As pointed out in one of the comments of my previous article, I was lazy lol Specifically, I took the second choices as given by Ekos and applied them, without adjusting for the obvious bias in Quebec. The problem is of course for the Bloc. When Ekos tells us that 4.7% of the Liberals voters (in Canada) have the Bloc as their second choices, these 4.7% all come naturally from Quebec. Therefore, within Quebec Liberal voters, the Bloc is at way more than only 4.7%. I was lazy in my first simulation because I wanted to focus on the CPC, not the Bloc. But since it doesn't take much longer to do it right, I decided to re-do the calculations.
Using the fact that the second choices for the Bloc all come from Quebec (I'm assuming no one in Ontario answered the Bloc as second choice, which would make no sense, anyway, I'm pretty sure the interviewer didn't even give this option to people outside of Quebec) and assuming some kind of proportionality for the remaining second choices (specifically, except for the second choices for the Bloc, the ones for the other parties are proportionally the same in Quebec and in the Rest of Canada). We get the following second choices:
| Quebec | First choices | ||||
| seconc choices | CPC | LPC | NDP | Green | Bloc |
| CPC | - | 13.1% | 8.9% | 15.6% | 7.6% |
| LPC | 17.9% | - | 26.0% | 20.8% | 14.9% |
| NDP | 17.8% | 36.2% | - | 21.1% | 42.2% |
| Green | 7.4% | 11.4% | 19.2% | - | 11.3% |
| Bloc | 10.2% | 19.2% | 27.8% | 17.1% | - |
A simple rule of thumb could be to multiply the Canada-wide second choices for the Bloc by 4, as Quebec usually represents 25% of the votes for a party. Naturally, the calculations differ for each party, as some take more or less votes from this province. Note: each column doesn't add up to 100% because I omitted the "others" and "no second choice" categories.
| ROC | First choices | ||||
| seconc choices | CPC | LPC | NDP | Green | Bloc |
| CPC | - | 16.2% | 12.3% | 18.8% | - |
| LPC | 20.0% | - | 36.1% | 25.1% | - |
| NDP | 19.8% | 44.8% | - | 25.5% | - |
| Green | 8.2% | 14.2% | 26.6% | - | - |
| Bloc | 0.0% | 0.0% | 0.0% | 0.0% | - |
So we can now re-do the simulation of having an instant run-off, but with better second choices, especially in Quebec where the previous ones were clearly underestimating the votes the Bloc could get from the other parties.
The results are here:
| seats | CPC | LPC | NDP | Green | Bloc |
| actual | 146 | 78 | 37 | 0 | 47 |
| I-RO | 136 | 87 | 38 | 0 | 47 |
The drop for the Conservatives is a little bit bigger than before. But the big difference is naturally for the Bloc, who would now not lose any seat.
By the way, some of you might wonder why the second choices for the ROC are also affected. It just make sense, if 15.4% of the Liberals voters, Canada-wide, have the Conservatives as second choices, when the Bloc is an option, it necessarily means that this percentage is higher in the ROC where Liberal voters don't have the Bloc as an option (again, assuming general proportionality). Of course, all these calculations have some assumptions and the only way to see if I'm right would be to have Ekos breaking down the second choices by province.
