Political Perspectives is produced by the students and faculty of Carleton University's School of Journalism and Communication, Canada's oldest journalism school.


Bits and pieces, week five begins

Posted by ealboim under All, Election 2011, Election 2011 Campaign strategy, Election 2011 Faculty links, Election 2011 Media commentary

Elly Alboim

It’s too early to be definitive about anything in this final week of the campaign – other than the trite observation that something is happening. So, some observations:


In some ways, there is a predictable pattern emerging. Leadership numbers and party preference are starting to come into consonance. Mr. Harper’s numbers have always been a fit – he’s ahead in both and fluctuations have been mostly minor. Mr. Layton’s high personal numbers have stayed high and NDP preference is rising to meet them. Mr. Ignatieff’s low personal numbers have stayed low and Liberal preference is dropping to meet them. As to Mr. Duceppe, there hasn’t been a lot of work done on his leadership numbers over time. But presumably, they’re heading into consonance as well.

Timing of the NDP surge

There are two primary choices – take your pick.

1) It has happened early enough for it to build and reverberate elsewhere. Underlying this will be the dawning realization for anti-Harper progressive voters that there is now a comfortable place to lodge a protest vote. And a protest vote it likely is. It may also be broad and deep, if not explicitly partisan. There’s no reason to believe that NDP policy has suddenly converted legions of people or that most have come to believe that Jack Layton is best to be Canada’s next PM. But the NDP and Layton do seem suited to play a rigorous accountability role. So snowballing protest votes could continue to roll and spread for the next six days. That’s the wave theory.

2) It has happened too early for the NDP because it will now have to withstand a furious onslaught for a full week – an onslaught that will target its two major vulnerabilities: its economic management capacity and its new French language policies. By week’s end, there may be some buyer’s remorse that will blunt and stall the NDP’s surge. At a minimum, the Conservative Get Out the Vote machine will probably be way behind the rush of conservative voters trying to try to stop the NDP.

The polling cacophony

There are probably too many companies doing too many polls without adequate resources. Almost no one is spending what needs to be spent. There are new companies trying to make their mark. There are established companies vying to be first with trends for the media clients. There are at least five varying mixes of methodologies. There are rolling polls and picket polls. There are polls with long questionnaires and some with very short questionnaires. There are telephone polls manned by people and robots. There are on line polls with self-selected panels and randomly selected panels. Sample sizes range from under a thousand to over three thousand. Margins of error vary from large to small to non-calculable. And this week, everyone will suddenly up their sample and spend much more to get the last one right so that two years from now, we’ll remember that and not the shrill cacophony.

It’s a helluva a way to run an industry.

Seat projections:

Seat projections work on the basis of an assumption that has been tested and proven mostly right over time. Vote swings tend to distribute themselves relatively evenly and by shifting the last vote base by the current swing you can guess the disposition of a riding. Well and good. The Brits even have weird swingometer graphics they use on TV to let people in on the secret. But, and it’s a big but, the larger the territory you apply it to, the larger the gross errors. Parties don’t do that because they can’t afford those errors as they decide which ridings to target offensively or defensively and how to shift resources as polling changes. They poll sub-regions within provinces and project only within those regions. Ontario, for instance, traditionally has six to seven non-overlapping regions that are used for projection purposes. Quebec has five. Further, they poll sample ridings to make sure their swing assumptions derived from regional polling are not anomalous.

But most current public projection models work on the basis of provincial numbers and in some cases, groups of more than one province. The margins of error are large (because the sample sizes of many polls are small) and even where they are manageable, the territory being covered is too large and variable. When parties cluster geographically, applying a provincial swing may understate or overstate the potential shift.

Further, we have little experience with gauging four party races and vote transfers as is currently happening in Quebec. We also have little experience with the kind of structural change that may be happening. Shifting the vote base of a party that had only a marginal number of votes and is suddenly increasing it many fold is at best an uncertain enterprise.

Finally, seat projections work on public opinion polling of the population at large. They do not take into account voter turnout. As we learned last time, roughly the same absolute vote can mean a very different final percentage depending on the final turnout.

The rush to estimate the seat distribution of every poll that is made public has led to another sort of confusing cacophony. It’s the curse of easy access to spread sheets.

Gathering the evidence

The numbers don’t lie but the on the ground evidence is sometimes hard to assemble. Journalists travel and try to assess change but that is a tad haphazard. In Quebec, it is unlikely that any of the three federalist parties have any sort of organized intelligence gathering and reporting presence, let alone canvassing capability, in whole swatches of the province because they had no expectation of being competitive in those areas. Today, they are relying on wholly inadequate anecdotal information to make strategic decisions in ridings to which they never thought they would have to pay attention.

Media narrative

This final week will be the Grimm brothers’ story book of election campaigns. The potential narratives are legion and becoming more and more compelling.

There is the potential Greek tragedy in Mr. Ignatieff and the Liberals. There is the obvious Cinderella story in Mr. Layton and the NDP. Mr. Harper may get his majority he has doggedly sought (the little engine that could) or keep rolling his ball up an endless hill. If you want an alternative that’s a bit more modern, he may finally kick the field goal or like Charlie Brown, have the football snatched away yet again.

Then there are the narratives about May 3 and what combination of leaders and parties can be put together to do what in what sequence. That one is a rabbit hole of speculative wonder.

From what had been a very boring set of narrative options, we now have a menu of stories to fill page after page after page.

Elly Alboim is an associate professor of journalism and a former CBC TV parliamentary bureau chief.