26 Sep Why Problem Gambling Just Doesn’t Add Up
Why Problem Gambling Just Doesn’t Add Up
Over the course of the last two decades, reported rates of problem gambling in Great Britain have doggedly refused to go down (or up) by very much. This finding should offer no comfort to the industry; but at the same time, it is not clear that it should be used as the basis for blame.
In this article we consider the pitfalls of misaligning activity with evaluation; and ask whether manipulation of statistics is acceptable – even in the pursuit of noble objectives.
What is the point of ‘Safer Gambling’? No seriously, what precisely is the point? Amidst all the hand-wringing, virtue-signalling and self-flagellation, what exactly are we hoping to achieve and how should we measure progress?
An easy answer to this question (and one favoured by the Gambling Commission) is that our aim should be to reduce the rate of problem gambling in Great Britain. Stable reported rates of problem gambling in this country (and indeed across a large number of jurisdictions) may be somewhat remarkable given the way that markets have changed over the last two decades (and in particular the migration of expenditure towards faster, more repetitive and less supervised forms of gambling) – but this is not good enough. The rate must come down – and woe betide licensees if it does not.
So how should we achieve this goal? First, we need to define precisely what we mean by “reducing the rate of problem gambling”. Considered in a scientific sense, this means that we should be trying to reduce the scores obtained from adults in Britain when answering the PGSI and DSM-IV derived problem gambling screens (contained within NHS Health Surveys). Following this logical thread, we ought therefore to be engaged in activity that is going to result in respondents endorsing fewer items on these screens (assuming that levels of self-report accuracy remain constant).
If we can agree that this is the aim, then what might be the most efficient means of achieving this reduction? Surely, efforts to address the most commonly endorsed diagnostic criteria will give us the best bang for our buck. According to the 2010 British Gambling Prevalence Survey (the last survey to provide sufficient detail in this area), loss-chasing is the most commonly observed criterion for problem gambling – with around 60% of ‘at-risk’ or ‘problem’ gamblers admitting this behaviour “sometimes”, “most of the time” or “almost always”.
On this basis, it would seem sensible to focus our energies on reducing loss-chasing – but is this wise and is this happening? One might say that the Government’s decision to reduce the maximum stake on FOBTs is consistent with this logic by truncating the scale of possible bets that a player can make on an EGM. If true, then the long delay in first making the decision and then implementing it makes the Government (in theory at least) culpable for sustaining rates of problem gambling at higher levels than they need be. However, this would appear to be an exception.
The National Responsible Gambling Strategy for example does not prioritise loss-chasing and amongst all the studies funded by GambleAware, only one of them (the 2016 Forrest and McHale analysis of slot machine play in some of Britain’s casinos) has attempted to identify this behaviour. It would appear that the broad thrust of safer gambling activity is not directly aimed at reducing reported rates of problem gambling in the most efficient fashion.
Perhaps this view is too literal – a bit narrow. After all, what we are really trying to do is achieve a reduction in harmful behaviours and problem gambling prevalence is simply an imperfect guide to this end. However, we may have an issue if our system of measurement is not aligned to the problems that we are trying to solve. How are we to measure progress accurately (and fairly) if our efforts are out of kilter with the items being evaluated?
Let us consider another example of this mismatch. In recent years, there has been increased activity in the area of ‘responsible gambling’ advertising. The Senet Group has led the way on this and now Public Health England and GambleAware are set to follow with a three-year, £21m ad campaign. Quite aside from the fact that we are spending a substantial proportion of our annual RET funds on activity that has questionable effectiveness, we ought to reflect on what these campaigns seek to achieve.
The Senet Group, in assessing the benefits of its ‘Bad Betty’ and ‘When the Fun Stops’ campaigns, cited the fact that one-in-ten British adults claimed to have been prompted by these slogans to warn others about their gambling (even if in jest). As someone who loves being told what I can and cannot do, I find all of this highly reassuring. There’s just one problem – what if this exercise in organised finger-wagging actually increases the reported rates of ‘problem’ and ‘at-risk’ gambling?
To understand how this might be possible we must go back to the PGSI. Two of the questions on the nine-item survey concern whether people have had their gambling criticised and whether they have felt guilty about their gambling (over the prior 12 months). It seems entirely plausible that an advertising campaign which seeks to encourage (non-clinically trained) members of the public to chide others about their gambling may therefore increase the number of people who endorse the ‘criticism’ and ‘guilt’ items on the survey. This is in addition to the more general point that efforts to raise awareness of any disorder may serve to increase reporting of the disorder – even if its incidence remains constant.
If we take Senet’s claim that its awareness campaign has resulted in 5.7 million people per year warning others about their gambling (and accept that certain individuals may have received warnings from multiple parties) then the organisation could in fact be responsible for a substantial inflation in the reporting of ‘at-risk’ and ‘problem’ gambling. I doubt that this is really the case but how should we reconcile the Senet claims with Health Survey data?
This does not mean that Senet’s approach is ‘wrong’ (although we ought to be very careful in how we use shame and prompt guilt) but it does suggest another imbalance between what we do and how we decide whether it works.
Going a bit further (and possibly more controversially), it may not be the case that every reduction in problem gambling status is necessarily and unequivocally a good thing. Returning to where we started, should we aim to eliminate entirely the chasing of losses? Regular loss-chasing (particularly where it is found alongside other diagnostic criteria) is a cause for concern; but what about the player who scores one point on the PGSI for sometimes trying to win back what he has lost? We must consider how much harm is likely to occur as a result of occasional loss-chasing and whether any benefits (in terms of enjoyment) might accrue from such behaviour.
The respected researcher, Paul Delfabbro (University of Adelaide) addressed this last year in his paper on the prevention paradox, when he wrote: “It is questionable whether some of the items identiﬁed [in diagnostic screens] are really indicators of harm. For example, chasing losses, gambling more to obtain the same excitement, or betting more than one could afford are really behaviours that might lead to harm if repeated too often.”
I am ashamed to admit that there have been times in my otherwise saintly existence when I have drunk just a little too much (and yes, even weeks when I have exceeded the Surgeon General’s 14-unit limit). I have no doubt that I am somewhat less healthy for these occasional episodes; but I recall too that some of the best times of my life have occurred when I have drunk just a little too much. The point here is that occasional indulgence in risky behaviour need not necessarily be a bad thing.
This leads us on to some of the statistical distortions that have become popular of late in policy discourse (and which too often are cited as self-evident truths).
The first of these is the assertion that there are two million people in Great Britain at risk of “developing a serious gambling problem”. This claim is based upon the finding from the 2015 Health Survey that around two million people are classified (under PGSI) as ‘low-risk’ or ‘moderate risk’ gamblers (the figure falls slightly if we look at the 2016 survey).
If we extrapolate ratios from the last British Gambling Prevalence Survey (the Health Survey reports do not provide the data required for the calculation), around three-quarters of those 2 million ‘at-risk’ gamblers are likely to be categorised as ‘low-risk’. These are people who will have “few or no identified negative consequences’ from gambling, according to the designers of the PGSI. Again, using 2010 prevalence survey ratios, more than two-thirds of these 1.5 million ‘low-risk’ gamblers are likely to have scored just one point on the PGSI (which runs from 0 to 27) – a group that may include a large number of those gamblers who sometimes chase losses and who endorse no other items.
Looked at in this way, is it really fair and proportionate to claim that there are 2 million of us “at risk of developing a serious gambling problem”? How worried should we be about these individuals and how much of our limited resources should we expend on them (compared with ‘moderate-risk’ or ‘problem’ gamblers)?
Another commonly cited claim is that “only about 2% of problem gamblers who need treatment get treatment”. Again, it is a notion that has some basis in fact but too distant a relationship with detail. The 2% figure is based on the number of people going through GambleAware funded treatment services (essentially GamCare and its partners, the NHS clinic and the Gordon Moody Association) as a percentage of the 450,000 or so people who may be problem gamblers. The statistics are fine; it is the interpretation that is faulty.
GambleAware does not have a monopoly on problem gambling treatment. We don’t know how many people attend Gamblers Anonymous groups every year but it is likely to be sizeable. Other people with gambling problems (particularly those from ethnic minorities) may seek help in less conventional places such as faith groups or community centres. Given how little we know about treatment effectiveness, we ought not to exclude these sources of help from our calculations.
While relatively few problem gamblers undergo formal treatment courses (and most self-correct), a somewhat larger number receive assistance from helpline and netline services (principally from GamCare). Given that recent research from New Zealand indicates that a single call to a helpline is a predictor of reduced problem gambling severity, we may need to include these callers in our thinking (indeed, one might argue that GambleAware undersells itself on precisely this point). Put simply, the numerator in the problem gambling treatment calculation is likely to be significantly understated.
Then there is the idea that all problem gamblers require formal treatment – and the very specific range of treatment services funded by GambleAware. There does not appear to be much evidence to support this contention. Given the heterogeneity of problem gamblers and the variable nature of gambling-related harms (including comorbid harms) and problem gambling diagnoses, it seems rather implausible that everyone requires precisely the same set of therapies. Thus, the denominator may be inflated.
We do need to spend substantially more on problem gambling treatment in this country – but do we need to make things up in order to get there? The distortion of facts (whether by the pro-gambling lobby or the gambling concern movement) has the potential to skew policy and ought to make us uncomfortable.
In the past, Britain’s gambling companies were sometimes guilty of being a touch blasé about problem gambling; but those days have gone. Given that there is now a strong (if imprecise) appreciation of the links between excessive gambling and harm, we should not need to make things appear worse than they are.
We should seek to expunge lazy analysis and attempt to gain greater clarity on the nature of the problems we are trying to solve – and how we are to measure progress. This requires an acceptance of the possibility that reported rates of problem gambling may remain stable even as levels of harm go up or down.
Stable reported rates of problem gambling do not justify industry complacence; but nor should they be used as a reason to browbeat licensees. If anything, an obsession with high-level data from prevalence surveys may serve to obfuscate our view of whether standards of consumer protection are being raised and what is happening as a result.