Britain’s new Conservative government has barely settled into office, but already the results of last week’s general election have got certain members of UK society fearing for their future. They are scorned by the tabloid press and social media; even serious observers are questioning whether the country has been in thrall to them for too long. An inquiry has already been announced.

Opinion pollsters, the media told everyone, were predicting the closest election for decades. Labour and the Conservatives were neck and neck; weeks of constitutional chaos would follow the election as mandarins and officials wrestled with competing and overlapping political claims to power. The small print says that opinion polls should always be taken with a decent pinch of salt. But who reads the small print when there is an election on and a 24-hour news cycle to fill?

It took a single poll of voters post-voting to reveal the truth, which was confirmed as the counted results flooded in: David Cameron’s Conservative Party had grabbed 37% of the vote (see page 134). That was nearly seven percentage points ahead of Labour and, crucially, well outside the margins of error of all the previous deadlocked polls.

Amid the fallout, a single polling firm revealed that it had correctly predicted — and then buried — the result. Gathered the day before the election, its poll results seemed so out of line with what everyone else was saying that the firm did not dare to publish them. “I chickened out of publishing the figures,” confessed Damian Lyons Lowe, the chief executive of Survation in London. “Something I’m sure I’ll always regret.”

Nature’s readers can surely sympathize. The question of how to deal with anomalous data is a centrepiece of research, and the results can make or break careers — or launch scientific revolutions. From the discovery of the ozone hole over Antarctica to the observation that some people seemed unaffected by HIV infection, unusual results — data that make you go ‘hmmm’ — have led scientists to question their methods, their knowledge and, ultimately, their understanding of the world.

The importance of anomalies in science has spawned its own sub-field of research into how researchers respond to them. In the mid-1980s, psychologists supported by US military funds went as far as constructing a bespoke computer program to recreate how Hans Krebs reacted to surprising results during his discovery of the urea cycle in 1932. Others conduct in vivo studies by filming astronomers and physicists as they wrestle with unexpected findings.

The ultimate test of anomalous data is, of course, to repeat the experiment. But that demands that scientists have the courage and insight to treat such results seriously in the first place. How many potential discoveries lie in the waste-paper bin of history because the cautious chickened out?