[Ottawa – October 20, 2019] After an exciting campaign, the 43rd Canadian general election has come to a close. We at EKOS believe we did a very good job in charting the direction in one of the tightest campaigns in recent memory. In the end, we correctly predicted that the Liberals would retain power (we noted that a minority government was the most probable scenario), although we acknowledge that our final estimate of the Conservative Party’s support fell slightly outside the margin of error. This was largely due to a sampling error in our final week of polling (more on this below).
We are particularly proud of our final seat projections. Not only did we have the overall national seat distribution correct, we accurately predicted 301 out of Canada’s 338 electoral districts (89 per cent), which is an extremely strong performance:
|
LPC
|
CPC
|
NDP
|
BQ
|
GP
|
PPC
|
OTHER
|
Final projection:
|
158
|
117
|
31
|
28
|
3
|
0
|
1
|
Oct. 21 results
|
157
|
121
|
24
|
32
|
3
|
0
|
1
|
Our main source of error was underestimating the Conservative vote and, to a lesser extent, the Bloc Québécois vote in Quebec. With respect to the Quebec vote, we had the Liberals and Bloc in a statistical tie (29 per cent and 30 per cent, respectively) one week out from the election, but we saw the Bloc recede in the final days of the campaign. The underestimate of the Conservative vote was largely due to a sudden eleventh-hour drop in the party’s support in the Prairies. Finally, we saw some unusual shifts in Atlantic Canada in our final poll and we warned readers that we suspected these figures were a product of the small sample sizes there. In Ontario and British Columbia, our final poll was within the margin of error.

In all three cases, the range of figures we observed in the final 10 days of the campaign were largely within the margin of error of the final results so we would attribute these erroneous late shifts to sampling errors, which we explicitly incorporated into our seat forecasts. For example, in both Quebec and Atlantic Canada, we gave the Bloc Québécois and Liberals significantly more seats than we would have from our final poll alone.
In closing, we would like to thank the 100,933 Canadians who generously responded to our surveys since January. We would also like to thank those who have provided advice, comments, and feedback throughout the election, particularly in the world of social media. Finally, we congratulate all those who participated in the political process and recognize the passion and effort that they bring to Canadian democracy.
Please click here for the full report.
A Brief Post-Mortem on Election 43
[Ottawa – October 20, 2019] After an exciting campaign, the 43rd Canadian general election has come to a close. We at EKOS believe we did a very good job in charting the direction in one of the tightest campaigns in recent memory. In the end, we correctly predicted that the Liberals would retain power (we noted that a minority government was the most probable scenario), although we acknowledge that our final estimate of the Conservative Party’s support fell slightly outside the margin of error. This was largely due to a sampling error in our final week of polling (more on this below).
We are particularly proud of our final seat projections. Not only did we have the overall national seat distribution correct, we accurately predicted 301 out of Canada’s 338 electoral districts (89 per cent), which is an extremely strong performance:
LPC
CPC
NDP
BQ
GP
PPC
OTHER
Final projection:
158
117
31
28
3
0
1
Oct. 21 results
157
121
24
32
3
0
1
Our main source of error was underestimating the Conservative vote and, to a lesser extent, the Bloc Québécois vote in Quebec. With respect to the Quebec vote, we had the Liberals and Bloc in a statistical tie (29 per cent and 30 per cent, respectively) one week out from the election, but we saw the Bloc recede in the final days of the campaign. The underestimate of the Conservative vote was largely due to a sudden eleventh-hour drop in the party’s support in the Prairies. Finally, we saw some unusual shifts in Atlantic Canada in our final poll and we warned readers that we suspected these figures were a product of the small sample sizes there. In Ontario and British Columbia, our final poll was within the margin of error.
In all three cases, the range of figures we observed in the final 10 days of the campaign were largely within the margin of error of the final results so we would attribute these erroneous late shifts to sampling errors, which we explicitly incorporated into our seat forecasts. For example, in both Quebec and Atlantic Canada, we gave the Bloc Québécois and Liberals significantly more seats than we would have from our final poll alone.
In closing, we would like to thank the 100,933 Canadians who generously responded to our surveys since January. We would also like to thank those who have provided advice, comments, and feedback throughout the election, particularly in the world of social media. Finally, we congratulate all those who participated in the political process and recognize the passion and effort that they bring to Canadian democracy.
Please click here for the full report.
Share this: