April 24th 2013: What would it take for a Green candidate to be elected?

If my model is relatively kind on the Conservatives (especially when compared to other models), it can seem incredibly hard on the Green party. Indeed, with the latest numbers, I don't project a single Green MLA and even the best case scenario would only see one Green candidate win his/her riding. That seems low since this party is polled around 10-12%.

People seem to think that one Green candidate has some real chances: Andrew Weaver in Oak-Bay-Grodon-Head. He's a well known candidate and definitely qualifies as a "star candidate". These candidates can perform significantly above expectations, or above what a "normal" candidate would. The best (and potentially more relevant) example is Elizabeth May. No matter where she ran (Ontario, NS or BC), she got a share of the vote much greater than the candidate before. During the last federal election, I spent a lot of time trying to estimate a "May effect" based on her past results. I found out that it could be as high as 20-25 points. However, during the actual eleciton in 2011, she seemed to have done even better. She indeed won the riding by increasing the share of votes for the green party from 10.45% in 2008 to 46.33% in 2011. All that while the Green party actually decreased by almost 2 points provincially! This is a typical examples of the effect of a star-candidate. However, and having done that for multiple elections, I can tell you that such a big effect is rare. Elizabeth May wasn't a typical star-candidate. She was a popular leader, known across the country. She even participated to the debate in 2008. She also seemed to have taken full advantage of the collapse of the Liberal party in BC.

So let's look at the results in Oak-Bay-Gordon-Head in 2009:

BC NDP: 44.3%
BC Liberals: 46.5%
Green: 9.1%

The current projections, without any bonus/boost/star-candidate effect are:

BC NDP: 47.1%
BC Liberals: 33.2%
Green: 22.6%

As you can see, the model already acknowledge that Victoria (and Vancouver Island in general) is a good region for the Green party. Indeed, it transposes a provincial swing of around 5-points (according to recent polls) into a riding-level swing of almost 13-points. So don't think that the model is biased against the Green or doesn't account for some regional effects. The problem however is the fact that the NDP isn't going down, not provincially at least. Therefore the model will predict a stable NDP in this riding as well.

But what if we account for the increased potential of Andrew Weaver? As opposed to May though, we don't have past data to estimate the magnitude of this effect. So we'll have to guess (until we get a riding-level poll for instance). Let's be generous and assume that the average effect is 15-points, taken mostly and evenly from the NDP and the Liberals. I say in average cause over the 1000 simulations that I run, this effect will sometimes be as low as 0 and sometimes be as high as 30-points. Again, we need to add some uncertainty here, especially due to the lack of previous data.

So with this bonus, out of 1000 simulations, the Green party is now projected to win this riding about 40% of the time, with the NDP winning the remaining 60%. Quite a difference!

So how should you interpret these numbers? Well, what they mean is that without a strong candidate-effect, the Green party won't win a riding, not at the current level this party is in the polls. In the mentioned riding, the effect has to be around 15-points for Andrew Weaver to have a real chance. And this effect better comes at the expense of the BC NDP, at least partially. So if you believe that Andrew Weacer can indeed pull that off, then there is a race in this riding. But I can guarantee you that getting a boost of 15-points is very hard and very rare. Usually, only the leader of a party could achieve that. So while I would love to see more parties elected at the legislative assembly, I have to remain skeptical of the chances of the Green party in BC.