Some updates

Today, I don't have a very structured post. Rather, I would like to share with you some updates.

1. The Atlantic: I talked about it in the past, but I would like to go back there a little bit. Using the recent polls published this week (or last week), I currently have the Tories at 32.4% in this region, the Liberals at 36.2% and the NDP at 25.4% (thanks to the recent surge). Given that some (if not all) of the increase for the Tories come from the ABC effect being gone, it actually means the Conservatives are DOWN in this region (except in NF-L of course). Indeed, adjusting for the ABC effect, the Tories were actually at 32.61% last time. On the other hand, the Liberals are up. Again, adjusting for the ABC effect being (halfway, according to my assumption) gone, the Liberals increased from 33.56% in 2008 to 36.2% today. As for the NDP, they are up slightly.

What this means is that the Tories, while gaining 1-2 seats in NF-L (Avalon for sure, maybe another one as well), they would actually lose seats elsewhere in the region. That could be important in order to secure a majority. On May 2nd, if the Tories don't get to 14-15 seats in this region, they will need huge gains in Ontario. And given that they are polled in average at "only" 41% in this province, that would unlikely lead to massive gains (especially since the Liberals seem to be up a little bit since 2008).

2. Ontario: Speaking of this province, I decided to work a little bit more on this province. In particular, I wanted to see if parties experienced different swings whether they won a riding by a close margin or lost by less than 5-points during the last election. I did that in order to see if the Tories could expect more gain in the GTA, especially in the West-905 where they lost a couple of ridings to the Liberals by less than 5-points (and where they target ridings aggressively). The results? Well, the Conservatives don't seem to increase more in ridings where they lost by a small margin. Yes they do a little bit (and they solidify more their ridings won closely as well), but we're talking less than 1-point. On the other hand, the Liberals dropped more in ridings they won closely. Between 2004 and 2008, they dropped 2.13-points more in those ridings. This is potentially important as it means that in the new model, some ridings such as Brampton-Springdale are now projected to turn Conservatives.

Still about these close wins/losses, surprisingly, the party who experienced the biggest differences in these ridings is actually the NDP. Indeed, they were up 2.26-points in ridings they won closely and up 3.7-points in ridings they lost closely! That alone explained a lot how the NDP managed to increases its seats in Ontario while actually experiencing a negative swing between 2006 and 2008. This is evidence of the increased efficiency in its vote. Will we see the same efficiency this time around? Or will the NDP waste ressources by spreading them everywhere, thinking they can win 25 ridings thanks to the surge? I don't know.

3. NDP in Quebec: I worked hard as well on this. At the end, since the NDP is so far away from the level of supports we've seen in the past, I actually simplified my model for this party. Specifically, I included less variables in the estimation process (I only take into account of the region now). And by tweaking slightly the coefficients for some regions, I now project 3-5 seats for the NDP in this province. Even more if the Bloc was to drop as low as 30%. I thus feel more comfortable using this mix of estimation/calibration for this party in this province alone.

[update] we have two polls for the two ridings in Abitibi. As usual, I'm not a big fan of riding-level polls as the sample size is too small and, by experience, they are usually way off. Nevertheless, these polls show two things: 1) the Bloc is down by a lot. I'm currently waiting for the new Crop-La Presse poll which is supposed to show some crazy things and I wouldn't be surprised to see the Bloc and NDP almost tied... 2) The swing experienced by the NDP in these two ridings seems to be mostly uniform, i.e: they are up 12 points in Quebec and 12-points in these ridings. It's a little bit amplified, but that's all. As for the projections, the "new" model is off 3.6-points and 5-points for the two ridings. For the latter, the difference could well come from the "star effect" of the new NDP candidate. Therefore reinforcing the fact that the swing is NOT amplified in the north for the NDP. So that's one region to cross out of the potential list lol

Please note that the changes to Ontario and Quebec/NDP haven't been implemented in the public version of the simulator yet. I'll do that later.