Introducing Volatility Ranges – Dealing with Volatile Voters

A short post on a new measure we have been working on with seat projector and poll analyst Bryan Breguet from  Last week we released our latest federal ballot tracking that found a close three way race between the Tories, NDP and Liberals.

We have been going back and forth for a while since the Alberta and BC elections discussion how pollsters could report polling results with a little less certainty.  One of the mistakes we (pollsters, those interested in politics, the news media) often make is being overly certain – or at least – how polls are reported by pundits and the news media.

While polls are snapshots in time, we know from a lot of academic research that Canadians are volatile when it comes to voting behaviour.  The concept of “partisan dealignment” advocated by Harold Clarke and his colleagues studying Canadian elections in the 1970s and 1980s and more recent 2004 and 2006 elections argues that short-term factors such as current issues, leadership, and campaign effects have a greater impact on voter behaviour than social factors such as class, race, language, or region (Clarke et al, 1979; 1991).

Bryan developed a way to measure the potential range of support of political parties using our scaled “likelihood to vote” survey questions.

We first introduced these scales during the 2011 Ontario election.  Basically, survey respondents are asked to rate their likelihood of voting for each of the main parties on a scale from 0 to 10 where 0 means they will absolutely not vote for the party and 10 means they will absolutely vote for the party.

Bryan explains how he turns our “likelihood to vote” question into probabilities using the traditional vote question over on his blog.  Using these probabilities he then runs simulations to determine the probability of different vote outcomes and what this has on the aggregate level.  This produces a range of popular vote scores that could happen based on the likelihood of voters to vote for each of the main political parties.

Using the latest data from our survey conducted at the end of August and early September, the chart below details the committed vote intention, the minimum and maximum level of support for each of the five main parties in Canada.


 As you can see, our latest poll had a three-way tie between the CPC, LPC, and NDP.

And when you look at the volatility ranges, the scores produced for the three main parties are basically the same meaning that two years out from the next election, anything can really happen.  There’s a scenario where the Tories could get 37% of the vote but also one where their support drops to 21%.  The same is basically true for the Liberals and the NDP.

If we move away from looking at vote intention as a static variable and one that can move – it is clear that the vote intentions of Canadians are volatile and any of the three main parties could conceivable win the largest share of the vote.

David Coletto is CEO of Abacus Data and leads its Public Affairs research practice. He has a PhD from the University of Calgary and is an adjunct professor at Carleton University.  He’s an avid road cyclist.

Contact David Coletto:

T: 613-232-2806 x. 248



Sean on Twitter David on LinkedIn David on Google Plus