I wrote last year about some ‘experimental statistics’ from the Gambling Commission that suggested that there are far more problem gamblers in Britain than we thought there were. They appeared to show that 2.5 per cent of the UK population have a gambling problem whereas every official survey in the last 25 years has found the rate to be around 0.5 per cent. The experimental statistics are now official statistics and anti-gambling campaigners are claiming that they prove that “the harms caused by gambling have been massively underestimated”.

When four out of five people refuse to take part in your survey, you no longer have a random sample of the population

But do they? To save money, the government is switching from face-to-face surveys and telephone surveys to online surveys despite it being well established that “online surveys over-estimate gambling harm”. Until recently, people’s gambling habits were tracked as part of the Health Survey for England. People were randomly selected to take part in the survey and those who accepted were interviewed in their home. Since people feel it is their civic duty to take part in a national health survey, its participation rate was a respectable 50 per cent.

The new Gambling Survey for Great Britain is very different. Letters were randomly sent to 37,554 addresses asking the residents if they would like to take part in an online survey about gambling. If they didn’t respond, they were sent another letter telling them that they could do the survey by post if they preferred. Despite the inducement of a £10 fee, most people did not respond. Only 19 per cent of those who were sent the letter ended up taking part in the survey, mostly online.

When four out of five people refuse to take part in your survey, you no longer have a random sample of the population. Unlike the Health Survey for England, the Gambling Survey for Great Britain is explicitly badged as being about gambling. Who is this most likely to appeal to? People who gamble a lot. And people who gamble a lot are more likely to problem gamblers than people who don’t.

Everyone knows this. When Professor Patrick Sturgis assessed the ‘experimental statistics’ earlier this year he warned that there was “a non-negligible risk that [the new estimates] substantially over-state the true level of gambling and gambling harm in the population.” The Gambling Commission knows this because it commissioned the Sturgis Review and spent last week warning campaigners and the media that “[t]here is a risk that the GSGB may overstate some gambling behaviours and therefore estimates should be used with some caution.”

Have campaigners and the media used the estimates with caution? Reader, they have not. Despite the Gambling Commission explicitly stating that the figures in the survey cannot be used “to gross up the prevalence of problem gambling”, the Guardian reports that a 2.5 per cent prevalence rate “would equate to 1.3 million people”. Will Prochaska from the Coalition Against Gambling Ads said: “This new data, even with the caveats that the industry have pressured the Gambling Commission into making [?!?], is a source of national shame. The government must now treat gambling as one of the greatest threats to public health in society.”

So which figure is nearer the truth – the new estimate of 2.5 per cent or the previous official estimate of 0.3 per cent? Cheerleaders for the new figure say that people are likely to under-report problem gambling symptoms in face-to-face interviews out of shame and embarrassment. This seems plausible, although the Health Survey for England addressed this by having interviewees write down their answers to the more awkward questions themselves. In any case, is it really credible that 7 out of 8 problem gamblers would lie in a face-to-face (or telephone) interview, but not in an online interview?

That is the one and only reason to prefer the new methodology over the old. The far more compelling reason to prefer the old survey is that it had a much higher response rate and it didn’t attract problem gamblers like moths to a lightbulb. Looking at the survey results in detail, it is clear that those who completed the survey were not representative of the general public. For example, 11.4 per cent of them “had thought about taking their own life or had attempted to do so in the past 12 months”. Among the 48 per cent of them who had gambled in the last month, the figure was almost identical at 11.6 per cent. This is an inconvenient fact for those who claim that gamblers are more likely to commit suicide, but 11.4 per cent is more than twice as high as the rate of suicide ideation among the general public. Perhaps people who are prepared to answer 200 questions about gambling in an online survey for a tenner are more depressed than the average person? It wouldn’t surprise me, but whatever the reason, the small minority who agreed to take part in the survey were clearly unusual.

The gamblers who responded were not even representative of gamblers. Their rates of participation in online betting, bingo and casino gambling were all much higher than that seen in previous surveys. The last Gambling Commission survey found that 7.1 per cent of Britons bought scratchcards. Among the gamblers in the new survey, the figure is 13.3 per cent. Previous surveys have found that around 2 per cent of Britons have played bingo in the last month. The figure in the new survey is more than twice that. Gambling companies know roughly how many customers they have and it is not as many as the new survey suggests. As the gambling expert Dan Waugh told me, “our analysis of hard industry data reveals systemic and substantial over-reporting of participation in betting and gaming in the Gambling Survey for Great Britain.”

Why has the Gambling Commission put out figures that it must know are wildly incorrect?

The new survey not only attracts more gamblers than the old survey, but attracts highly engaged gamblers who play lots of different games. These are the people who are most likely to display problem gambling traits. Need more proof? Fine. In the new survey, one per cent of gamblers had “sought help from gambling support services” in the last 12 months. That amounts to around 360,000 people a year. GamCare, the main provider of gambling support, says it treated 9,009 people in 2022/23. The National Gambling Helpline took 39,000 calls from gamblers. Gamblers Anonymous and private health services will have spoken to a few more, but the final tally would still be a fraction of the 360,000 implied by the new survey. It is obvious that the gamblers who took part in the Gambling Survey for Great Britain were more troubled than the typical gambler and it is equally obvious that you cannot extrapolate the figures across the whole country.

Why has the Gambling Commission put out figures that it must know are wildly incorrect? The Commission has attempted to argue that it is a toss-up between the Health Survey and the GSGB as to which is more accurate, but is this what it really thinks? Last year, the Commission’s lead adviser, Professor Heather Wardle sent an e-mail to the Commission explaining that the main problem with the GSGB was “topic salience”, i.e. the fact that it was advertised as a gambling survey. She went on to say that the issue of under-reporting in face-to-face surveys was “marginal” by comparison. The Commission may be peddling a line it doesn’t actually believe.

Meanwhile, the increasingly professional anti-gambling groups are cock-a-hoop. Asked on breakfast television what policies he hoped this “better number” would inspire, Will Will Prochaska immediately said “the statutory levy”. Charles Ritchie from Gambling with Lives said: “The new government must immediately bring in a statutory levy on gambling industry profits”. I have written before about how anti-gambling groups are fighting like rats in a sack to get a slice of the £100 million pounds that a gambling levy could raise and how Will Prochaska teamed up with Jolyon Maugham’s Good Law Project to knock GambleAware out of the race. Those who are familiar with Jolyon’s career in activism will not be surprised to hear that his complaint to the Charity Commission about GambleAware was rejected this week, but while he might have lost the battle, he may yet win the war.