Written evidence from the National Centre for Social Research (GA 38)

Key points

· Assessing the impact of the Gambling Act upon problem gambling rates should be based on sound empirical evidence.

· The main source of data about gambling behaviour in Britain is provided by the British Gambling Prevalence Survey (BPGS) series.

· In 2010, this study demonstrated that there were some changes in gambling behaviour: more people were gambling regularly, people were taking part in a greater range of activities, attitudes to gambling were changing and becoming more positive than previously and there was some evidence that the prevalence of problem gambling was increasing, although this was at the margins of statistical significance.

· While the observed change in problem gambling prevalence should be approached with caution, other changes are clearer, such as the increase in the number of regular gamblers. This is important to monitor going forward as increased levels of gambling involvement are associated with problem gambling.

· The BGPS series is a nationally-representative survey that has been conducted in 1999, 2007 and 2010. However, it was not designed to measure the impact of the Gambling Act and can not provide evidence about why changes have occurred. It was not an impact study; it simply compares behaviour pre and post implementation of the Act and we do not have sufficient evidence to adequately measure counterfactual change (i.e., the changes that would have been observed if the Act was not introduced).

· There is insufficient evidence to robustly measure the impact of the Act upon problem gambling rates. Further research is needed to assess whether the changes observed in 2010 are a short-term fluctuation or a longer-term trend.


1. We welcome the opportunity to contribute evidence to the Culture, Media and Sports Committee Inquiry about Gambling in Britain , with specific focus on the implementation and operation of the Gambling Act 2005. We believe that policy and regulatory decisions should be informed by robust empirical evidence . O ngoing research and evaluation based on sound scientific principles is critical to providing such evidence .

2. The National Centre for Social Research is Britain ’s leading independent social research institute in the UK . We are experts in gambling research having designed, conducted and analysed all three studies in the British Gambling Prevalence Survey series and conducted a number of important related studies focusing on exploring gambling behaviour in Britain in depth. We therefore have a great deal of expertise and knowledge about gambling in Britain today and are keen to share our insight based on this.

3. This response focuses on the inquiry question to which our experience is most pertinent; the impact of the Act on levels of problem gambling. Our response specifically raises the challenges of answering this question with the current evidence base.

Our response:

4. Any assessment of the impact of the Gambling Act on levels of problem gambling should draw on robust empirical evidence. The British Gambling Prevalence Survey (BPGS) Series provides national data about gambling behaviour. It was conducted 1999, 2007 (prior to the full implementation of the Act) and again 2010. The aims and objectives of the 2010 study were to:

· Measure the prevalence of participation in all forms of commercial and private gambling

· Estimate the prevalence of problem gambling

· Investigate socio-demographic and other factors associated with gambling and with problem gambling

· Explore attitudes towards gambling

· Where appropriate, provide comparisons pre and post implementation of the Gambling Act 2005.

5. The study is a large scale survey of the adult population in Great Britain living in private households. To date, more than 23,000 adults aged 16 and over have taken part in it. Despite the rich data the study provides about gambling behaviour in Britain, it was not designed to measure the impact of specific terms of the Act upon problem gambling rates and, as a repeat, cross-sectional survey, it can not say why any changes have occurred.

6. The picture is further complicated because the study has only been conducted on three occasions, meaning that it is difficult to detect underlying trends and patterns in gambling behaviour and that any changes observed between survey years have to be carefully interpreted.

7. The BGPS 2010 included two instruments designed to measure problem gambling prevalence rates: the Problem Gambling Severity Index (PGSI) and an instrument based on the American Psychiatric Association Diagnostic and Statistical Manual of Mental Disorders (version four: DSM-IV). The instruments were developed and designed for different purposes. The PGSI was developed to measure levels of gambling related harm at a population level and included more focus on the harms associated with gambling (such as ill health). The DSM-IV was initially designed as a list of criteria to be used by clinicians in diagnosis. Each instrument therefore uses different questions and thresholds to categorise problem gamblers. As such, two different problem gambling prevalence rates are available in each survey year: one based on the PGSI and one based on the DSM-IV. [1]

8. The 2010 results showed the there was no statistically significant change in the prevalence of problem gambling when measured using the PGSI criteria. However, when problem gambling was defined using the DSM-IV, rates in 2010 were significantly higher than 2007 and 2010, rising from 0.6% in the earlier studies to 0.9% in 2010. As noted in point 6, the interpretation of this result needs careful consideration. This finding was at the margins of statistical significance. The p-value was 0.049, the threshold for significance being 0.05. A number of further statistical tests were performed to assess whether changes in the underlying profile of respondents affected this result. The difference remained significant.

9. There is some debate about how to interpret this finding. While we recommend caution for the reasons outlined above, when viewed in the context of other findings within the survey, the increase seems plausible. For example, BPGS 2010 showed that a greater number of people than ever before were gambling on activities other than the National Lottery. There was a significant increase in the proportion of people who reported gambling regularly, that is once a month or more often, and those who were gambling regularly were taking part in a greater range of activities. The 2010 data also showed that attitudes towards gambling were more positive in 2010 than in 2007 and that people, by and large, reported that they gambled for fun and for the chance of winning money.

10. This evidence shows indications of a potential shift in gambling behaviour with more people taking part more often and diversifying the types of gambling they engage in. What this evidence does not show and cannot say is:

a) whether these observed differences are a direct result of changes introduced by the Gambling Act 2005; and

b) whether the results observed in 2010 are a short term fluctuation or part of a broader trend.

11. To illustrate point a), we can use the example of betting on sports and other events. Changes in the restrictions of advertising were arguably, for the general public, one of the most visible changes introduced by the Act, with the sports betting sector in particular embracing new forms of advertising permitted since 2007. Between 2007 and 2010, the prevalence of betting on sports and other events increased significantly from 6% to 9%.

12. However, increases in the prevalence of sports and other betting were evident prior to the implementation of the new advertising laws in 2007, rising from 3% in 1999. Therefore, it is not possible to say whether the increase observed in 2010 is the direct result of the Act as there is evidence of a pre-existing trend. Likewise, we do not know what the result would have been if advertising had not been introduced and we do not have a sufficient evidence to be able to estimate what proportion of the increase is attributable to advertising (if any) and what proport ion is a reflection of the pre-existing trend. Much of the evidence presented in the 2010 suffers from these problems and any evidence drawn from the BGPS series should consider this.

13 . Turning to point b ) , what the BGPS series does show is that in 2010 gambling behaviour in Britain was different compared with gambling behaviour previously. We see more women and older people becoming involved with gambling and an increase in the proportion of people who gamble regularly . The latter is particularly important as higher l evels of gambling involvement are associated with problem gambling. It will be important to continue to monitor this going forward as this has the potential to effect problem gambling rates observed in Britain .

14. This is not to assume that increased involvement automatically means increased problem gambling prevalence, as there are a number of factors which may mitigate or ext enu ate the experience of gambling-related harm among a population . It is possible that estimates observed in 2010 are a short-term fluctuation and that over time the population will adapt to changes and rates will return to those observed previously (adaptation theory). It is also possible that changes in 2010 represent a broader shift in gambling behaviour and attitudes which may signal the turning point of a new trend. With just three surveys conducted to date and the BGPS series unlikely to continue, we do not have the evidence to confirm which of these forces are at work. We believe it is important to continue to monitor the direction of travel and to assess the ongoing relationship between levels of gambling involvement, problem gambling rates and regulatory actions and princip les which may influence both.

How we can help :

1 5 . NatCen conducts a wealth of research into gambling issues. Our response is based on evidence from the BGPS series but our expertise extends beyond this. We have completed three high-quality in-depth studies of gambling behaviour (including a focus on the gambling careers of problem gamblers), produced methodological advice relating to the implementation of a longitudinal study of gambling, conducted evidence reviews on many topics varying from the effectiveness of gambling regulation, the impact of advertising, motivations for gambling and triggers for changing gambling behaviour and we work closely with , and are highly regarded by , many leading international scholars in the field. NatCen is therefore delighted to have the opportunity to contribute to this assessment and would be very happy to expand upon this written submission if that would be helpful .

June 2011

[1] This is true for BGPS 2010 and 2007. In BGPS 1999 pre-dated development of the PGSI and that survey used an alternative instrument called the South Oaks Gambling Screen along with the DSM-IV.

Prepared 1st August 2011