The Gambling Commission’s flagship prevalence survey has overstated participation by up to 694% for some activities, according to new research from Dan Waugh of Regulus Partners, which compares the regulator’s statistics against real operator data.
The Gambling Survey for Great Britain (GSGB) was launched in 2023 as a bespoke online survey targeting 20,000 respondents per year, billed by the Commission as the source of “one version of the truth” on gambling participation and problem gambling in the country.
But research presented by Waugh at the UNLV Eadington Conference 2026 suggests the survey may be significantly overstating how many people gamble.
Waugh compared GSGB participation estimates for three activities against administrative data obtained directly from operators, finding the survey figures were substantially higher in every case.
Speaking to NEXT.io, Waugh said: “The data reveals substantial over-reporting of gambling participation in the GSGB compared to what is actually happening in the regulated market.
“There are only two possible explanations for this. Either the survey sample is skewed or there are massive – and hitherto undetected – black markets for land-based casinos, betting exchanges and the football pools.
“Whichever explanation one applies, it is clear that the GSGB cannot be considered a reliable source of data for measuring gambling behaviours in the regulated market.”
The flagship survey’s data on gambling harms – which is significantly higher than previous estimates – has been highlighted by gambling critical voices as proof further restrictions on the UK market are needed.
Survey figures dwarfed by operator data
For casino visits, Waugh obtained actual visit figures from the Betting and Gaming Council, which collects data covering 100% of the licensed casino market.
The GSGB estimated that between around 1.75 million and 2.25 million visits took place for the period between July 2023 and February 2024 for casino table games alone.
Actual recorded visits to casinos for all purposes over the same period were just over 1 million, and Waugh suggested survey estimates had landed between 69% and 176% higher than real-world visits.
Waugh further noted that BGC casino operators estimate that between 40% and 80% of casino visits do not involve playing table games at all, meaning the true gap is likely to be considerably larger.
On that basis, the GSGB estimates for casino participation were between 408% and 628% above the administrative data figures across 2023 and 2024.
The discrepancy for football pools was starker still. The Football Pools Limited holds a monopoly on the product, making its customer data a definitive measure of participation.
The GSGB estimated around 863,000 unique players in the past four weeks for a given period in 2023, against just 109,000 actual customers recorded by the operator – suggesting a survey overstatement of 694%.
For betting exchange use, Waugh used data from Betfair, which accounts for an estimated 85% of the exchange betting market.
Scaling up Betfair’s figures to reflect the full market, the derived estimate of exchange bettors for January 2024 to January 2025 was around 172,000, compared with the GSGB’s estimate of around 612,000, with Waugh suggesting a gap in the figures of 225%.
Waugh said the findings were consistent with known biases in survey methodology. He pointed to the risk of “topic salience bias”, under which the people who respond to a survey about gambling are more likely to be gamblers than the general population, alongside the issues of low response rates and “selection bias” more broadly.
Previous academic research by Williams and Volberg (2009) and Sturgis (2025) has identified similar patterns.
The GSGB had acknowledged the risks of bias before launch and committed to proceeding only if those risks could be addressed, Waugh noted.
The Gambling Commission has stated that questions in the survey are interpreted by respondents as intended.
Waugh’s paper also raised the question of whether the GSGB figures might reflect gambling activity taking place outside the regulated British market rather than overstatement.
However, the implied per-visit or per-player spending figures produced when cross-referencing the GSGB data against the Commission’s own industry gross gambling yield statistics also raised coherence concerns: for non-remote casino games, the implied spend worked out at between £13.69 and £20.80 per visit.
‘Systemic bias’
Waugh concluded that his analysis pointed to “systemic bias” in the GSGB that affects both participation and frequency estimates, and warned that regulatory policy risks being based on inaccurate data.
He called for a range of data sources to be considered rather than relying on any single survey, and said administrative data from operators had an important role to play in testing survey reliability.
Regarding the GSGB, a Commission spokesperson previously told NEXT.io: “The GSGB is the largest survey of its kind in the world and has been subject to significant external scrutiny by those with recognised expertise in survey design.”
This is not the first time the survey has come under fire.
The Office for Statistics Regulation’s review of the GSGB last year recommended the Commission should communicate in its statistical releases that the data risks overestimating gambling behaviours.
However, the Commission opted not to go down this route, highlighting an LSE review of its work that it claims showed the overestimation fear was not as high as previously thought.
This assertion was later challenged in a report by Waugh who argued the UKGC misrepresented the findings, which he said actually demonstrated mixed results which should have weakened confidence in the survey.
Share