David Spiegelhalter: ‘Data does not tell you what to do’

In a keynote interview, Spiegelhalter reflected on how statistics and data have been communicated during the Covid-19 and warned that data alone cannot make decisions. 

Spiegelhalter, like other scientists, has seen an influx of media invitations in the past year, and he has learned to stick to his own knowledge when being interviewed. He said: “I’ve had to learn the hard way – in the beginning I felt that I should give an answer to questions beyond my realm of expertise.”

“I have learned that one thing is you should stick to what you know about and shut up about everything else. I am a statistician, but nobody understands what statisticians are – they ask you about what people feel about it. I don’t know, because I’m not a psychologist.”

He added: “Basically, I just try to explain things – I haven’t got a side or an agenda, for example, I’m not saying we need to lock down more. I haven’t got an agenda, except use the data in the best way possible and make that data accessible in a clear and transparent way.”

Spiegelhalter said it was “not good for the scientific community” when it’s unclear for audiences whether scientists are discussing the state of things or their opinion. He said: “I don’t mind people having an opinion, but just make it clear when you’re moving on from the facts to what you think should be done.”

Despite frustrations over some media reporting – the “golden oldies of horror stories” that are the daily death figures being one – Spiegelhalter said: “I think the media as a whole have done very well indeed, and are getting better – the BBC, the Financial Times, the Guardian, the Spectator – they’ve had some really good data journalism going on with a huge number of explainers.”

However, he added: “When political journalists try to explain the specificity of a diagnostic test, I just fall about laughing. That’s not their job – it’s a challenge for anybody.”

Spiegelhalter criticised the paucity of information on the Covid-19 vaccinations rollout, saying availability of data had been “shocking”.

“We get some big figure and some detail but nothing like enough to understand what’s going on in different parts of country, or with different priority groups and different communities.”

Clearly sharing the data and being transparent is key, he said, “not just government holding it to themselves and allowing bits to dribble out”.

He added: “Open data is a deeply important and ethical principle for government.”

The UK’s speedy development, procurement and rollout of the vaccine itself has been a “staggering success”, Spiegelhalter said. 

“The prioritisation programme and the decision to approve the AstraZeneca vaccine for over-65s were bold, courageous decisions made by scientists without political interference and they’ve stuck to it – they were the right decisions. I find it moving that this group of people in the back room came up with these bold, courageous decisions.”

The pandemic has been full of unknowns, but Spiegelhalter said trustworthiness, rather than certainty, should be the aim of communication.

“Our research suggests very strongly that if you’re confident about your uncertainty, the audience does not lose trust in you – there’s clearly a strong feeling we have to be certain but that’s completely wrong, as far as we can see.

“The main thing is to demonstrate trustworthiness. Onora O’Neill [philosopher and member of the House of Lords] emphasises that it’s not to do with ‘how I can get myself trusted, but how I can demonstrate trustworthiness'.”

This is particularly important for statisticians generating large volumes of information, he added.

Statisticians also have to listen to people, he said. “You have to listen. You don’t just communicate, you have to know the audience, know their concerns and treat them with respect.”

Spiegelhalter expects the pandemic to change how evidence is used by governments, but he cautioned against relying too much on data: “Don’t follow the data. It does not tell you what to do. This idea of ‘following the science’, that data will offer up its secrets, is crass. Data does not tell you what to do – but I wouldn’t want to make decisions without it.”

Book today to join the event and access all content via the MRS on-demand service – mrsannualconference.com/book