Final Projection Prior to the 2008 Election

The following is an application of the LISPOP seat projection model applied retrospectivrely using the actual election vote proportions provided regionally as presented in the Globe & Mail website. Two tables are presented below. The first is a postdicted seat distribution using the actual regional percentage splits (with the actual seat distribution in parentheses). The second reports the actual percentage regional splits from the election. In parentheses are the poll-based percentage splits (from the final Globe & Mail regional aggregations reported the morning of Oct.14) which formed the basis for the final Oct.14 LISPOP projection.

Table 1 - 2008 Seat Postdiction based upon actual regional vote splits

 
Other
Canada
147
78
34
49
1
2008 Election Results
143
76
37
50
2
Atlantic provinces
9
20
3
--
--
2008 Election Results
10
17
4
--
1
Quebec
11
14
1
49
--
2008 Election Results
10
13
1
50
1
Ontario
55
33
18
--
--
2008 Election Results
51
38
17
--
--
Manitoba, Saskatchewan, and territories
21
6
4
--
--
2008 Election Results
23
3
5
--
--
Alberta
28
--
--
--
--
2008 Election Results
27
--
1
--
--
British Columbia
23
5
8
--
--
2008 Election Results
22
5
9
--
--

Table 2 - Actual regional percentage splits with Oct.14 Globe and Mail regional aggregates

 
Ind.
Canada
38%
26%
18%
18%
8%
Globe and Mail Aggregates
35%
27%
19%
10%
--
Atlantic provinces
29%
35%
26%
--
--
Globe and Mail Aggregates
24%
37%
28%
--
1
Quebec
22%
24%
12%
38%
--
Globe and Mail Aggregates
21%
21%
12%
41%
1
Ontario
39%
33%
18%
--
--
Globe and Mail Aggregates
34%
35%
21%
--
--
Prairies
54%
15%
25%
--
--
Globe and Mail Aggregates
52%
18%
21%
--
--
British Columbia
45%
19%
26%
--
--
Globe and Mail Aggregates
40%
23%
25%
--
--

Observers will note that the greatest differential was in Ontario where the Conservatives had a 6% lead over the Liberals, but the pre-election estimate estimate was a 1% Liberal lead. those who would like to work it out for themselves, can simply apply the change from the 2006 election to the current figures and apply it to each seat in the designated region. For example in Ontario in 2006 the Liberals led the Conservatives in Ontario 40% to 35%, a 5% difference. By subtracting the 2008 difference of a 6% Conservative lead, one arrives at an 11% movement toward the Conservatives in the region. Apply that difference to each of the 2006 constituency results and do the same thing for the NDP to obtain an estimate of the results in Table 1. There is a bit more to the model than that , but this is the core, should you wish to verify the above figures. This translated into a 12 seat difference in Ontario from the final pre-election projection of 132-90-51-33-2, which is broken down elsewhere on the website.

Note:

The "regional swing model" is more fully explained in a paper originally prepared and presented by Dr. Barry Kay to the 1990 annual meeting of the Canadian Political Science Association, entitled "Improving Upon the Cube Law: A Regional Swing Model for Converting Canadian Popular Vote into Parliamentary Seats". It should be noted that the application of the model above does not make use of the "incumbency effect" described in that paper. In tests for past elections, using late campaign polls to project electoral outcomes, the model has proved to be accurate within an average of four seats per party since 1963. Readers interested in post-dictions for past federal elections dating back to 1963, for projections using pre-election polls dating back to the 1980 federal election and for three Ontario provincial elections, may contact me at bkay@wlu.ca.

Level of Government: