In 2024 we carried out research for Bible Society that showed 12% of people attended church once a month or more, a significant increase from 8% in a previous 2018 study. In particular, the poll showed a rise in church attendance amongst younger people.

The 2024 survey generated considerable public attention. When it initially attracted media interest, the study’s research team went back to the data and found no methodological issues at that time.

Due to the ongoing scrutiny this work received we have re-analysed the data using new and additional better tools and techniques. This review, carried out by our Platform and Data Sciences teams, now found that specific demographic groups particularly emphasised in this survey contained a number of respondents who we can now identify as fraudulent. This occurred at a higher rate than typical amongst other polls at the time and enough to make a few points difference to the key result. This was because some of the anti-fraud measures available in 2024 were not administered in the optimal way.

YouGov takes full responsibility for this. Online market research has come under increasing attack by fraudsters in the past several years. YouGov have greatly strengthened our procedures to defend against potential fraud using new technologies, within-survey fraud detection, identity verification, and duplicate detection. We have reduced the percentage of UK panellists who fail these tests to under 1%, which we believe is among the lowest in the industry.

We track every panellist across time, platform, and at every step of their lifecycle – building a unique profile that evolves with every interaction. We own our proprietary panel; we don’t rely on third-party networks or single-survey snapshots. This gives us the right context to decide who to trust, who to verify, and when to intervene. Our approach is built around three principles – collectively they inform how we design policies, collect signals, detect threats, and ensure high-quality data. These measures are constantly evolving and improving over time.

The other factor affecting the key result was that for the 2024 Bible Society survey we used a unique and complex sample. Given the high level of correlation between ethnicity and religious behaviour, it included quotas interlocking age and ethnicity, with a focus on obtaining a representative sample of younger people by ethnicity. This is the audience most likely to be attending new, evangelical and charismatic churches. It also included an oversample of regular church attendees who were weighted back to the level of church attendance in the main sample to allow more in-depth analysis of those who attend church.

There is a tendency for fraudulent respondents to have a bigger impact amongst harder to reach groups. Problematic accounts are more likely to be found among these demographics, as there are fewer available genuine respondents and bad actors target those groups where they think they will receive more surveys and more incentives. This means this specific survey was particularly vulnerable to sample quality issues. The client did ask us about methodological issues related to this survey but did not have access to YouGov’s fraud detection tools and systems.

Not all of our latest anti-fraud measures can be backdated and so we cannot tell what the full impact would have been had they been available and correctly used at the time. To get the most robust picture possible we will carry out an updated study later in the year with fieldwork taking place by early summer.

YouGov CEO Stephan Shakespeare said: “YouGov takes full responsibility for the outputs of the original 2024 research, and we apologise for what has happened. We would like to stress that Bible Society have at all times accurately and responsibly reported the data we supplied to them. We are running the survey again with Bible Society to get robust data on this topic.”

Subscribe to the YouGov newsletter