UBC-ESM Election Forecaster
developed by Prof. Werner Antweiler
The UBC-ESM Election Forecaster is an analytical tool
developed for researchers of Canadian elections as well as traders on
the UBC Election Stock Market (UBC-ESM).
The UBC-ESM Election Forecaster's predictive ability
is based on a "voter migration matrix". A user of this software can
guess the probability with which a voter of a particular party will vote
for the same or another party in the forthcoming election. Once all
probabilities have been guessed and entered into an on-line form,
the voter migration matrix is applied
uniformly in each electoral district. This leads to predictions of
the votes for each party, including a prediction of the winner in each riding.
The sotware also identifies "safe" and "marginal" seats for each party.
The UBC-ESM Election Forecaster is available for the following elections
(past and forthcoming):
- Canadian Federal Elections
- Alberta Provincial Elections
- British Columbia Provincial Elections
- Manitoba Provincial Elections
- Newfoundland and Labrador Provincial Elections
- Nova Brunswick Provincial Elections
- Nova Scotia Provincial Elections
- Ontario Provincial Elections
- Princed Edward Island Provincial Elections
- Saskatchewan Provincial Elections
Because the federal and provincial election systems
are based on "first-past-the-post", prediction of the election
result in terms of the seats distribution in Parliament or Legislative
Assembly must be based on a prediction of the election result
in each constituency.
The use of a voter migration matrix reflects the notion that
voters change their opinion about candidates and parties in
a similar way across the entire province. That is, a party that
loses in one city tends to lose similarly in another city.
In the simplest form, this "swing" can be applied
across all constituencies.
There are several caveats with applying a simple voter migration
matrix to a previous election result. First, this method does not
reflect the change in population. Some children have reached voting
age, some (mostly older) people have died, and other people have moved
to Canada or away from Canada. Second, it does not reflect the
possibility that voters in the previous election opt to abstain in the
forthcoming election, or abstainers in the previous election choose to
vote in the forthcoming election. Third, the voter migration matrix
is applied identically across all constituencies, thus ignoring
important local factors. However, to account for the significant
differences between provinces in the federal election, the forecasting
program can be applied to individual provinces to obtain more accurate
predictions. Fourth, when the boundaries of electoral districts
have changed between the last and the fortcoming election, or when
the number of constituencies has been altered, the voter migration
matrix cannot be applied. (For example, this was the case in Ontario's
1999 election. In the federal election in 1997, a few constituency
boundaries were changed as well.)
2017 Sauder School of Business,
University of British Columbia.