Wednesday, September 14, 2011

Quebec, and "how to" do simple projections.

Teddy here. A look at recent Quebec polls gave me an idea. The current polls looked good to me to do a simple projection.

Projections, the way I do them (and I do note that I got closer to anyone else - seat by seat - in the last federal election) uses a ratio method. While I have difficulty explaining this, it is shockingly simple to do.

First, I "rounded" the polls to make the projection as easy as possible. Here are the numbers I've used.
PLQ - 31.6%
PQ - 26.4%
ADQ - 16.3%
QS - 11.4%
PV - 4.4%

There is a reason that I've used these specific numbers. They are easy to use in a ratio. For the PLQ and PQ, this is .75 This means that I am presuming each party will lose one quarter of it's vote in each and every riding.

The ADQ is stable. The QS is up threefold, while the Greens have doubled. Going riding by riding, I get these results, and, I've provided some examples.

Green Gains:
none

QS Gains:
Laurier-Dorion (PLQ)
The QS is sitting at 2,963, and the PLQ at 9,769. The PLQ is reduced by one quarter to 7,327 while the QS gets three times it's vote and rises to 8,889 thus taking the seat.
Sainte-Marie--Saint-Jacques (PQ)
With the PQ at 9,135 and the QS at 3,009, is is clear the QS can win easily using these mathematical estimations.
Gouin (PQ)

ADQ Gains:
Montmorency (PLQ)
Portneuf (PLQ)
The ADQ here is at 9,388 and will remain stable. The PLQ is at 11,055 and loses a quarter of their vote, resulting in a new total of 8,291
Vanier (PLQ)
Beauce-Sud (PLQ)
Drummond (PQ)
Levis (PLQ)
Nicolet-Yamaska (PQ)

The ADQ is the big winner, gaining seats the PLQ and PQ no longer have the strength to hold on to.

Results:
PLQ - 60
PQ - 47
ADQ - 14
QS - 4

The great thing about these projections is that those of us with math talents, can usually eyeball the results. We can easily half or double numbers in our heads. While the accuracy of such "quick" projections might not be 100% reliable, they are certainly closer than anything you could get, and faster as well, than any other "simple" method.

No comments:

Post a Comment