Anti-Corbyn groupthink and the media: how pundits called the election so wrong

Reporting based on the current consensus will always vastly underestimate the possibility of change, argues James Fox

July 26, 2017 · 11 min read

The Times’ prediction of a Tory landslide
When prime minister Theresa May called an unexpected election, nearly all commentators believed that the move would significantly enlarge her majority.

After all, when the election was called, the polls showed May 10 points ahead of Labour, with predictions of a 100-seat majority. The Guardian among others doubled down on this prediction after the May local elections.

As late as three weeks before the election, the polls projected that the Tories would win between 391 to 415 seats (depending on which poll you were looking at), while Labour ranged from 152 to 175. The actual results had the Conservatives at 317 (down 13) against Labour’s 262 (up 30).

The assumption of no change

My intention in this article is not to disregard the role and value of polls, which have a worthy cause in chronicling the mysterious thinking of our national mind. I was certainly fascinated by the poll results throughout the campaign. However, as Cliff Zukin, professor of public policy at Rutgers University, has written, ‘polls don’t predict, they describe the situation at the moment’.

True, polling companies do us a disservice by claiming predictive or ‘forecast’ power, but the true responsibility for accurately conveying what polls can and cannot tell us lies with the major news sources who report on them throughout the electoral campaign.

It was the media’s responsibility, at least, to remind the public that the polling predictions were incorrect for the Brexit referendum the previous year, as well as being enormously wrong about the general election the year before.

In all polls, the raw data doesn’t tell you much. Polling relies on ‘weighting’ – turning a small sample into a sample that is thought to be representative, using statistical data from previous elections. This makes polling good at recognising tendencies and trends that remain consistent over time, but leaves them almost useless at predicting sudden or subtle shifts in our political orientation.

At this election, the pollsters’ assumption – backed, to be fair, by most of recorded political history – was that young people would not turn out to vote in the numbers they said they would. So the polls were in turn used to ‘prove’ that young voters would not make a difference, when in truth this was an assumption in their statistical model. Electoral impartiality rules do nothing to stop broadcasters taking polls as fact.

Backing the wrong horse

It’s easy to forget episodes in which the industrial media backs entirely the wrong horse with total confidence, but there are plenty, particularly in the run up to wars (like Iraq’s supposed ‘weapons of mass destruction’ in 2003). They frequently get it massively wrong. To get an idea of why they failed so badly, and why they do so often, we have to look at the way the industrial media presents information in general.

Two rules of thumb might be useful when thinking about how the industrial media presents information, and this is inculcated into every news outlet, from the Guardian to RT, from the BBC to Fox News.

Firstly, views that pundits persistently express are presented as the ‘consensus’ viewpoint, or the most mainstream and least contentious views.

Secondly, media organisations tend to pride themselves on ‘neutrality’, which means that when debate takes place the news organisations generally refrains from taking up one definitive side of the debate, instead trying to act as a mediator.

The grounds of the debate is whatever the media presents as the consensus. The two sides of the debate must fight it out upon those grounds. Debate takes place within the constraints of the consensus. The basic agreement is established first, then the precise implications of that agreement are fought out during the debate.

These two principles seem quite obvious at first glance, perhaps even inherent. Many would question whether journalism could be organised in any other way. But when one considers the significant impact that this structure has upon the content of information presented by the media, problems become apparent.

By the end of this article I hope to have demonstrated that this media structure is neither favourable nor inevitable.


The consensus viewpoints are always opinions and interpretations that are presented as the most respectable, reasonable, acceptable, or mainstream to the organisation and its investors.

Which views can be presented as mainstream in this way are always, of course, determined by people within the organisation who have some degree of editorial power. Those with stakes either have a direct role of influence in the organisation (managers, CEOs, editors, etc) or have financial investment which allows them to have some influence (like governments, investors, PR groups, etc). Consensus views are agreed to by a large enough proportion of the stakeholders that the organisation can present them as unanimous, whether they are or not, without causing discord amongst the organisation’s investors.

Examples of consensus views in the run up to the election are easy to find: there was consensus that May would win; that she had called the election to gain strength for Brexit negotiations (not to bury Tory expenses fraud); and that Corbyn was ‘unelectable’ and unpopular.

There were rarely even suggestions that May might lose her majority – such a claim would very likely have resulted in being laughed off the show

The assumption that the polls would remain relatively stable and May would win an improved majority was questioned by some Labour-supporting guests, but this represented a small voice fighting against an overwhelming tide of consensual opinion and analysis.

May’s strength was presented as the consensus viewpoint, and the debates leading up to the election often debated the details of the results, but rarely questioned the assumption that the polls would stay approximately as they were. Pundits instead speculated on how much May would win by, whether it would be the landslide she expected, and which of the other parties would make gains or losses. They rarely questioned the fundamental assertion that May would win an improved majority.


All the major media organisations are proud to present themselves as neutral. In debates they seek to provide samples of the most widespread opinions, to attain the most ‘balanced’ coverage. Some outlets are more neutral than others but, regardless of their bias, they all try to project a sense of neutrality.

The appearance of neutrality emerges by proportionally representing the two sides of an argument without obviously taking one side or the other. They get the antagonists to face one another and fight it out for about 4 minutes. The news organisation’s responsibility is to be a fair mediator, to make sure each has their time to speak, each gets a fair hearing, and each gets to make their case.

The guests and commentators have, almost without exception, already made up their minds. The viewer’s job is to choose a side. The option of developing your own interpretation in response to an open set of information is never really given. The experts present their cases then you are encouraged to pick which of the two acceptable options you agree with.

During the early weeks of the electoral campaign there were rarely even suggestions that May might lose her majority – such a claim would very likely have resulted in being laughed off the show. In fact, even when the polls had tightened significantly and things were altogether less certain, Damian Lyons from Survation was publicly mocked on the Daily Politics the day before the election, even though he was one of the only pollsters to correctly predict a hung parliament.

Pollster mocked for predicting a hung parliament

Debate which pushes disagreement beyond the tight limits of the consensus view is quickly marginalised and put in its place as a minority or ‘fringe’ viewpoint. It is not de-legitimised because it is poorly researched, biased or disprovable, but because it undermines the consensus viewpoint that determines the limits of ‘reasonable’ debate.

This is not a conspiratorial policy to silence dissent, it is simply the normal way the media operates, it can only integrate views which fall somewhere within its region of consensus. Views which exceed the edges of the accepted debate are not thought of so much as threatening and subversive, but rather as silly, fringe, extreme or irrational.

What’s the alternative?

At this point, one might ask, ‘What other options do news broadcasters have? Yes polls can be imprecise and wrong, but what other factors can we go by?’ Particularly when the polls are as clear as they were a few weeks before the election, showing huge leads for the Tories.

Yes, polls can change dramatically in a few weeks but, you might argue, the broadcaster has to go with the information it has. Therefore, they have to make educated predictions based on the information available, right?

Well no, the broadcaster does not have to make predictions at all – that’s the job of its viewers, and perhaps some of its guests. The job of the broadcaster is to provide clarity to the information they present and to broadcast information which is relevant and important for viewers to develop an informed opinion. They should cover party manifestos and policies in preference to the daily ‘who’s up, who’s down’.

Of course they should say what the polls are projecting, but it is equally their responsibility to give context to those results: how they have changed in the past, what might they be underestimating or failing to take into account, what assumptions the results are built on, and so on.

The best way to predict elections, or any other complex and unpredictable phenomena, is not to use just one metric as gospel just because it presents data in a scientific and empirical way, but rather to review many complex factors and variables and to assess which of these will prove to be most significant.

This is not an easy task. However, insofar as they seek to represent our political reality in an unbiased way, it is the responsibility of media organisations to try to cope with reality’s complexity.

Our level of understanding develops especially when we expose ourselves to unheard or previously unconsidered perspectives. These perspectives don’t necessarily come from people who reject the dominant narrative, but people who attempt to add something new to it – viewpoints from outside the consensus viewpoint, not within it.

Media organisations need to reverse the logic they have become so used to. Rather than starting with what is agreed, then constructing a debate on one side and the other, they need to begin with disagreement: what is conjectured, what knowledgeable people think is changing or trending, overstated or understated. From that point of disagreement, some kind of productive common ground or agreement might eventually become clear. If clarity remains elusive this will not tell us that we are doing a bad job at predicting the results, only that the results are highly unpredictable.

Consensus should be the result of open and contentious debate – it should not represent the limits of that debate or the ground upon which it is established. In the case of the election, factors that might have been investigated include the previous accuracy (or not) of polls, a consideration of what polls can’t capture or explain, and the global trend in recent elections away from elites and towards populist ‘anti-establishment’ candidates. Instead commentators devoted much energy to actively explaining away phenomena such as the rallies for Corbyn as irrelevant.

The most important responsibility of media organisations is to give the audience as much information they can so as to arrive at their own assessment of what might happen: this means fundamentally beginning from what experts don’t know, what they disagree about, what is unclear. Beginning from consensus will inevitably underestimate factors of change, and in our times of great change, instability and unpredictability, failure to account for what might change will leave us baffled by the outcome of just about any contentious and complex issue.

The media wants to appear to be certain, competent, knowledgeable and informed, but as long as it deceives itself and its audiences with a contrived sense of consensus, it will not only demonstrate that it is antiquated and incorrect, but will also exacerbate people’s feelings of confusion, turbulence, and a complete lack of clarity from major sources of political information.

Momentum responds to the Labour Muslim Network Report

Labour seems eager to ignore its Islamophobia problem. The Party is making a grave mistake, explain Solma Ahmed, Sonali Bhattacharyya and Mish Rahman

No holding back?

Labour’s road to recovery must bring together grassroots organising with a recognition of changing class composition, argues Christine Berry

Red Pepper interviews Momentum’s NCG

Join us on Friday 27 November from 5pm as we talk to Momentum NCG members Sonali Bhattacharyya and Deborah Hermanns about what's next for the left

Lynne Segal

Being Jewish in North Islington Labour Party

Calling for Jeremy Corbyn's reinstatement, Lynne Segal looks back on her experience of 40 years as a party member in his constituency

In and against, and outside, the party

Following major defeats, the left on both sides of the Atlantic must urgently get stuck into community organising, movement building and political education, argues Joe Guinan

The downfall of Robin Hood Energy

The sale of Robin Hood Energy doesn’t mean public ownership doesn’t work, but that we need to be more ambitious, argues Edward Dingwall