How political bias blinds us Ideological cocoons prevent experts from seeing, and engaging with, the wider world. 

Notwithstanding the general hysteria and catastrophism that has followed every twist and turn of the Johnson premiership, YouGov notes that “the ongoing Brexit chaos hasn’t adversely affected the PM’s favourability figures — which are, in fact, slightly higher now than they were when he first moved into Downing Street”.

In fact, despite the furore over Johnson’s suspension of parliament, the dismissal of special advisors, the expulsion of 21 moderate Conservative MPs and the PM’s allegedly dictatorial tendencies, support for the Conservative Party has increased.

The week before Johnson became Prime Minister, his party averaged 24% The week after he entered Downing Street it jumped to 29%. Today, it is 33%. Groupthink tells us that Johnson is making his party unelectable. The data tells us that Labour has only been ahead in one of the last 40 opinion polls.

There is, of course, a lot that could still go wrong for Johnson. With the SNP still dominant in Scotland, Britain’s electoral map is more constrained. Volatility is high and so it is much harder for either of the two main parties to win a majority.

But if Johnson’s strategy is to mobilise Brexit Britain while exploiting a divided Remain camp, then, so far, it seems to be working. All he needs to do now is capture between one-third and a half of what remains of Nigel Farage’s vote. If he does that,  then he will almost certainly cruise to a comfortable majority and his stay in No 10 will be longer than many of his critics think.

But all of this speaks to a deeper point. It underlines why — with a fresh election on the horizon — we should be far more critical of herd thinking than we currently are.

Think for a moment about all of the shocks that have rocked British politics. The triumph of UKIP at the European elections in 2014. David Cameron’s surprise majority in 2015. The rise of the SNP in Scotland. Jeremy Corbyn’s election as Labour Party leader. The Leave victory in 2016. The Trump victory a few months later. The hung parliament in 2017. And the strong result for Corbyn’s Labour.

Most of the political Twitterati herd did not see these shocks coming. The experts didn’t help much either.

In 2015, hundreds of academics, journalists and pollsters were asked in a survey to predict the outcome of the general election. This was based on the popular idea that there is “wisdom in crowds”. These crowds predicted a hung parliament in which Ed Miliband and the Labour Party commanded the largest number of seats while the Liberal Democrats and SNP held the balance of power. The reality was the first Conservative majority since 1992.

More complex election forecasts didn’t offer us much insight either. In fact, few, if any, successfully forecast the result. The most generous prediction gave the Conservatives 296 seats while the polling ‘guru’ Nate Silver was also way off, predicting 278. In the end, Cameron won 330 seats. Political scientists were left to reflect on what one described as their “substantial over-confidence in predictions of the election outcome”.

More recently, experts would tweet out Jeremy Corbyn’s dismal leadership ratings like clockwork every month: “Look how terrible he is!” But then Labour presided over the highest share of the vote since Tony Blair’s second landslide. The experts went quiet. As one academic rightly argued, the response to Corbyn reflected a wider problem: a tendency among experts to prioritise not objective analysis but their own normative positions and a desire to conform to the established ideological orthodoxy.

Much the same could be said about Brexit.

When the referendum arrived in 2016, another survey of nearly 600 experts delivered a remarkably clear view. Nearly 90% expected Remain to win. Just 5% felt that Brexit was the most likely outcome. Journalists were noticeably worse: 97% predicted Remain and just 3% Brexit. And most expected Remain to win by at least 10-points. There were simply too few people willing to question and challenge the herd thinking.

And then came the 2017 general election. This time around “[t]he prevailing view [of another expert survey]…was that the Conservatives will secure a large majority, with an implied probability of 65% that they will win a majority of more than one hundred seats”.

Once again this view was bolstered by even more technical election forecasts, all but one of which confidently predicted a solid Conservative majority (take a look for yourself). The largest majority of 124 seats was forecast by Ian Warren, an expert on the Labour vote.

It is easy to bash election predictions and take this argument too far, of course. We should remember that the forecasters had a better election in 2010 (although Nate Silver was still wide of the mark as he was, again, in 2016). And while, last week, Ipsos-MORI revealed that only 23% of people think that pollsters are trustworthy, the polls have actually got more right than people think.

The problem is that some of the things that they have called correctly — some online studies showed a swing to Leave, there were some state-level polls in the US that pointed to a Trump victory, or the MRP model that predicted a hung parliament in 2017 — were simply never accepted by the herd. Too many people only saw what they wanted to see.

Like the much-derided economists who failed to see the financial crisis coming, it seems only fair to ask why so many political analysts have failed to diagnose a succession of shocks. Often, it has been those who are closest to the world of politics who have been consistently outflanked by political change.

I suspect that several factors are at play.

The sharp liberal Left slant that runs through the Twitter and university communities ensures that few people have strong links to the values and voters that run against the dominant ideological orthodoxy. Ideological cocoons prevent experts from seeing, and engaging with, the wider world.

Rather than challenge our priors, it is much easier to let motivated reasoning — a bias toward an outcome or decision that confirms with what we already know — win the day. As a result, anything or anybody that deviates from the consensus and the herd is challenged, undermined or simply ignored.

Furthermore, and as Philip Tetlock, author of Superforecasting argues, experts tend to suffer from a potent interplay of over-confidence, binary thinking, a herding instinct and, to be blunt, hubris. All of this undermines their ability to gauge the public mood in an increasingly volatile political environment. This seems especially likely when most of the movements that are challenging the status quo are anchored in communities that most highly-educated and privileged experts simply do not understand.

To this we can add the dominance of the polls in shaping much of our media commentary, which is deeply problematic. Inaccurate polls feed inaccurate forecasts while many journalists, understandably, lack the time and skills to question the methodological choices being pursued. Equally, it’s easy to disregard those polls that simply don’t fit your requirements.

All of this underlines why — with a fresh election on the horizon — we all need to be far more sceptical of what we are told and more willing to challenge the consensus.

Tetlock suggested that the art and science of prediction is a skill that can be improved with practice. The jury is still out on that one. For the rest of us, one might suggest that another lesson can be taken from recent history: always challenge the herd.

---

Matthew Goodwin is Professor of Politics at the University of Kent. His new book, Values, Voice and Virtue: The New British Politics, is out on March 30.

GoodwinMJ