The UK gambling industry has faced some heavy criticism in the media in the past few months. Most of it began when 888 was rightly castigated for its failure in its social responsibility with no-one in the industry, or at 888, dissenting at the Gambling Commission’s judgment. But a good deal of criticism has been sown longer-term by anti-FOBTs campaigners and a media willing to regurgitate negative claims.
Regardless of the strength of the stories printed, the clear narrative in the media is that there has been an explosion of problem gambling.
This has been reflected at a political level, with Labour deputy Tom Watson announcing his own review into gambling citing a ‘hidden epidemic of problem gambling’ and it is these kind of claims that the DCMS is having to look at in its bloated Triennial Review of Stakes and Prizes (and advertising) which is due to be published any day now.
So SBC took a look at what official figures are out there on the problem gambling rate. The method used to measure problem gambling is constantly evolving, but the one constant in the last 20 years has been using the DSM-IV scale. While earlier studies (and some international measures) used SOGS and there is a general trend towards PGSI, DSM-IV at least allows us to compare like for like over a long period.
Some of the original Prevalence studies are difficult to find online, so we have located the official ones and have made them available to download at the foot of the page to allow stakeholders to discover and investigate themselves without viewing things through the lens of the media.
According to the 1999 Gambling Prevalence Survey, the prevalence of problem gambling using DSM-IV was 0.6% – the rate being higher among men (0.9%) than women (0.3%).
Statistics being what they are the confidence interval around this estimate is 0.4% to 0.8% – this means we are 95% sure that the figure lies between these two values. This may not sound too accurate, but it is a byproduct of there being quite a low number of problem gamblers identified from the study’s sample size.
Despite the fact that a major review of gambling took place in 2001 which eventually became the 2005 Gambling Act, the next national Gambling Prevalence Survey was not until 2007.
The prevalence of problem gambling in the population, based on the DSM IV was 1.0% for men and 0.2% for women (0.6% overall). The confidence interval around this estimate is 0.5% to 0.8% (meaning that we can be 95% confident that the true value lies between these two figures).
This suggests very little movement from the figures from 1999, even though FOBTs started to emerge in 2001 and the online betting industry was beginning to make its presence felt.
The 2007 Survey also included a new measure for identifying problem gambling – the Problem Gambling Severity Index (PGSI), which is not directly compatible with DSM-IV but a different indicator. The prevalence of problem gambling in the population, based on the PGSI was 1.0% for men and 0.1% for women (0.5% overall). The confidence interval around this estimate is 0.4% to 0.8% (meaning that we can be 95% confident that the true value lies between these two figures).
Things change for the 2010 (and last) Gambling Prevalence Survey with figures seemingly on the rise. DSM-IV problem gambling prevalence rates for the whole population were 1.5% for men, 0.3% for women and 0.9% overall. The confidence interval around the total estimate is 0.7% to 1.2%, meaning we can be 95% confident that the true estimate falls between these two values.
This range overlaps with the range from 2007 (0.5% to 0.8%) but is obviously on the upper end of the scale. It was at the time described as being on the ‘fringes of statistical significance’.
It was a similar story for problem gambling prevalence estimates for the whole population according to the PGSI were 1.3% for men, 0.2% for women and 0.7% overall. The confidence interval for the total estimate was 0.5% to 1.0%, meaning we can be 95% certain that the true estimate lies between these values. In 2007 the range was 0.4% to 0.8%.
The report itself suggested that we need to be cautious with comparisons. It said: “Differences between surveys may appear statistically significant but could be the result of some other underlying difference rather than demonstrate actual change in behaviour. For example, changes in the age profile of the population, in the profile of the responding population for each survey and sampling error between each survey year all have the potential to affect the observed estimates.”
This is why the DCMS’ slashing of the budget for the Prevalence Study also came at a particularly bad time given that many figures for the 2010 study were on the fringes of being significantly different from those in 2007, including the all-important problem gambling rate, and a further study would have thrown fresh light on whether the changes were a trend or a blip.
This left the regulator settling for a ‘decoupled approach’ through national health surveys, which separated gambling activity levels from problem gambling data. While not directly comparable to the Prevalence Study, it’s the best data available.
According to the DSM-IV, problem gambling prevalence among adults living in private households in England and Scotland was 0.5%. Men were more likely than women to be classified as a problem gambler according to the DSM-IV (0.8% and 0.1% respectively). The confidence interval around the total estimate is 0.3% to 0.7%.
According to the PGSI, problem gambling prevalence among adults in England and Scotland was 0.4%, with men again being more likely than women to be classified as a problem gambler (0.7% and 0.1% respectively). The confidence interval around the estimate for all adults was 0.2% to 0.6%.
Now these figures are much more comparable to the figures in 1999 and 2007, which initially suggests that the 2010 figures may have been an outlier and unnaturally high. However this data was just for England and Scotland and didn’t take Wales into account the report also warns that the change in methodology could end up in under reporting issues.
It concluded: “Overall, based on this evidence, it appears that problem gambling rates in England and Scotland are broadly stable.”
NatCen’s report into the three national Health Studies in 2015 found that problem gambling prevalence among adults was 0.7% according to the DSM-IV. The confidence interval around this estimate is 0.5% to 1.0%, meaning that taking into account sampling error we can be 95% confident that the true estimate falls between these two values.
Again men were more likely than women to be classified as problem gamblers according to the DSM-IV (1.3% and 0.2% respectively).
According to the PGSI, problem gambling prevalence among adults was 0.6%. The confidence interval around the estimate for all adults is 0.4% to 0.9%. As with the DSM-IV, men were more likely than women to be classified as problem gamblers (1.1% and 0.1% respectively).
In the Gambling Commission’s 2016 report, its section covering problem gambling, unfortunately, made no mention of DSM-IV and relied completely on PGSI.
The report said that overall 0.7% of respondents identified as problem gamblers, with men more likely than women to be categorised as such (1.2% compared to 0.1%).
The report also lays out the problem gambling rate for 2013 (0.5%), 2014 (0.5%), 2015 (0.5%) and 2016 (0.7%), albeit with no real detail in the data other than it used the PGSI screen. In fact the figures for 2013-2015 are lower than those reported in the Health Surveys for that period.
It added: “Again, despite percentage changes in the observed rates of problem gambling, trend data should be treated with caution as none of the changes in the problem gambling are significant at the 95% level.”
It’s hard to disagree with the Gambling Commission which said in August that ‘overall problem gambling rates in Britain have remained statistically stable’. When considering the huge increase in gambling participation during the time period, it is almost counter-intuitive that the recorded levels of problem gambling have remained more or less the same.
However the figures produced in 2010, while looking like an outlier at the moment, should be a warning signal to the industry and given the advances in problem gambling screening, the industry should be striving to actually lower this rate over a long period of time.
What it also highlights is what a poor decision it was from the Conservative-Lib Dem Coalition Government to pull the funding for the regular Gambling Prevalence Surveys. The idea to save £500,000 across three years by pulling the funding for these impartial studies means that the decision making around the industry is compromised by having to compare studies with different methodologies. The best thing the government could do now is to reintroduce the Prevalence Studies to make sure that the industry trends are properly and independently measured.