conservative MP Rob Clarke was 22.2 points up on his NDP opponent, former FSIN vice-chief Lawrence Joseph. I remember seeing those results and thinking: wow, the NDP has 30% excluding the reserves? That seat is in play! Project Democracy's conclusion: that the Conservative was so far in the lead, no recommendation was worth making. The two wound up less than 800 votes (some 3.6%) apart.

The Saskatoon-Humboldt poll did not test for the one condition that had caused people to believe the poll might be required: namely the possibility that former MP-turned Independent candidate Jim Pankiw could have an impact on the result. The last time he ran as an Independent, the riding was a four-way race. While Pankiw did not wind up scoring a significant tally on Election Day this time, his exclusion from their poll meant that informed locals could not have trusted the results.

In other cases, ridings did not make the cutoff for polling, but had previously been listed with the wrong recommendation for weeks.

Riding Expected Winner Recommendation Actual Outcome
Rationale / Discussion
Essex TCTC NDP, changed from Liberal? Cons-NDP, by 12.9%
In this last weekend it appears the NDP candidate Taras Natyshak has pulled into 2nd and has a good shot at defeating the Conservative with a little help from Liberal and Green voters.
Saskatoon – Humboldt TCTC NDP, changed from Liberal NDP-Cons, by 17.7%
Given local considerations that could potentially impact the race in this riding, we have invested (with your generous financial support) in local polling to help determin the party/candidate best positioned to defeat the Conservative candidate. The poll did not test support for Independent candidate Jim Pankiw who drew 7,000 votes when he last ran as an independent in 2004, but shows the opening for the NDP who are running a clear 2nd to Conservative Brad Trost. We recommend voting for NDP Deise Kouri.
Esquimalt – Juan de Fuca TCTC NDP, changed from Liberal NDP-Cons, by 406 votes
Given local considerations that could potentially impact the race in this riding, we have invested (with your generous financial support) in local polling to help us identify the party/candidate best positioned to defeat the Conservative candidate. The results of the poll are now reflected in the chart to the left. We recommend voting for NDP candidate Randall Garrison. Please read the News section.
Desnethé–Missinippi–Churchill River Cons Vote your pref, changed from Lib Cons-NDP, by 3.6%
Given local considerations that could potentially impact the race in this riding, we have invested (with your generous financial support) in local polling to help determine the party/candidate best positioned to defeat the Conservative candidate. Results show the Conservatives with a strong lead — vote your preference.
Kenora No Call No Call, changed from Liberal Cons-NDP, by 18.7%
This is a rematch between Liberal Roger Valley, NDP Tania Cameron and Conservative Greg Rickford. Last time Conservative Rickford ran right up the middle and defeated Valley. The Greens received 1,000 votes, the NDP's Cameron just over 5,000 votes and was 2,000 back from the Liberal. The Conservatives won by just 2,000 votes. This was three-way vote splitting pure and simple that elected the Conservative. It would be tragic to repeat this situation when a Conservatives can be un-seated. Until this week the indications were that Liberal Roger Valley has the best chance. This week's regional polling results have moved the NDP upwards, making this race is too close to call. Unfortunately this riding was not in our roster for local polling. We are not making a pick for now and will be looking for any/all means to decipher whether a viable call is possible over the next few days to avoid a Tory win.
Abitibi–Baie-James–Nunavik–Eeyou NDP No Call, changed from BQ NDP-Cons, by 22%
We don't feel comfortable making a call in this riding based on the model results.

Didn't meet criteria (1 case)

This riding changed hands by nearly 10 points last time. By the site's own methodology there was no chance of really defeating Conservative Environment Minister Peter Kent in the current election either.

He was included on a policy basis, but no similar policy screen was applied against any of the other candidates recommended by the site. Nor were recommended Liberals asked whether they would have supported an NDP minority government over a Harper Conservative minority government.

Riding Expected Winner Recommendation Actual Outcome
Rationale / Discussion
Thornhill Cons Liberal Party Cons-Lib
Peter Kent is perhaps the worst Environment Minister in a long line of terrible Environment Ministers. The Liberals have Mock the best chance to send Kent packing if progressive voters cooperate. We recommend voting for Liberal Karen Mock.

There were another 74 cases of "Key Contests" where the "correct" call was made, 36 of them settled by less than 10%, 14 settled by between 10%-15%, and 25 of them settled by more than 15%. Of those:

  • 52 were won by the Conservatives (33 holds, and 19 gains), but where Project Democracy correctly recommended the party which placed second,
  • 22 were won by parties other than the Conservatives (7 Liberal holds, 7 NDP holds, and 7 NDP gains from Conservatives or André Arthur, plus 1 Green gain from the Conservatives)

The sponsoring group has yet to publish a full post-mortem, and has already pulled down its Twitter account which made all the recommendations in a repeating cycle. Founder Alice Klein did post an immediate debrief at Now Magazine online, where she claimed the group had "a success rate of 87 per cent on 84 picks", and promised a more complete analysis later on. Its Facebook page continues, posting items relating to electoral reform.

So, Project Democracy proved unable to rally sufficient voters to prevent 19 Conservative gains, completely missed the opportunity to even recommend strategic votes in 10 other cases where the Conservatives gained a seat, and got the recommendation right in just 8 cases where Conservatives were actually defeated, leaving 33 targeted Conservative MPs in place.

And the Conservatives were reelected with a majority government. No wonder they loved Project Democracy.

Catch 22 – A Fishnet Full of Holes

by Alice Funke and Stephen Yardy

This analysis will go a bit lighter on the Catch-22 campaign, which at least published a very frank post-mortem immediately after the election admitting that "Results did not reach the goal".

They claim to have defeated 4 Conservatives, and defended 11 opposition ridings, and so give themselves a success rate of 15 out of 60 ridings.

However, they only recommended a candidate against 2 of the 4 Conservative MPs defeated, rating the other two (Beauport-Limoilou and Pontiac) "too close to call".

They did concede that in 8 of their recommended picks, the candidate they endorsed actually came third. In the 34 other cases where they made a pick, their pick placed second, and Essex, Kenora and Nunavut (which were all deemed "too close to call") were all held by the Conservatives.

Here is the Catch-22 campaign's strategy and purpose, in its own words:

What is Catch 22's campaign strategy? (Under review)

Our strategy is to work in a limited number of ridings (about 30 to 40) where the Conservatives are vulnerable enough to be defeated. As the campaign progresses, we will continuously assess the relative strength of the opposition candidates in our target ridings. Once an election is called, we will ask voters in each target riding to support the candidate (Liberal, New Democrat, Bloc or Green) that we believe has the best chance of defeating the Conservative incumbent.

A Catch 22 Campaign Team, headed by a local coordinator, will be set up to work in each targeted riding. Individuals and organizations will be recruited in each riding to assist with our campaign. Local volunteers will distribute Catch 22 brochures in their riding that (a) explain why we feel the Conservative incumbent should be defeated, (b) identify the candidate with the best chance of defeating the Conservative incumbent, and (c) advocate for proportional representation.

How will Catch 22 decide which candidate to endorse?

Catch 22 is party-neutral in that we do not seek to favour one particular opposition party. Our target ridings were chosen in order to maximize the effectiveness of our campaign (i.e. we are targeting the most vulnerable Conservative incumbents). In each target riding, we will endorse the opposition candidate with the best chance to win. As such, our endorsements simply reflect each parties' chances of winning the target ridings.

While our central campaign has identified winnable ridings, the final decision for endorsements rests with our local Catch 22 groups. Endorsements will only be made after all the opposition parties have completed their candidate nominations. There will be a discussion period followed by a decision. We hope to reach consensus around our candidate picks. We want to ensure that everyone has their say on this important matter.

In reality, they made a recommendation based on the riding's projection at (whose own gloomy post-election post-mortem is found here), and then partisans joined the local riding groups and argued for their candidates unless or until something looking like solid data came along. As co-founder Gary Shaul told the Toronto Star, they used "common sense and historical results". Neither of which was worth much, in the end.

In General

by Alice Funke

The maddening thing about the seat projection sites and strategic voting campaigns, from the perspective of party riding campaigns, is that it was nearly all they wound up hearing about at the doorstep if they were one of the targeted seats, and the strategic voting canvasses also sucked up a lot of volunteer resources for fool's errands, taking them away from the actual candidates' campaigns.

Local newspapers routinely reported ThreeHundredEight's projections for individual ridings as "polls", so you'd get a story saying "the NDP has moved ahead of Peter MacKay in a poll" when no poll had been conducted, and MacKay wound up winning handily.

The strategic voting campaigns also printed material that was distributed in the ridings (shown is an example from Edmonton Centre), leafleted people on their way into the subway (seen here), and conducted phone blasts in the days leading up to Election Day urging people to vote for the endorsed candidate as "the only way to stop Harper" (this happened in Brant), which detracted from the campaigns' ability to get their own vote out, as they parried numerous calls from confused voters.

[Click on images to open full-sized versions.]

Project Democracy Flyer for Edmonton Centre, 2011 General Election, front

Project Democracy Flyer for Edmonton Centre, 2011 General Election, back

Note that the Edmonton Centre "projections" were made on the basis of a Nanos Poll from April 14 (the flyer was photographed on April 24). Nanos includes Alberta, Saskatchewan and Manitoba together for regional sub-samples, which have a very high margin of error on their own. Nik Nanos has repeatedly called riding-level seat projections "a photocopy of a photocopy of a photocopy", and would certainly not have endorsed this use of his work. The actual election results in Edmonton Centre were Cons-48%, NDP-25%, Lib-22%. The projection was completely wrong.

The Canadian Auto Workers union (CAW) used a hybrid of Catch-22 and Project Democracy recommendations, leaving it in the position of endorsing a number of anti-union business Liberals whose support for an NDP minority government rather than a Conservative one might not have been assured, and who had voted against key demands of the house of labour (anti-scab laws, stopping free trade with Colombia, I could go on).

"Does Strategic Voting Work?", the Catch-22 campaign rhetorically asked in its FAQ. The only cited Canadian examples were 1999 provincially in Ontario where it *nearly* worked, Edmonton-Strathcona in 2008, and the Anybody But Conservative campaign in Newfoundland & Labrador, which probably didn't affect any more seats than the three on the Avalon peninsula.

I think it's time to say that these projects are not politically sophisticated enough to get their calls correct, and while they get a lot of people engaged in our democracy, which I can't ever be opposed to, they do so under false pretenses: namely that you can know the outcome in a riding ahead of time, and game the system to your own ends.

The record of the two main strategic voting campaigns in 2011 proves that you cannot.

Now, someone could easily review my list of guesses in the "Roadmap to May 2" series of blogposts, where I tried to handicap the races riding by riding across most of the country, and rate me. I guessed the contest in most cases, but not all, and some seats that I called one way turned out quite differently. My own private seat projection was low on the Conservative side, correct enough for the NDP (but without calling the number of Quebec seats properly, which means it doesn't count), and being too generous to the Liberals in Ontario and the Bloc in Quebec.

But here's the thing: I never used any of it to tell anyone else how to vote.

Your vote is your opportunity to express your point of view. Don't let others try and divert it for their purposes, and don't believe strategic voting snake oil salesmen and soothsayers. They actually have no idea which way a riding is going to go, nor what is the best way to vote to stop that.

If you vote in favour of something or someone, you will never have to second-guess whether the choice was correct. And you may find out that many others had actually agreed with you all along. And, at least the politicians will have received the correct mandate.

Thus ends my rant on strategic voting … at least for the current election.


Thanks to Stephen Yardy for collecting all the data that made this post possible.

Why the Conservatives Love the “Strategic” Voting Sites

April 18th, 2011 | 63 Comments

[Welcome, National Newswatch readers!]

Every election campaign lately seems to mark the return of so-called "strategic voting" websites. No matter how carefully debunked their methodology, they keep coming back like elephant bamboo in the garden that you always think you've gotten out by the roots, but keeps popping back up to block the sunshine and steal all the nutrients from the other plants.

At least three strategic voting websites have popped up claiming to be using "local polling data" (in quotation marks, because there is no such data), in order to recommend which opposition party candidate people should vote for to "get rid of Stephen Harper" in 22, 33 or 50 ridings.

You would think that with all these sites ganging up on Conservative candidates, that party would be up in arms for fear of losing its shot at a majority, right?


To the contrary: they are gleeful, as evidenced by the fact that they have not said one peep about the three sites publicly since the campaign began.

Here's why:

  1. The sites' entire raison d'être validates the concept that people who voted for the Conservative Party in 2008 can't be appealed to further to change their vote now, and thus discourages people from even trying. This is a fundamentally defeatist proposition for the sites' founders to take, one that also underlies the decision by the Liberal Party not to bother making appeals in that marketplace, but to turn its attention towards other competitors instead. It also implicitly discourages people from voting at all where things seem "hopeless" based on previous election results, which feeds precisely into a vote suppression strategy for the Conservatives, and in fact does at least part of that suppression for them.
  2. Their dubious methodology is based on assumptions dressed up as hard data, using the results of seat projections down to the riding level, a technique that pollster Nik Nanos has recently likened to "a photocopy of a photocopy of a photocopy". I will explain below what the methodology is, and why it's so unreliable, but suffice it to say: the impression is being left that it's local polling when it is no such thing, and the sites' sponsors are either lying about that or at the very least omitting to tell the truth, because the opposite answer makes them feel important.
  3. The Conservatives can't believe their luck that these dubious recommendations from the academic salons of Yorkville (the same ones whose brilliant strategies inadvertantly helped elect Rob Ford Mayor of Toronto) have convinced trade unions like the Canadian Auto Workers to endorse candidates so right-wing they would easily fit into a Conservative government. If you want proof, look no further than two of the very last recruits to the Conservative Party's slate of candidates: former Liberal Joyce Bateman in Winnipeg South Centre, and former Liberal Sandy Lee in Western Arctic. Had either of those two been nominated under the party banner they sported mere weeks ago, they would have been hailed as a "progressive choice to stop Stephen Harper" by the strategic voting site sponsors. Certainly David Emerson was so endorsed in 2006, less than six weeks before he crossed the floor to take a cabinet job in the new Conservative government. In fact, if the Conservative Party falls a few seats short of a majority on May 2, Stephen Harper's first few calls will almost certainly be to certain opposition party members who could be enticed to cross the floor for a seat in cabinet. All of them with the Catch22 and/or Project "Democracy" seal of approval.
  4. The sites' obsession with who can win has virtually eliminated issue-based politics from either election coverage or debate at the riding level. This is a perfect state of affairs for a party such as the Conservatives which is consciously trying to move the ideological centre of the country a few inches to the right. Of course, if you weren't sure which party best reflected your point of view on the issues, you could always consult the CBC Vote Compass. Oh wait.
  5. They also promote confusion in cases where the party likely strong enough to win on its own is not the party that placed second in the last election. In fact, since the birth of "strategic voting" campaigns, the number of candidates who came up from third to win has dropped dramatically, but used to be rather large and changes according to the electoral cycle. I'll show the figures below. It is also borne of the belief by these sites that they need to appear "fair" to all the opposition parties, so they have to endorse at least some from each one, implying that the endorsements in fact are not those who can win, but merely a politically acceptable range of candidates from every party they didn't want to piss off. The best example of this error in 2008 was South Shore-St. Margaret's in Nova Scotia where a number of strategic voting guides recommended the Liberal candidate, who wound up placing third, and leaving the NDP candidate just shy of the required votes to achieve their stated objective of defeating Conservative M.P. Gerald Keddy. Keddy should put a link to these websites on his homepage.

Outside of the academic seat projections produced by the Laurier University Institute for the Study of Public Opinion and Policy (LISPOP) which takes great pains to point out that they are a snapshot in time, and are not accurate in predicting the outcomes in individual seats, amateur seat projections and strategic voting sites should be given a wide berth if they claim to be based on local polling, or to be able to predict a result in an individual riding, and how it might be changing from one day to the next during the current election campaign.

They should be particularly suspect when coming from people with no background in the polling industry, nor in statistics, or in managing election campaigns, and where they don't have even a single federal general election's track record behind them. Or need to build their list of "star" candidates by asking on Twitter. Or refuse to publish their detailed methodology.

Let's look at an example of how this works in a riding we'll call "RidingABCDEF". Below you will see the "Riding" results from 2004, followed by the results for that riding's "Province", and the "Offset" of the former from the latter. Next, the provincial sub-sample results from two "Polls" I could find from approximately the correct dates 60 and 30 days before voting day, their "Average", and the seat's "Projection" using the offset calculcated earlier.

Party 2004 E-60 E-30
Rid Prov Offset Prov Polls Avg Proj Prov Polls Avg Proj
A 49.7% 48.9% +0.9% 59% 54% 56.5% 57.4% 48% 52% 50% 50.9%
B 25.6%  33.9% -8.3% 23% 30% 26.5% 18.2% 33% 29% 31% 22.7%
16.0%  8.8% +7.2% 10% 8% 9.0% 16.2% 11% 8% 9.5% 16.7%
4.1%  4.6% -0.5% 4% 7% 5.5% 5.0% 5% 8% 6.5% 6.0%
E 3.4%  3.2% +0.2% 3% 2% 2.5% 2.7% 3% 3% 3.0% 3.2%
1.3%  0.7% +0.6% 1% 0.5% 1.1% –& 0% 0.6%

E-minus-30 is around the time the strategic voting sites start making their recommendations. Note that all candidates are not known until E-minus-21.

So, assuming it's Party A we want to defeat, we recommend party B on E-minus-30, right? Or maybe even say the riding is hopeless, worse still.

Well, let's see how that worked out by E-minus-15, and on E-Day.

Party 2004 E-15 2006
Rid Prov Offset Prov Polls Avg Proj Rid Prov Offset
A 49.7% 48.9% +0.9% 53% 26% 39.5% 40.4% 37.9% 42.1% -4.2%
B 25.6% 33.9% -8.3% 20% 12% 18% 9.7% 10.0% 20.8% -10.8%
C 16.0% 8.8% +7.2% 15% 47% 31% 38.2% 39.5% 24.6% +14.9%
D 4.1% 4.6% -0.5% 9% 3% 6.0% 5.5% 8.0% 7.5% +0.5%
E 3.4% 3.2% +0.2% 3% 3.0% 3.2% 4.1% 4.0% +0.1%
F 1.3% 0.7% +0.6% 0% 0.6% 0.5% 1.1% +0.6%

Whoa! Look at Party C. It came out of nowhere on E-minus-15, but only in one of the two polls picked here at random. Plus all our materials endorsing Party B as the best way to beat Party A have already gone to the printer and are in the mail, and resources have already been applied to the ridings according to the earlier ranking!

Suddenly having too many people following the E-minus-30 recommendation to vote for Party B (communication being imperfect, particularly when correcting the record) would have inadvertantly reelected Party A, and completely missed the last minute surge of Party C. And even on E-minus-15 you'd have to be prescient to know which poll was right.

Can anyone guess which riding this is, by the way? All the above numbers are real.

As a bonus question, which party's leader: A, B, C, D, or E is meeting with the mayor of the city this riding is in today (Monday) on their leader's tour?

Now let's look at the number of seats won from third place in recent elections:

Seats by Rank of Previous Winner (either By-election or General Election), 1993-2008

Election 1 2 3 4 5 n/a* Total
* Winning party did not exist, or winner had been/became Independent, etc.
2008 GE 268 38 1   1   308
2006 GE 253 46 7 1   1 308
2004 GE 175 39 61 30 2 1 308
2000 GE 269 25 7       301
1997 GE 248 35 12 5   1 301
1993 GE 93 86 26 24 5 61 295

Interestingly, even though the government didn't change until 2006, the 2004 election already demonstrated a lot of shifts, due mainly to the merger of the two conservative parties. And in 1993, with two new major parties contesting seats, incumbency was worth less than ever.

If you didn't read my plea not to vote strategically in the last election, I urge you to take another look now. A vote "against" someone or something is a vote in favour of nothing. It gives no mandate to elected officials, creates all the wrong incentives for the politicians who are elected that way, and guarantees that Parliament will descend even further into the partisan barking we see there now. Indeed the perverse problems with the methodology itself have led respected website Democratic Space author Greg Morrow to stop publishing his "strategic voting guide" from previous elections.

In this election, read the platforms, watch the debates, take a measure of the leaders and the candidates, and vote your heart. If everyone did that, who knows what we might come up with together.

Hill Times Column on Vote Switching and Non-Voters

November 10th, 2009 | 6 Comments

While we were all busy covering the by-elections, the Hill Times was kind enough to publish a column of mine yesterday. It’s reprinted here with kind permission.

Byers’ formula fails in his own province

Michael Byers recently urged the Liberals and NDP to strike a one-time deal where each party would stand down from running candidates in ridings in which the other party placed higher. But his vote switching theory doesn’t fly.


OTTAWA—University of British Columbia professor Michael Byers has studied climate change, the war in Afghanistan, and Canada’s North. But after a provocative op-ed in last Monday’s Toronto Star, I think he should go back and study the actual numbers—at least the fields of voter behaviour and vote switching.

Byers recently urged the two main English Canadian opposition parties, the Liberals and NDP, to strike a one-time deal where each party would stand down from running candidates in ridings in which the other party placed higher, with the objective of unseating the Conservative government and installing a form of proportional representation.

The proposal has been roundly criticized by columnists and bloggers in the NDP, Liberal and even small-c Conservative ranks, who deemed it unwise, undemocratic, and/or politically unfeasible.

But would it even work? A look at switches in party voting in Byers’ own province of British Columbia, from 1988 to 2008, suggests not.

In a multi-party system, dissatisfied voters have the choice of staying with their previous party, switching to another party, or staying home. New seats can be won in one of two ways: either by increasing your own share of the electorate, or by decreasing your most serious opponents’ share, whether by causing them to switch to a third party, or just stay home.

Using riding-level data from the database, and calculating each party’s vote and the number of non-voters as a percentage of the number of electors in each riding, it was found that, over the seven general elections held in the last two decades, many previous party supporters would rather stay home than switch.

And the switchers there were, did not usually move where Byers assumed they would. Because the Liberal and NDP shares of the electorate both dropped in most seats in 2008, those voters either moved to the Conservatives or Greens, or stayed home. Only in Saanich-Gulf Islands is it clear that a large number of previous NDP voters switched to the Liberals. And only in Kamloops-Thompson-Cariboo, Burnaby-New Westminster and Victoria did at least some previous Liberal voters switch to the NDP, although more switched to the Conservatives in the Kamloops seat, while more stayed home in Victoria or Burnaby-New Westminster.

There are currently 36 seats in B.C. The Conservatives hold 22, of which 13 were won with more than half the ballots cast. For Byers’ strategy to work at all, the agreement would have to focus only on the seats the two opposition parties have the best chance of winning from the Conservatives. Because if the parties were to stand down in each other’s held seats, many would be at risk of falling to the Conservatives instead, undermining the whole point of the exercise.

[For the purpose of this exercise, I’m also ignoring the campaign effects of trying to run both together and against each other at the same time. But it’s safe to assume there would be a strong reaction from the Conservatives.]

The NDP finished ahead of the Liberals in 23 of the 36 B.C. ridings. It is arguably in striking distance of winning about five of them next time, but would either need to win over virtually every remaining Liberal supporter (unlikely given historical patterns), or would need former Liberals to leave the Conservatives, and then return to voting Liberal, vote NDP, or stay home.

2008 vote swings in five B.C. seats that would run only NDP candidates under the Byers plan

Percentage of Electors NDP Lib Grn Cons NV
Kamloops – Thompson – Cariboo + 2.8 - 9.8 + 2.0 + 3.9 + 1.0
Surrey North - 6.6 - 3.1 + 1.3 + 4.9 + 3.7
Pitt Meadows – Maple Ridge – Mission - 2.5 - 9.0 + 2.4 + 5.4 + 3.9
Nanaimo – Alberni - 1.6 - 7.3 + 4.1 + 1.8 + 4.0
Vancouver Island North - 1.3 - 5.9 + 1.9 + 2.2 + 2.6

Meanwhile, the Liberals finished ahead of the NDP in 13 of the 36 B.C. ridings. In its next five best seats currently held by the Conservatives, the NDP share of the electorate fell in 2008, but the Liberal share fell more. Three of the five were Conservative pickups from the Liberals last time, won on the basis of Liberal-Conservative switchers. To win those seats, the Liberals would find more likely supporters amongst those who switched to the Conservatives last time, while in Fleetword-Port Kells the two opposition parties are nearly evenly split.

2008 vote swings in five B.C. seats that would run only Liberal candidates under the Byers plan

td class="lib attrib">- 8.8

Percentage of Electors Lib NDP Grn Cons NV
Saanich – Gulf Islands + 8.6 - 15.4 + 0.1 + 3.3 + 2.8
North Vancouver - 4.8 - 3.0 + 1.9 + 2.4 + 3.4
West Vancouver – Sunshine
Coast – Sea to Sky Country
- 4.6 + 4.9 + 3.6 + 5.0
Richmond - 8.0 - 1.7 + 0.8 + 4.1 + 4.3
Fleetwood – Port Kells - 4.3 - 2.3 + 2.1 + 4.9 + 3.9

The riding of Saanich-Gulf Islands, in which Byers now lives, is a special case inasmuch as almost 80 per cent of NDP supporters last time abandoned their party when their candidate resigned. Of the 19.4 per cent share of the electorate earned by the NDP in 2006, 8.6 per cent voted Liberal in 2008, four voted NDP, 3.3 per cent voted Conservative and 2.8 per cent stayed home. This move was still insufficient to change the outcome, and in fact it strengthened the Conservatives’ hold on that seat.

But perhaps there’s another approach.

Using the per cent of the electorate rather than per cent of the vote allows us to look at the fastest-growing “party” of the last 20 years in British Columbia: non-voters. In fact, if the non-voters were a political party, they would have won every B.C. seat in 2008 except Saanich-Gulf Islands and Surrey-White Rock-Cloverdale.

Looking at elections results this way, we notice that Conservative support has remained fairly constant in most B.C. federal ridings since 1988. It’s just that the turnout is down.

A big drop in turnout occurred in 1993 across every B.C. riding. Although the Liberals picked up five seats, in fact they typically held their share of electorate while other parties’ supporters stayed home, or split across two other parties.

A recent master’s thesis (opens PDF) from the University of Waterloo (by Maria Mavrikkou) has found that NDP support is most likely to drop when turnout drops, since their support correlates with low-income status, and that correlates with lower voter participation. And in 1993 as the NDP vote plummeted in virtually every B.C. riding, about half of the drop could be accounted for by big increases in non-voters. But in 2004, the B.C. seats showing the biggest gains for the NDP were also seats where turnout went up.

So, Byers may want to go back to the drawing board to find a more successful formula, and perhaps a different dance partner.

It might not be the one he originally had in mind.

Alice Funke is the publisher of

Coalition Math: The McGregor Hypothesis

December 9th, 2008 | 2 Comments

I’ve been trying to avoid commenting on the coalition-prorogation-confidence situation, because it’s not largely numbers-based and revolves around a parliamentary rather than an electoral coalition.

However, one reader asked me to calculate what the outcome of the recent election would have been if, outside Quebec, only the higher of the Liberal and NDP candidates ran against winning Conservatives, assuming their votes could be combined. I did this calculation, with the following results:

  • Of 31 ridings won by the Conservatives outside Quebec, combining the votes of the two coalition partners would have shifted 26 seats to the Liberals and 5 seats to the NDP, for a House of Commons composition of 112-Cons, 103-Lib, 49-BQ, 42-NDP, 2-Ind.
  • But the coalition had also been endorsed by the Green Party, so I also tested adding in the Green vote: which would have put an additional 23 Conservative seats into play, giving 3 more to the NDP, plus now 2 to the Green Party (Central Nova, NS and Bruce – Grey – Owen Sound, ON), for a Commons composition of 89-Cons, 121-Lib, 49-BQ, 45-NDP, 2-Grn, 2-Ind.

At first blush then the Liberals would appear to benefit from an electoral coalition more than the NDP, although they would be facing a very divided House.

But I just don’t buy the assumptions that have to be made in order to run that kind of analysis. As Bloc Québécois leader Gilles Duceppe said last week, echoing the famous quote of former Prime Minister Jean Chrétien, “if my grandmother had wheels, she’d be a tractor“.

Here’s why:

  • Turnout matters too. Turnout dropped significantly in the most recent general election. Early speculation is that it was Liberals who stayed home for whatever reason (leadership, policy reasons, regional reasons, negative advertising, who knows). Would those voters have been more or less likely to vote under other conditions?
  • Would Liberal voters and non-voters have supported a coalition candidate, or would they have been more likely to vote Green or Conservative?
  • Would all NDP voters have supported a coalition candidate, or might some of them have voted Green or stayed home?
  • Who would have led the coalition, because clearly that would have mattered when answering the above questions?

Meantime, Glen McGregor ran a similar analysis for this morning’s Ottawa Citizen, but extended it to account for recent opinion surveys by using second choice preferences reported in the recent Ekos survey to redistribute the votes of the junior coalition partner in each riding.

This approach tries to address the 2nd and 3rd points above, but also fails to take turnout into account. Also, while it’s probably highly unsafe to rely on polls while leadership issues are in such flux, the approach suffers doubly from using a national result with a sample much too small to extrapolate to riding-level results, which was one of my major criticisms of the methodologies used by the so-called strategic voting websites during the last election.

Look, it’s easy to criticize, and I love the work Glen McGregor has been doing with data analysis for the Citizen (he’s the author of the definitive compilation of voting and non-voting records in the last Parliament, and a very impressive recent cross-reference of party fundraising and results by riding). If an election were held today, the results might indeed be similar to those McGregor projects using his methodology. But if we’ve learned no other lesson from the massive effort invested in strategic voting in the last election and the results it produced: voters will make up their own minds when the times comes, a week is a lifetime in politics, and if my grandmother had wheels …

Think Twice About Voting Strategically

September 30th, 2008 | 26 Comments

The Pundits’ Guide is a non-partisan site and does not endorse any political party, and tries by the selection and application of quantifiable criteria to treat all major parties the same. I’ve even tried to redress the situation faced by smaller parties by starting to add in their party affiliations on riding and candidate profile pages.

However there is one issue I’m going to weigh in on during this election, and that’s the advocacy of so-called “strategic voting”. It’s based on faulty assumptions, is frequently clumsily calculated and executed, often by people with hidden but vested interests, and most seriously: it can lead to quite perverse outcomes.

Faulty Assumptions: Many of the strategic voting sites are based on the assumption that past performance can predict future outcomes. This is no more true in electoral politics in a multi-party system than it is in the stock market (say no more). Last February, struck by the tendency of the Hill Times to use “close races” in the last election as a proxy for “swing ridings”, I decided to test that hypothesis. As it turns out — and this is true for each of the last 3 elections — more of the seats that changed hands in an election had previously been won by margins of more than 5%, than had been won with margins under 5%. Unfortunately, this “close contest” shorthand for determining what are the swing ridings was picked up and adopted in this campaign by no less than CTV, the Globe and Mail and their pollster the Strategic Counsel. The resulting polling numbers have been a mish-mash of Liberal-Conservative contests with Conservative-Bloc contests, Liberal-NDP contests, and NDP-Conservative contests. How can you possibly tell who’s winning when you roll them all up together?

Clumsy execution: Consistent criteria are not applied, and incomplete data is used to determine the “likely” outcome in a riding or the “best” way to vote. This first came to my attention when folks were linking to riding profile pages on this website, to question how in the heck one of the strategic sites could have come up with the conclusion it did in a given constituency. In other cases, the site’s authors were claiming it made no difference how a person voted since the race was so close between the two front-runners from the last election, completely ignoring the fact that the riding in question had been a 3-way race in the last campaign (that entry has since changed). Still further cases claimed that a party had no chance of victory, since it had no history of success in the seat, even though it had finished 2nd by just over 300 votes in a by-election within the last decade, but on the old boundaries.

There are some major limitations in using the current batch of previous results to predict future outcomes: first of all, most of us are only using data from 1997 forward, since that’s all that’s available in easily downloadable form from Elections Canada. But voting patterns completely changed in 1993, and with them the composition of Parliament. Ditto for 1984, and to a lesser extent 2006. Secondly, past voting behaviour is not a straight-forward indicator of future behaviour in elections where the government changes. Heck, polls two weeks out from Election Day can still swing widely. Incumbency matters, as does the absence of an incumbent, whether they are first-time incumbents, and whether they are on the government side or the opposition side going in (and perceived to be, going out). And candidate recruitment matters significantly (a point that pretty much makes itself in this campaign, to be sure).

Like it or lump it, the first-past-the-post system of parliamentary democracy is what we’ve got now, and it has to produce a number of contradictory outcomes in each election. First, it demands that voters consider and return the best representative for their community to send to the House of Commons. Second, it takes a collective reading of those constituency votes to select a Prime Minister and determine the composition of the Commons. And finally, it gives a “mandate” on one or more issues of the day to those elected representatives. All with one vote.

The mandate politicians are given, or believe they’ve been given, is paramount. The reelection of a government is taken by the members of that government as an endorsement of its policies. Holding an incumbent M.P. to a smaller margin of victory after their voters raised holy heck about a local issue is read very clearly by that M.P. as a signal to shift priorities. Public engagement on an issue during an election campaign is heard loud and clear by any candidate whose name is on the ballot.

I am sorry to say this, but people who claim to be “voting for the environment” are spending more time poring over past voting statistics (as prepackaged and interpreted for them by others) and daily rolling-polls, than they are actually spending poring over the environmental platforms and proposals of the political parties they are considering voting for. In the Skinnerian world of electoral mandates, these voters are unfortunately and unwittingly rewarding cynically manipulative electoral strategies, rather than thoughtful platform development, in our politicians. And afterwards they’ll be more disappointed than ever.

If you really want to vote based on the environment, then get yourself informed on the issues, and vote for the policy you support. Do you believe climate change is a problem? If it exists, is it man-made or natural in origin, and if man-made can it be reversed. If reversal is possible and desirable, how much change is required and how quickly, and at what other cost. Which approach would be the most effective: a carbon tax (whether revenue neutral to individuals or to government), a cap-and-trade system of emission permits, carbon capture and storage, emission regulation, or something else. The answers you arrive at should determine your vote, because that’s the message that will be received by the politicians you elect.

If all a politician has to do to get your vote is convince you they’re more likely to win, just think what kind of Parliament you’re going to get: an even scrappier place filled with people who’ve learned the only way to survive is to win at all costs.

What would have been really helpful is if, in urging voters to vote for the environment or any other issue, the authors of those websites had spent their time documenting the issues, and the past behaviour of politicians on them. Where were their records in presenting legislation, their speeches and other activites on environmental issues, or their voting records? The past behaviour of politicians is by far the more accurate predictor of their future behaviour, but is nowhere to be found amidst all the poll results.

If I were any kind of expert on any of these issues, that’s the kind of website I would have produced. Unfortunately the only thing I’m qualified to do is crunch the historical voting numbers.

But I’m doing my reading, and I’m going to vote passionately and with conviction. That’s my strategy for voting.