Prediction (#nanos, April 25, 2011): #CPC 166 #LPC 50 #BQ 47 #NDP 45 #elxn41 #cdnpoli #polls41

The title of this blog appears on Twitter, so I have tagged it with the appropriate hashtags to ensure that it gets placed correctly.  Unless you are a Twitter user, it won’t make much sense to you (like when Jack Layton said “Hashtag Fail” at the leaders’ debate) but for those of you who do follower Twitter, you’ll get it.

Throughout the election campaign, I have been monitoring the polls and posting seat predictions to Twitter.  These have bounced all over the place as the polls have moved, which has made the campaign a bit of an emotional roller coaster.

Things are, in a nutshell, getting very interesting.

Nanos Research released their latest daily tracking poll this morning, so I fed the data into my seat predictor and this is what I came up with.  Remember a Parliamentary majority requires 155 seats:

Conservative 166
Liberal: 50
Bloc Quebecois: 47
NDP: 45

When I factor the margin of error into my calculations, we get a spread of seats which each party is likely to win.  I prefer to use the spread rather than a definite number because polls, and specifically seat predictions, are notoriously volatile:

Conservative: 160-196
Liberal: 50-64
Bloc Quebecois: 47-48
NDP: 0-48

My methodology is fairly simple, and is based on a number of assumptions:

(a) That the same number of people will vote in this election as voted last time.
(b) That party support is uniformly spread across the geographic regions published by the polling firm.
(c) That party support is not affected by local demographics, such as ethnicity, faith, or other facts which drive votes one way or another.
(d) Independent candidates are not considered.

The calculation process involves breaking the country down into geographic areas similar to those used by the polling firm in their data gathering, and working out the party support in that particular area as a percentage.  Then, the change in support from last time is applied on a riding-by-riding basis.  A new winner is then calculated.

On a national scale, and, to a lesser extent, a provincial scale, this is fairly accurate.  I ran models during the campaigns of 2000, 2004, 2006, 2008 and each time, I was reasonably close at a national level, however it can not – and does not – come close to predicting individual riding results.

Things are looking very well for the Conservatives, who are in clear majority territory now.  Michael Ignatieff’s political career is pretty much over, Jack Layton, if the trends continue, may just become leader of the Official Opposition, and Gilles Duceppe seems to be on track to retirement.

About Steven Britton

Steve is a freelance programmer, partial billionaire, dad, Recovering Atheist, Conservative, and occasionally prolific blogger.