1
Copyright 2015. No reproduction without permission.
Presentation to the MRIA Forum on Public Opinion Po lls from the 2015 Federal Election
November 26, 2015www.ekos.com
The 42nd Election: A Polling RetrospectiveBy Frank Graves
Ottawa, Ontario
2
Copyright 2015. No reproduction without permission.
Conclusions3.
Overview of the 2015 Election1.
Notes on Polling2.
3
Copyright 2015. No reproduction without permission.
Methodology
» The Canadian IntelligentVoter Information System(CIVIS) is EKOS’ innovativepolling service
» Features include:
• Large sample sizes(130,000+ cases in 2015)
• Probability-basedsampling that covers theentire population
• Unique fusion of:
1. High Definition Interactive Voice Response (HD-IVRTM)
2. Live Computer Assisted Telephone Interviews (CATI)
3. EKOS’ unique hybrid online/phone panel, Probit (probability sampling)
• Dual-mode sampling (i.e., both cellphone and landline)
4
Copyright 2015. No reproduction without permission.
0
10
20
30
40
Jan-12 Jun-12 Nov-12 Apr-13 Sep-13 Feb-14 Jul-14 Dec-14 May-15 Oct-15
Liberal Party Conservative Party NDP Green Party Bloc Québécois Other
Prologue:2011
Election 2015 Election
Apr. 14, 2013:Trudeau becomes LPC leader
Oct. 22, 2014:Parliament Hill shooting
Aug. 2, 2015:Writ dropped
Act I:Descent into progressive darkness
Act II:Crimson tide rising
Act III:Slow dance of the
promiscuous progressives
Act IV:Coming to public
judgement
May-11
A four-part play: the reinstatement of progressive Canada
5
Copyright 2015. No reproduction without permission.
Prologue & Act I: Descent into progressive darkness
» The play begins on Election Night 2011 when Harper secures a majority
• Shocking in light of clear disconnect with values and interests of most
Canadians
• Fashioned from uneven turnout across generational lines
» A sense of buyer’s remorse was evident almost immediately as barometers
of trust, approval, and confidence in national direction plumbed new lows
0
10
20
30
40
May-11 Aug-11 Nov-11 Feb-12 May-12 Aug-12 Nov-12 Feb-13
Liberal Party Conservative Party NDP Green Party Bloc Québécois Other
March 24, 2012Mulcair becomes NDP leader
October 2, 2012Trudeau enters LPC leadership race
May 2, 201141st Canadian general election
6
Copyright 2015. No reproduction without permission.
Act II: Crimson tide rising» A year after assuming power, Trudeau seemed to be on an
unstoppable path to victory
• In a poll one year out, the Liberals found themselves almost exactly
where they were on Election Night 2015
• Success forged from focus on middle class renewal and a more optimistic and progressive tone linked to appetite for change
0
10
20
30
40
Apr-13 Jun-13 Aug-13 Oct-13 Dec-13 Feb-14 Apr-14 Jun-14 Aug-14 Oct-14
Liberal Party Conservative Party NDP Green Party Bloc Québécois Other
November 5, 2013Duffy suspended
July 17, 2014Duffy formally charged
April 14, 2013Trudeau becomes LPC leader
7
Copyright 2015. No reproduction without permission.
Act III: Slow dance of the promiscuous progressives» The shooting on the Hill reignited concerns with security and terror
• These concerns were shrewdly exploited by Harper
» As Harper began to rise, Trudeau began to sink
» Trudeau’s Liberals went from 12-point lead to third place in three-way contest
» Contrary positions on C-51 were linked to NDP rise and Liberal decline, particularly with university educated
» Notably, the NDP rise predates the Notley effect, which amplified it
0
10
20
30
40
Oct-14 Nov-14 Dec-14 Jan-15 Feb-15 Mar-15 Apr-15 May-15 Jun-15 Jul-15
Liberal Party Conservative Party NDP Green Party Bloc Québécois Other
May 5, 2015Alberta NDP wins majority government
March 26, 2015Liberals propose amendments to Bill C-51October 22, 2014
Parliament Hill shootings
8
Copyright 2015. No reproduction without permission.
Act IV: Coming to public judgement
» At the outset of the 42nd election campaign, NDP in 1st,
Conservatives 2nd, Liberals 3rd
» Two critical events altered the course of the election:
1. Mulcair committed to balanced budget, Trudeau
took vividly different position
2. The image of the drowned Syrian child ignited a
broader contest about values; temporary boost to
Harper, but advantage shifted as Canadians reflected on which values should define Canada in
the future
» An important election about the economy morphed
into an historical election about values
9
Copyright 2015. No reproduction without permission.
0
10
20
30
40
Aug 2 Aug 9 Aug 16 Aug 23 Aug 30 Sep 6 Sep 13 Sep 20 Sep 27 Oct 4 Oct 11 Oct 18
Liberal Party Conservative Party NDP Green Party Bloc Québécois Other
August 2, 2015Writ dropped
2015Election
September 2, 2015Alan Kurdi’s body found
August 27, 2015Trudeau proposes deficit spending
August 25, 2015Mulcair promises balanced budget
September 15, 2015Conservative Party announces plan to banniqabs at citizenship ceremonies
Act IV: Coming to public judgement
10
Copyright 2015. No reproduction without permission.
Meaning of the 2015 Election
» Harper enjoyed high ground on niqab, but inferior
position on the broader values contest
» Election reflects public judgement rooted in the
growing normative tension between Harper and dominant societal values
» Also reflects rising discomfort with the withering of middle class progress and a rejection of minimal
government, austerity, and trickle-down economics
11
Copyright 2015. No reproduction without permission.
42
24
23
23
11
17
12
11
45
57
63
64
Q.Which of the following choices best describes your ideal vision of Canada?
Insistence on adherence to Cdn values
Moral certainty
(1-3)
(1-3)
(1-3)
(1-3)
Defense
Minimal federal government
(5-7)
(5-7)
(5-7)
(5-7)
Accommodation, Tolerance, & Respect
Reason and evidence
Humanitarianism and development
Active federal government
Neither (4)
Neither (4)
Neither (4)
Neither (4)
BASE: Canadians; October 8-12, 2015 (n=1,124), MOE +/- 2.9%, 19 times out of 20
Preferred vision for Canada
12
Copyright 2015. No reproduction without permission.
Other factors
» Engagement very high, which bolstered turnout
• Turnout advantage of 2011 was erased for the
Conservatives
» 2011 was about inertia, this election was about motion
• Major shifts throughout the campaign which continued to final days
» Final shifts reflected a public judgement that Justin
Trudeau represented best bet for achieving twodominant objectives:
1. To retire Mr. Harper
2. To install a clearly progressive government
13
Copyright 2015. No reproduction without permission.
Critical ingredients of the late majority
» Quebec
» Younger voters and cellphone-only households
» Seniors’ defection from Conservatives
» New Canadians
» University educated
» Liberal ground game
» The polls
14
Copyright 2015. No reproduction without permission.
14
15
14
3
9
16
5
6
8
1
2
2 1
1
2
0 5 10 15 20 25 30 35 40 45
Other
BASE: Canadians who voted in 2015; October 20-23, 2015, n=1,636, MOE +/- 2.4%, 19 times out of 20
26%
Q.Did you vote in the most recent federal election, either in an advance poll or on Election Day on Monday, October 19th?
[If Yes] How did you vote in this election?
Loyalists(those who made their choice BEFORE the campaign and stuck to it)
Promiscuous Voters(those who switched from an earlier choice)
One-Timers(those who made their choice DURING the campaign and stuck to it)
33%
42%
Note: Figures adjusted to exclude those who skipped the question(s)
Non
-sw
itche
rsS
witc
hers
TOTAL
Three-way division of voters
15
Copyright 2015. No reproduction without permission.
Conclusions3.
Overview of the 2015 Election1.
Notes on Polling2.
16
Copyright 2015. No reproduction without permission.
39.5
38.8
35.8
31.9
30.4
31.9
19.7
18.5
20.4
3.4
5.6
4.71
1
1
6.7 4.3
4.9
Liberal Party Conservative Party NDP Green Party Bloc Québécois Other
3-day roll-up
2015 Election Results
2-day roll-up
BASE: Canadians; October 16-18, 2015, n=2,122, MOE +/- 2.1%, 19 times out of 20
BASE: Canadians; October 17-18, 2015, n=1,260, MOE +/- 2.8%, 19 times out of 20
Q. Thinking about the upcoming federal election on October 19th, have you already voted either at an advance poll or by special ballot?
[If Yes] How did you vote in this election?
[If No] How do you plan to vote in the upcoming federal election on October 19th?
Impact of final Liberal surge
17
Copyright 2015. No reproduction without permission.
38.8
44.3
35.2
30.4
29.6
30.9
18.5
15.7
20.3
6.7
5.8
7.2
4
3
5
1
1
1
Liberal Party Conservative Party NDP Green Party Bloc Québécois Other
HD-IVR™
Live CATI
Impact of interview mode (final 2 -day roll-up)
BASE: Canadians; October 17-18, 2015
n=1,260, MOE +/- 2.8%, 19 times out of 20
n=896, MOE +/- 3.3%, 19 times out of 20
Q. Thinking about the upcoming federal election on October 19th, have you already voted either at an advance poll or by special ballot?
[If Yes] How did you vote in this election?
[If No] How do you plan to vote in the upcoming federal election on October 19th?
Note: Figures adjusted to exclude undecided voters, those not eligible to vote, and those who skipped the question.
Mergedn=364, MOE +/- 5.1%, 19 times out of 20
18
Copyright 2015. No reproduction without permission.
35.4
31.4
31.1
38.4
21.4
20.6
7.3
5.8
3
2
2
2
Liberal Party Conservative Party NDP Green Party Bloc Québécois Other
Version 1
Version 2
Impact of ballot question
Note: Figures adjusted to exclude undecided voters, those not eligible to vote, and those who skipped the question.
Q. If a federal election were held tomorrow, which party would you vote for?
Q. If a federal election were held tomorrow, please tell me what your first and second preferences would be in your riding.
BASE: Canadians; October 2-8, 2015
n=639, MOE +/- 3.9%, 19 times out of 20
n=589, MOE +/- 4.0%, 19 times out of 20
19
Copyright 2015. No reproduction without permission.
35.1
42.5
35.8
36.8
20.9
31.9
17.9
23.8
20.4
4.3
9.2
5.6
5
3
5
1
1
1
Liberal Party Conservative Party NDP Green Party Bloc Québécois Other
Overall
Cellphone-only households
Impact of cellphones
BASE: Canadians; October 16-18, 2015
n=1,356, MOE +/- 2.7%, 19 times out of 20
n=2,122, MOE +/- 2.1%, 19 times out of 20
Households w/ both a cellphone & landlinen=256, MOE +/- 6.1%, 19 times out of 20
» In 2011, sampling CPO
increased our prediction error
» In 2015, omitting CPO would
have produced an even more
significant error
» If we had had a larger sample of CPOs, we would have been
somewhat closer
20
Copyright 2015. No reproduction without permission.
Conclusions3.
Overview of the 2015 Election1.
Notes on Polling2.
21
Copyright 2015. No reproduction without permission.
Likely voter models
» The idea that there is some kind of likely voter model
that will work consistently across elections is something of a chimera; there is no unified theory of voter turnout
» The need for likely voter diminishes as turnout increases
» The dynamics of turnout were dramatically different in this election than in last election and this problem will
afflict any turnout models
22
Copyright 2015. No reproduction without permission.
Seat projections
» EKOS most accurate of all main seat projections
» Discrepancies are almost entirely due to highly unusual
late shift
11
1016
4454
99116
184151
2015 Election ResultsEKOS Projection(October 19, 2015)
Party
Accuracy = 82%Correct: 276 Incorrect: 62
23
Copyright 2015. No reproduction without permission.
Conclusions (i)
» In some respects, capturing the election result is less
important than accurately revealing the state of voter intention earlier
» The real impact of polls reported the night before the election is probably modest (and we’ll know the right
answer less than 24 hours later)
» As the polls did influence late shifting, it’s actually more
important to be providing voters with valid data in the final stages rather than what’s going to happen on
Election Day
» For the first time, we see clear evidence of the polls at a
macro level influencing voter behaviour
24
Copyright 2015. No reproduction without permission.
35
42
44
60
72
62
58
55
40
28
0 10 20 30 40 50 60 70 80
No impact (1) At least some impact (2-5)
Q. To what extent did public opinion polls affect your final decision on Election Day?
Before the election campaign
After the leaders’ debate
In the last week of the election campaign
On Election Day or at the ballot booth
Before the leaders’ debate
BASE: Those who followed the polls; October 20-23, 2015, n=1,327, MOE +/- 2.7%, 19 times out of 20
Impact of political polls
25
Copyright 2015. No reproduction without permission.
Conclusions (ii)
» This means there is a co-agency and slippage across
subject matter and the process of scientific observation
» If polling is to be instrumental in shaping voter behaviour, it is critical that voters have access to
reliable and valid estimates in the final stages, not just
the final day
» We would posit that the campaign did not begin in a
tie, that the Conservatives clearly led for a considerable period, and that the majority was only clear in the final
weekend
26
Copyright 2015. No reproduction without permission.
New yardsticks for polling?
» A fairer yardstick would include nearness of one’s final poll to the election,
but also other factors:
• Did the polling show the major shifts in the electorate over the
campaign and why they occurred?
• Did the polling accurately identify the forces at work and the range of
options that might occur on Election Day?
• How accurate was the late (but not final) polling?
• What is the level of transparency and disclosure?
• What are the quality of the indicators used to diagnose and track the
election?
• What was the total number of observations drawn to allow more
accurate estimates of the above?
• Did the polls use accepted methods for achieving scientific
accuracy?1
• Did the pollsters conduct an immediate post-election survey to assess
what happened?
1. American Association for Public Opinion Research, “Report of the AAPOR Task Force on Non-Probability Sampling”, June 2013. Available online at: https://goo.gl/VHHyPy
27
Copyright 2015. No reproduction without permission.
Frank Graves
EKOS Research Associates
(613) 235-7215
www.ekos.com
For more information: