The curious tale of the 1992 BBC Exit Poll

For us in Northern Ireland today is Polling Day Eve, a seemingly annual occurrence just like Christmas itself. As at Christmas, for the nerdier among us at least, it brings memories of polling days past.

One of the first I recall clearly was the 1992 UK General Election, notable in my mind precisely because it is so often misremembered in retrospect. It is indeed a classic case of how popular memory often clashes with how the event was experienced at the time. Unlike the popular memory, the opinion polls actually had the election too close to call on the eve of polling and, most notably of all, the BBC Exit Poll absolutely did not forecast a Labour win.

Yet almost everyone thinks it did. Indeed a Professor of politics stated, immediately before the 2005 Exit Poll, that it did; which is particularly curious, because he was actually in the studio in 1992.

So why do people misremember things?

Proof

What is most striking to me about this misremembering of the BBC Exit Poll on that evening of 9 April 1992 is that it is so easy to disprove. Here is a screen shot of the Exit Poll, which appeared immediately after polls closed at 10pm on BBC1. Plainly, not only does it not predict a Labour majority, but it does not even have Labour as the largest party.

Even this obvious proof, however, does not stop people stating that the BBC Exit Poll forecast a Labour win. Why?

Peer pressure

The first lesson here is that people tend to surrender to peer pressure. So many people believe that the BBC Exit Poll forecast a Labour win that it becomes very difficult to be the odd one out and insist that it didn’t – even if you can “google” proof of that fact in a matter of seconds.

It is an odd feature of human nature that few will thank you for googling that. Ultimately, we prefer the narrative that the “experts” forecast a Labour win but the people delivered a Tory triumph. It is much more dramatic than a narrative that the Exit Poll was on the right lines but fell short of predicting the overall Conservative majority which actually occurred.

Extremism

The other aspect tied to this is straightforward extremism – extreme stories make for better stories. A slightly wrong Exit Poll makes for a much lesser story than an extremely wrong one.

What is strange is this just demonstrates how narrow the margins are between a devastating embarrassment and a glorious triumph. In 2015, with opinion polls similarly showing both main parties more or less tied, the BBC Exit Poll similarly projected an outcome of the Conservatives as the largest party but short of a majority. In actuality, as in 1992, the Conservatives in reality had an overall majority. The 2015 Exit Poll was closer to reality than in 1992 – it had the Conservatives closer to an overall majority and in reality that overall majority was smaller – but only just. Yet 2015 is still seen as a glorious triumph for the pollsters who were perceived as basically bang on, and 1992 lives on in ignominy. Small margins…

Late change

It is still curious that a Professor who was literally in the studio just after 10pm on 9 April 1992 would, thirteen years later, have stated a blatant untruth about the content of the Exit Poll released at that time. Part of this is for the above reasons – it is easier and more dramatic just to go along with the narrative, even if you are a Professor who knows it to be untrue.

Another reason, however, is that the Exit Poll headline was changed at the last minute – literally. Presenter David Dimbleby subsequently said that as he began the countdown to the announcement of the Exit Poll, the screen in front of him was indeed displaying a picture of Neil Kinnock with the headline “Narrow Labour majority”. This was changed, less than a minute before the announcement, and he heard in his ear only that “everything had changed” and that he should follow what was placed in front of him.

So in fact the Professor probably had seen a screen forecasting a Labour victory; he was one of very few who did.

Late voters?

In fact, the broadcast – which is available in its entirety on BBC iPlayer and YouTube – went on to change the Exit Poll forecast again before a single result was in – shifting to “Conservatives short by 21” as its central forecast and “Conservatives short by 6” within range. This was not that far from the final outcome, an overall Conservative majority of 21.

This opens up a curious point that it may well be that Conservative voters genuinely turned up at the polls later in the day. Exit Polls announced at 10pm are typically based on votes cast (well, recast in the mock ballot box presented by the exit pollster) until 9pm, but they may subsequently by updated with information from the final hour. This should make little difference – few voters wait until the final hour – but occasionally it can matter.

“Forecasts”

The truth also is that there was an extent to which the “computer” and the experts rather embarrassed themselves in 1992.

This had been a constant risk since the first Exit Poll in 1970, In the same way that the American networks only got caught out by calling Florida too early in 2000 because it turned out to be decisive (as early as 1960 they declared victory for Kennedy when they called California for him, but he didn’t actually win California – on that occasion it just so happened that, by the time they withdrew the call, Kennedy had won enough other states to be elected President so few really noticed), UK networks only got caught out in 1992 because they maintained a “forecast” too heavily based on the original Exit Poll rather than real results for too long.

This is again a constant feature in human psychology – we acquire information and then we assess future information based on that original information, but we still give too much credence to the original as the benchmark. In 1992, the very first result from Sunderland South showed that the Conservative vote share had barely fallen; the third result, from the marginal seat of Basildon, showed a tiny swing to Labour meaning it missed out on a key target seat required even to force a hung parliament. Yet for some considerable time afterwards, interviews and commentary on the programme remained based on the assumption that a hung parliament was likely; indeed, the official “forecast” did not change to a Conservative overall majority for several hours after the Exit Poll and indeed for several hours after it was entirely obvious that it would be the outcome.

Disbelief

Of course, one reason it took so long to recognise the likelihood of a Conservative overall majority (and, thus, why there is a historical reflex that the experts got it so wrong with their Exit Poll) was that it was so utterly baffling. It remains baffling to this day. In the midst of a recession, the Conservatives received what remains the highest ever popular vote in UK electoral history, with their vote share barely falling at all from that received in the previous two elections when they won landslide majorities. The only reason that they lost seats at all was the distribution of Labour and Liberal Democrat votes (including an element of tactical voting), rather than the total number of them.

This brings us to the modern day. We often hear of election “experts”, but the truth is no such people exist. Election outcomes are consistently baffling. As one newspaper put it on 10 April 1992, if elections were decided by electrons they would be entirely predictable; the fact they are decided by electors means they are entirely unpredictable.

That’s why we love them…