If I could design a leadership race from scratch, maybe I'd design a US-style primary system just for the drama of it.
My basic principle, though, is that whoever leads the party should have the backing of caucus — be it directly by having them choose, or indirectly by requiring potential candidates to be signed-off on by a majority of caucus members.
I very much like the format of this new election prediction. I found that the broad ranges in other reports are unhelpful, even if an accurate reflection of the data. You have qualified your opinion, and the result is helpful for me to visualize. I wonder whether running a few dozen 'Monte Carlo' simulations and taking the average might be a worthwhile comparison? Maybe reject the outliers and keep within a standard deviation (or two)?
That's starting to sound like a statistical model! I could include some ranges that take into account the marginal seats, but I want to keep things simple. When I dive into the numbers from time to time I'll get more into the potential ranges for each party.
The reason I like your method (and not thge ranges) is that marginal seats tend to swing in one direction. A 2% shift one way in one riding is often accompanied by a similar shift in the next one, and so on. Partly why Quebec (and others) swing so dramatically.
Great point — when these sorts of projections are off, the individual errors tend not to always cancel each other out. One party tends to over-perform at the expense of another.
I like the election prediction format. It gives a snapshot of the parties in each jurisdiction. Those of us who are not immersed in this every day can lose track of what is going on in each place. We also may not see how things are evolving.
I think it could be improved by including a comparison with the present standings - as it appears you intend to include. I also would like to see some visual representation of how these standings are changing over time.
If I could design a leadership race from scratch, maybe I'd design a US-style primary system just for the drama of it.
My basic principle, though, is that whoever leads the party should have the backing of caucus — be it directly by having them choose, or indirectly by requiring potential candidates to be signed-off on by a majority of caucus members.
I very much like the format of this new election prediction. I found that the broad ranges in other reports are unhelpful, even if an accurate reflection of the data. You have qualified your opinion, and the result is helpful for me to visualize. I wonder whether running a few dozen 'Monte Carlo' simulations and taking the average might be a worthwhile comparison? Maybe reject the outliers and keep within a standard deviation (or two)?
That's starting to sound like a statistical model! I could include some ranges that take into account the marginal seats, but I want to keep things simple. When I dive into the numbers from time to time I'll get more into the potential ranges for each party.
The reason I like your method (and not thge ranges) is that marginal seats tend to swing in one direction. A 2% shift one way in one riding is often accompanied by a similar shift in the next one, and so on. Partly why Quebec (and others) swing so dramatically.
Great point — when these sorts of projections are off, the individual errors tend not to always cancel each other out. One party tends to over-perform at the expense of another.
I like the election prediction format. It gives a snapshot of the parties in each jurisdiction. Those of us who are not immersed in this every day can lose track of what is going on in each place. We also may not see how things are evolving.
I think it could be improved by including a comparison with the present standings - as it appears you intend to include. I also would like to see some visual representation of how these standings are changing over time.
Thank you for asking and Happy New Year.
That's a good suggestion about tracking charts, I will try to include them now and then when some trends are developing.