Back To Listing

Wentworth by-election means ‘death’ of cheap robo-polling & ‘leaked’ private party polls!

Roy Morgan CEO Michele Levine two days before the Wentworth by-election said on Sky News’ new political show ‘Conroy & Kroger’ when asked why she was so critical of cheap polls said: ‘They may be cheaper, but I can’t really see any point in doing a poll that is going to get the wrong answer’.

By Gary Morgan, Michele Levine & Julian McCrann, Roy Morgan, Melbourne

Roy Morgan CEO Michele Levine interviewed on 'Conroy & Kroger' about the Wentworth by-election on Thursday October 18, 2018
Roy Morgan CEO Michele Levine interviewed on Sky News 'Conroy & Kroger' about the Wentworth by-election. Thursday October 18, 2018.


Roy Morgan CEO Michele Levine two days before the Wentworth by-election on Sky News’ new political show ‘Conroy & Kroger’ when asked why she was so critical of cheap polls said: ‘They may be cheaper, but I can’t really see any point in doing a poll that is going to get the wrong answer’.

Conducting a worthwhile poll is not as simple as interviewing 500, or a 1,000, people in a given area, and filling up gender and age quotas to match the area’s basic demographic population profile and assuming you have an accurate poll.

The recent Wentworth by-election provided a stunning example of how simply filling up quotas in this way can produce a prediction that is so far removed from reality to be useless.

The ‘wrong’ Greenpeace commissioned ReachTel robo-poll results  were reported by Carly Earl for the Guardian last week prior to the Wentworth by-election (the actual results are in brackets for comparison):

“The Liberal party is in serious danger of losing the seat of Wentworth this weekend according to a new ReachTel poll that shows Liberal candidate Dave Sharma’s primary vote has slumped to 32.7% (actual vote 43.2%). The vote of high profile independent and local GP Kerryn Phelps has surged to 25.8% (actual vote 29.1%).  Labor’s Tim Murray has increased his share of the primary vote to 21.6% (actual vote 11.5%), compared with 19.5% in a ReachTel poll two weeks ago. The Greens’ Dominic Wy Kanak has 9.1% (actual vote 8.6%) while independent Licia Heath has 5.6% (actual vote 2.3%)”.

The Greenpeace ReachTel robo-poll primary vote predictions were significantly in error for several of the leading candidates. Out by more than 10% for both Liberal Dave Sharma and ALP candidate Tim Murray and out by more than 3% for the two Independent candidates Kerryn Phelps and Licia Heath.

The ReachTel robo-poll’s prediction for the critical two-candidate preferred result was even further out. Not only did the two-candidate preferred result compound the error made by being over 10% out on the primary vote for Liberal candidate Dave Sharma and ALP candidate Tim Murray but the preference flows were a further source of error and meant the poll was completely wrong when predicting the two-candidate preferred result.

The ReachTel robo-poll had a ‘ridiculous’ preference flow of 90% to Kerryn Phelps compared to 10% for Liberal Dave Sharma which led to a two-candidate preferred prediction of Phelps (62.4%) cf. Sharma (37.6%). This predicted two-candidate preferred figure was well over 10% off the correct result and will be even more incorrect as postal votes continue to favour the Liberal candidate to produce a close final two-candidate preferred result – likely to be close to 51% Phelps cf. 49% Sharma.

On Thursday October 18, 2018 – two days before the Wentworth by-election, Roy Morgan CEO Michele Levine was interviewed on SkyNews new political show ‘Conroy & Kroger’ and warned about relying on cheap polls (including robo-polls such as the Greenpeace commissioned ReachTel poll) that rely on a cheap methodology rather than striving for any degree of accuracy.

The following exchange between host Stephen Conroy and Roy Morgan CEO Michele Levine addresses the issues with relying on ‘cheaper polls’:

Stephen Conroy, Host of Conroy & Kroger: "Alright, look that’s fair enough, I mean, the parties now are faced with this dilemma, But the cheaper polls are done in a way that doesn’t always give you a really solid result.  What is behind that?  Why are you so critical of those polls?”

Michele Levine, CEO Roy Morgan: “I’m critical of them, they may be cheaper, but I can’t really see any point in doing a poll that is going to get the wrong answer.

So for instance, Newspoll is now owned by YouGov and YouGov got Brexit wrong.  Now if there was ever anything that really mattered, if it was really important to get the truth, then I think Brexit would have been it.

And this is not, as somebody was saying earlier, about people not wanting to tell the truth or in fact lying when they get polled, it’s not that at all, it’s just that these modern new silly ways of polling which are either robo-polls or telephone polls which only get a small proportion of the population or online polls with commercial panels of people who volunteer to be part of it, they only get a really small piece of the population.

They never get, as in the case of Brexit or in the case of Trump being elected in America, they never get the disenfranchised people.  They just miss them completely and when they come out and vote they’re surprised.  So why people choose cheap polls, I don’t know.

As well as the Greenpeace commissioned ReachTel robo-poll there was also a ‘leaked’ internal Liberal poll (reported in the Australian) released in the days before the Wentworth by-election which had Sharma’s vote “in the mid-30s” and Phelps “well into the 20s”. Of course no details of the poll’s sample, methodology or questioning were made available.

The internal Liberal poll showed Independent Kerryn Phelps on 55% of the two-candidate preferred vote well ahead of Liberal Dave Sharma on 45% and prompted Prime Minister Scott Morrison to announce that the Liberals were going to lose the Wentworth by-election. The ‘leaking’ of the results of this poll by the Liberal Party to The Australian is suspect as it significantly over-estimated the support for Phelps while downplaying support for Liberal candidate Dave Sharma. It is well known the Liberal Party’s preferred position prior to an election is ‘behind’ – the thinking is that it will galvanise their supporters and discourage a ‘protest vote’.

These erroneous predictions are very much like the predictions from UK pre-Brexit pollsters such as YouGov (who in Australia conduct Newspoll) and most pre-Trump Presidential election polls!


Results of 2018 Wentworth by-election and pre-election polls

AEC Official Results

Wentworth*

Greenpeace

Liberal ‘Internal Poll’

2016 Federal

Election

Wentworth

Change since 2016 Federal Election

ReachTEL

Poll

Error

Internal

Poll**

Error

October 20,

2018

October 15,

2018

October 14-15, 2018

July 2,

2016

Sample

79,760

661

n/a

88,641

Method

Robo Poll

n/a

Primary Vote

%

%

%

%

%

%

%

Dave Sharma (LIB)

43.2

32.7

+10.5

34-36**

+6-9

62.3

-19.1

Kerryn Phelps (IND)

29.1

25.8

+3.3

25-27**

+2-4

n/a

+29.1

Tim Murray (ALP)

11.5

21.6

-10.1

n/a

17.7

-6.2

Dominic Kanak (GRN)

8.6

9.1

-0.5

n/a

14.9

-6.3

Licia Heath (IND)

2.3

5.6

-3.3

n/a

n/a

+2.3

Others

5.3

5.2

+0.1

n/a

5.1

+0.2

TOTAL

100

100

Two-Party Preferred

%

%

%

%

%

%

%

Dave Sharma (LIB)

48.9

37.6

+11.3

45

+3.9

67.7

-18.8

Kerryn Phelps (IND)

51.1

62.4

-11.3

55

-3.9

32.3 (ALP)

+18.8

TOTAL

100

100

100

100

*Official Results at 5pm Wednesday October 24, 2018. **The Liberal internal polling showed Sharma’s vote in the ‘mid 30s’ and Phelps ‘well into the 20s’ Source – The Australian (October 16, 2018: https://www.theaustralian.com.au/national-affairs/kerryn-phelps-to-trounce-liberals-in-wentworth-party-polling-shows/news-story/04a7ff7057c9172fd99d055294a37696).

The final Wentworth by-election result is not yet known because an unknown number of postal votes are yet to be returned to the Australian Electoral Commission. The AEC issued 12,788 postal vote envelopes for Wentworth. By Wednesday October 24, the AEC had processed 7,623. To be included in the final result postal votes must be received by Friday November 2, within two weeks from the date of the election and be post-marked as mailed by the date of the by-election.

Analysing the postal votes shows the Liberals gaining about 61.7% of the postal votes already counted on a two-party basis. If another 4,000 postal votes are returned by next Friday November 2 and the same preference flows apply, with 61.7% going to Mr Sharma on a two-party basis, the result will be close although Phelps will still win with just under 51% of the two-candidate preferred vote.

Surely this must mean the ‘death’ of cheap robo-polling & private party polls!

Roy Morgan SMS Polling was extremely accurate at both the last two Federal Elections: See Accuracy Results here.

Results analysed by Roy Morgan Helix Personas are available on a subscription basis. www.HelixPersonas.com.au.


For further information:

Contact

Office

Mobile

Gary Morgan:     

+61 3 9224 5213  

+61 411 129 094

Michele Levine:       

+61 3 9224 5215  

+61 411 129 093


For comments or more information please contact:
Roy Morgan - Enquiries
Office: (+61) (03) 9224 5309
askroymorgan@roymorgan.com