Since the model seems to be working pretty well, it is now time to make manual adjustements to some ridings. For instance, we now know that some "high profile" independant candidates are running. So here is a lit of some of the changes that have been made to the model (the changes have also been applied to the simulator).
1. Saanich-Gulf-Island: Someone pointed out to me that the NDP got only 5% of the vote in 2008 only because of an horrible campaign. It indeed seems to have been the case, as the NDP, who was at 21.6% in 2004 and 26.5% in 2006, fell to 5.7% in 2008. The provincial swing of the NDP can't account for that, even with a region-specific coefficient. So instead of using the actual results of 2008 as the baseline, I'm now using the projected results in 2008. Specifically, here is a comparison. See the estimated results as what should have happened, given the provincial swing of each party.
| Actual | Estimated | |
| CPC | 43.4% | 44.2% |
| LPC | 39.4% | 24.9% |
| NDP | 5.7% | 20.7% |
| Green | 10.5% | 14.0% |
As you can see, the Liberals indeed benefit from a boost due to the drop of the NDP.
Using the estimated percentages as the baseline, I then apply the current provincial swings, as well as the Elizabeth May effect. The end result, currently, is still to project the Conservatives to win and the Green to be around 30%, but the Liberals are now 3rd and the NDP is not at 0% anymore, although they are definitely last.
2. Simcoy-Grey: Helena Guergis is a former Conservatives who was forced to resign because of a RCMP investigation. Even though nothing wrong was found at the end, she was excluded from running for the CPC and is thus running as an independent. It's always tough to estimate how much of the vote such a candidate can steal. A similar candidate in Alberta in 2008, in Edmonton-Sherwood Park, managed to get 32.45% and almost won the riding. Given the local association wasn't necessarily too happy to have the PM office preventing Guergis to run again, we could well see a real threat to the Tories. It's almost impossible to assess exactly what score Helena Guergis will get, but I'll give ger 30% of the Conservatives vote. It'as arbitrary and I might change this decision later on. I don't display the result for Guergis in the pdf though, not at the moment, since by construction, she can't win since she's getting only 30% of the vote).
3. Edmonton-Sherwood-Park: speaking of this riding, the same ex-CPC independent candidate is running again. Given that he failed to beat the official Tory candidate in 2008, I'm simply ignoring him. My guess is that it has been too long for him to win. Yes he will still votes, but no he won't prevent the Conservatives from winning.
4. St-John East: in 2008, the NDP got a huge boost in this riding, because of the return of a star candidate (and the ABC effect as well). Here I'm assuming that half og his boost is gone (and half of the ABC effect is gone as well). Therefore, the baseline results are 25.6% for the CPC, 21.3% for the LPC and 47.1% for the NDP. After that, I apply the calculations using the regional swing. If you think that cutting 50% of the St-John effect is too much, let me know.
5. Port-Neuf Jacque-Cartier: Here it's pretty simple, I'm assuming André Arthur is the Conservative candidate. After all, he votes with them 99% of the time and this party doesn't run a candidate there...
Let me know if you could think of other ridings that should be manually tweaked. I always appreciate your input.
