Back To Listing

Morgan Poll Very Accurate at Recent Elections

Election Accuracy at recent Elections: 2007 - 2012.

Electors Do Change Their Votes in The Lead Up to an Election

Recently, the accuracy of the Morgan Poll has been questioned - most often this claim relies on referring to the 2001 Federal Election at which a Bulletin-Morgan Poll conducted on the weekend prior to the Election predicted an ALP victory and the 2004 Federal Election at which the final Morgan Poll said the election was “too close to call”, but with the ALP ahead.

It is true, the Morgan Poll was not the most accurate pollster at these two elections, however, the Morgan Poll learnt valuable lessons about the timeliness of polling in determining accuracy and has put those lessons into practice in more recent elections.

We reflected heavily on our polling methodology after the 2001 Federal Election and put out two press releases dealing with what happened: “Did the debate over refugee boat people in the last two days change the electorate or was the Morgan Poll wrong?” — November 13, 2001 and “What happened on November 10? Did the 'race card' (border protection) swing the electorate?” — November 27, 2001.

The conclusions we reached were based on the on-the-ground results we found. The results of the face-to-face Morgan Poll conducted on the day of the 2001 Federal Election (November 10, 2001) showed a remarkable accuracy for that day’s election results. This led us to the obvious conclusion that when the poll is conducted is crucial to determining its accuracy. The discussion about refugees in the week leading up to the 2001 Federal Election clearly swung that election for the Howard Government.

In recent Federal Elections (2007 & 2010) the Morgan Poll has interviewed as closely as possible to the Federal Election to take account of any late swings that have the potential to change the election result.

At the 2007 Federal Election (November 24, 2007) the Morgan Poll conducted our final poll on Friday evening, November 23, 2007. The Morgan Poll was the most accurate major poll for predicting the voting results for the major parties (ALP & L-NP) and all parties (ALP, L-NP, Greens, Family First, Ind./Other).

2007 Federal Election

  • Average error on major parties: Morgan Poll — 0.35%; Newspoll — 0.50%; Galaxy Research — 0.65%; ACNielsen — 3.3%.
  • Average error on all parties: Morgan Poll — 0.52%; Galaxy Research — 0.64%; Newspoll — 0.9%; ACNielsen — 1.92%.

  • Average error on 2PP vote: Newspoll — 0.7%; Galaxy Research: 0.7%; Morgan Poll — 0.8%; ACNielsen — 4.3%.

    At the 2010 Federal Election (August 21, 2010) the 7News Morgan Poll once again conducted our final poll on Friday evening, August 20, 2010. The 7News Morgan Poll was the most accurate major poll for predicting the voting results for all parties (ALP, L-NP, Greens, Ind./Others) and just behind Newspoll for predicting the voting results for the major parties (ALP & L-NP).

2010 Federal Election

  • Average error on major parties: Newspoll — 1.0%; 7News Morgan Poll — 1.05%; Galaxy Research — 1.30%; Nielsen — 1.55%.
  • Average error on all parties: 7News Morgan Poll — 0.7%; Newspoll — 1.05%; Nielsen — 1.1%; Galaxy Research — 1.3%.

  • Average error on 2PP vote: Newspoll — 0.1%; 7News Morgan Poll — 0.9%; Galaxy Research — 1.9%; Nielsen — 1.9%.

These results clearly show that any assertion that the Morgan Poll is “not accurate” has no basis in regards to recent Federal Elections. In fact, at the most recent Federal Election the 7News Morgan Poll was the first poll to predict Australia’s first ‘Hung Parliament’ since the 1940s — a week before the Federal Election, and this was backed up by the final Morgan Poll conducted on the eve of the Federal Election.

The most recent Federal Elections show that the Morgan Poll ‘learnt our lesson’ from earlier Federal Election predictions and understands that, particularly in close elections, late swings do happen that can change the election result. This ‘phenomenon’ was also borne out in the most recent Victorian State Election on November 27, 2010.

The ALP was clearly ahead during the bulk of the election campaign, however special Roy Morgan Reactor research picked up an adverse reaction to a Labor campaign ad targeting Liberal Leader Ted Baillieu in the last week of the campaign that contributed strongly to a late swing to the L-NP in the final week of the campaign. The final Morgan Poll (conducted November 22-25, 2010), picked up the late swing to the L-NP and predicted a ‘Close L-NP win’. As it turned out the L-NP gained a majority by only one seat (45-43).

2010 Victorian State Election

  • Average error on major parties: Morgan Poll — 0.5%; Galaxy Research — 0.5%; Nielsen — 1.2%; Newspoll — 1.7%; Sky News Exit Poll — 2.2%.
  • Average error on all parties: Morgan Poll — 0.9%; Galaxy Research — 1.4%; Nielsen — 1.75%; Newspoll — 2.0%; Sky News Exit Poll — 2.0%.

  • Average error on 2PP vote: Newspoll — 0.2%; Morgan Poll — 0.3%; Nielsen — 0.7%; Galaxy Research — 1.3%; Sky News Exit Poll — 2.7%.

The most recent State Election, the Queensland State Election earlier this year (March 24, 2012) was widely predicted to be a landslide LNP victory — as it turned out to be. Once again, the Morgan Poll very closely and accurately predicted the result — despite using only a small sample size of interviews.

2012 Queensland State Election

  • Average error on major parties: Newspoll — 0.8%; Morgan Poll — 1.3%; Galaxy Research — 3.0%; (Galaxy Research Exit Poll — 1.0%).
  • Average error on all parties: Morgan Poll — 1.2%; Newspoll — 1.6%; Galaxy Research — 2.5%; (Galaxy Research Exit Poll — 0.7%).

  • Average error on 2PP vote: Morgan Poll — 1.1%; Newspoll — 2.3%; Galaxy Research — 3.1%; (Galaxy Research Exit Poll — 0.1%).

The most accurate pre-election poll for all parties and the 2PP vote was the Morgan Poll, while Newspoll was the most accurate pre-election poll for the major parties. The Galaxy Research Exit Poll was also very accurate, once again showing that the most accurate polls tend to be those that are actually conducted on Election Day.

So what conclusion can be drawn from recent election polling? Clearly it is not correct to assert that the Morgan Poll is inaccurate with regards to our polling. Our methodology compares very well with our competitors in terms of recent elections in the past five years, and our accuracy across this period has been second-to-none on an overall basis.

Available below are more in depth tables that look at the results of polling in relation to the recent elections quoted above.

2007 AUSTRALIAN FEDERAL ELECTION

FEDERAL VOTING INTENTION —Primary Vote Predictions

 

 

Nov 24

Nov 23

Nov 20-22

Nov 19-21

Nov 20/21

 

2007

Telephone

Telephone

Telephone

Telephone

 

Federal election

Morgan Poll

Newspoll

ACNielsen

Galaxy

 

Result*

%

error

%

error

%

error

%

error

 

 

 

 

 

 

 

 

 

 

Sample Size

 

(2,115)

 

(2,615)

 

(2,071)

 

(1,186)

 

 

 

 

 

 

 

 

 

 

 

L-NP

42.1

41.5

-0.6

43

+0.9

40

-2.1

42.5

+0.4

 

 

 

 

 

 

 

 

 

 

ALP

43.4

43.5

+0.1

44

+0.6

48

+4.6

42.5

-0.9

 

 

 

 

 

 

 

 

 

 

The Greens

7.8

8.5

+0.7

7

-0.8

8

+0.2

8.5

+0.7

 

 

 

 

 

 

 

 

 

 

Family First

2.0

2.5

+0.5

NR

n/a

1

-1.0

2.5

+0.5

 

 

 

 

 

 

 

 

 

 

Ind/Other

4.7

4

-0.7

6

1.3

3

-1.7

4

+0.7

 

 

 

 

 

 

 

 

 

 

Av. error on major parties (L-NP, ALP)

 

 

0.35

 

0.5

 

3.3

 

0.65

Av. error on all parties

 

 

0.52

 

0.9#

 

1.92

 

0.64

FEDERAL VOTING INTENTION —Two Party Preferred Predictions

 

L-NP

47.3       

46.5

-0.8

48

+0.7

43

-4.3

48

+0.7

 

 

 

 

 

 

 

 

 

 

ALP

52.7

53.5

+0.8

52

-0.7

57

+4.3

52

-0.7

 

 

 

 

 

 

 

 

 

 

ALP Lead

5.4

7

 

4

 

14

 

4

 

 

 

 

 

 

 

 

 

 

 

Ave error

 

 

0.8

 

0.7

 

4.3

 

0.7

* Australian Electoral Commission (http://results.aec.gov.au/13745/Website/HouseStateFirstPrefsByParty-13745-NAT.htm); full results, 2007

# Newspoll did not release a Family First vote (they included Family First in “Ind./Other”) so the Newspoll average error is based on four estimates not five.

 

2010 AUSTRALIAN FEDERAL ELECTION

FEDERAL VOTING INTENTION —Primary Vote Predictions

 

 

Federal election

August 21,

2010

Aug 20

Aug 17-18

Aug 17-19

Aug 17-18

 

Telephone

Telephone

Telephone

Telephone

 

7News

Morgan Poll

Newspoll

The Australian

Nielsen

The Age & SMH

Galaxy

Herald-Sun

& Daily Telegraph

 

Actual Result*

%

error

%

error

%

error

%

error

 

 

 

 

 

 

 

 

 

 

Sample Size

(12,402,364)

(1,872)

(2,507)

(2,040)

(1,200)

 

 

 

 

 

 

 

 

 

 

ALP

38.0

39

+1.0

36.2

-1.8

39

+1.0

38

0

 

 

 

 

 

 

 

 

 

 

L-NP

43.6

42.5

-1.1

43.4

-0.2

41.5

-2.1

41

-2.6

 

 

 

 

 

 

 

 

 

 

Greens

11.8

11.5

-0.3

13.9

+2.1

13

+1.2

14

+2.2

 

 

 

 

 

 

 

 

 

 

Ind/Other

6.6

7

+0.4

6.5

-0.1

6.5

-0.1

7

+0.4

 

 

 

 

 

 

 

 

 

 

Av. error on all parties

 

 

0.7

 

1.05

 

1.1

 

1.3

FEDERAL VOTING INTENTION —Two Party Preferred Predictions

 

ALP

50.1

51

+0.9

50.2

-0.1

52

+1.9

52

+1.9

 

 

 

 

 

 

 

 

 

 

LN-P

49.9

49

-0.9

49.8

+0.1

48

-1.9

48

-1.9

 

 

 

 

 

 

 

 

 

 

ALP Lead

0.2

2

 

0.4

 

4

 

4

 

 

 

 

 

 

 

 

 

 

 

Ave error

 

 

0.9

 

0.1

 

1.9

 

1.9

Australian Electoral Commission (http://vtr.aec.gov.au/HouseStateFirstPrefsByParty-15508-NAT.htm). Final AEC Election results.

 

2010 VICTORIAN STATE ELECTION

VICTORIAN VOTING INTENTION —Primary Vote Predictions

 

Victorian

Election

Nov 27,

2010

Nov 22-25

Nov 23-24

Nov 24-25

Nov 23-25

Nov 27

 

Telephone

Telephone

Telephone

Telephone

Exit Poll

 

Morgan Poll

Galaxy

Herald-Sun

The Age

Nielsen

Newspoll

Sky News

 

Actual Result*

%

error

%

error

%

error

%

error

%

error

 

 

 

 

 

 

 

 

 

 

 

 

Sample Size

(3,164,731)

(990)

(800)

(1,533)

(1,451)

(3,000+)

 

 

 

 

 

 

 

 

 

 

 

 

L-NP

44.8

44.5

-0.3

44

-0.8

45

+0.2

45

+0.2

48

+3.2

 

 

 

 

 

 

 

 

 

 

 

 

ALP

36.2

35.5

-0.7

36

-0.2

34

-2.2

33

-3.2

35

-1.2

 

 

 

 

 

 

 

 

 

 

 

 

Greens

11.2

13

+1.8

14

+2.8

14

+2.8

15

+3.8

12

+0.8

 

 

 

 

 

 

 

 

 

 

 

 

Ind/Other

7.8

7

-0.8

6

-1.8

6

-1.8

7

-0.8

5

-2.8

 

 

 

 

 

 

 

 

 

 

 

 

Av. error on

all parties

 

 

0.9

 

1.4

 

1.75

 

2.0

 

2.0

PREDICTED OUTCOME

 

Close

L-NP win

Close

ALP win

Clear

L-NP win

Hung

Parliament

Big

L-NP win

VICTORIAN VOTING INTENTION —Two Party Preferred Predictions

L-NP

51.3

51

-0.3

50

-1.3

52

+0.7

51.1

-0.2

54

+2.7

 

 

 

 

 

 

 

 

 

 

 

 

ALP

48.7

49

+0.3

50

+1.3

48

-0.7

48.9

+0.2

46

-2.7

 

 

 

 

 

 

 

 

 

 

 

 

L-NP Lead

2.6

2

 

-

 

4

 

2.2

 

8

 

 

 

 

 

 

 

 

 

 

 

 

 

Ave error

 

 

0.3

 

1.3

 

0.7

 

0.2

 

2.7

Victorian Electoral Commission (http://www.tallyroom.vic.gov.au). VEC Election results at 4pm, Monday December 13, 2010.
*The Two-Party preferred vote is yet to be released and the figure quoted here is based on past voting trends at the 2006 Victorian election.

 

2012 QUEENSLAND STATE ELECTION

Primary & Two-Party Preferred Vote Predictions

 

State Election

March 24, 2012

Mar 20-21

Mar 20-22

Mar 15-16

Mar 24

 

Telephone

Telephone

Telephone

Exit Poll#

 

Morgan Poll

Newspoll

Galaxy Research

Galaxy Research

 

Actual Result*

%

error

%

error

%

error

%

error

 

 

 

 

 

 

 

 

 

 

Sample Size

2,320,327

205

1,538

800

xxx

 

 

 

 

 

 

 

 

 

 

LNP

1,154,587 — 49.7

51

-1.3

50

-0.3

47

+2.7

51

-1.3

 

 

 

 

 

 

 

 

 

 

ALP

619,842— 26.7

28

-1.3

28

-1.3

30

-3.3

26

+0.7

 

 

 

 

 

 

 

 

 

 

KAP (Katter’s

Australian Party)

266,849 — 11.5

8.5

+3.0

9

+2.5

8

+3.5

11

+0.5

 

 

 

 

 

 

 

 

 

 

Greens

170,913 — 7.5

7.5

0

6

+1.5

9

-1.5

7

+0.5

 

 

 

 

 

 

 

 

 

 

Others

108,136 — 4.6

5

-0.4

7

-2.4

6

-1.4

5

-0.4

 

 

 

 

 

 

 

 

 

 

Average error on

ALL PARTIES

 

 

1.2

 

1.6

 

2.5

 

0.7

 

 

 

 

 

 

 

 

 

 

Average error on

MAJOR PARTIES

 

 

1.3

 

0.8

 

3.0

 

1.0

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

LNP

63.1

62

+1.1

60.8

+2.3

60

+3.1

63

+0.1

 

 

 

 

 

 

 

 

 

 

ALP

36.9

38

-1.1

39.2

-2.3

40

-3.1

37

-0.1

 

 

 

 

 

 

 

 

 

 

Average error on

2PP RESULTS

 

 

1.1

 

2.3

 

3.1

 

0.1

*Final tally. 2PP result estimated by Antony Green. Queensland electoral laws mean voters do not have to allocate a preference
for all parties when filling out a voting slip. #Galaxy Research Exit Poll was taken on the day of the Queensland Election.