Back To Listing

Victorian Election: Roy Morgan Most Accurate In Estimating ALP & L-NP Primary Vote

The Nine Network Morgan Poll was the most accurate of all the polls in estimating the 4.8% swing from the ALP in the recent Victorian election.  The Morgan Poll showed a 5.4% swing — estimating the ALP vote at 42.5% (0.6% below the actual result).  Galaxy and AC Nielsen underestimated the ALP vote by 1.1%, while Newspoll overestimated Labor’s vote by 1.9%.  McNair Ingenuity underestimated the ALP vote by 4.1%, while overestimating the L-NP vote by a massive 6.4%.

The Morgan Poll estimated the L-NP primary vote to be 40.5%, which is just 0.9% above the actual election result of 39.6%.  In comparison, Newspoll underestimated the L-NP vote by 2.6% and Galaxy by 0.6%, while AC Nielsen overestimated the L-NP by 1.4%.

As AC Nielsen and McNair Ingenuity did not publish separate estimates for the Liberal and National Party votes, comparative accuracy tables have been produced only for the average error for the major parties (ALP and L-NP).

Gary Morgan says:

“In the days following the November 25 election there was a vast amount of media misinformation circulating about the result of the Victorian election; an article published in The Sunday Age the day after the election suggested the swing against the ALP was as small as 2%.  Headlines such as: ‘Only a 2 per cent swing? Just too easy for Labor’ (Tim Colebatch, The Sunday Age, November 26) and ‘Newspoll, Galaxy prove most accurate’ (The Australian, November 27) distorted the true facts.

“The facts are that Labor’s primary vote is down 4.8% from the 2002 election (43.1% cf. 47.9%) and its two-party preferred vote is down 3.4% to 54.4% (Malcolm Mackerras final two-party preferred estimate).

“The claim in The Australian that ‘Newspoll and Galaxy came the closest to correctly tipping the weekend’s election result’ is false.  Of the four main polling companies, Newspoll was the least accurate in estimating the primary vote for the two major parties.

“The Morgan Poll was the most accurate in predicting the primary vote for the major parties; estimating the ALP and L-NP votes within 1% and an ALP swing of 5.4%.  The Morgan Poll estimated the ALP vote at 42.5% (0.6% below the actual result).  Galaxy and AC Nielsen underestimated the ALP vote by 1.1%, while Newspoll overestimated Labor’s vote by 1.9%.

 

“The McNair Ingenuity Poll, which added Gallup after their name without approval, was just wrong.  The Sunday before the election McNair forecasted 50.5% - ALP, 49.5% - L-NP.  No doubt Dr. George Gallup has turned in his grave!”

 

2002

electionresult*

2006

electionresult**

Nine Network

Morgan Poll

The Age

AC Nielsen

The Australian

Newspoll

Herald-Sun

Galaxy

Sunday Herald-SunMcNair Ingenuity

 

%

%

%

% error

 

 

%

% error

%

% error

%

% error

Sample Size

   

956

2,023

1,500

942

609

 

 

 

Primary Voting Intention

ALP

47.9

43.1

42.5

-0.6

42

-1.1

45

+1.9

42

-1.1

39

-4.1

Total L-NP

38.2

39.6

40.5

+0.9

41

+1.4

37

-2.6

39

-0.6

46

+6.4

Av. Error

     

0.75

 

1.25

 

2.25

 

0.85

 

5.25

Liberals

33.9

34.4

36

+1.6

 

 

32

-2.4

35

+0.6

 

 

National

4.3

5.2

4.5

-0.7

 

 

5

-0.2

4

-1.2

 

 

The Greens

9.7

10

12.5

+2.5

11

+1

9

-1

12

+2

 

 

Family First

^

4.3

3

-1.3

2

-2.3

2.5

-1.8

4

-0.3

 

 

Other

4.2

3.0

1.5

-1.5

4

+1.0

6.5

+3.5

3

0

 

 

 

 

 

Two-Party Preferred

ALP

57.8

54.4^

53

-1.4

53

-1.4

56

+1.6

55

+0.6

50.5

-3.9

L-NP

42.2

45.6^

47

+1.4

47

+1.4

44

-1.6

45

-0.6

49.5

+3.9

ALP lead

 

8.8

6

 

6

 

12

 

10

 

1

 

Avg error

 

 

 

1.4

 

1.4

 

1.6

 

0.6

 

3.9

* Final results according to the Victorian Electoral Commission

** Victorian Election Commission (VEC website, February 13, 2007)

^ Malcolm Mackerras final two-party preferred estimate

For further information:

Gary Morgan:               Office +61 3 9224 5213   Mobile +61 411 129 094  Home +61 3 9419 3242