A Critical Review of the Harm-Minimisation Tools Available for Electronic Gambling

High-intensity commercial gambling has evolved relatively recently in comparison to other legalised, hazardous, and consumptive behaviours, such as tobacco and alcohol use (Adams et al. 2008). Gambling products and their advertising are now almost unavoidable and the promotion of gambling has arguably become as a social norm (Parke et al. 2014a, b). The presence of gambling has become ubiquitous, inextricably linked with national and international sporting events on television, omnipresent in towns and cities in the form of licensed betting offices, casinos, bingo halls and amusement arcades, and remote gambling, including gambling via the internet, mobile phone and interactive television (Griffiths et al. 2014).

Of particular importance is the evolution of gambling products into sophisticated, electronic platforms that possess structural features that interact with the gambler to produce ego-dystonic and maladaptive effects (see e.g., Breen and Zimmerman 2002), which may broadly be described as ‘gambling-related harm’. The strategic approach to tackling this harm is of great importance, as is the focus on efforts to reduce such harm. Adams et al. (2008) argue that in a society demonstrating relatively stable consumption, it is justifiable that attention should be directed towards the treatment of those suffering with a gambling problem. However, such concentration of effort as Adams et al. (2008) go on to argue, is less urgent in a rapidly changing environment that is demonstrating escalation of risk. Instead, effort would be best directed towards attending to the situation itself:

…when a submerged rock pierces a hole in the bottom of a boat, it makes little sense to attend solely to those who have been injured and it makes considerably more sense to focus a good deal of energy upon stemming the flow of water through the hole (Adams et al. 2008; pp. 869).

This analogy may be particularly relevant given the evolving view that the Theory of Total Consumption (Lederman 1956) is valid for gambling behaviour (Lund 2008). In the field of alcohol studies, it has long been accepted that there is a positive association between mean alcohol consumption among a population and the relative proportion of heavy or problem drinkers in that society (Babor et al. 2003). Such a relationship, originally proposed by Lederman (1956), is known as the total consumption model, or the single distribution theory. Emergent evidence suggests the total consumption model is valid in a wide variety of phenomena (Lund 2008). This has included gambling behaviour, with several studies finding evidence of increased gambling participation as gambling accessibility increases (e.g., Room et al. 1999; Turner et al. 1999), with such evidence being taken as support for the application of the theory of total consumption to gambling.

One assumption of the theory is that when individuals along the entire consumption continuum increase their gambling, this will also include those gambling at a level below or just below the limit for heavy or excessive gambling (Lund 2008). Consequently, increased gambling participation in this subgroup is enough to shift them towards the heavier gambling group. This is particularly important given the figures that demonstrate that in addition to a 0.5 % prevalence estimate for problem gambling in the UK, an additional 4.2 % of adults can be classed as ‘at-risk’ for developing a gambling problem (Wardle et al. 2014), equating to around 2.5 million people. From a total consumption perspective, increased gambling consumption has the potential to shift those at risk into the problem gambling category, as well as converting those who gamble recreationally, problem-free, to at-risk gamblers. Furthermore, for every problem gambler there are a number of family, friends and individuals in a community who are negatively impacted by problem gambling (Dickson-Swift et al. 2005) although the number of individuals affected is fewer for adolescent problem gamblers (Griffiths 1995). This provides strong argument for problem gambling to be tackled from a public health perspective.

The question remains as to how to tackle the promotion of responsible gambling (RG) and the prevention of problem gambling. This has led to the introduction of many RG and harm-minimisation initiatives. For example, one harm-minimisation approach has been to restrict the availability of gambling by reducing opening hours in licensed gambling premises (Wohl et al. 2010), as well as reduce the quantity of gambling products by restricting the number of electronic gambling machines (EGMs) in licensed betting offices in the UK to four (Association of British Bookmakers 2015). Similarly, voluntary self-exclusion programmes allow individuals who feel they have a problem with gambling to identify themselves to the gambling venue and mutually agree upon a venue exclusion for a predetermined or indefinite period of time. It is important to note that such a decision to voluntarily self-exclude may also be viewed in a positive light and from a preventive approach, as voluntary self-exclusion is available to those who may not yet have developed a gambling problem, but feel they may be at risk or simply feel like they do not want to gamble anymore.

The above examples represent the ‘supply reduction’ type of harm-minimisation. Other approaches include ‘demand reduction’, by adopting policies that make gambling less attractive, such as limiting or banning in-house smoking or the consumption of alcohol (Williams et al. 2004). Other demand reduction approaches may aim to educate customers about the true nature and odds of specific gambling games (e.g., Wohl et al. 2010), in the hope that this may enlighten gamblers that, statistically speaking, they are likely to lose money, or dispel cognitive myths relating to illusions of control or specific ‘winning’ gambling strategies, in the hope that this may reduce the desire to gamble.

The final type of harm-minimisation initiative—and the focus of the present paper—is ‘harm reduction’, which operates more from a ‘restrictivist’ philosophical and moral standpoint in tackling problem gambling. As Collins et al. (2015) identify, a restrictivist view operates somewhere in the middle of the continuum between prohibitionists and libertarianism. Unlike prohibitionists, restrictivists disagree that gambling should be banned outright, and unlike libertarians, they identify that gambling is not like any other leisure or entertainment business (Collins et al. 2015). This view argues that while gambling should be allowed, restrictions should be put in place to ensure that gambling is done so as safely and responsibly as possible.

As gambling products become more technologically sophisticated, the same technological innovation can be used to facilitate the development of harm-minimisation tools to assist gamblers in maintaining self-control and make rational and controlled gambling-related decisions. Harm-minimisation tools aim to make the time spent gambling safer, without reducing the uptake of gambling per se. Such tools have taken on a variety of forms, and while harm-minimisation as a research field within psychology is on the rise in terms of volume and quality of empirical research, the evaluation of such tools remains in its infancy. The aim of the present paper is to conduct a systematic literature review to synthesise and critically evaluate the empirical evidence available that tests the efficacy of current harm-minimisation tools. To our knowledge, while some now dated reviews have been undertaken assessing the evidence for specific harm-minimisation tools, no literature review exists that examines the collective evidence from across the harm-minimisation literature as a whole.

Gamblers often enter into states of dissociation (Jacobs 1986) that leads to a loss of control over time and money spent gambling. RG initiatives that temporarily stop gambling allow dissociative states to be broken and the re-evaluation of one’s gambling behaviour. Indeed, the use of enforced breaks in play as an RG tool is derived from robust theoretical underpinnings. Anderson and Brown (1984) hypothesised that arousal produced within a gambling session narrows a gambler’s attentional focus and facilitates a secondary reward of escaping psychologically distressing stimuli and wider distressing life situations. Jacobs (1986) extended this concept with his general theory of addiction, and proposed that those vulnerable to addiction were either chronically hypo-aroused or hyper-aroused. Engagement in an addictive pattern of behaviour is therefore seen as a way of maintaining homeostatic balance of arousal through generated dissociative experiences.

The use of enforced breaks in play, in the absence of supporting mechanisms such as presentation of self-appraisal messages as a RG tool (e.g., Monaghan and Blaszczynski 2010a, b), may be challenged on theoretical grounds, which indicate that breaks in play may actually have an adverse effect on the gambler. For example, the Behaviour Completion Mechanism Model (McConaghy 1980) posits that driven behaviours (includes pathological gambling), build a neuronal model of behaviour facilitated by conditioning effects. Exposure to a conditioned stimulus or cue results in the activation of the neuronal model, and any interruption to the expression of the behaviour results in an aversive state, or a state of craving, which drives the individual to the completion of the behaviour (Blaszczynski et al. 2015).

Recent research testing the efficacy of imposing short breaks in play as an RG tool challenges the use of breaks in play as a standalone RG approach. Blaszczynski et al. (2015) tested the effects of breaks in play of varying lengths in terms of their impact on cravings to continue gambling and subjective negative arousal, and compared this to a control condition featuring no break in play. Their study comprised 141 university students (78 female) who played a simulated electronic blackjack game, and were randomly assigned to an 8-, 3-min, or no break condition. Results showed that self-reported craving, as measured by the Gambling Craving Scale (Young and Wohl 2009), was significantly higher in the longer break condition, compared to the shorter break and no break condition. Significantly higher craving was also reported in the shorter break condition compared to the no break condition. It was also predicted that forcing breaks in play should reduce levels of dissociation, which has been theoretically proposed as a mechanism promoting extended play. However, no relationship between break condition and feelings of dissociation, as measured by the Dissociative Experience Scale (Jacobs 1988) was found. Therefore, no evidence for the use of breaks in play as a way to combat dissociation was found. However, there was a significant and positive correlation between feelings of dissociation and cravings to continue play, which supports the theoretical position for the role of dissociation in continuation of gambling within a session. Furthermore, the effect of the break condition on craving was mediated by levels of subjective negative arousal.

Given these findings, caution must be taken when implementing breaks in play as a standalone RG strategy. Breaks with accompanying RG messages show a certain level of positive efficacy, however, breaks alone may have unintended effects. Such effects include the promotion of cravings and desire to continue to gamble, rather than breaking dissociative states often experienced by gamblers. Conversely, limited evidence exists to give indication as to the appropriate length of break required to produce positive effects. Consequently, the efficacy of breaks should not be disregarded based on one study alone. For example, the long break condition applied in Blaszczynski et al.’s (2015) research was only 8 min long, and is open to interpretation as to whether or not this constitutes a ‘long’ break. For example, a much longer period of time may be required before maladaptive cravings dissipate and the positive effects of a break may begin to surface. However, recommendations as to what this length of time should be needs to be empirically based, but is likely to differ on an individual-by-individual basis. In addition, given differences in responses between university and real life gamblers (Gainsbury et al. 2014), it remains to be determined if the findings have external validity in terms of how such effects are applicable to real gamblers in real world gambling environments.

Gamblers frequently spend more time and money than initially intended (Monaghan and Blaszczynski 2010a). Furthermore, exceeding financial time and monetary limits within a gambling session has been identified as a key risk behaviour for the development of problem gambling. Failure to stick to pre-set limits arguably reflects a loss of, or impairment in, self-control and self-regulation, which can be undermined by a variety of factors (Parke et al. 2014a, b). Such factors include an inability to regulate emotion (Scanell et al. 2000), and the use of emotion in the decision-making process over the use of problem-focused strategy (Blaszczynski et al. 1990).

Limit-setting is a harm-minimisation strategy that allows gamblers to set time and monetary limits prior to commencement of a gambling session. Limit-setting is based on the principles that decisions concerning time and monetary limits (a) should be made in a state of non-emotional arousal, and (b) once made, must be adhered to for the remainder of the gambling session (Ladouceur et al. 2012). Limit-setting represents an RG tool designed to prevent excessive expenditure in individuals prone to impaired self-control, as well as those who wish to use the feature as a positive, pre-emptive measure. The intention of limit-setting is to promote rational decisions regarding expenditure in advance of play, and, by imposing barriers, to ensure compliance with such decisions when emotionally aroused after losses (Ladouceur et al. 2012), or indeed, wins. Evidence for its use also comes from the natural recovery literature, where it has been shown that 40–82 % of individuals with a gambling disorder recover without professional help (e.g., Abbott et al. 1999). One of the primary techniques adopted by such self-recovery populations was the use of self-imposed time and/or money limits (Blaszczynski and Nower 2010).

Setting limits on gambling time and monetary expenditure may also be viewed as a form of public commitment, and past research indicates that publicly committing to a goal will increase the chances of that goal being reached (Mussell et al. 2000). Outside of gambling, such public commitment strategies have been successfully applied in other areas of health research such as weight loss programmes (e.g., Nyer and Dellande 2010).

Broda et al. (2008) examined the effects of enforced betting limits on gambling behaviour and analysed the behaviour of those gamblers who typically exceeded limits in comparison to those who adhered to monetary limits. Two years of sports gambling behavioural data were analysed from 47,000 subscribed users of the online gambling company bwin. Only a very small proportion (0.3 %) exceeded deposit limits at least once. Gamblers who did were shown to have a higher than average number of daily bets and higher average bet sizes, compared to those who did not exceed limits, indicating that exceeding limits may be indicative of the most intense gambling sub-group. Furthermore, behaviour after exceeding limits showed that average bet sizes steeply increased, though the number of bets reduced. Results indicated that the setting of limits, accompanied by a reminder once limits have been reached, was enough to deter the vast majority of gamblers from exceeding those limits. However, the small majority of those who exceeded limits may represent the most heavily involved gamblers, and arguably, the most in need of help, suggesting the use of limit-setting may be best placed as a preventative RG tool, rather than an intervention for those who may already be exhibiting gambling problems.

Wohl et al. (2010) applied the principles of the Health Belief Model (HBM; Janz et al. 2002) to an animation-based educational video designed to facilitate adherence to pre-set limits. The HBM predicts that healthy and adaptive behaviour will be adopted by individuals when an intervention has a targeted and specified impact on the knowledge, attitudes, and perceptions of target group members. This was applied in a gambling context, more specifically, during slot machine gambling, where the HBM suggests that risk behaviours will be reduced if players come to understand: (1) the true odds of winning, (2) that odds do not improve with persistence. (3) that the consequences of exceeding financial limits can be serious and difficult to reverse, (4) that staying within affordable limits eliminates the chances of developing gambling problems, and (5) that low-risk practices can be used to stay within affordable limits.

A total sample of 242 non-problem gamblers were recruited. Those exposed to an educational animation video applying the principles of the HBM, designed to dispel cognitive distortion, and promote the use of and adherence to time and monetary limits, reported a significant reduction in erroneous cognitions, an effect that was retained at 24-h and 30-day follow-up. Exposure to the video also resulted in participants being more likely to strongly endorse ‘low risk’ gambling practices, including the use of limit-setting, but this effect was not retained at the 30-day follow up. In addition, the video promoted greater behavioural intention to use the ‘low-risk’ practices, but again, this effect was not retained at the 30-day follow-up. Finally, participants exposed to the video reported exceeding their self-set limits less often (8 vs. 25 % for a control group), but again, the effect was not retained at 30-day follow-up.

Clearly, the self-report method applied is subject to inaccuracies, and behavioural intention does not always lead to behavioural execution, particularly in situations where demand characteristics may be working to provide positive outcomes. Alternatively, the effects of the animated video may be more subtle and not noticed by participants, meaning the failure to find a lasting effect at 30-day follow-up may simply be a failure for participants to experientially detect a change, and not necessarily portray a lack of change. What is required is empirical behavioural gambling data to measure pre-and post-intervention effects. It must also be noted that the effects of the video on cognitive distortions were long-lasting, which may equip individuals well in the long run as a protective factor against developing problems with gambling, but longitudinal evidence is required to test such a proposition.

Using a virtual reality gambling environment, Wohl et al. (2013) examined if there was an interaction effect between the use of educational videos dispelling erroneous cognitions and promoting safe-play, including the use of limit-setting, and pop-up messaging reminding participants when they had reached their pre-set limit. Participants were 72 young adults (mean age = 19.69 years, SD = 1.82) with recreational gambling experience, and were predominantly female (70.8 %). Participants played an EGM in a virtual reality environment, gambling with a total of 80 credits ($20). Results showed participants exposed to the educational animation video adhered to pre-set limits more than those in a control video condition (97 vs. 77 %). Those exposed to monetary limit pop-ups also showed greater adherence to pre-set limits (97 vs. 77 %). However, these two main effects were qualified by an interaction effect, with results showing that of the participants who were not given a pop-up reminder, the ones who were exposed to the educational animation video stayed within their pre-set monetary limits more than those in a control condition (94.1 vs. 61.1 %). However, no difference was found in limit-adherence among the participants who all received monetary pop-up reminders, but either saw or did not see the education animation video. The authors concluded that from an RG perspective, there was no additive effect of exposure to both RG tools, and thus, pop-up messages reminding gamblers when they have reached their pre-set limits would be the most effective and efficient RG tool.

It should also be noted that only the education video had a significant effect on reducing erroneous cognitions, and in the absence of pop-up messages, exposure to the video had an effect on gambling behaviour in terms of limit adherence. This shows the potential for education animations as an RG tool, but that it may not be as effective as other measure such as pop-ups in terms of their efficacy in influencing gambling behaviour during play. There is potential for strategies such as educational animations, or education in general, to be applied where pop-ups may not be feasible, for example, in literature in and around gambling venues, or as part of a mathematics curriculum in schools. However, the effect on problem gamblers remains unknown.

Auer and Griffiths (2013a, b) examined the efficacy of limit-setting among high intensity gamblers, across a variety of gambling activities, in a real-world online setting. Data were initially collected from a representative random sample of 100,000 players, of which 5000 had opted to use the voluntary time and/or monetary limits. The top 10 % most intense gamblers, as derived via theoretical loss (house advantage multiplied by amount wagered; see Auer et al. 2012), were taken from each of the sub-gambling type groups (i.e., poker, lottery, and casino games). Results showed that theoretical loss significantly decreased among the top 10 % most gaming-intense lottery players in the 30 days following all kinds of voluntary limit-setting (time and money) compared to the total theoretical loss in the 30 days prior to the implementation of limits. The impact of the cash-in limits on theoretical loss was higher than playing duration limits. Similarly, limit-setting decreased the theoretical loss for the top 10 % most intense casino gamblers. However, time limits had no significant impact on theoretical loss for this subgroup. It was also noted that casino gamblers showed the biggest significant change among the general gambling population, with 77 % of the theoretical loss being spent in the 30 days following limit-setting compared to theoretical loss in the prior 30 days. Among the top 10 % most intense poker players, the amount lost in the poker rake decreased in the 30 days following limit-setting, but this was only the case for those who set weekly spend limits and daily time limits. Overall, time limits had the greatest effect on rake loss for poker players, with those setting daily time limits losing 73 % of the loss in the 30 days prior to the setting of limits. As expected, the setting of daily time and session length limits had a highly significant effect on overall play duration. This is important given the fact that excessive time spent gambling, and not just excessive monetary spent, can have deleterious impacts on the lives of gamblers.

The behavioural tracking paradigm used in this study only gives information about gamblers on one particular gambling site and does not identify the overall profile and behaviour of a particular gambler. This is important as the most problematic gamblers have been shown to play multiple types of gambling platforms concurrently (McCormack et al. 2013), which may mean that reaching monetary or time limits on one site, on one platform, does not necessarily mean cessation of gambling until such limits are reset. It may simply mean that gamblers switch from one site to another once a self-set limit has been exhausted. Pairing (or grouping) of online gambling accounts may be a way around this issue, much like the facility afforded by gaming operators such as Pokerstars and Full Tilt. Of course, this relies on cooperation among competing gambling operators to be a viable option, but it would allow the potential for ‘central’ limits to be set across all of an individual gambling accounts, rather than several isolate limits set at each of the sites where and gambler has an account.

The focus on the most intense gamblers is certainly of relevance given the fact that this sub-group is most likely to benefit from limit-setting. However, the results provided do not tell us how the majority of gamblers, falling more centrally in the distribution curve, interact with limit-setting. As limit-setting is often viewed as an RG tool with preventive utility (see Wohl et al. 2014), such large scale, real-world, behavioural tracking techniques should also be applied to those gamblers below the threshold for problem gambling criteria.

Using the principles of Human Computer Interaction (HCI) and Persuasive System Design (PSD), Wohl et al. (2014) aimed to improve the efficacy of monetary limit-setting as an RG tool, by improving the way that gamblers interact with such features in electronic gambling. HCI principles suggest that for technology to be user-centred, potential users must be involved in the design, testing, and evaluation process. Consequently, they conducted a series of focus groups involving non-problem gamblers discussing their views one existing limit-setting tools, as well as discussing potential design improvements that may increase the tools RG utility.

Using information gained from the focus groups, Wohl et al. then designed new monetary limit-setting with pop-up message reminder, and compared this to older, more simple iterations of such a design. New monetary reminder pop-up message features included a traffic light visual display, informing participants of their spend relative to their limits (i.e., green light ‘safe’, amber ‘close’, red light ‘limit reached’), this was to allow self-monitoring of behaviour, one of the principles of PSD. Once limits had been reached, a 1-min delay was enforced before players could opt to continue to play. Fifty-six EGM gamblers (37 females) were recruited and participated in an EGM simulation in a virtual reality environment. They gambled with 80 credits ($20) and any money left at the end of experiment was kept by the participant. Gambling outcome was controlled for by the experimenter to ensure all participants reached their limits.

Only seven participants (three from the HCI/PSD condition, and four from the standard monetary limit-setting condition) failed to reach their limits and were thus excluded from subsequent analysis. Results showed that those exposed to the HCI/PSD tool were significantly more likely to adhere to their pre-set limits compared to the standard monetary limit tool (62.2 vs. 2 % respectively). Also of importance was the fact that two participants stopped prior to reaching their limits immediately after viewing their player statistics. Self-report data also indicated that participants perceived more engagement with the HCI/PSD tool. However, encouragingly, mean ratings for both the HCI/PSD and old design were above the mid-way point of the scale, showing perceived engagement in both conditions.

Using an EGM simulator in a virtual reality environment, Kim et al. (2014) assessed the impact of prompts encouraging the setting of time-based limits on both the uptake of setting such limits, and the impact this had on session duration. Forty-three, non-problem/low risk Canadian university student gamblers were recruited and given $20 to gamble with in the experiment. Analysis revealed that participants who were explicitly asked to set a time limit did so with a 100 % compliance rate (20/20), compared to just one out of 23 for those participants not prompted to set limits. Those prompted to set a limit prior to engaging in play gambled for significantly less time than those who were not asked to set a limit (5 vs. 9.48 min respectively). Of note, 11 out of 20 of participants in the limit-setting group gambled for less time than their self-set limit.

Several limitations exist, including then potential for demand characteristics in the experimental paradigm to drive the high percentage of participants setting limits. In such a laboratory environment, many structural and situational characteristics of real gambling environments are lacking, all of which may draw attention away from the available RG tools. In addition, participants were only exposed to a single RG tool, and thus, the study cannot report the relative additive (or deleterious) impact that multiple available tools can have in moderating gambling behaviour. However, the results indicated that setting limits on gambling session duration may be effective as an RG tool by reducing the amount of time an individual spends gambling. The authors note that while some gambling activities may benefit from the use of monetary limits, some activities may benefit from time limits. This is perhaps particularly relevant for gambling platforms such as EGMs, where there may be a tendency to dissociate and lose track of time (see Diskin and Hodgins 2001), or poker, where tournaments are typically long and cash games have no defined end as such.

Research indicates that providing gamblers with personalised feedback helps them to better understand their behaviour and change it if necessary (Auer and Griffiths 2013a, b). Digital technology affords the opportunity to track behavioural player data, which in turn, allows the opportunity to profile gamblers, assess behavioural change that may be indicative of a problem developing, and thus, provide gamblers with personalised feedback to facilitate awareness of such behavioural change. Behavioural tracking also produces datasets that allow identification of behavioural markers that may be indicative of harm, which in turn, further allows the development of understanding related to both responsible and problematic gambling practices.

Auer and Griffiths (2013a, b) argue that personalised messages can be applied using the principles of motivational interviewing, where behavioural tracking allows the delivery of personal, transparent, and motivational feedback. They argue that the target population for behavioural tracking tools should be those who are ‘at-risk’, or those who are developing a problem. Behavioural tracking tools may therefore provide motivation for change via the use of personalised feedback, and for this reason, personalised feedback via behavioural tracking is in line with the Stages of Change Model (SCM; Prochaska et al. 1994). The SCM has been applied to a broad range of behaviours, including weight loss and alcoholism, where the idea is that behaviour does not change in one step, rather, change occurs through a series of steps, starting from pre-contemplation, all the way through to maintenance of a behavioural change (for example, see Prochaska et al. 1994).

Auer and Griffiths (2015a, b) assessed the effectives of the behavioural feedback system mentor, in terms of its ability to influence the amount of time played and theoretical loss experienced by gamblers. Behavioural data were obtained from a European online gambling site, with a sample of 1015 gamblers who had used the mentor system. A matched pairs design was used to compare behavioural change of gamblers who opted into use the mentor behavioural feedback system, with behaviour of gamblers who did not use the mentor system (n = 15,216), and were matched for age, gender, playing duration, and theoretical loss in the 14 days prior to uptake of the mentor system for the experimental group. The mentor system also applied the principles of HCI and PSD (see Wohl et al. 2014), and provided players with visual feedback in the form of a graphs with the amount of time and money spent gambling in comparison to normative behaviour of other gamblers in the database. Results indicated that of the 1015 gamblers using the mentor system, 625 (62 %) showed a smaller theoretical loss ratio and 60 % showed a shorter playing duration ratio in comparison to theoretical loss and playing duration of matched control group ratio (12 and 10 % above chance level respectively). The findings indicated that overall, gambling behaviour of those using a personalised behavioural feedback system decreased more than control group members.

While a difference in behaviour as a consequence of the personalised feedback system was found, the effects were small, which means a degree of caution is required before a full endorsement of behavioural feedback is made. In addition, a limitation of the study includes the fact that no information about the gambler’s level of risk or problem gambling status was obtained. Consequently, it is not known whether the tool was most effective for those players with problem gambling tendencies, or whether the tool was most effective in moderating the behaviour of those gamblers who already gambled responsibly. In addition the study was unable to determine if the gamblers were concurrently using any other gambling sites or platforms during the evaluation period.

Wood and Wohl (2015) assessed the efficacy of the PlayScan behavioural tracking tool, which provided gamblers with behavioural feedback about their gambling, in terms of its impact on gambling behaviour. A sample of 779 gamblers (694 males) who opted into use Playscan was obtained from the online gambling site Svenska Spel. Gambling behavioural data was compared for those who opted into use the Playscan system with matched controls who did not opt in. The behavioural feedback utilised an algorithmic system which provides players with a colour-coded risk rating according to their expressed behaviours, with green indicating no issues, yellow being at-risk, and red being problematic. Gambling expenditure data (deposit and wager amounts) were gathered for the week in which players enrolled on Playscan, as well as the subsequent week and 24 weeks later. These data were also gathered for the matched pairs control group.

Results showed that at-risk players (‘yellow’ players) who used the feedback tool significantly reduced the amount of money deposited and wagered compared to players not utilising the RG tool. Furthermore, this effect was obtained for both the week following enrolment and at 24 weeks later. Results indicated that those gamblers who received behavioural feedback showed a significant reduction in deposited amounts compared to the control group in the week after enrolment. However, ‘red’ and ‘yellow’ players (i.e., those showing signs of problematic or risky play) did not significantly reduce their deposit amounts in this period compared to a control group. Only the ‘green’ group showed a significant deposit reduction for this period, relative to the control group. However, deposit reductions were noticeable over time, with ‘green’ and ‘yellow’ gamblers showing a significant deposit reduction from week of enrolment to week 24 compared to the control group. There was no such reduction over this period of time for red players.

In terms of wagering amounts, while ‘red’ players reduced their wagering between enrolment and 24 weeks later, this amount did not differ compared to the control group. However, for the same period, ‘yellow’ and ‘green’ gamblers significantly reduced their wagering amounts compared to a control group. This suggests that behavioural feedback via behavioural tracking may have a positive impact in keeping controlled gamblers safe, as well as positively impacting at-risk players, while the effects on those gamblers already exhibiting problematic symptoms appears minimal. This supports the notion of behavioural feedback as an RG tool aimed at preventative measures, rather than an intervention for problem gamblers. However, as the authors noted, the extent to which the colour classifications actually relate to more standardised measures of problem gambling is unknown.

One method that had been implemented in Norway as a way to reduce gambling expenditure and gambling-related harm is the prohibition of note acceptors on slot machine, which produced a 40 % reduction in the turnover produced by slot machines (see Norwegian Gaming Authority 2006, 2007). The prohibition or restriction of note acceptors appears to be a valid avenue of exploration in RG, particularly given evidence suggesting problem gamblers more frequently use high denomination bank notes when gambling compared to non-problem gamblers (Sharpe et al. 2005). Despite evidence from Australia that (1) suggests problem gamblers prefer to use note acceptors while gambling (Australian Productivity Commission 1999), and (2) there is a strong correlation between problem gambling and use of note acceptors (McMillen et al. 2004), there is very little empirical evidence demonstrating the efficacy of prohibition/restriction of note acceptors in reducing problem gambling among EGM players.

Sharpe et al. (2005) tested the effects of several modifications to gaming machines, including a restriction on note acceptors to allow a maximum of a $20 note. The research was carried out in an ecologically valid environment, with 779 participants of varying problem gambling severity playing on the modified gaming machines in hotels and bars. Several proxy measures of gambling behaviour recorded, including time spent gambling, number of bets, net loss, and lines per wager. However, machines with restrictions on note acceptors failed to have any significant impact on any aspect of gambling behaviour compared to control machines.

The authors highlighted several limitations of the research, including the fact that a large proportion of gamblers approached to take part in the study declined, bringing into question how representative their sample was. Other limitations included the potential part that demand characteristics played on participant gambling behaviour, due to the fact that participants were being observed by the experimenter to record gambling behaviour. In addition, there were an insufficient number of probable problem gamblers in the sample to compare whether the machine modifications had differential efficacy in modifying behaviour for problem gamblers in comparison to non-problem gamblers.

Hansen and Rossow (2010) explored the impact of prohibition of note acceptors on slot machine players in terms of its impact on gambling behaviour and problem gambling measures (SOGS-RA and Lie/Bet) in adolescent-aged gamblers. The samples comprised 20,703 students in 2004 (pre-intervention); 21,295 in 2005 (pre-intervention); and 20,695 in 2006 (post-intervention). Respondents were mostly 13–19 years old with an average age of 15 years and there was an approximate 50/50 gender split. Importantly for the efficacy of note acceptor prohibition as an RG measure, results showed no significant changes in gambling behaviour and problem gambling at time points one and two (pre-prohibition). However, significant differences were found at time point three following prohibition. Following prohibition, and controlling for potential confounding variables, slot machine gambling frequency was reduced by 20 %, the proportion of ‘frequent’ slot machine gamblers was reduced by 26 %, and overall gambling frequency was reduced by 10 %. In addition, the proportion of problem gamblers was reduced by 20 %. No significant gender differences were found.

Only one-third of adolescent gamblers reported noticing the removal of bank note acceptors, and two-thirds reported either stopping gambling or reduced gambling following the prohibition. Hansen and Rossow (2010) reported that only a small fraction of participants attributed the changes in their gambling behaviour to the removal of bank note acceptors. Importantly, no compensatory behaviour in terms of transition to other forms of gambling was observed after the intervention, and decreases in gambling behaviour were also observed for both at-risk and problem gamblers.

A limitation of the research is that it does not offer explanatory value in terms of the mechanisms of change. One argument proposed by Hansen and Rossow (2010) stated that an inability to use notes slows down the speed of play, where speed of play has frequently been implemented as a problematic characteristic of electronic gambling (McCormack et al. 2013). In addition, it is possible that the need to transfer notes into coins may break up the rhythm of play, which may have the added effect of breaking dissociative states and raising levels of self-consciousness regarding gambling time and monetary expenditure. The time taken to transfer notes to coins, or the associated increased time it takes to load a machine with coins, may be sufficient time to allow any increased levels in stress and arousal to dissipate, allowing gambling decisions to be made rationally in a ‘cold’ (as opposed to ‘hot’) emotional state (Parke et al. 2014a, b).