The latest polls give plenty of fodder to suggest that there is a Liberal surge overtaking the race.  There are a number of supportable points on this:

• Nightly poll tracking by Nanos has been reinforced by Innovative and Ekos.  The narrative is that the Liberals are pulling away from the Conservatives with the NDP far behind.
• The “not ready” line of attack has been embraced by the Liberals; they have played it differently than “just visiting” and “not a leader” attacks from previous campaigns.
• The heat of the anti-Harper passion is far beyond that what has been seen in previous elections.  Various poll metrics (eg. “time for a change”) suggest voters are ready for a new government.
• Advance voting turnout is high which could mean higher overall turnout favouring the Opposition.

Yet, there is that gnawing feeling that there could be another polling surprise just around the bend.

Look at the UK election last May.  The intensity of British media coverage and polling exceeded that of the Canadian election yet no one saw a Conservative majority coming.  Even famed predictor Nate Silver blew it badly.  When BBC forecasted a majority moments after polls closed – based on results from exit polls – pundits were absolutely gobsmacked.  Not only that, two of the party leaders were caught with their pants down by their ankles and resigned by morning.  The prime minister (“Bluedini”) was likely as surprised but had the winning strategy on his side.  (Exit polls were based on interviews with voters immediately after they voted, not pre-voting surveys)

Yes, there is the litany of Canadian surprises too.  Mainly favouring incumbents – Christy Clark, Alison Redford, Greg Selinger, Dalton McGuinty, and Kathleen Wynne to name some plus Jean Charest who just missed re-election when pollsters had him in third.  Remember Doug Ford?  The final polls in the Toronto mayor’s race had him dead and buried but he only lost by 6.5%.  Overall, it’s certainly not a sterling track record.  I’m speaking about media polls here.  Sure, some parties have got it wrong too, but clearly some (the winners) are getting it right.

But they called it in Alberta, right?  A quick check from the 308 poll aggregator site shows that most pollsters (not all) overstated NDP support and understated PC support.  It didn’t matter since the NDP won handily.  But in a close election, some of the pollsters were off by a considerable margin when you look at the NDP-PC difference.  The poll aggregator had that gap at 20 points (it was 13%).  That’s a big difference in a close election and an error of similar magnitude in this election would lead to a different outcome.

Three things I’m watching

1.  The engagement level of voters

Last week, I wrote about Greg Lyle’s extensive research.  He updated his research this week with a research deck of over 100 pages.  There is enough data here to keep Wikileaks busy for a month.  Overall, his survey reported a 35-30-24 race (Lib – CPC – NDP).

Greg has built a respondent profile based on consistency versus ambivalence.  If the respondents have a consistent pattern to their responses, they are at the ‘consistent’ end of the scale.  If they answer “don’t know” to a lot of questions, they are on the ‘ambivalent’ end of the scale.  Conflicted respondents are in the middle.

About 15% of respondents are ambivalent.  Most will not vote.The ‘perfectly consistent’ are primed to vote.

This week, Greg posted the vote results by each of these clusters.  Bearing in mind the Liberals had an overall five-point lead on the CPC, here’s how that broke down among the consistency segmentation:

The race is tied among the ‘perfectly consistent’.  I would throw out the 15% of ambivalent respondents, which shaves a fraction off the Liberal lead.  Among the 18% of ‘conflicted’ voters, the Liberals have a substantial lead over CPC, which appears to be based on NDP switchers (hence the fact they are ‘conflicted’). These respondents are a lot more likely to vote than ‘ambivalent’ but less likely than consistent voters.

Therefore, the Liberals have more work to do to mobilize this voter group in order to realize a five-point win.

         2.  Inconsistencies between pollsters in age and gender

There has traditionally been a gender split with the Conservatives doing better among men and the Liberals doing better among women.  Ekos, which had the Conservatives ahead two points, shows the Conservatives leading among women.

Meanwhile, Nanos in the field at the same time, has the Liberals with an 11.6% lead among women.  That’s a pretty big gap between pollsters.

Is Ekos overestimating CPC support through a blip in female support, or is Nanos overestimating Liberal support among males (below).  Nanos has shown a consistent Liberal uptick.

Oh, and did you see the Ekos age split?  48-30 for CPC among seniors.  Nanos has the Liberals with a 1.5% lead among plus 60 voters.  Again, another big difference.  Who’s right?  I dunno.

        3.  Regional races

There is considerable variance in the regional horse race numbers, especially in Ontario and Quebec.

In Ontario, it is either a death-duel between the Blues and the Reds OR it’s the Reds walking away.

In Quebec, it is either a four-way collision or the NDP retain enough support to win the majority of seats, keeping it in national contention.  When you see numbers like this, the party above 30% can harvest a lot of seats.  The difference in Quebec between 27% and 32% could be 40 seats.  The latest poll findings are inconclusive, to say the least.

	308	Leger	Nanos	IRG	Forum	ARI
NDP	30%	28%	33%	29%	25%	31%
LIB	27%	28%	29%	26%	29%	24%
CPC	19%	20%	14%	19%	22%	17%
BLOC	21%	23%	23%	22%	21%	27%

Finally, there are the usual warnings:

• 10-second Tories (in BC, we called them 10-second Socreds).  Voters that decide in the final analysis to hold their nose and vote for the incumbent.
• “Cranky won’t says” – the 6-8% of voters who won’t and don’t cooperate on surveys are a statistical wildcard.
• The final weekend – voters have a sixth sense that is not entirely detectable.  The Redford win in 2012 manifested from an unease about Wildrose in the final week.  A combination of not-likely voters, strategic voters, and strange bedfellows changed the game.
• Advance and special votes – Perhaps up to 20% of votes are already in the bag.  Who do they favour? Will the winner on October 19th lose the overall election?

I’m not forecasting anything here – only caution.  It is clear that the Liberals have won the campaign thus far.  They started with low expectations and have exceeded them, and have to this point eclipsed the NDP in the ‘primary’ that established which party had the best chance to defeat the Conservatives.

Having said that, if I was a Liberal strategist, I would be tempering my grassroots’ naturally-occurring public-poll-based-optimism with Eeyore-like gloom and insist they are still running from behind.  I wouldn’t want to be like the gobsmacked Brit who couldn’t read the tea leaves – even at tea time.  If I was a CPC strategist, I wouldn’t assume the poll numbers will necessarily improve – it is going to be a gruelling week but a plurality is very much possible if they have a strong finish, particularly with the likeliest of voters.  If I was an NDP strategist, I would move mountains to move vote in Quebec.  If they lose Quebec, all is lost.

Ultimately, the great thing about campaigns is that it’s up to the voters.  The strategic voting organizations, the media outlets, and the pundits are not inside the voting booth.  It’s between the voters and the names on the ballot.  And that is the greatest variable – voters just damn well choose who they want to, sometimes with surprising results.