How do I interpret UK gambling statistics reports when a new survey sets a new baseline?

How do I interpret UK gambling statistics reports? Find the series first

How do I interpret UK gambling statistics reports? Start by finding which survey series a figure belongs to. Then find which year of that series it comes from.

The Gambling Survey for Great Britain (GSGB) is the Gambling Commission's current official survey. Its first year of official data collection began in July 2023, using a new method. The Commission calls its results a new baseline.

So a GSGB figure can be compared with earlier GSGB years. It should not be lined up with older survey figures as if they made one trend. This page shows why, and what you can say instead.

Sources: Gambling Commission: GSGB 2023 introduction; Gambling Commission: GSGB guidance.

A new baseline — the Commission's own words

The Commission states the rule in three places. Its 2023 annual report says its findings 'represent the first year of a new baseline against which future changes can be compared'. Its 2025 report calls that year 'the third year of a new baseline'.

Its guidance on using GSGB statistics is more specific. It lists trend work as a proper use, measuring change against the 2023 baseline and the 2024 baseline. It lists direct comparison with other gambling or health surveys as a use to avoid.

Two baselines are named because the first year was short. Year 1 ran from July 2023 to February 2024. From the second year, data collection moved to a calendar-year basis.

Sources: Gambling Commission: GSGB 2023 introduction; Gambling Commission: GSGB annual report 2025; Gambling Commission: GSGB guidance.

Why the older surveys sit on the other side of the line

Before the GSGB, the Commission measured gambling in several ways at once. The table below sets them beside the new survey. The Health Surveys were face-to-face surveys that carried gambling questions roughly every two years.

In December 2020 the Commission consulted on replacing them with one method. It published the results in June 2021. The new survey is push-to-web: letters invite people to answer online, and a paper questionnaire follows for those who do not.

The 2023 report is direct about the consequence. Its findings 'are not directly comparable' with the Health Surveys or the British Gambling Prevalence Survey series. The reason it gives is the different approach of the new method.

Sources: Gambling Commission: GSGB 2023 introduction.

9,742, then about 20,000 — three survey years side by side

The first GSGB year had 9,742 adults taking part. The next two had 19,714 and 20,775. The 2023 year held only six months of fieldwork, which is why it is smaller.

Response was lower than planned in every year. The technical report gives an address-level response rate of 19% in 2023 and 2024 and 18% in 2025, against a target of 22%. Boost samples were added in 2024 and 2025 to lift the number taking part.

A bigger sample narrows the margin of error. The guidance cites the larger samples from 2024 onwards as one reason the Commission removed some of its earlier caution in October 2025.

Sources: Gambling Commission: GSGB technical report; Gambling Commission: GSGB guidance.

Mode — the design choice that moved the problem-gambling score

A new method can move a figure even when behaviour stays the same. The Commission tested this. It asked the London School of Economics and a team at NatCen to run an experiment on three design choices.

Mentioning gambling in the invitation letter raised the participation estimate, but not the estimate on the PGSI, a nine-question screen for problem gambling. A longer list of gambling activities had no detectable impact on the estimates.

The way people answered did matter. People who completed the survey online scored significantly higher on the PGSI than people answering by telephone. The Commission reads this as interview surveys under-reporting harm, because people soften answers when someone is asking.

Sources: Gambling Commission: GSGB guidance.

A survey year is a fieldwork window, not a calendar year

Each GSGB year is built from four waves of fieldwork. For the year labelled 2025, wave 1 began on 13 January 2025. Wave 4 closed on 22 January 2026.

The waves overlap, so a household reached in late 2025 may reply in January 2026. When you quote a GSGB figure, quote the year label and the fieldwork dates together.

The Commission also publishes each wave on its own. Its guidance says the annual data should be the default. Wave data suits questions about one period within a year, or questions asked only in some waves.

Sources: Gambling Commission: GSGB technical report; Gambling Commission: GSGB guidance.

Inside the series — 48% then 47%, and what that means

Within the GSGB, change can be measured. The Commission's trends report compares 2023 with 2025 for six key estimates. It found few differences in the gambling behaviours it examined.

Participation in the past four weeks was 48% in 2023 and 47% in 2025. The share with a PGSI score of 0 was 76.1% in 2023 and 76.9% in 2025. The report calls this a pattern of stability across most measures.

It did find some notable differences, listed below the table. None of them is a change in the overall rate.

These are population figures; they say nothing about any one reader. If gambling is causing harm to you or someone near you, GamCare is on 0808 8020 133, free, 24 hours.

Sources: Gambling Commission: GSGB trends 2023 to 2025.

Across surveys — what you can say instead of a trend

Suppose you have a GSGB figure and an older Health Survey figure for the same measure. The guidance says not to present the gap as change over time. Differences in design, method and sampling can mislead.

You may still point out the difference. The guidance allows it if you make clear that the methods differ. Its own worked example says the GSGB estimate 'is higher than estimates produced by other studies which use different methodologies'.

That is the whole of the permitted claim. It does not say harm has risen or fallen. For the two headline rates and the reasons behind the gap, see why Britain has more than one official problem gambling rate.

Sources: Gambling Commission: GSGB guidance.

The base — 47% of adults, or 27% without lottery-only players

Inside the GSGB itself, one figure can be quoted on several bases. The guidance names three: all participants, people who gambled in the past 12 months, and people who did more than take part in lottery draws.

The third base exists because lotteries are far more popular than any other form of gambling. In 2025, 47% of adults had gambled in the past four weeks. With people who only took part in lottery draws left out, the figure was 27%.

Both numbers are correct. They answer different questions, so name the base every time you quote one.

Sources: Gambling Commission: GSGB guidance; Gambling Commission: GSGB annual report 2025.

The range around a figure — 2.4%, or 1.9% to 2.8%

Every GSGB figure is an estimate from a sample, so it has a margin of error. The guidance gives an example. About 2.4% of adults in Great Britain had a PGSI score of 8 or more in 2025, the band UK statistics call problem gambling.

The same example says the true value is likely to fall between 1.9% and 2.8%. The guidance also gives a simple formula for a 95% confidence interval, which the calculator below uses. On 20,775 people it gives a narrower range, about 2.2% to 2.6%.

The published range is wider because the survey's design, such as its weighting, also affects the interval. Use the published range where there is one. When the annual report comments on a difference, it is one the Commission treats as statistically significant.

These are population figures; they say nothing about any one reader. If gambling is causing harm to you or someone near you, GamCare is on 0808 8020 133, free, 24 hours.

Sources: Gambling Commission: GSGB guidance; Gambling Commission: GSGB technical report.

How to read the next GSGB release

In a typical year the GSGB publishes four wave releases and one annual report. The steps below turn any of their figures into a sentence you can defend. They add to the wider checklist for reading gambling statistics on this site.

Most misreadings come from the base, not the number. A share of all adults and a share of people who gamble are different claims, which is why the denominator changes the story.

Sources: Gambling Commission: GSGB guidance.

What the statistics cannot tell you

The GSGB is careful work, but it has limits that the Commission sets out itself. The PGSI is not a measure of gambling addiction, and the guidance says not to describe it as one.

Some questions are asked only online, and those figures should be labelled online-only. The questions on wider consequences cover only the past year. They miss harm that carries on after someone has stopped gambling.

Published figures can also move after release. A correction amended some 2023, 2024 and early 2025 estimates. Most moved by between 0.1% and 1%, and the Commission says the overall narrative stayed the same.

The underlying data can be checked independently. Annual datasets go to the UK Data Service, and the technical report says the Year 3 data will be submitted by November 2026.

Sources: Gambling Commission: GSGB guidance; Gambling Commission: GSGB technical report.

Sources checked 23 September 2026

We read every source below on 23 September 2026 and quote them as published. All five statistical sources are Gambling Commission pages. The helpline number was read on the GamCare site on the same day.

The 2.2% to 2.6% range is our own arithmetic, using the formula in the guidance. Nothing on this page was measured by us.

Sources: Gambling Commission: GSGB technical report; Gambling Commission: GSGB guidance; Gambling Commission: GSGB annual report 2025; Gambling Commission: GSGB trends 2023 to 2025; Gambling Commission: GSGB 2023 introduction; GamCare: talk to us now.