If the Liberals and NDP were to both put up terrible candidates and disorganized campaigns, it seems unlikely that they’d together pull less than 25% or even 30% of the vote (heck, 6% still voted NDP even when the party no longer had a candidate). That leaves at most 75% up for grabs between other parties and Lunn would have to get 37% or less (similar to what you project in your best case, Greg) for May to win. That’s likely only if the Conservatives fall apart during a campaign or if someone other than Lunn stood as the Tory candidate — both unlikely scenarios.

In other words, the Green Party would need a perfect storm of running their own beautiful, high-profile, well-organized campaign (bringing in a strong pro-green voter turnout and securing party switchers) while seeing all three of their main competitors run their campaigns into the ground.

This overlooks the importance of the categorically different ‘base credibility’ of the Green brand in Nova Scotia versus BC [or Southern Ontario].

May came into a situation where the Green brand is a fringe phenomena. So you insert her media presence and credibility into that and of course you are going to get a much bigger bump, then you would in BC [or Ontario] where the GPC already has a significant core of voters and leaners.

In Nova Scotia May could cream some of those same demogaphic, lifestyle, and ideological ‘types’ who vote or lean Green in BC, but who in Nova Scotia had simply never had reason to consider the possibility.

The Greens do draw from all 3 parties, but by far more heavily from the NDP, then the Liberals and only a very small amount from the CPC. That’s solidly backed up by recent internal data I have. If you have something concrete to refute that in LNC, feel free to post it.

You either missed or disagreed with what I said about by-elections: “while a candidate can deliver a much higher premium in a by-election (as May did in London North Centre), since the government is not on the line and more votes are personal votes, but not in a general election”. LNC was a by-election. And as you note, there was no Liberal candidate in CN.

I projected 33-36% for May in Central Nova in 2008 and she got 32.2%. So if anything, I slightly overestimated the number of personal votes she would get. People said that CN would be too wild to predict, but my range was pretty much what she got.

Doesn’t look like May can expect much help from Penn.

First, SGI has a history of strategic voting, that is, many voters in the LIB/NDP/GREEN parties will look to see who is likely to be number 2, and swing their vote to that candidate. As you noted in your analysis, Penny benefited from that swing in the last election and you corrected for that. However, the amount of that swing is uncertain, and the amount of additional loss to Liberal party if another candidate was the perceived leader is not considered in your analysis.

Second, you have quoted polling which suggests that the vote attrition from the Conservative party is minor. While I have no doubt that your study indicated this outcome, it is certainly at odds with the London North Centre outcome, and with historical studies that have been done which suggest that equal numbers migrate to the Greens from teh Conservatives, Liberals and NDP.

Finally, I believe that your estimate of Elizabeth’s “personal votes” is less than half of what we have seen, first in LNC, and lately in Central Nova. While CN was affected by the absence of a Liberal candidate, it is also clear that a lot of Liberals simply didn’t vote.

In short, I don’t think that a regular kind of projection is going to give us anything close to a reliable result. Please note that I say this while acknowledging the amazing job you do with your projection model, and the great results you have achieved with it. I just think that if EM ends up in SGI, it will too wild a ride to predict.