Friday, March 25, 2011

Canadian Elections:: Seat Projections by WLU &

2011 Seat Projections Based on March 2011 data, LISPOP-WLU
Update:: 13 April 2011, new blog listing seat projection sites.

Seat projections in a parliamentary system is tricky business, when based on opinion polling data at the regional or provincial level. I've been going through the WLU LISPOP site, based on Barry Kay's statements on the latest projections. The projections are based on polls taken in early March, before the scandals started to rock the Conservative's boat. The most recent Ipsos poll has the Conservatives with a substantial 20 point lead, although the Ipsos "house effects" favour the Tories. I agree with Kay that it's Harper's election to lose and that wherever the Conservatives are when the writ is dropped is likely to be a high water mark. I found the LISPOP methodology interesting [pdf], trying to increase predictive power by improving the so-called "cube law".

The LISPOP projections::

are consistent with those of, 2011 seat projections
Éric at threehundredeight has posted popular vote projections by riding in a table, which is useful, as is his thorough detailing of his methodology, which is based on a proportional swing model.

I think one of the limitations of such modeling is that it's hard to integrate riding effects, as evident in Éric's work. One of my interests is characterizing ridings in order to better target campaigning efforts, both across and within ridings. While punditry likes to talk of "swing ridings", so much of it is armchair handwaving, such as this phoned-in version by the National Post. I've been working on some indices that may be useful in characterizing party strength in a riding, using historical data and cluster analysis. I'll post this when I get around to crunching those numbers.

As for these predictions, I think it's early days. I've blogged about the fragmentation of Canadian politics and I've heard rumblings that support for Harper within Conservative circles is strained. It will be interesting to see how the polls shift over the next few weeks and how that affects the seat projections. I sense the election may be a bit like beach volleyball. Momentum at critical points can dramatically affect the outcome.

Ontario Urban Ridings, 2011 Seat Projections Based on March 2011 data, LISPOP-WLU

Twitterversion:: [blog] Canadian parliament seat projections by LISPOP/WLU & based on early March polling data.  @Prof_K


Blogger said...

NeoBux is the best work from home site.

Blogger said...

Get daily suggestions and methods for earning $1,000s per day ONLINE for FREE.