Statistics Watchdog Tells Gambling Commission to Flag GSGB Bias in Its Releases
By Antonina Tupikova · Founder, iGaming Times2 min read
The Office for Statistics Regulation says the Gambling Commission has met six of its nine recommendations on the Gambling Survey for Great Britain, but that the risk of overestimation is still missing from the statistical release itself. It has added a tenth recommendation on how the Commission used the Sturgis research to raise its confidence in the survey.
- The Office for Statistics Regulation (OSR) published a follow-up review of the Gambling Survey for Great Britain (GSGB) on 19 November, finding six of its nine May 2025 recommendations met and three partially met
- The OSR says the statistical release itself should set out the biases that may affect GSGB estimates, including the risk that some gambling behaviours are overestimated
- A new tenth recommendation asks the Commission to explain how research by Sturgis et al (2025) raised its confidence in the survey and its decisions since, including allowing PGSI estimates to be grossed up
- The OSR found no evidence of an intention to deliberately mislead, but told the Commission to reconsider its decision not to engage with unpublished academic work before peer review
- The Commission welcomed the review and says it will publish a progress update by February 2027, the deadline the OSR has set
The OSR Signs Off Six Recommendations and Keeps Three Open
The Gambling Commission has welcomed a follow-up compliance review by the OSR, the independent regulator of UK official statistics, published on 19 November. It concludes that the Commission "has carried out a significant amount of work" and has met six of the nine recommendations the OSR made in May 2025, after the Commission asked in 2024 for the survey to be assessed against the Code of Practice for Statistics.
Three remain partially met: communicating potential biases within the statistical release, comparing the GSGB with the Health Survey for England (HSE) and the Adult Psychiatric Morbidity Survey (APMS), and remaining open to challenge. The push-to-web survey's Annual Report 2025, published on 16 July 2026 from 20,775 adults, put the share with a PGSI score of 8 or more at 2.4%.
Bias Is Explained in the Technical Report, Not the Release
GSGB estimates of gambling and harm are higher than those from interviewer-led surveys, the OSR notes. Research commissioned in 2025 from Sturgis, Kuha, Howe and Maxineanu attributed "a large part of the variation" to sample composition driven by the subject of the invitation and to "socially desirable responding due to interviewer presence". In the OSR's view, the risk of overestimation for problem gambling estimates "is not as high as previously thought". But naming gambling in the invitation raised reported participation by 4 percentage points, and the OSR concludes that "bias does exist".

The Commission's technical report says "the survey overestimates the true level of gambling in the population". In January 2026, however, the Commission said the risk was "not as high as initially thought" and added no warning to the release. The OSR says users who read only the release "will not be aware of this important information".
Industry Data and the Peer-Review Position
A small number of users shared with the OSR an unpublished academic paper finding higher GSGB participation estimates than industry data for in-person casino table games, football pools and betting exchanges. The Betting and Gaming Council (BGC) gave the OSR industry data on casino visits, which is not public. The Commission said it would not normally comment on academic work before peer review; the OSR says it should reconsider, since concerns "should be given proper scrutiny and investigated thoroughly". The Commission has added a question on in-person casino games and met the BGC.
Some users also argued that the Commission misrepresented the Sturgis research when, in October 2025, it dropped its advice not to gross up PGSI estimates to population numbers. The OSR says that decision "does not directly follow on from the research findings" but is reasonable if the figures are described as estimates.
Commission executive director Sarah Fox said the GSGB "continues to provide a robust and authoritative source of evidence for gambling in Great Britain". The Commission's response letter says it will review how it presents the Sturgis research, and that APMS raw data is "now available".
Harm and Participation Estimates Carry Different Warnings
The OSR says the 2025 annual report's wording implies that oversampling of gamblers is offset by the absence of an interviewer. The review separates the two. On harm, it accepts that GSGB estimates "may be more accurate" than interviewer surveys. On participation, it points to a measured invitation effect.
The Industry Challenge
The OSR does not endorse the unpublished paper and says there may be valid reasons for the gaps with industry data. Its recommendation concerns how the Commission engages with such challenges. The Commission also publishes industry statistics on operators' gross gambling yield.


