Much of the debate around the data underpinning the quiet revival centres on the differing survey methods used by different researchers.

The BSA Survey is known as the “gold standard” in the polling industry because of its longevity, and its use of random probability sampling, which means everyone in Britain has an equal chance of being picked to take part.

In comparison, YouGov is an “opt-in” survey company, meaning people volunteer themselves to take part, in exchange for points, which can be converted to cash.

Hackett says that opt-in polling companies can be victim to “bogus respondents”, skewing their data.

“And it’s not random,” he says. “The distortion tends to be highest among younger respondents.”

He says bogus respondents could include people taking surveys very quickly to get rewards, pretending to be younger than they are so they can get access to a broader range of polls, or even those in poorer countries using virtual private networks (VPNs) to take surveys as if they are in richer ones, as the money is worth far more to them.

It was the Bible Society that wrote the Quiet Revival report based on the YouGov figures it commissioned. How is it responding to the idea of bogus respondents?

“Sometimes, especially in smaller, quick and rough surveys, you just get these bonkers answers,” says Dr Rhiannon McAleer, director of research and impact at the Bible Society.

She says that having looked in detail at the responses to what was a lengthy survey of around 100 questions, she is satisfied the respondents were genuine and their answers consistent with people of faith.

Artificial intelligence (AI) is another problem for the polling industry – with respondents potentially using chatbots, which can carefully mimic the way a human would complete a survey, to do the work for them.

YouGov told the BBC that unlike other pollsters, it recruits and maintains its own panel of respondents.

A spokesperson said: “Our approach to dealing with bot respondents is therefore layered across the panellist lifecycle, from registration to participation to reward, combining identity checks, device fingerprinting, multi-source geolocation, real-time threat scoring, and payout oversight to ensure bad actors do not slip through the net.

It says it is going to look again at the topic when it repeats the study later this year.