Facts on regulation
Article series
European Gambling RegulationPart 3 of 4Start with Part 1: Balancing the Odds: British and Dutch Gambling Regulation
Measuring Gambling Harm in Europe
What British survey methods and Dutch market monitoring can—and cannot—tell us
Gambling harm and the size of the unlicensed market are related policy concerns, but they are not the same measurement problem. A survey of gambling experiences asks different questions from a market monitor estimating where money is spent.
Britain’s Gambling Survey for Great Britain (GSGB) and the Netherlands’ official market monitoring illustrate why those distinctions matter. Read together, they can help identify questions for policy. They cannot be turned into a simple league table of which country is protecting players better.
Measuring Gambling Risk — Data Insights from the UK and the Netherlands
Market Surveillance and Regulatory Oversight
Two sources, different purposes
The GSGB collects information about participation, gambling experiences and adverse consequences among adults in Great Britain. It uses a push-to-web approach with an option to complete a paper questionnaire, rather than relying entirely on interviews conducted in the home.
The Dutch Kansspelautoriteit’s market monitor combines information from licensed operators with estimates of unlicensed activity. Its channelisation measures concern the distribution of players and gross gambling revenue. Neither measure is a direct estimate of gambling harm.
That difference should shape how the findings are used: a fall in licensed revenue does not, by itself, establish a rise in harm, and a change in a survey estimate does not necessarily establish a change in behaviour.
Making sense of the numbers
The GSGB’s first annual report, covering 2023, estimated that 2.5% of adults aged 18 and over in Great Britain scored eight or more on the Problem Gambling Severity Index (PGSI). That is a screening threshold, not an individual clinical diagnosis.
Earlier health surveys used different methods, age ranges and geographical coverage. Their results cannot be placed alongside the GSGB as if they formed an uninterrupted time series. The Gambling Commission’s August 2025 research summary explains both the differences and the continuing uncertainty.
METHOD MATTERS
The same questionnaire can produce different estimates when the invitation, sample or way of answering changes. A higher estimate is not, on its own, evidence of a sudden increase in harm.
A 2022 study by Patrick Sturgis and Jouni Kuha found that online surveys in its comparison tended to include more frequent and online gamblers, pointing to selection bias. It should not be read as a definitive explanation of every later GSGB result.
Randomised experiments published in August 2025 examined additional mechanisms. Online self-completion produced higher PGSI scores than telephone interviewing, consistent with people being less willing to disclose sensitive experiences to an interviewer. Mentioning gambling in the invitation also affected reported participation. The experiments helped explain differences between methods; they did not establish a single, error-free population estimate.
EU flags outside a contemporary government building, symbolizing coordinated policy efforts on gambling harm and market oversight.
The limits of European comparison
Before comparing countries, readers need to check the population, age range, reference period, screening instrument, sampling method and mode of interview. A percentage among people who gamble online is not a percentage of all adults. A past-year estimate is not a past-month estimate.
Methodological differences can explain some variation, but it would be equally misleading to assume that all differences between countries are statistical artefacts. Real differences in products, participation and harm may also exist. Identifying them requires comparable evidence.
Independent scrutiny is part of the evidence
In May 2025, the Office for Statistics Regulation reviewed the GSGB. It recognised the survey’s frequent and comprehensive coverage while recommending improvements in transparency, quality assurance and communication of limitations.
The accompanying statement called for clearer warnings about possible overestimation and further investigation of biases. It did not endorse the GSGB as a statistical gold standard, nor conclude that the figures were unfit for publication. Those are materially different positions.
READ THE DENOMINATOR
People, active accounts, stakes and gross gambling revenue measure different things. A channelisation percentage is meaningful only when its denominator and period are stated.
Submitting a rating on a mobile device, with casino chips and a roulette wheel in the background—highlighting the importance of consumer experience in gambling regulation.
The Dutch policy dilemma
The Ksa’s autumn 2025 monitoring report estimated that about 94% of online players gambled exclusively with licensed operators, while licensed providers accounted for an estimated 49% of online gross gambling revenue in the first half of 2025. These are different measures, not contradictory descriptions of the same percentage.
The regulator also reported lower average losses among players in the legal market following protective measures. It warned that some players might move to unlicensed sites. The combination illustrates the policy challenge: protection within the licensed sector and movement outside it must be monitored together. A before-and-after observation alone cannot isolate the effects of a particular rule.
Later revision: The Ksa subsequently revised its estimate of licensed GGR channelisation for the first half of 2025 from 49% to 56%. The figures above describe what was reported in October 2025 and should not be mistaken for the latest estimate of that period.
What would improve the comparison?
Consistent definitions and transparent methods would make cross-border discussion more useful. Reports should identify who funded the work, explain how respondents or operators were selected, publish uncertainty where possible and distinguish observed data from modelled estimates.
For policy evaluation, repeated measurements using stable methods are especially valuable. Regulators also need evidence on outcomes: whether interventions reduce harmful losses, whether people obtain support, and whether those at risk move between licensed and unlicensed providers.
These are analytical priorities, rather than evidence that European regulators have agreed to adopt one survey model.
Conclusion
Better gambling data does not eliminate the need for interpretation. The GSGB makes the effects of survey design visible; Dutch monitoring shows how player-based and revenue-based channelisation can tell different stories.
The useful question is not simply which percentage is highest. It is what the number measures, how it was produced and which policy conclusion the evidence can support.
Sources and further reading
- Gambling Commission: New research on gambling survey estimates, 14 August 2025.
- Sturgis, Kuha, Howe and Maxineanu: Three Experiments on the Causes of Differences in Estimates of Gambling and Gambling Impacts in General Population Surveys, August 2025.
- Sturgis and Kuha: survey-method comparison, Public Health 204 (2022), pp. 63–69.
- Office for Statistics Regulation: Review of the GSGB, May 2025.
- Kansspelautoriteit: autumn 2025 monitoring report, 13 October 2025.
Correction — 14 October 2026: We clarified the distinction between gambling-harm surveys and market monitoring, corrected the description of survey-method research and replaced an incorrectly identified academic reference. We removed an unverified operator quotation, unsupported claims about other regulators’ intentions and conclusions that went beyond the evidence. This article retains its original October 2025 context.
A user interacts with a digital chart projected from a tablet, reflecting how both the production and interpretation of data shape policy responses in gambling regulation.