A major report claiming a significant Christian revival among young people in England and Wales has been withdrawn after it was revealed to be based on fraudulent survey data, casting a harsh light on the growing vulnerability of online polling to automated deception and “survey farming”.
The Bible Society’s “Quiet Revival” report, published in April 2025, had painted a striking picture of resurgence. It stated that monthly church attendance among adults had risen from 8% in 2018 to 12% in 2024, with the most dramatic increase reportedly among 18 to 24-year-olds. These findings, based on a 2024 YouGov survey, were widely circulated, suggesting a notable cultural shift. However, YouGov has since informed the Bible Society that the data sample was flawed. The polling company acknowledged that key quality control technologies were not activated during that specific survey due to human error, allowing a significant proportion of fraudulent respondents to corrupt the results.
The Broken Assumption and the AI Threat
While there is no direct evidence that artificial intelligence caused the fraud in this specific case, experts warn that the incident is symptomatic of a deep crisis in online survey research. Sean Westwood, an associate professor of government at Dartmouth College, states that the foundational assumption of survey research—that respondents are real people giving coherent answers—is now “broken.”
The proliferation of “survey farmers,” individuals or groups paid to complete questionnaires at scale, is a growing global problem. David Voas, an emeritus professor of social science at University College London, notes participants can generate substantial revenue this way. The emergence of cheap, accessible AI tools has dramatically exacerbated this vulnerability. Westwood warns that AI can be “weaponised,” programmed with a single instruction to systematically bias answers on topics like politics or religion while maintaining a plausible demographic profile, making the manipulation invisible to standard screening. He has developed an AI tool capable of bypassing 99.8% of current bot detection methods, illustrating the scale of the challenge.
YouGov states it employs a range of countermeasures, including identity checks, device fingerprinting, and proprietary systems like its Response Quality Score. A spokesperson said, “The rise of organised survey farms, bots, and now AI-assisted responses makes detection a vital, continuous and constantly evolving discipline.” However, the company’s admission that human error disabled core safeguards in the 2024 survey underscores how fragile these defences can be.
Why Youth Figures Are Particularly Suspect
The now-retracted report had placed young people at the heart of the alleged revival, but methodology experts highlight that this demographic is especially prone to distortion in online opt-in surveys. Courtney Kennedy, vice-president of methods and innovation at the Pew Research Center, explains that opt-in estimates for under-30s often contain high levels of error and are more likely to be influenced by “click farms.”
Kennedy notes that individuals skilled at concealing their identity online tend to be younger, and because young adults are traditionally hard for pollsters to reach, surveys frequently need such respondents. “It is advantageous to self-present as young because surveys tend to need such respondents,” she said. Furthermore, bogus respondents frequently exhibit “positivity bias”—a tendency to answer “yes” to any question—which artificially inflates estimates of behaviours like church attendance.
This flawed data stands in stark contrast to other established metrics. Analysis of the gold-standard British Social Attitudes Survey, which uses rigorous probability-based sampling, indicates a continued long-term decline in churchgoing, including among Generation Z. The Church of England’s own published statistics also report falling attendance over the same period.
David Voas criticises the withdrawn report not only for the fraudulent data but for a failure to critically engage with these existing sources. “If you were doing serious scholarly research, you would need to review the literature and see what other evidence was out there,” he said. The episode leaves a corrected public record but a lingering problem: the immense difficulty of undoing a compelling narrative. As Voas observes, “The amount of effort required to correct it is an order of magnitude higher than the effort needed to disseminate it in the first place.”
