23rd July, 2018
Fake opinion polling is a useful tool in covert and clandestine operations, and can be used in a variety of scenarios for a variety of purposes. It can be used, for example: to provide cover for election fraud; to sway the votes of parliamentarians; and to manipulate true public opinion.
Polling can be faked outright, or, as is more often the case, subtly manipulated via seemingly minor changes in methods and formulas. David Moore, a former executive of major US polling company Gallup, said in his book The Superpollsters:
“The views that people express in polls are very much influenced by the polling process itself…Slight differences…can have profound consequences.” 1
The ‘sampling’ process & the 2016 US Presidential election
One area of polling where differences can have profound consequences is the ‘sampling’ process. ‘Sampling’ is an attempt to overcome the impracticality of polling an entire electorate made up of millions of people. Instead, a relatively tiny subset of people are chosen who are supposed to be representative of the whole. The process by which this subset is chosen is called ‘sampling’, and it can be deliberately manipulated to produce particular results.
During the 2016 US Presidential election, opinion polling from establishment polling companies consistently pointed to a Hillary Clinton (Democratic Party) victory, but some people studying the data in detail pointed out that registered Democrats were being “oversampled”, meaning too many registered Democrats were included in the polls relative to their prevalence in the entire electorate.
There is direct evidence that this sort of “oversampling” is deliberate.
In October 2016 Wikileaks released private emails from Democratic Party operatives discussing how they manipulate polls by “oversampling” demographics more likely to vote Democrat.
It should be noted that ideological/political operatives are often involved in polling companies, for instance, Hillary Clinton’s chief political strategist for her 2008 campaign for the Democratic Party’s nomination was Mark Penn, the founder of ‘Penn Schoen Berland’, a international polling company.
In one of the aforementioned emails released by Wikileaks, a Democratic Party operative is writing to two of his colleagues saying:
“Hey, when can we meet? I also want to get your…folks to recommend oversamples for our polling before we start in February. By market, regions, etc. I want to get this all compiled into one set of recommendations so we can maximize what we get out of our media polling.”
Attached to that email was a 37-page guide containing specific instructions on how to oversample to achieve desired results.
One of the polling companies accused of oversampling Democrats was YouGov, an international polling company based in the UK. In the lead up to the 2016 U.S. Presidential election they conducted weekly opinion polls on voter intentions, every one of which had Hillary Clinton winning comfortably. Political commentator Bill Mitchell, one of very few who predicted Trump’s election victory, says, for instance, a YouGov poll taken in August 2016 had a sample that over-represented registered Democrats by around 16%.
The significance of YouGov’s activities will become apparent as you read further.
Polling fraud also happens in Australia
In October 2017 The Australian newspaper reported that fake polling was being circulated about the federal electorate of New England in the days before the High Court decision on New England MP Barnaby Joyce’s eligibility to sit in parliament. The polling was bad for Joyce, and was apparently meant to pressure him into not re-contesting the seat in a by-election if found ineligible. Joyce though, did re-contest when he was found ineligble and won the seat in a massive landslide.
In August 2013 Clive Palmer was interviewed by Emma Alberici on the ABC’s Lateline program and the following exchange took place regarding his activities during his time as Campaign Director for the National Party in QLD:
Palmer: When I was a former party director there were polling companies that I used to give large donations to, and they’d write the results for them.
Alberici: Are you saying the polling’s rigged?
Palmer: Well, of course the polling’s rigged in Australia.
We even had an blatant public attempt at polling fraud recently, with Waleed Aly, a host of Channel 10’s The Project, using the program to encourage his Labor and Green-voting audience to lie to pollsters to prevent Turnbull losing his 30th Newspoll in a row.
Even when honestly undertaken, the political opinion polling carried out by companies like Galaxy Research, Essential Media, Reachtel, and Ipsos, is often presented as having far greater weight than it deserves by agenda-driven media outlets. Up-ticks and down-ticks can be used by the media to create public hysteria, and build political momentum. Polls can thus feed back into themselves and become self-fulfilling, as in the ‘bandwagon effect’.
The public release of polls can also be timed by the agenda-driven media outlets who commission them, in order to achieve a particular desired political effect. Sky News host Paul Murray highlighted an example of this in December 2017.
Turnbull’s Poll Problem: The YouGov Solution?
Malcolm Turnbull cunningly used poll hysteria to launch his putsch against Tony Abbott, citing Abbott’s 30 consecutive two-party preferred Newspoll losses. But by the end of 2016 Turnbull himself had lost his 6th Newspoll in a row. As his losing streak continued, commentators began recalling the polling benchmark he established to attack Tony Abbott, saying his position would be untenable if he replicated it.
The pressure was now well and truly on. It was estimated that, if he continued his Newspoll losing streak he would reach the 30 milestone at some point in early 2018.
What to do? What to do?
In December 2017 came the announcement that YouGov, the aforementioned international polling company, had purchased Galaxy Research, the company that runs Newspoll.
YouGov had only been in Australia for a short time at that point. In fact they seem to have started operations in Australia only a couple of months after Turnbull’s ascension to the Prime Ministership in 2015. In late June 2017, just as Turnbull was approaching his 15th consecutive Newspoll loss, YouGov started running an opinion poll, under the YouGov brand, on the federal voting intentions of Australians. That poll was notorious among experienced poll watchers for its outlier status, heavily favouring the Coalition above all other polls (see graph below). Veteran polling analyst William Bowe, author of the Poll Bludger blog, even refused to put YouGov’s two-party preferred numbers in his post headings, which is his normal practice for other polls, preferring instead to post them amidst disclaimers. Bowe called the YouGov poll “very unusual“, “highly unorthodox“, and “out on a limb”. According to Bowe YouGov are “weighting heavily by past vote” and their samples don’t have enough respondents from the youngest age group.
Below i’ve drawn a graph showing all of YouGov’s two-party preferred polling results for their original YouGov-branded poll, which ran bi-weekly from late June 2017 until mid-December 2017. For comparison, on the same graph i’ve plotted the two-party preferred results from 4 other major opinion polling companies, including Newspoll results prior to December, when it was announced that Newspoll had been acquired by YouGov.
With these uniquely eccentric results, and no brand name recognition in Australia, YouGov was not taken seriously by anyone of any influence. It was still all about Newspoll, Newspoll, and Newspoll. That was the brand name with the excellent reputation that politicians, especially Liberals, took very seriously. Indeed, that’s why Turnbull had cited it in the first place.
“With YouGov’s operating model and panel, and Galaxy’s reputation and foothold, the combined business will be in a strong position…”
What wasn’t announced in December 2017 though was that the methodology for calculating Newspoll’s two-party preferred numbers was radically changed at this very time. It was only months later, on the 30th of April 2018, that any change was publicly acknowledged by YouGov. The change apparently involves an increased preference flow from One Nation to the Coalition based on the results of the WA and Queensland state elections but, as polling analyst William Bowe said:
“…the exact split of One Nation preferences being used by Newspoll remains a trade secret.”
According to the 30th of April newspaper article that first revealed the methodology change, the last 8 Newspoll results, at that point, had been conducted using it. That would mean the first Newspoll using the new methodology would be the one released on the 4th of December, which saw the Coalition up-tick from 45 to 47 in the two-party preferred measure.
Below i’ve drawn a graph of Newspoll results before and after the methodology change and, as you can see, the change coincided with the beginning of an upward trend for the Coalition.
The “coincidences” don’t stop there.
Lately, whenever Turnbull is under pressure over policy or leadership there seems to be an up-tick in Newspoll just in time to pacify the backbench, many of whom are money-grubbers and careerists who are happy to sell out the values of Menzies if they think their seat is safe.
There are also down-ticks, but only when Turnbull’s leadership is looking secure.
In early April Turnbull’s leadership was under serious pressure as he was about to lose his 30th consecutive Newspoll, thus failing his own standard. Although this did indeed eventuate, there was a convenient up-tick from 47 to 48, which took the wind out of the sails of any challenge. Turnbull could use a “trajectory” argument to protect his leadership through this dangerous phase. When the next Newspoll up-ticked to 49, the whole 30-in-a-row issue went quiet. These up-ticks had come at the perfect time for Turnbull.
The Newspoll remained on 49 until the 28th of May, when it went down to 48 just a few days after the date for a series of by-elections was announced. The knowledge of an upcoming election tends to temporarily quell internal party divisions.
The Newspoll stayed at 48 for a while, but in late June Turnbull came under serious pressure over his “National Energy Guarantee” policy. A large segment of the Liberal backbench were pressuring him to get approval from the party-room before finalising a deal with Labor state premiers at the next Council of Australian Governments (COAG) meeting in August. The Newspoll published on the 2nd of July provided a much needed up-tick for Turnbull.
If Turnbull comes under more pressure over the “National Energy Guarantee” (which is just an emissions trading scheme with better public relations) Newspoll might even produce a 50-50 result.
At this point there might be some ardent coincidence theorists who have a few questions:
If a politician were committing polling fraud, why would they allow their numbers to down-tick at all?
A strategic drop can be useful, because it can help remove suspicion and make polls look more realistic. If the key issue is maintaining the Liberal leadership, and the poll drop happens at a time when that leadership is safe due to other factors, then it isn’t a risk.
But the 30 straight losses actually eventuated. Wouldn’t stopping that be the whole point?
The answer here is essentially the same as the question above. The point of the enterprise would not be to prevent 30 Newspoll losses in a row. It would instead be to save his leadership so he can continue implementing his policies. Had Turnbull not reached 30, suspicions would have been raised about the new ownership and the methodology changes. Indeed, the ownership and methodology changes did trigger some commentary along those lines. In January 2018 Sky News host Paul Murray said:
“There was one poll that, all the way through last year was the best news for the government…It is this poll. The YouGov poll…that throughout most of 2017 had the government either level or in front…This would otherwise be known as an outlier. But, it’s not an outlier anymore because of this news that alot of people missed towards the end of the year. YouGov…they now own Newspoll. So it’ll be fascinating to see…Has the methodology changed? I don’t know, but there’s certainly a new owner in town and they’ve been far more favourable to team blue than team red. The Prime Minister will be desperately hoping for a change.”
Had Turnbull not reached 30, there would have been even greater scrutiny on Newspoll’s ownership and methodology changes. Newspoll might no longer have been widely trusted, and up-ticks would not have triggered endorphin release in the brains of Liberal backbenchers.
The important thing for Turnbull was that, even as he hit 30 Newspoll losses in a row, there was no serious threat to his leadership, and that’s because there was an up-tick after the 3 previous Newspolls had the government seemingly stagnant on 47. Newspoll’s undamaged reputation meant the up-tick was taken seriously, thus it pacified the backbench and eliminated the possibility of building momentum for a leadership change.
The next Newspoll was also a loss for Turnbull, and had him breaking Tony Abbott’s record. But again, because he up-ticked to 49 all the talk was positive for him. He was now within “striking distance” and so forth.
Anyone claiming this is all just a coincidence should keep in mind just how many simultaneous coincidences they’re proposing. They’re saying that YouGov just happened to enter the already bloated Australian polling market a couple of months after Turnbull’s ascension, and they just happened to start an outlier poll favouring the Coalition as Turnbull hit the halfway point to his 30-in-a-row Newspoll benchmark, and they just happened to acquire Newspoll when nobody was taking much notice of the YouGov-branded poll, and at this very time there just happened to be a change to the methodology for calculating the very poll result Turnbull was interested in, and that change just happened to benefit Turnbull, and recent up-ticks in Newspoll just happened to occur exactly when Turnbull needed them to secure his leadership and push his policies through the backbench.
For those confident in this coincidence theory, please contact me. I have some ocean-front property to sell you in Alice Springs.
Back to Homepage
1. Moore, D. (1995). The Superpollsters: How They Measure and Manipulate Public Opinion in America (2nd ed.). (pp. xiii, 325). New York: Four Walls Eight Windows.