One interesting, and perhaps unanticipated, finding was that the nature of the relationships between private betting and betting on horses and gambling problems changed when we considered the influence of involvement: This data reduction replicated that employed by Griffiths et al. This document contains the following information: For others who do not have high rates of involvement, playing these types of games might reflect social setting characteristics e. After controlling for involvement, individuals who engaged in private betting or betting on horses were significantly less likely to have gambling-related problems than people who did not. We used weighted data for all analyses. Taken together, these two sets of findings suggest that researchers and others use caution when interpreting results showing that people who play specific types of games have a higher rate of gambling-related problems than others.
Survey data on levels of gambling participation
We operationally defined gambling involvement as the number of types of gambling for which an individual reported being involved during the past 12 months.
We used weighted data for all analyses. Specifically, the BGPS created a weighting variable correcting for dwelling and household selection probabilities, age, gender and individual non-response within participating households. We conducted three primary sets of analyses. First, we calculated for the full sample, and by gender, participation rates for each game type.
We used chi-square analyses to determine whether those rates varied by gender. Third, we conducted a series of logistic regressions, which used participation in each gambling type to predict disordered gambling status among past 12 month gamblers. We conducted these logistic regressions first without controlling for involvement and then added involvement as a control. Table 1 shows the participation in all types of gambling by gender and for the full sample.
The top five gambling types with respect to participation were: Also popular were private betting, bingo and other sports betting other than online betting or betting on horses or dogs. We observed a number of gender differences for gambling participation. Table 2 shows gender stratified prevalence rates for gambling-related problems reported by individuals who participated in various types of gambling.
These types of games also were associated with the highest mean numbers of DSM gambling symptoms and mean number of types of gambling during the past year i. The top five prevalence rates of any DSM gambling symptoms by types of game were: The top five types of games for the mean number of DSM gambling symptoms were: The top five types of games for the mean number of types of gambling during the past year i. Among the full sample, participants engaged in an average of 1. Table 3 shows a series of logistic regressions illustrating how well each type of gambling contributes to the prediction of gambling-related problems i.
Bivariate analyses showed that all types of gambling, except for the National Lottery, contributed significantly to the prediction of gambling-related problems and all increased risk for gambling-related problems. The top five odds ratios were for: Logistic regression analyses predicting disordered gambling status from type of game, with and without controlling for involvement. Subsequent regressions that added involvement i. The addition of involvement greatly reduced the contribution of games to the prediction of gambling-related problems in each model.
For almost all games, the addition of the involvement variable rendered the significant positive association between gambling type and gambling-related problems non-significant. The exception was virtual gaming machines, which maintained a significant positive relationship to disordered gambling status after adjusting for involvement.
Two games, private betting and betting on horses, had a reversal of association. After controlling for involvement, individuals who engaged in private betting or betting on horses were significantly less likely to have gambling-related problems than people who did not. In this study, we provide a comprehensive analysis of participation with different games among British residents aged 16 and older.
We placed a special emphasis on the nature and strength of the associations between types of games and gambling-related problems. The types of games that had the strongest associations with gambling-related problems did not include all of the games that the conventional wisdom might expect.
Virtual gaming machines had the strongest association with gambling-related problems, but few people i. These findings suggest that popular perceptions of risk associated with specific types of gambling for the development of gambling-related problems might misrepresent actual risk. Regardless of the type of game, past month participation was associated with disordered gambling; however, for the most part, such associations disappear, or at least become weakened, when statistical analyses control for the range of gambling involvement.
Our findings with a primarily adult British sample are consistent with Welte et al. Taken together, these two sets of findings suggest that researchers and others use caution when interpreting results showing that people who play specific types of games have a higher rate of gambling-related problems than others. In fact, these studies reveal that some games might be indicators of unhealthy involvement, rather than critical factors for gambling-related problems themselves.
One interesting, and perhaps unanticipated, finding was that the nature of the relationships between private betting and betting on horses and gambling problems changed when we considered the influence of involvement: This finding suggests that the apparent risk between gambling activities and developing gambling-related problems resides, perhaps primarily or even entirely, among individuals who have high rates of involvement. For others who do not have high rates of involvement, playing these types of games might reflect social setting characteristics e.
These findings hold some disparate possibilities for theories of gambling exposure. On one hand, these findings might imply that more opportunities to gamble create more opportunities for involvement and, therefore, might yield more gambling-related problems.
On the other hand, these findings might suggest that more opportunities to gamble will have little to no impact on the prevalence and incidence of gambling-related problems because individuals are more or less prone to involvement. Increases in gambling opportunities will not influence individuals who are less prone to involvement, but only those likely to become, or who already are, involved.
There is some evidence to support the latter view because the rate of gambling disorders has changed little during the past 35 years despite the extraordinary growth of gambling opportunities and access around the world.
Notable strengths of this study include the analysis of multiple game types simultaneously and the incorporation of a measure of involvement into analyses that examine the association between type of game and gambling-related problems. Controlling for involvement allows a more sophisticated understanding of the risk unique to some types of games and provides a level of analytic sophistication more advanced than the majority of available research.
In fact, controlling for involvement drastically reduces the ability of games to discriminate statistically individuals who have gambling-related problems from those who do not.
Another strength of this study is that it advances this more sophisticated methodology and line of inquiry from a US adolescent sample to a British primarily adult sample.
This broader study sample helps to avoid problems related to legal access to different types of gambling observed among the US sample and concerns about different gaming interests by age cohorts. Nevertheless, this study is not without limitations. First, the analyses rely on self-report data and not actual gambling activity. Self-report is vulnerable to weaknesses, including faulty memory, factual errors and self-presentation biases.
Second, we only included one measure of involvement i. Other measures of involvement e. Third, this study relied on retrospective reports of behavior and therefore cannot establish any causal patterns. Fourth, many of the game-type variables represent multiple types of games by definition e. This approach is conservative and only presents as a limitation because of the inability to distinguish the effects of subtypes of games.
Fifth, a small number of people played some types of games; consequently, increases in the sample size might alter the findings for games played by small numbers of people. Future research should include the longitudinal assessment of real-time gambling data and multiple measures of involvement to yield a better determination of whether involvement is a moderator, mediator or both, of disordered gambling. Other important directions include examinations of game clustering, to determine whether subtypes of involvement are possible, the determination of whether there might be a critical level of involvement e.
The range of gambling involvement frequently is a better predictor of disordered gambling status than type of gambling. This research does not suggest that differentiating between types of games is completely unimportant; clearly, there are differences in the popularity of games. These and similar results 7 suggest the need to reconsider the conventional assumptions related to the influence of game types and direct more attention toward global behavioral characteristics, such as the range of involvement.
The Division on Addictions also receives financial support from: During the past 5 years, the authors of this paper have received honoraria and fees for a variety of activities, including gambling-related symposia, speaking events, education events, research, and consulting. HJS has served as an expert witness for gambling and other addiction-related legal cases.
The authors have no other personal or employment-related competing interests. The sponsor had no role in this research. We conducted this research independently from the sponsor.
Please tell us what format you need. It will help us if you say what assistive technology you use. The aim of the survey being to provide data on participation in all forms of gambling in Great Britain, the prevalence of problem gambling, attitudes to gambling and to explore a range of associations with gambling behaviour.
The study is the first in this series to be conducted after the full implementation of the Gambling Act Therefore a further objective was to, where possible, provide comparisons pre and post implementation of the Act.
Findings include that 73 per cent of the adult population participated in some form of gambling activity within the past year, representing a return to rates observed in The most popular activities were the National Lottery Draw 59 per cent , a slight increase from