Earlier, I said that I could use 3rd place results to determine what things would look like without the Saskatchewan Party. I've been able to do so. It became quite simple, actually, once I was able to fill in the gaps in the PC Candidate list.
There are a number of ways I was able to do this, and I wish to detail the method(s) I've used so that you may do so yourself.
1 - Use a swing from another party, or other parties. This is most difficult, as you need to figure out exactly where the swing is coming from. It's also most likely that the swing is not coming from just a single party, but multiple parties; compare, for example, the Reform vote in 1993, and you'll notice it does not match the PC vote, but rather, a combination of PC and NDP vote, but with the NDP weighted less in areas where the NDP did well, and vice versa in certain rural areas.
2 - Check the past. In the last election the party ran in Moose Jaw, but it did not this time, as such, one way to fill in the gaps is to simply pen in the last-time results, with possible adjustments for vote growth (or shrinkage) based on the ratio of the last-time vote, with further adjustment for current or past star candidates. This is also the sole method I used to fill in slots for the Green Party. You can also use results from another level of government (Federal or Provincial) but it is very important to keep in mind unique circumstances, (like the BC Liberals), or that strengths are not equal (like the Tories right here in our example of Saskatchewan)
3 - Check nearby ridings, or similar ridings. One simple way to fill in a gap is to take the average of all neighbouring ridings. This, clearly, has some faults to it, but it can do in a crunch. Always be sure to reduce the result by 10% or 20% when the gap was created by the party being unable to find a candidate, as opposed to an error (like a death or being slightly late with the paperwork) as being unable to find a candidate tends to indicate weakness in any particular area.
4 - Use the area's average. This is more useful in cities. For example, the PC Party missed a few spots in the "South and East" region I use, and as such, using the area average (with the same reduction noted above) is a good way to achieve the result you need.
5 - Randomness. At some point, especially for the tiny parties, picking a random number actually provides you with the most accurate. Note this only works for tiny parties, parties taking about 1% of the popular vote in the ridings where they run, and, who only run in a handful of ridings. You can also use this to decide upon a random factor; for example, I've decided that 0.1% of the vote is an accurate number for the "Other" parties in Saskatchewan, based on randomness alone.
6 - Update Update Update. If you plan to keep using your file, you will want to keep things up to date. Any of these methods can work for a growth of about 2X (double the last-time result) plus about a raw 2% of the vote. As such the PC Party, which took 4.5% in the ridings they ran. This is too much for Randomness. As such the "area average" can be used to determine the entire PC vote province-wide, and, accurately work even if the PC Party takes 11% of the vote. (double plus 2%) If, however, they are polling at 12%, you will want to use the next method up on the list. There is no exact math to determine when you need to use #1 from the list, but I am personally comfortable that using accurate swings will work even if a party is up 100 fold (0.5% last time, 50% this time)
Remember that these numbers are a baseline, to which you can add further adjustments later on.
I've also divided the province into a number of regions, as outlined on this map:
The reason for these regions focuses almost exclusively on the PC Party, especially the rural split. This split is the most useful for using method 4 as outlined above, and that was the first method I used.
One important thing I noticed was a negative swing compared to the Liberals; that is, where the Liberals did well, the PC Party did poorly. I also noticed that where the Sask Party did well, the Pc Party did well. This helped me determine which ridings would lean more Liberal (where the Liberals and Sask Party did well) and which were more Conservative (where the Sask Party did well, but the Liberals did not)
Either way, when I ran the math, this is what I ended up with:
The NDP's weakness becomes clear, as even with the Saskatchewan Party split in two, they fail to win a majority, or even a plurality. What is most interesting is there has been a change, over time, in areas of strength. The Liberals previously had done in the central-east areas, near Yorkton, but now find themselves doing better near the Battlefords. Federal results over the last two decades do indicate a slight shift in this direction as well; this seems to have been instigated by conservative and Conservative strength in the eastern area or the province, which has prompted former Liberals to choose between the NDP and the Tories, while the opposite has happened in the areas around Battleford, where the NDP's weakness has allowed more votes go go Liberal.
It's difficult to say how such an alternate election would have turned out. The parties have been "merged" as the Saskatchewan Party since 1997. You'd need to re-run the 1999, 2003, 2007, and 2011 elections to determine a more "real" result. In 1999 in particular, it's likely the NDP would have faced a minority, but also faced a Liberal party unwilling to deal. 2007 also may not have been won by either the Liberals or Tories.
Regardless, this is a potentially useful way to indicate ridings (IE Liberal ones on the map) that the Saskatchewan Party could lose if it decides to move too far to the right, or, ridings they could lose to a resurgent PC Party (the blue ones) if it moves too far to the centre.
As it stands, Brad Wall is too smart to pull his party off it's successful track, and it would take a significant change for any of this to actually be of any impact to the Saskatchewan Party.
No comments:
Post a Comment