The year ahead

Most people turn away from politics over the holiday period when the…

Cutting Your Power Bills In Half And Other…

A few years ago there was a scam where people were promised…

Values Based Capitalism: The Imperative of Defining Commitment…

By Denis Bright Editorial insiders at The Weekend Australian (28-29 January 2023)…

A walk in the forest

Bayerischer Wald can be just as hard to get to than it…

An Emergent Premier Chris Minns - Uniting Sydney…

By Denis Bright After more than a decade in Opposition, NSW Labor is…

Forget Australia Day And Celebrate: Rum Rebellion Day…

After pointing out for a number of years that January 26th isn't…

Whither Constitutional Change?

Within a very short space of time, we are going to be…

Distracted by hate, we are robbed

We are at a crossroads. The Ultra High Net Worth Individual (UHNWI)…


Day to Day Politics: Making Sense (or Nonsense) of Political Polling

Tuesday 28 March 2017

Anyone interested in politics who says they don’t take notice of polling is just kidding themselves. Trying to make any sense out of them can be trying at times but they are always full of conjecture and intrigue. Watching Insiders on Sunday the Newspoll of last week was mentioned and spoken of as if it were the only poll in existence, or at least the only one you should take notice of. Frankly that’s nonsense. The only way to make sense of them is to put them all together over a period of time and evaluate the trends or lack thereof.

I have said before that the Newspoll was a rouge one. They do throw them up now and then. Mondays IBIS Fairfax Poll had Labor a clear 10% in front and I predict Essential will be the same. If they were both true it means that millions of people would vote differently one to the other.

The Crickey Poll Bludger which frequently does an analysis of them all had this to say:

The Fairfax papers have an Ipsos poll that belies last week’s improvement for the Coalition in Newspoll. The report identifies Labor’s two-party lead as 55-45, but it features a chart showing Labor to be leading 56-44 “by overall preference flows”, whatever that means. The primary votes are Labor 34%, Coalition 33% and Greens 16% – a high Greens apparently having become a feature of this series. Malcolm Turnbull is down five on approval since November to 40% and up three on disapproval to 48%, while Bill Shorten is down two to 35% and steady at 53%, while Turnbull’s lead as preferred prime minister is down from 51-30 to 45-33. The poll was conducted Wednesday to Saturday from a sample of 1400.

Now if you are as concerned with Australian politics as I am then no doubt you take an interest in polling. What do they tell us? Are they legitimate? How do pollsters arrive at their conclusions? What about their accuracy? Can we have confidence in them or are they overrated?

Whether you like them or not it seems that Australians have a great appetite for them. The media recognises this and promotes them as the definitive measurement of the political standing of all parties.

The Australian newspaper even creates stories around its Newspoll. Often an exercise in how to just make things up. But how do we know the difference between Australia’s most popular pollsters and the diverse techniques they use? However, one thing for sure is that we do have a morbid fascination with polling. Political leaders may brush them off as being meaningless but surreptitiously they are taken very seriously.

Modern political polling began in 1936, with two polls attempting to predict the outcome of the American presidential election. The Literary Digest conducted its poll by sending out 10 million post cards asking people how they would vote. They received almost 2.3 million back and said that Alfred Landon was leading Franklin Roosevelt by 57-43 per cent. In contrast, market researcher George Gallup employed a much smaller sample of only 5,000, but because he ensured that it was representative of the American voting public, he predicted Roosevelt to win by a landslide. In the event, Roosevelt won 60% and Landon just 37%. The Literary Digest lost credibility and subsequently merged with Time magazine in 1938.

We can be assured that the polls are not going to go away. They will in fact intensify as we get closer to the election even if its more than two years away

So it is important that they are of the best quality possible. We need to know the differences between the good ones and the bad ones. Or at least how each arrives at its conclusions. But how can we? Australia has a number of national pollsters. The Australian newspaper publishes Newspoll. It usually publishes a poll on federal voting intentions every two weeks but is more lengthy when an election is years away. It does not disclose the criteria it uses on its website but rather espouses its expertise and track record. It is known that it polls landline phone accounts only.

The problem with this is the diminishing access to a variety of people of younger age groups. Telstra estimates that it will lose 60% of its landline users over the next few years. Even now one would have to question where they source enough younger folk to participate. Are people on landlines, the stay at homes, truly representative of the populace?

Fairfax’s The Melbourne Age and The Sydney Morning Herald parted, after a 19 year association, with Neilson in June of 2015 this sought a new partner and now publish under the title IPIS using a similar formula.

Roy Morgan is the only poll that takes in landlines, mobile telephony, face to face and the internet. It delivers its findings fortnightly via email. It usually delivers a sharp contrast to the other polls.

Galaxy is an independent research company operated by Jamie Briggs. It doesn’t outline its methodology for political polling but its web site does indicate a broad spectrum of media usage.

The Essential Report is based on a weekly rotating sample of 1000 and is usually accompanied by a survey of questions of social interest that usually gives a fair indication of how people are thinking on various issues.

Then, there is Reachtel. It relies on both landline and mobile telephony for both marketing and polling using recorded messages. It’s a relatively new innovation that it yet to be fully tested over time. People press numbers (interactive voice response) in relation to answers but they aren’t asked age or sex.

All of those listed use different methodology for the vital ingredient of preferences. The two party preferred is the figure that matters most. They all allow for a margin of error using different criteria. You can add to these the preponderance of online polls like The Drum that saturate the internet and poll every policy and controversy. Of course there is not an online newspaper that doesn’t poll something every day which usually just reflect the readership of the newspaper.

Then we have the television news polls that are meaningless, only reproducing the views only of those with an earnest interest in whatever the subject is. Should I mention audience polling? The media are obsessed with polls. Donors, supporters and political parties really do fret over, or use them tactically to create impressions about how well their party is doing. Conversely, polls are criticised as biased, inaccurate, or simply wrong if they don’t support your view. So what’s in a poll? How do we measure the veracity one to the other? The fact is we cannot. Polls need to be put into perspective.

They are snapshots that reflect public opinion at a specific point in time. Of utmost importance to polling is the credibility of the sample and we are not able to test this. It is the same as trying to understand why the voting on talent shows is never revealed. There are however, some things we do know. We know that by historical evidence poll samples under 1000 are untrustworthy and inaccurate. We also know that the way in which a question is framed will often determine the outcome. And we also know that they can be very accurate if measured over a long period. They can also be a precise gauge of public opinion at the time, but in terms of predicting an election result two years away, to be utterly useless. They can tell us, for example, that the Rudd/Gillard governments were very unpopular, but their policies were popular, and that the Abbott government never received the usual honeymoon period gifted to a new government.

Or that an unpopular leader can benefit from a national crisis or at least the perception of it. They can show us how different age groups are thinking. For example most polls showed that young people hated Abbott with a vengeance. Some polls like Morgan poll over a two-week period rotating between face to face interviews and telephone. The rotation might include a particular event or controversy in the cycle or miss it altogether.

The best way of to evaluate what the polls are saying at any given time is to take an average over several different polls on a continuous basis. This has the advantage of combining different sample sizes and methodologies to give greater precision.

As I said earlier Newspoll is considered the benchmark of polls. Not necessarily for its accuracy, but rather because of its ability to create a story around its figures. Often a misleading one I might add. Because they only poll land lines there are often not enough young people at home when they ring. So they weight the responses to equalise ages and sex. Weighting polls is trying to equalise the demographics. Of course polling is a competitive business and pollsters won’t disclose how they weight. That would be like giving away your favorite recipe. Getting too complicated. How about this then: ‘The only poll that matters is the one on polling day’.

My thought for the day

“We exercise our involvement in our democracy every three years by voting. After that the vast majority takes very little interest. Why is it so?”



 544 total views,  2 views today


Login here Register here
  1. Barry

    The Poll Bludger is an excellent site for political wonks with some serious arguments that happen in the comments thread, also as a home grown Tasmanian site I’d recommend Kevin Bonham’s site
    A comprehensive round up of not only federal polls but also state, not as regular as PB but very interesting

  2. Möbius Ecko

    Couldn’t let this go through. Wow a red blush poll, you learn something everyday.

  3. Max Gross

    The LNP are not governing. They are clinging to office by the skin of their teeth (Or donkey skin, or something).

  4. David1

    Morning John, thoughts with the folk in Queensland weathering the storm, hope the animals are also all cared for. Re the Insiders panel, with the makeup of Sundays blatant right wing Coalition bias, they were never going to touch the ‘Labor ahead by 10 points’ in the Ipsos and Essential’ polls.
    Savva is a disgrace and the only discussion group she is qualified to be a member of is on her home territory, Murdoch’s Sky. My admiration for Kristina Keneally grows as she battles the rabid right practically single handed, in Murdoch’s domain.
    You are correct of course when you write ‘the only poll that counts is on polling day’ except they tend to influence who leads political parties outside polling day. These days Labor are an exception of course.
    Good article thanks

  5. Richard Creswick

    Nikki Savva a disgrace? Can it be so, the recipient of a lifetime achievement award from Her peers? Sadly, despite her obvious attempts to appear even-handed, the liberal bias, learned at the knee of John Howard, shines through. Cassidy, being a former media adviser to Bob Hawke, has the difficult job of avoiding potential allegations of pro Labor bias. The show’s ratings seem to indicate it gets the mix about right.

  6. Roswell

    Sorry to be off topic but this was too good not to share:

    Saw a cartoon of Trump sitting at his desk in the Oval Office with the caption “The Ruble stops here”. ?

  7. Terry2

    The former Newspoll – now Galaxy – was a fortnightly poll of around 1150 people with landline telephones, conducted by live interviewers reading out questions and entering responses. The new Galaxy-run ‘Newspoll’ adopts a mixed methodology and a larger sample size. Respondents will either be ‘robopolled’ or surveyed online with the two sources of data being combined to produce a result. Robopolling is fully automated : a computer makes calls to households automatically and a pre-recorded questionnaire is played to respondents who enter their answers using the keypad on their phone.

    Since Robopolling was introduced it has become evident that certain of us – including moi – hang-up as soon as they hear an automated voice and this tends to be the case with the more progressive people in our society, so the poll is already impaired and is likely to favour people like Pauline Hanson. Although, this is balanced out to some extent as research suggests that One Nation voters have a problem with digital devices and tend to have their keypads jammed up with Cheezels and the like and this distorts the results …..or so I’m told !

    Robocalling doesn’t have a good name, do you remember when John Howard was thrown out of office and lost his seat in parliament ? He had adopted robocalling to get people to vote for him but it rebounded as households reacted unfavourably, children answering the phone were scared, pets hid under beds and there was a higher incidence of indigestion in the community than is normal.

    Have a good day !

  8. kerri

    When studying geography at tecaher’s college, we did a unit on surveys. It should be compulsory for all citizens who vote! We learned all the tricks and mistakes. Like how a question is phrased often leads to the answer you want rather than a realistic answer. eg; Brandis’s phrasing of same sex marriage. A loaded question.
    When teaching I told my year 12 kids that a survey was useless if the subjects were less than 100!
    Deep groans! But seriously some kids would survey 25 people and multiply by four meaning 2 people represented 8% of the cohort! Completely ridiculous.
    But by far the most outstanding and long lived maxim from that college unit was
    There will always be one section of society that will not be represented! The people who don’t do surveys!
    I suspect, many, like myself, hang up when the robo voice is on the other end!

  9. David1

    Richard you say “The show’s ratings seem to indicate it gets the mix about right”. What are the ratings? How do they compare against competition from other channels? How do we know the mix of the audience is a good balance of political persuasions? All factors that for me would indicate the ‘mix is near enough to be balanced’
    #Insiders hash tag on twitter shows otherwise.

  10. Wayne Turner

    The most pointless and BIASED LOADED question that should NOT be asked in ANY of these polls,in Australia is:-

    “Preferred Prime Minister” – We the public do NOT elect,and do NOT vote directly for the prime minister.They are chosen by the parties.Plus is used by the BIASED MSM to focus on if it’s result favors their leader/party of choice,when the 2 party preferred is bad news.

    Also,it’s a LOADED & BIASED question,because using the term “Prime Minister” plants in peoples minds and LEADS them to think of who the current “Prime Minister” is.Especially for the political ignorant.

    If they want/must ask about leader/s.A better worded question to ask is: Preferred leader?

    Finally,the lazy and biased MSM love to use polls as horse race style journalism,often to be biased and lazy eg: How the biased MSM continued to report non-stop with “HEADLINE NEWS” the bad one’s for Labor when Gillard was PM.IE: Self fulfilling prophecy and to use against Gillard and Labor.

  11. Graeme Henchel

    Shallow Mal part 2

    Shallow Mal has been returned
    but now more weak than ever
    The double D has shown us all
    Shallow Mal is not too clever

    Shallow Mal has had a fright
    The ego had a bruising
    A tantrum on election night
    Now a morbid fear of losing

    Shallow Mal is on the edge
    The right still have his balls
    More policy paralysis
    Avoiding party brawls

    Shallow Mal has made a call
    A dump on Kevin Rudd
    Another weak decision
    Showing Shallow Mals a dud

    Shallow Mal is trying to look
    all upbeat and excited
    But pumping fists and waving arms
    Won’t keep this mob united

    Shallow Mal can thank himself
    For a senate that is worse
    Eleven on the cross bench
    Including Howard’s curse

    Shallow Mal must watch his back
    The Thug is sure to plot
    The Delcons on the right still dream
    Of returning to the clot.

    Shallow Mal is getting hit
    By Ley and Centre link
    As right wing nutters move to split
    Shallow Mal is on the brink

    Shallow Mal is near the end
    His future has been Trumped
    This whole charade will fall apart
    Shallow Mal will soon be dumped

    Shallow Mal was once a climate prince
    He is now the king of coal
    The people know he’s on the take
    As he digs a deeper hole.

    Shallow Mal has jumped the shark
    He is going the full Tony
    A petulant tanty aimed at Bill
    The last resort of a phoney

    Shallow Mal can’t wash his hands
    As penalty rates are cut
    The workers will be taking aim
    At kicking Malcolm’s butt

    Shallow Mal watched the west dissolve
    As deals with phonies backfired
    The polls continue to show that
    Mal’s use by date is expired

    Shallow Mal is in so much shit
    Even Dutton thinks he’s hot
    Mal is stalked by a potato
    You couldn’t make up this plot.

    Shallow Mal is changing 18c
    Except he knows it will come to nothin
    It’s just a bone to the rabid right
    And another nail in Malcolm’s coffin

    Shallow Mal is walking dead
    His Faustian pact is f*cked
    So break out the popcorn now
    As we watch him self destruct

  12. jimhaz

    I don’t care about non-end of term polls – other that when the reporting is being used as political propaganda by the media, as they certainly were during the Rudd and Gillard years. Seemed to be a new poll every few days.

    I do care about the Toxic Oozes (Trumps) polls as when it gets high enough that should trigger an in-house assault on him.

  13. helvityni

    Graeme Henchel,

    ‘Shallow Mal was once a climate prince
    He is now the king of coal’

    I suppose it’s the very shallowness that allows you to be anything; today you praise multinationalism, tomorrow you make it easier to be a racist…

  14. Terry2


    Bowen and surrounding areas affected by cyclone Debbie have lost power : Malcolm Turnbull and Josh Frydenberg have blamed the Queensland government for its unrealistic renewable energy targets ; Peter Dutton has laid the blame at the feet of HRC President Gillian Triggs.

  15. Matters Not

    Terry2 – Pauline Hanson blames Islam. Apparently there’s one Muslim who lives in Mackay but picks tomatoes in Bowen.

  16. helvityni

    David 1, you are right, there was no mention of Labor being ahead in the polls by last Sunday’s Insiders, ‘our’ ABC is leaning towards the Right alright…

    I’d love Kristina to come back into politics, she’d make a good PM one day…

  17. helvityni

    …oops, again, multiculturalism …not ‘multinationalism’

Leave a Reply

Your email address will not be published. Required fields are marked *

The maximum upload file size: 2 MB. You can upload: image, audio, video, document, spreadsheet, interactive, text, archive, code, other. Links to YouTube, Facebook, Twitter and other services inserted in the comment text will be automatically embedded. Drop file here

Return to home page
%d bloggers like this: