Appearing on Bourque is this Montreal Gazette headline; Confessions of a Pollster.  A title like that doesn’t not get the attention of those who work in the industry.

In the sixty or so seconds it took me to move my mouse over the link, click on it, allow the page to load, and read enough of the story to get the gist of it, I was entirely conscious of the fact I was expecting something worse, much worse.

No, I wasn’t expecting a story about a number-crunching propeller-head who’d figured out how to lure unsuspecting victims using reams of data as sultry foreplay. However it did occur to me how someone may finally be fessing-up to how, everyday, survey data is used and misused to tell one version of a story or another.

If I had a nickel for every time I saw a survey-backed story that somehow didn’t feel right, and upon review of the methodology including especially the sample or questionnaire design discovered a bias – accidental or otherwise – I’d be, well, a little bit richer. But that’s another story for another day.

Despite the attention grabbing title, the Gazette piece turns out to be nothing too salacious or incriminating. It’s an interview and a peek-a-boo look into the polling perspectives of the well-respected Christian Bourque of Quebec based Léger Marketing. Léger & Léger as I used to know it, is a great firm that includes in its claims to fame, some of the most accurate polling data money can buy, especially in Quebec.

The interview is a good piece, a recommended read, and I can relate to many portions of it. The assertion that politicians who say they don’t pay attention to polls being liars is especially true.

However, where I slightly differ in opinion from my colleague is on the matter of governments not governing by polls. Mr. Bourque says “…fortunately for all of us, governments do not govern by poll. They will use public opinion to help them fine-tune and find the best timing to talk about a policy initiative. But if people governed by polls we would have never had the GST, we would pay no taxes, etc., etc.”

Polls maybe, but public opinion absolutely governs the way governments govern. Just because some polls miscalculate or don’t catch early enough the full impact of public opinion, is a knock against the science of polling that purports to measure opinion. It does not mean that public opinion isn’t the principle guiding factor that ultimately shapes the way governments govern, especially today and even more so in the future.

Two factors didn’t exist in 1991 when the GST was allegedly foisted upon Canadians; first, the Internet, and second what one can do with public opinion using the Internet if one knows what he or she is doing.

To prove my point, examine a recent issue that erupted and consider how it might have played out in 1991. A few weeks ago Ontario Premier Dalton McGuinty tried to introduce legislation which, had it passed, would significantly restrict the freedom of young drivers. Almost overnight thanks to what highly charged and motivated opinion can do using, among other tools, Facebook, the legislation has been drastically rewritten if not abandoned altogether.

This never would have happened in 1991. In 1991 Mr. McGuinty’s legislation would likely have passed without much objection or fanfare.

Was public opinion about young drivers any different in 1991? If anything, it was likely more lax. Today, the only difference is how public opinion – if it is sufficiently massed, charged-up and motivated – can be made to bite a politician’s behind before he/she knows what hit him.

The recent case against Steven Harper is not much different. The difference is once the full impact and realization of Mr. Harper’s mistake was served to him on a platter, Mr. Harper knew how, with a little think time – and trust me, with mountains of public opinion data – to fully exploit the situation and completely turn the tables on his opponent. Mr. Dion still doesn’t know what hit him.

In a matter of one week Mr. Harper and Mr. Dion have gone from the very real prospect of trading places, to Mr. Dion, late fuzzy video and all, is no longer able escape the long-ago case against him as shaped and measured by public opinion. 

This is why Liberal’s will choose Mr. Ignatieff over Mr. Rae. Not because Mr. Rae isn’t a great guy and charismatic, he is. But Liberal’s now understand how the negative vs. the positive aspects of public opinion about Mr. Rae can and would successfully be used against the Party if Mr. Rae was chosen leader. Liberal’s aren’t about to make the same mistake twice. Mr. Rae can only ad to the carnage if he, like Mr. Dion, doesn’t immediately move to accept reality. 

For decades, the youth of our country have been labeled voter apathetic. In the recent Federal election the efforts of young adults such as Peter Cleary didn’t pan out as they hoped. However, I predict that during 2009-2011, if not sooner, we will see a turn-around in younger voter apathy including the impact this group has on the way governments govern. It’s already happening.

More to the point, garnered public opinion doesn’t just shape the way governments govern around election time. Indeed, election day is heretofore only to be known as judgment day for those who think to govern against the tide of public opinion.

What does this spell for visionaries and initially unpopular but worthwhile initiatives such as the GST, NAFTA, or even graduating licensing? Going forward, the ability to shape public opinion in advance, of which polls are only one part, will be a great deal more important to the way governments govern than simply reactively using polls after-the-fact to measure public opinion in order to figure out what to do next. By then, it’s way too late.

This is both an art and a science, especially of the opinion that forms subconsciously, that will define the battleground between Mr. Harper and Mr. Ignatieff. Mr. Harper has a slight head start both in terms of historical data mining and money.  Mr. Ignatieff, if he chooses the right combination of methodology and strategy, could very quickly catch-up.

Being as impatient as I am, and I am, I didn’t want to wait the “up to 28 days” it may take for Elections Canada to post final-validated results for last week’s federal election.

So, tapping into the Type A portion of my personality, I downloaded Election Canada’s horribly organized text-file which is a mix of both preliminary and validated results as at October 18, 2008. And from that created a series of spreadsheets to serve as reference points for the analysis below.

First, a 22 inch monitor snapshot of all 308 ridings sorted by seats won by the five major parties and two independents, appears here.

At one time, there was a plan for that spreadsheet to contain a list of seats the Green Party would win. Instead, Elizabeth May decided to run against Peter McKay in Central Nova.

The next interesting analysis consists of the 21 ridings Liberals would have won if the Liberal and Green vote were combined. Presumably these are among the 21 ridings Elizabeth May is ‘no longer able to live with herself for having done something to help re-elect Stephen Harper.’ 

In addition, who’s to say that a concerted campaign effort where the message wasn’t so confusing could not have resulted in more seats going Liberal, notwithstanding the man at the helm. For that, we add the 18 ridings where the Liberal + Green vote was within 5% of the winning margin. Surely 5% isn’t too far of a stretch for having clear communication. Note the appearance on this list of John Baird, Environment Minister within 2.5%. 

Finally and in fairness, Jack Layton never said “anyone but Harper” so it should come as no surprise by that narrower view of the world, that picking up eight more seats is likely the NDP’s preferred outcome. However, combining Liberal, Green, and NDP votes results in 63 ridings that would not have gone Conservative (54) or Bloc (9).

I have argued at length on this space that vote-splitting would re-elect the Conservatives, and it did as shown above.

Taking the analysis a step further, on October 15 I posted a graphic showing the differences between the 2006 election, the number of seats for each party Canadians would have preferred on October 14, as well as the recent election outcome. So long as Canada does not have some sort of Proportional Representation, why not consider an outcome that combines Liberal, NDP, and Green votes under one party banner, assuming the Bloc would have none of it. Here is what that graphic would look like:

Safe to assume the Left prefers, as Tom Flanagan so eloquently put it, “10-12 years of Harper-land” before it considers a merge.

The good news about a minority is uncertainty keeps political strategists and analysts in business. The bad news about a minority is it keeps political strategists and analysts in business

In the days, weeks, and months ahead many will opine on the meaning of this past election and take stabs at the direction of the Canadian political landscape.

What happened and why? Which leaders will survive, which wont? What impact did vote-splitting have and will the Left finally get it together, or could a figure such as a Jean Charest do for Canada’s natural governing party that which Mr. Dion hadn’t a hope?

It is this observer’s intention to keep adding to the discussion, hopefully always from a different angle and only when I have something important to say as opposed to a set schedule.

Speaking of different and important;

A couple of weeks ago when I wasn’t talking about vote splitting – which I admit wasn’t very often – it occurred to me to field the Core Strategies “Seat 100″ question just as Canadians were heading to the polls. So we did.

To refresh your memory, the question asks not for whom you will vote but rather how many seats would you like each of the five main political parties to have in the house of commons. In short, it’s a question about Proportional Representation by other means.  

The point of the exercise is to showcase the divide between what kind of representation Canadians want versus what our electoral system gives us. Something is wrong when a system that espouses Democracy delivers something very different from what that Democracy expressly says it wants.

Those who oppose some form of Proportional Representation – usually the larger power-based parties – will point to examples such as the recently defeated referendum in Ontario as reason to ignore PR. For the record, Ontarians didn’t reject the concept, they simply didn’t understand the question as posed.

Generally speaking, the same is true of most Canadians. It’s easy to fear and reject the unknown and we have no better proof than the election just witnessed to see how fear of the unknown affects behaviour and election outcome.

The seat distribution analysis isn’t one that is meaningful only on aggregate. On the contrary, analysis by voter intention is even more interesting and revealing. The chart below, while perhaps a little difficult to get your head around, depicts the number of party seats by voter intention. For example based on the survey of 1,200 Canadians we conducted Oct 9-12, Conservative voters want the Conservative Party to end up with 193 seats. However, that same group of Conservative voters wants 48 Liberal seats, 33 NDP, 19 Green, and 14 Bloc.

In other words even diehard Conservatives want the other parties to have a certain number of seats.

It’s also an interesting analysis to compare the results by voter intention and ask questions like why is it Liberal voters only want their own party to have 157 seats, whereas Conservative voters want their party to have 193 seats? And what does the answer to that question say about trust in one’s own party, or party leader?

Here’s another good question; why is it that between December of 2007 and October 2008, Liberals who wanted 177 seats now only want 157 for themselves, whereas all other parties want more? Perhaps this is an indication of how much weaker Liberal faith is within its own party.

In percentage terms the “Seat 100″ line of questioning yields identical voter intention results to the traditional vote question. For example 103 of 308 seats Conservative is 33%, and 83 Liberal, 57 NDP, 35 Green, and 30 Bloc is 27%, 19%, 11% and 9% respectively. These figures are nearly identical to aggregate of Harris-Decima, Strategic Counsel, Nanos, Ekos, and Ipsos.

It’s what the Seat 100 data tells us in addition that is so darn interesting. Almost as interesting as the fact first-past-the-post gives Canadians what they don’t want.  

~~~

Survey Methodology

The study was conducted via CATI from dialing facilities in Toronto, Montreal, and Ottawa. Data collection for this study was conducted between October 9th through October 12th, 2008. The sample size was 1200 interviews completed among randomly selected adults 18 years of age and older. The data is weighted in tabulation to replicate actual population distribution by age and sex within region according to the 2006 Census data. The margin of error on a sample of 1200 is +/- 2.8% 19 times out of 20.    

The question posed was:

Imagine that there are 100 seats in the House of Commons.  How many would you like to see each of the following parties win in the October 14^th federal election?  {Read and Rotate, and ensure adds to 100} Liberal, Conservative, Bloc Quebecois, NDP, Green

            a.       Conservative

b.       Liberal

c.       Green

d.       NDP

e.       Bloc Quebecois

Last year at this time a seemingly insignificant piece of data surprised me; exactly fifty percent of Canadians wanted a majority government in Ottawa, and the other fifty percent wanted a minority.

I’d never seen that ratio before, only ever recalling a preferential leaning toward a majority status. And seems to me prior minority governments resulted not because minority was what voters wanted, but because competing majority forces couldn’t squeak one out.

For a couple of years running, public sentiment has been different and last year I began to wonder how, in what way, and why. Focus group junkies are obsessed with these types of questions.

By definition, a person’s desire for a minority government means they want someone else to have seats too. To that point in time last year, I’d never really thought about political opinion in that way. I’d only ever considered that a person either voted for choice A, B, or C, and more recently, possibly D. The trick was always to know for whom, and how opinion could be shifted.

For the most part, pollsters and political parties have paid little if any attention to the number of seats a voter for party A wants party B to have and vice versa. Why would they? That’s not how the voting system works, right? Perhaps. But that is how the brain works, subconsciously at least, and that’s what gave me inspiration to ask the traditional vote question differently.

Last spring Core Strategies did not simply ask Canadians who they would vote for. Instead, and in view of Canadians significant desire for a minority government, we set out to understand how Canadians wanted a minority to look. Again, I was less interested in a minority look following a failed majority attempt, but was intently interested in the number of seats voters (even those desiring a majority) wanted other parties to have. This, in my view was a vastly different and more meaningful analysis, and not at all the same thing as asking voters for their second, third, and fourth choice.

Not to belabour the point with superfluous example, but, if at one point in time a person who intends voting Conservative also wants Liberals to have 60 seats a few months later reports they still intend voting Conservative but now want Liberals to have 80 seats, that tells us something very significant about the political landscape as it shifts beneath us. This methodology can also serve as a precursor of things to come. After all, there comes a time when the number of seats a person wants the other party to get increases so much, that one begins to question ones own vote.

I maintain, that he or she who best understands and pays attention to this particular type of ebb and flow to the political landscape, is in a better position to either explain current support levels, prognosticate voting intention, or manipulate voting intention depending on which business one is in.

If imitation is the sincerest form of flattery, then I can’t say enough that Harris/Decima, arguably one of the best full service research operations in Canada, last month picked up the methodology. Earlier this week Harris/Decima reported Canadians Prefer Polyglot Parliament using essentially the same question Core Strategies investigated last spring, namely: “Imagine that there are 100 seats in the House of Commons.  How many would you like to see each of the following parties win in the next federal election?  Liberal, Conservative, Bloc, NDP, Green.”

Similar to Core Strategies results in May and June 2007, the Harris/Decima results from December 2007 furthers the inference that many Canadians still favour a minority, one based on some sort of proportional representation or “Polyglot Parliament” as referred to by Harris-D.

While the wording of the question is not identical, it is certainly close enough to permit a look at the seat distribution data as a time series as depicted below.     

	All Voters					CON			LIB			NDP			GREEN			BQ
	Dec	June	May		Dec	June	May	Dec	June	May	Dec	June	May	Dec	June	May	Dec	June	May
CON	96	91	97		174	151	171	61	68	69	56	54	53	66	61	67	71	62	59
LIB	111	88	81		73	70	58	177	136	125	90	67	66	107	71	64	61	65	61
NDP	48	56	49		31	39	32	37	50	45	115	110	97	42	50	51	36	45	39
GREEN	25	38	42		12	26	24	19	32	39	21	46	51	72	83	92	20	34	37
BLOC	28	34	39		18	24	23	14	23	30	26	32	41	20	43	35	121	102	114

Unfortunately there aren’t data between June and December. However, my hunch is Conservative fortunes under this analysis were probably at an all-time-high in September-October, prior to Mulroney and Bali. Nevertheless the chart above suggests that, by-and-large, support at the end of the year came back to mirror data from the spring with improved figures for the Liberals, not dissimilar to pattern uncovered by the traditional vote question.

Where this methodology allows the analysis to be taken a step further, is in understanding from whom support is being taken and where it’s ending up. More important, the data provides an analysis of the intensity of voting intention.

For the most part Conservative voters have remained the most consistent in the support they have for other parties and themselves; in May at 171 seats, down to 151 in June, but rebounded to 174 by December. Conservative support for the other parties varied insignificantly save for the Green party where support fell to 12 seats.

The biggest gainer appears to be the Liberal support base. Liberals themselves want their party to have 177 seats, up from 136 and 125 in June and May. However, as Harris/Decima astutely points out “…because strategic voting has been an important factor in recent elections, it is useful to note how supporters of the smaller parties feel about the outcomes they would like to see for the larger parties.” I could not agree more.

As I have previously strongly argued in this space, within the context of minority sentiment the manner in which the vote splits on the Left is key not just to the eventual minority seat count, but to whether or not a minority actually occurs, or results in a “default majority.” Pertinent to this point is how NDP and Green party support for the Liberals has risen between June and December. From 67 to 90 seats for the former, and from 71 to a whopping 107 seats for the latter.

In fairness, NDP support for the NDP and Bloc support for the Bloc also intensified between June and December, 110 to 115 seats, and 102 to 121 respectively. Only the Green party support softened among its own supporters, 83 to 72 seats.

Those wanting to be critical of this methodology will be quick to point out that it deals in abstracts, talks about party support in terms of unrealizable seats counts, and assumes some form of proportional representation. All fair observations, but missing the larger picture. Just as stock indices are not indicators of individual stock values, they serve well in terms of market analysis, and that’s how this methodology should be viewed and worked with, as an analytical tool.

The Economy vs. Environment debacle hinted to in my last post has barely begun to grab hold, but early indicators are the analysis is correct and is only going to intensify. As short term economic concerns possibly begin to overshadow longer term environmental issues, will faith continue to flow toward Liberal support, or will traditional thinking set in that Conservatives are more fiscally responsible and trustworthy?  

That last question isn’t prognostication. We badly need mid-January and early February data to learn in which way the underlying sentiment is heading.

My hope? That Harris/Decima keeps asking the new question week-after-week, month-after-month.