The Record So Far


338Canada Seat Projection Record So Far


The 338Canada / Qc125 model has thus far covered five general elections in Canada: Ontario 2018, Quebec 2018, Alberta 2019, Canada 2019, and New Brunswick 2020. In total, 723 electoral districts were projected by the model and has correctly identified the winner in 651 districts - a success rate of 90%. Here is the complete breakdown of safe, likely, leaning and toss up districts:
The 338Canada / Qc125 model has thus far covered five general elections in Canada: Ontario 2018, Quebec 2018, Alberta 2019, Canada 2019, and New Brunswick 2020. In total, 723 electoral districts were projected by the model and has correctly identified the winner in 651 districts - a success rate of 90%. Here is the complete breakdown of safe, likely, leaning and toss up districts:

Districts Correct winner Wrong winner Correct %
Safe 279 275 4 99%
Likely 208 200 8 96%
Leaning 114 97 17 85%
Toss up 122 79 43 65%
Total 723 651 72 90%
Districts Correct winner Correct %
Safe 279 275 99%
Likely 208 200 96%
Leaning 114 97 85%
Toss up 122 79 65%
Total 723 651 90%



As mentioned above, the correct winner was identified in 90% of all districts of these five general elections. Among the 72 remaining districts, 45 of the winners (6%) obtained a share of the vote that was within the projection's margin of error (moe). Only 27 districts (4%) were complete misses.
As mentioned above, the correct winner was identified in 90% of all districts of these five general elections. Among the 72 remaining districts, 45 of the winners (6%) obtained a share of the vote that was within the projection's margin of error (moe). Only 27 districts (4%) were complete misses.





2020 New Brunswick election - Sept.14, 2020


See how pollsters fared in this election

Parties Final 338Canada Seat Projection Election results
PC 27 27
LIB 17 17
GRN 4 3
PA 1 2
NDP 0 0
Most likely outcome PC majority PC majority
Parties Final 338Canada Seat Projection Election results
PC 27 27
LIB 17 17
GRN 4 3
PA 1 2
NDP 0 0
Most likely outcome PC majority PC majority



The New Brunswick general election of September 2020 was the first election held during the covid-19 pandemic in Canada. It was also the first projection of a "small" province for the 338Canada model, and it turn out to be its most successful to date. Overall, the model correctly identified 47 of 49 winners (96%) even though there were only a few polls during the campaign.
The New Brunswick general election of September 2020 was the first election held during the covid-19 pandemic in Canada. It was also the first projection of a "small" province for the 338Canada model, and it turn out to be its most successful to date. Overall, the model correctly identified 47 of 49 winners (96%) even though there were only a few polls during the campaign.

Districts Correct winner Wrong winner Correct %
Safe 26 26 0 100%
Likely 14 14 0 100%
Leaning 4 4 0 100%
Toss up 5 3 2 60%
Total 49 47 2 96%
Districts Correct winner Correct %
Safe 26 26 100%
Likely 14 14 100%
Leaning 4 4 100%
Toss up 5 3 60%
Total 49 47 96%





2019 Canadian Federal election - Oct.21, 2019


See how pollsters fared in this election

Parties Final 338Canada Seat Projection Election results
LPC 142 157
CPC 125 121
BQ 33 32
NDP 35 24
GRN 2 3
PPC 1 0
IND 1 1
Most likely outcome LPC minority LPC minority
Parties Final 338Canada Seat Projection Election results
LPC 142 157
CPC 125 121
BQ 33 32
NDP 35 24
GRN 2 3
PPC 1 0
IND 1 1
Most likely outcome LPC minority LPC minority



The most challenging endeavour so far has been the 2019 Canadian federal election. Although most polls generally hit close to the mark, they collectively underestimated the Conservatives and overestimated the NDP. Nevertheless, the model identified 299 of 338 correct winners. Out of the 39 wrong winners, 27 were within the model's margin of error. Only 12 were complete misses.
The most challenging endeavour so far has been the 2019 Canadian federal election. Although most polls generally hit close to the mark, they collectively underestimated the Conservatives and overestimated the NDP. Nevertheless, the model identified 299 of 338 correct winners. Out of the 39 wrong winners, 27 were within the model's margin of error. Only 12 were complete misses.

Districts Correct winner Wrong winner Correct %
Safe 88 88 0 100%
Likely 106 103 3 97%
Leaning 60 53 7 88%
Toss up 84 55 29 65%
Total 338 299 39 88.5%
Districts Correct winner Correct %
Safe 88 88 100%
Likely 106 103 97%
Leaning 60 53 88%
Toss up 84 55 65%
Total 338 299 88.5%





2019 Alberta election - April 17, 2019


See how pollsters fared in this election

Parties Final 338Canada Seat Projection Election results
UCP 60 63
NDP 26 24
ABP 1 0
LIB 0 0
Most likely outcome UCP majority UCP majority
Parties Final 338Canada Seat Projection Election results
UCP 60 63
NDP 26 24
ABP 1 0
LIB 0 0
Most likely outcome UCP majority UCP majority



In the Alberta general election of April 2019, the 338Canada model correctly identified 82 of 87 district winners. Out of the five misses, three were within the confidence intervals.
In the Alberta general election of April 2019, the 338Canada model correctly identified 82 of 87 district winners. Out of the five misses, three were within the confidence intervals.

Districts Correct winner Wrong winner Correct %
Safe 38 38 0 100%
Likely 30 29 1 97%
Leaning 11 10 1 91%
Toss up 8 5 3 63%
Total 87 82 5 94%
Districts Correct winner Correct %
Safe 38 38 100%
Likely 30 29 97%
Leaning 11 10 91%
Toss up 8 5 63%
Total 87 82 94%




2018 Quebec election - Oct.1, 2018


See how pollsters fared in this election

Parties Final 338Canada Seat Projection Election results
CAQ 63 74
LIB 44 31
PQ 11 10
QS 7 10
Most likely outcome CAQ majority CAQ majority
Parties Final 338Canada Seat Projection Election results
CAQ 63 74
LIB 44 31
PQ 11 10
QS 7 10
Most likely outcome CAQ majority CAQ majority



In October 2018, I published a final Quebec projection in L'actualité magazine showing the CAQ led by François Legault was going to win - not assuredly, but most likely a majority. However, the final polls of the election missed the mark significantly by underestimating CAQ support (see for yourself here). The poll average for the CAQ was somewhere between 31-33%, and, when all the votes were counted, it had won 37%.

Nevertheless, the model correctly identified 112 winners out of 125 districts. Out of those 13 misses, four were within the margin of error. Nine districts were complete misses, most of those going to the CAQ at the expense of the Liberals, largely due to the polls underestimating CAQ support (and also a very low Liberal turnout).
In October 2018, I published a final Quebec projection in L'actualité magazine showing the CAQ led by François Legault was going to win - not assuredly, but most likely a majority. However, the final polls of the election missed the mark significantly by underestimating CAQ support (see for yourself here). The poll average for the CAQ was somewhere between 31-33%, and, when all the votes were counted, it had won 37%.

Nevertheless, the model correctly identified 112 winners out of 125 districts. Out of those 13 misses, four were within the margin of error. Nine districts were complete misses, most of those going to the CAQ at the expense of the Liberals, largely due to the polls underestimating CAQ support (and also a very low Liberal turnout).

Districts Correct winner Wrong winner Correct %
Safe 74 70 4 95%
Likely 21 19 2 90%
Leaning 18 14 4 78%
Toss up 12 9 3 75%
Total 125 112 13 90%
Districts Correct winner Correct %
Safe 74 70 95%
Likely 21 19 90%
Leaning 18 14 78%
Toss up 12 9 75%
Total 125 112 90%




2018 Ontario election - June 7, 2018


See how pollsters fared in this election

Parties Final 338Canada Seat Projection Election results
PC 70 76
NDP 47 40
LIB 6 7
GRN 1 1
Most likely outcome PC majority PC majority
Parties Final 338Canada Seat Projection Election results
PC 70 76
NDP 47 40
LIB 6 7
GRN 1 1
Most likely outcome PC majority PC majority



In June 2018, I published an Ontario projection that showed Doug Ford's PC was most likely going to win a majority government. On its first ever general election, the preliminary version of the 338Canada model identified the correct winner in 111 of 124 electoral districts in Ontario. Out of the 13 misses, 11 winners had results within the model's margin of error. Only two districts were complete misses.
In June 2018, I published an Ontario projection that showed Doug Ford's PC was most likely going to win a majority government. On its first ever general election, the preliminary version of the 338Canada model identified the correct winner in 111 of 124 electoral districts in Ontario. Out of the 13 misses, 11 winners had results within the model's margin of error. Only two districts were complete misses.

Districts Correct winner Wrong winner Correct %
Safe 53 53 0 100%
Likely 37 35 2 95%
Leaning 21 16 5 76%
Toss up 13 7 6 54%
Total 124 111 13 90%
Districts Correct winner Correct %
Safe 53 53 100%
Likely 37 35 95%
Leaning 21 16 76%
Toss up 13 7 54%
Total 124 111 90%