Across the line
Gambling Commission prepares to deceive
On Thursday this week the Gambling Commission crosses a line. The British gambling market regulator will publish results from its Gambling Survey for Great Britain (GSGB) in the knowledge that the results are inaccurate; and that they will be used to agitate for tighter controls on consumers and higher taxation for licensees.
First published in 2024, the GSGB reports substantially higher rates of ‘problem gambling’ (as measured by the Problem Gambling Severity Index or ‘PGSI’) than all other sources of official statistics, stretching back almost two decades. These include the ‘gold standard’ NHS Health Survey for England, the NHS Scottish Health Survey, the NHS Adult Psychiatric Morbidity Survey and the British Gambling Prevalence Survey. Last year, the Commission was provided with hard evidence that the GSGB over-reports rates of participation in landbased casino gaming, exchange betting and the football pools – three sectors where the availability of reliable administrative data can be used to test self-report survey estimates. The landbased casino gaming, exchange betting, and football pools sectors are clearly discrete modes of gambling, which means the over-reporting in the sectors that can be tested strongly implies a problem endemic to all sectors in the survey as well as participation as a whole.
The Gambling Commission does not dispute the quality of the administrative data presented (indeed, it formerly collected and published the casinos data itself); but it has proved distinctly unwilling to admit that its survey might be wrong despite glaring and incontrovertible evidence that it is. In supplementary evidence provided to the House of Lords Liaison Committee last month, the regulator set out a two-pronged defence. First, it asserted that the GSGB had been “developed by experts, reviewed by experts and approved by experts”. Second, it drew attention to conclusions from three experiments conducted last year by researchers at the London School of Economics (LSE), which it seemed to think put it in the clear.
The first of these arguments is clearly unscientific. It implies that the Commission is willing to disregard cold, hard evidence on the basis that it is inconsistent with expert say-so. Neither is it the case that all experts are fans of the GSGB – many are not. Some of those who have endorsed the survey are heavily involved in anti-gambling activism. The regulator’s decision to elevate the opinions of researchers whose goal is to ‘denormalise’ gambling (and who liken betting to tobacco and asbestos) over hard, empirical evidence reveals just how little interest it has in evidence-based policy or carrying out its duty as a balanced and objective regulator.
The second argument is more complex than the Commission would like us to believe. Last year’s ‘three experiments’ conducted by researchers at the LSE produced two main findings. First, survey description matters. In one experiment, a ‘gambling’ survey elicited much higher rates of gambling participation and higher PGSI scores than an otherwise identical ‘health’ survey. In a second experiment online data collection resulted in higher rates of gambling participation and higher PGSI scores than interviews conducted by telephone.
These results are unsurprising. They reinforce findings from studies in Canada and elsewhere dating back a decade-and-a-half. The first supports the theory of topic salience bias - that the GSGB is likely to over-sample gamblers (and highly engaged gamblers in particular) because it is advertised as a gambling survey. The second indicates that telephone surveys may under-report gambling and problem gambling. The relevance of this latter finding to the question of NHS data reliability is unclear. Health Surveys are conducted in-person (rather than by phone) and use self-completion booklets specifically to guard against the effects of social desirability bias. Previous studies have shown that in-person interviews are much less likely to be affected by this bias than telephone surveys. The Health Survey was therefore deservedly considered a 'Gold Standard' within the parameters of the possible; the GSGB represents more than a few steps backwards in terms of reliability, but in a direction that seems to suit the Commission and the 'experts' it likes to fund.
Taken together, the results of these experiments reinforce concerns that the GSGB overestimates participation and prevalence. The second experiment might indicate under-reporting in the Health Survey and other NHS surveys – but this is a matter of speculation (because the experiments did not test for this) and the size of any effect cannot be guessed at. If the experiments tell us anything about the GSGB, it is that it inflates estimates. The Commission believes that we should ignore these results and focus instead on subjective interpretation – once again encouraging policymakers to ignore the science and focus on the opinion. It is therefore relevant that one of the researchers involved in the study described the experiments as being, “not very informative about general population surveys”.
The Gambling Commission has expressed confidence that GSGB over-reporting cannot be explained by respondents misunderstanding the questions. This means that the over-reporting can only be the consequence of skew in the survey or the existence of massive and hitherto undetected unlicensed markets for landbased casinos, exchange betting and the football pools. Under either scenario, the survey results cannot be considered reliable for understanding consumer activity in the licensed markets. This however is precisely how the results are used – by campaigners, by politicians and by the Commission itself.
There should be only one course of action for the Gambling Commission. As well as the ethical responsibility to do the right thing (one of the regulator’s stated values), the Commission knows that the GSGB is being used by those who wish to subvert the Gambling Act 2005 (ie, the legislation that it is charged to uphold). As a producer of statistics, it is bound by the UK Statistics Authority’s Code of Practice, which requires it to “prominently explain the quality of the statistics, including strengths and limitations, and communicate the uncertainty in estimates”.
In recent years, the Gambling Commission has funnelled millions of pounds into anti-gambling activism, suppressed evidence on the basis of convenience and promoted illiberal speech codes. It has skirted with conduct beyond the pale. On Thursday, by publishing statistics it knows are not to be trusted, it crosses a line.