2012-Prevention of problem gambling.pdf

Similar low utilization rates have been reported in Australia and the United States 22

In some jurisdictions, ‘casino’ self-exclusion also includes other gambling venues such as racetracks with slot machines and/or bingo halls. Some jurisdictions also allow lottery self-exclusion (e.g., Iowa since 2005). Casino self-exclusion is primarily a ‘tertiary’ prevention initiative in that it occurs after problems have developed, and would not normally be covered in this review. However, it has been included because a) it is used as a preventive measure by some nonproblem gamblers; and b) it can be included on a continuum with some of the on-site interventions used for at-risk gamblers (described later).

47

(South Australian Centre for Economic Studies, 2003). One European jurisdiction with significantly higher rates is the Netherlands. This is due to the proactive nature of their program, where individuals with high rates of casino patronage are proactively approached to see whether they wish a ‘visit limitation’ or casino exclusion contract (Bes, 2002; De Bruin et al., 2001; Nowatzki & Williams, 2002; Goudriaan, de Bruin, & Koeter, 2009). Thus, “offering” appears to be more effective than “making available”. Other measures to determine the effectiveness of self-exclusion programs include a) the percentage of self-excluded people who do not actually re-enter the casino(s) during the contracted period of exclusion, and b) the impact self-exclusion has on overall gambling behaviour. Several studies have investigated these issues (see Gainsbury, 2010 for a review):  Ladouceur, Jacques, Giroux, et al. (2000) studied 220 individuals who self-excluded from a Quebec casino (that had no penalty for breaches). A subset of 53 went back to renew or reestablish a self-exclusion contract. Of this group, 64% reported they had not entered the casino during their previous exclusion period. However, the 36% who did return reported going back a median of 6 times.

Only 30% of these 53 people reported they had stopped gambling completely during their previous contract (which had typically been for a period of 6 to 12 months) (Ladouceur, Jacques, Giroux, et al., 2000).  Steinberg and Velardo (2002) studied a small subset (n = 20) of the 294 self-excluders at the Mohegan Sun Casino in Connecticut. Eighty percent reported they did not return to the casino during the period of exclusion, but the majority of the 20% that did return went back more than 9 times.  A review of self-exclusion in Victoria, Australia also concluded that a significant number of self-excluders re-enter casinos without being detected (O’Neil et al., 2003).  A survey of 135 self-excluders from New South Wales by Croucher, Croucher & Leslie (2006) reported that 70% of participants reduced their gambling expenditure by 50%. However, 75% started gambling within 6 months of self-excluding, and 67% of those who breached their agreements indicated they were detected.  A total of 169/239 Quebec first time self-excluders agreed to be contacted every 6 months by Ladouceur, Sylvain, & Gosselin (2007). Participants were divided into the length of exclusion they chose: 6 months, 12 months, or 24 months. Retention was low, falling to 116 at 6 months, 83 at 12 months, 60 at 18 months, and 53 at 24 months23.

Follow-up evaluations found self-excluders to have significant reductions in the urge to gamble and problem gambling symptomatology. At the 6-month follow-up, 40.5%, 42.3%, and 22.2% of the self-excluded patrons had re-entered a casino (6, 12, and 24 month groups respectively). At the 12 month follow-up 55.3% and 10.5% of the 12 and 24 month groups had breached. Finally, at the 18 month follow-up, 26.6% of the 24 month group had breached.  An Alberta Gaming and Liquor Commission (2007) study contacted a random sample of 300 Alberta self-excluders who had previously been in or were currently in the self-exclusion

23

With such attrition rates, results must be accepted with caution as it would be reasonable to anticipate that dropping out is more likely to occur because of breaching the self-exclusion contract.

48





 



program. 24 The majority of people were satisfied with the program with 67% reporting it was somewhat or very effective in doing what they expected it to do. However, 47% spontaneously indicated that the main weakness of the program was lack of enforcement. A total of 50% reported having re-entered during the time of their contract, breaching 33 times on average. Only 48% indicated that they had ever been detected, with the ones never being detected entering an average of 51 times, but the ones who had been detected only entering 13 times. A nonrepresentative sample of 76 self-excluders from 7 Canadian provinces was surveyed by the Responsible Gambling Council (2008). Participants reported their gambling was reduced following self-exclusion and 30% did not gamble at all during the contracted period. Nonetheless, 59% gambled at venues not covered by their agreement. Another 33% breached their agreements (70% of these breaches were undetected). Hayer & Meyer (2011a,2011b) contacted banned gamblers from 18 different casinos in Austria, Germany, and Switzerland (baseline sample = 152). Follow-up surveys at 1, 6, and 12 months found significant decreases in gambling. However, the participants who were actually followed up constituted a very small (presumably nonrepresentative) percentage of the total group.

Fifty percent of problem gamblers who had banned themselves from Holland Casinos reported they had gambled elsewhere during the period of the ban (e.g., illegal casinos, arcades, or abroad) (Goudriaan et al., 2009). Nelson et al. (2010) surveyed 113 Missouri self-excluders (from 419 selected to be interviewed) 4 to 10 years after receiving lifetime bans. Although only 13% had not gambled at all since their ban, the large majority reported gambling less than before the ban. Among these, there was a notable reduction in symptoms of problem gambling. Seventy four percent of people who continued to gamble went to casinos in other jurisdictions. Only 16% attempted to re-enter a Missouri casino, with 50% being successful, some re-entering a large number of times. Self-excluders who engaged in complementary treatment had more positive outcomes than those that did not. In British Columbia, 169 self-excluders (roughly 6% of eligible self-excluders) participated in the baseline longitudinal evaluation of the self-exclusion program. The sample decreased to 43 individuals by the fourth assessment 2 years later (BC Centre for Social Responsibility, 2011). The majority of individuals continued to gamble after exclusion, however, only 35% reported re-entering a British Columbia casino during the period of exclusion.

The majority of these individuals re-entered without being detected, often multiple times. There was rarely any penalty imposed for those who were detected.

Casino self-exclusion programs have lower potential to reduce harm compared to other initiatives, as they are designed to minimize the harm in existing problem gamblers rather than preventing problem gambling in the first place. However, it is clear they have some value as a 24

The Alberta self-exclusion program had been in operation for 6 years at the time of the study. Ban length was 6 months, 1 year, 2 years, or 3 years. 15% of the sample enrolled in 2006; 21% in 2005; 20% in 2004; 20% in 2003; 8% in 2002; 11% prior to 2002; 5% didn’t report when they enrolled. In Alberta, the penalty for breach is the possibility of up to a $10,000 fine and up to 6 months imprisonment (however, enforcement is uneven).

49

tertiary initiative. The most unambiguous impact is that most people who enter into these programs have a significant reduction in their gambling and problem gambling symptomatology. Undoubtedly, a good portion of this effect is due to the fact that people taking this step have recognized they have a problem, are highly motivated to do something about it, and have made a public proclamation that they do not intend to reenter casinos. The subsequent behavioural changes observed in self-excluders are not fundamentally different than what is observed in people presenting themselves to any form of gambling treatment. The additional utility of self-exclusion lies in its potential to provide additional external constraints on the person’s gambling when his/her motivation falters. This, in turn, is related to the following factors (Nowatzki and Williams, 2002): 1. Whether the bans are irrevocable. Revocable bans (which are common) defeat the purpose of self-exclusion, which is to set up enduring external constraints for people attempting to curb their gambling, usually after efforts at self-control have failed. Irrevocability helps manage periods of internal conflict, where the urge to gambling predictably becomes high. Evidence also indicates that patrons prefer irrevocable bans. 2. Whether longer ban lengths are available.

Many jurisdictions only have short ban lengths (e.g., 6 months to 2 years), which sends the wrong message to self-excluders, the majority of whom will have lifelong propensities for excessive gambling. The appropriate or optimal length of abstinence required to prevent problem gambling relapse is unclear. However, for substance abuse, abstinence of 3 to 6 months is commonly achieved but has almost no predictive value in preventing relapse. Rather, periods greater than 2 years are necessary to prevent relapse in most cases. Even then, 25% of alcohol-dependent people will relapse after four years of abstinence and 7% will still relapse after 6 years (Vaillant, 1995). Hence, from a clinical perspective, it would be prudent to encourage ban duration of at least 5 years and to have a lifetime option available. However, a short ban length of 6 months should also be available to encourage the use of self-exclusion in people uncertain whether they wish a long term ban. 3. How widely applicable the ban is. Self-exclusion has limited deterrent value if it is only applies to one venue or one type of gambling. Jurisdiction-wide exclusion is common in Canada (except Quebec) and Europe (although sometimes limited to casinos), but is less common in the United States and Australia. In Canada, self-exclusion also applies to racinos and bingo halls in some provinces.

Only a few countries apply land-based self-exclusion to their online gambling services (and vice versa). The ability of the self-excluder to readily access gambling in neighbouring jurisdictions is also related to the net deterrent value of self-exclusion. 4. The perceived and actual chances of casino reentry being detected. Most Western countries have always required some form of identification to enter a casino.25 This includes Australia (clubs only), Austria, Belgium, Bulgaria, Croatia, Czech Republic, Denmark, Estonia, Finland, 25

The ID required is usually some form of government-issued photo ID (e.g., passport, driver’s license). A few jurisdictions have used casino-issued non-photo ID (e.g., Missouri), which has less utility, as it is much more likely to be given away or loaned. Of note, when Missouri ended their ID requirement in 2008 an 18% decrease in the number of detected self-excluders occurred in the first half of 2009 (Young, 2009).

50

France, Germany, Greece, Hungary, Ireland, Italy, Luxembourg, Malta, Macedonia, Moldava, Monaco, Netherlands, Poland, Portugal, Slovenia, Spain, Sweden, Switzerland, and the United Kingdom (clubs only). Requiring ID to enter a casino is also common in most Asian, South American, and African casinos. Automated scanning of ID against a self-exclusion list provides much more effective detection than a system that requires casino security staff to visually recognize a self-excluded individual. Visual detection is currently used in Canada, the United States, Australia, New Zealand, and a few other countries. 26 Computerized identification checks further enable more proactive ‘visit limitation contracts’, as opposed to total bans, as occurs in the Netherlands (Bes, 2002) (see section on Intervention for ‘At Risk’ Gamblers). In some countries that rely primary on staff recognition of self-excluders, supplementary detection systems have been implemented that have modestly improved these low detection rates (i.e., detecting self-excluders through their Player Card usage, biometric facial recognition, license plate recognition, paying staff to detect selfexcluders)27. 5. The consequences of detection.

Even though penalties to the gambler are often stipulated in self-exclusion agreements (e.g., fines, confiscation of winnings, trespassing charges), with the exception of forfeiture of winnings, they are rarely applied (Nowatzki & Williams, 2002). Mandatory identification for wins over a few hundred dollars combined with automatic confiscation of winnings for barred self-excluders would eliminate much of the incentive to violate bans. 26

The difficulty of this task is illustrated by the fact that in 2007 over 10,000 people were self-excluded in the province of Ontario, and reference files for security staff filled 22 binders (Canadian Broadcasting Corporation, 2007). Schrans, Schellinck & Grace (2004) found in a Nova Scotia evaluation, that of 150 people posing as selfexcluders (i.e., they had formally registered in order to test the effectiveness of detection), 77% of breaches were not detected. Although North American casino owners often contend that patrons would not accept an ID requirement, showing ID is a common practice in many contexts, such as cashing a cheque, boarding a plane, etc. Furthermore, since 2003 many bars throughout Canada and the United States have been using an ID scanning system to exclude undesirable patrons (http://www.entersafe.com/) (formerly known as Barwatch and Barlink). 27

Illinois casinos pay employees $250 for every self-excluded gambler they detect) (Illinois Gaming Board, 2008). Facial recognition was introduced by some North American casinos in the late 1990s for identification of card counters, cheats, and ‘high rollers’ (Market Wire, 2000). In subsequent years it was adopted by many casinos around the world (e.g., U.S., Canada, Macau, Germany, and Australia) in an effort to also identify self-excluders. Historically, facial recognition biometrics has been beset with technical problems, making its utility quite limited (EPIC, 2005; Woodward, Horn, Gatune, Thomas, 2003; Zhao, Chellappa, Phillips, Rosenfeld, 2003). The main problem is that maximum effectiveness requires a full frontal screen of the person’s face under good lighting (e.g., locating cameras at a brightly lit entrances). The more typical use of overhead video cameras does not provide adequate facial capture, and thus, yields many ‘false negatives’ (i.e., people not successfully detected) and some ‘false positives’ (people incorrectly identified as self-excluders).

In June 2009, British Columbia abandoned facial recognition because of continuing poor results and replaced it with 'license plate identification' (beginning in 2008), whereby self-excluders provide their vehicle license number which is then detected by parking lot cameras at casinos (1/3 of detections are now due to this). Improvements in facial recognition continue, however (Moore, 2007; Williams, 2007), and casinos employing the most recent technology in combination with entrance cameras report good results (e.g., Zielke & Wolfer, 2008). One of the most advanced systems is used in Ontario, but a April 2009 evaluation report (obtained by Freedom of Information legislation by the authors) of the iGWatch Face Recognition System shows it to have a correct identification rate of only 15%, a false rejection rate of 85% (i.e., failed to detect a self-excluder), and a false acceptance rate of 2% (i.e., misidentified someone as a self-excluder).

51

6. The consequences of failing to detect. Gambling venues rarely suffer consequences for failing to exclude people they indicated they would.28 Indeed, most gambling venues consider themselves absolved of legal responsibility when a self-exclusion contract is breached. Their perspective is that these are ‘agreements’, rather than legally binding ‘contracts’ (many have also introduced clauses that stipulate they are absolved from responsibility should breaches occur). To date, courts in the United States, Australia, and the United Kingdom (Armstrong, 2008) have agreed with this position. However, in 2003 the Appeals Court of Austria ruled against Casino Austria, stating that the casino had an obligation to refuse entry to players whose financial solvency was in question (Rhea, 2005). At least three cases in Germany have been settled in favour of the self-excluder (Glűcksspiel & Recht, 2009). In 2007, German law unambiguously established a legal requirement for all casinos to take ‘reasonable efforts’ to ensure that casino patrons are not self-excluded gamblers, who may be reimbursed for their losses if casinos fail to meet this duty (Gambling Compliance Ltd, 2007; Meyer & Hayer, 2009). Since 2000, Swiss law has required casinos to detect gamblers ‘at risk of addiction’ and to then take appropriate action (Hafeli, 2009).

The Ontario Lottery and Gaming Corporation have settled 13 self-exclusion cases out of court in recent years in favour of the patron (Canadian Broadcast Corporation, 2007; Rhea, 2005). Many observers believe it is only a matter of time before the legal ‘duty of care’ established for alcohol providers (e.g., U.S. Dram Shop liability laws; Canadian Supreme Court 1973 Menow ruling) is applied under common law to gambling providers (Hillyer, 2003, as cited in Sasso & Kalajdzic, 2006). 7. The availability of complementary treatment. Problem gamblers who sign exclusion contracts have taken an important step, but many would further benefit from a formal treatment program (Sani, Carlevaro, & Ladouceur, 2005). Accordingly, treatment options should be offered and encouraged to everyone who enters into a self-exclusion contract. The province of Manitoba has long required self-excluded patrons to attend a responsible gambling awareness seminar prior to re-entry (topics include: review of past gambling history, information on how gambling works, plan for returning to gamble). Quebec is using a similar approach and Alberta instituted its own system in July 2009 (Responsible Gambling Council, 2008). In Illinois, the self-excluded person must receive certification from a psychiatrist that controlled gambling is possible before casino entry is re-instated.

28

Exceptions to this rule exist. The Iowa Racing and Gaming Commission fines casinos $20,000 - $30,000 each time they admit customers who have self-excluded (Gearino, 2008). Singaporean casinos can be fined up to $1 million and lose their license (Brunei Fm World News, 2009).

52

RESTRICTIONS OR ALTERATIONS ON HOW GAMBLING IS PROVIDED Modifying Electronic Gambling Machine (EGM) Parameters In recognition that EGMs are currently associated with the highest problem rates in Western countries, a number of research studies have investigated initiatives to alter machine features in attempts to mitigate harm. In most gambling locales, machines with the highest revenue generation continually replace less lucrative machines. Hence it can be expected that contemporary machines have evolved to employ a wide array of attributes that maximize revenue generation. Following is a summary of research that has endeavored to ‘unravel’ some of the feature modifications intended to minimize EGM harm. Reviews of this issue are also contained in Parke & Griffiths (2007), Griffiths (2008), and Livingstone & Woolley (2008). Win/Loss Patterns In one of the first laboratory studies of gambling, Lewis and Duncan (1956, 1957, 1958) found that a lower percentage of wins during the trial period produced longer periods of persistent EGM play after the machines no longer delivered wins. Lewis and Duncan (1957) also found that the larger the size of the wins, the longer it took to extinguish play. Levitz (1971) exposed university students to an EGM variable ratio schedule that either produced a net win or a net loss over 22 trials.

Both groups were then subjected to a net loss schedule. Participants with the prior exposure to the net win schedule persisted significantly longer in the second phase. Dickerson, Hinchy, England, Fabre, and Cunningham (1992) observed the natural play of 10 high frequency EGM gamblers and found that small wins (fewer than 50 credits) were associated with an increase in play rate, but big wins (more than 50 credits) caused a temporary decrease in play rates (termed ‘post-reinforcement pauses’). Similar results in similar naturalistic studies were reported by Dickerson (1993), as well as findings that the number of big wins during the session was strongly predictive of duration of the session. Delfabbro and Winefield (1999a), also using a naturalistic observational study, obtained the same findings regarding the effects of small versus big wins on play rates immediately following wins, but did not find win size made a difference in overall rate of play. A post-reinforcement pause in laboratory EGM play following wins was also found by Schreiber and Dixon (2001) and Dixon and Schreiber (2002). In addition, the latter study found that the speed of play increased as the number of non-reinforced trials (i.e., without wins) increased.

In contrast to Dickerson (1993), Kassinove and Schare (2001) did not find that a ‘big’ win in a laboratory setting resulted in greater persistence to extinction, although they suggest the size of the big win ($10), may have been insufficient for the anticipated effect. Similarly, Weatherly , Sauter & King (2004) found that laboratory participants who experienced a big win on their first play quit playing the simulation earlier

53

than participants who experienced a large win on the 5th play. Dixon, MacLin, & Daugherty (2006) found that participants had a preference for slot machines with more frequent payouts compared to ones with more intermittent payouts. Game Play Speed The speed at which play is permitted is also a reinforcement parameter, as it directly relates to frequency of rewards. Accordingly, some countries mandate a minimum time gap between games (e.g., 3 seconds in casino-based EGMs in Belgium, Druine, 2009; 5 seconds in Spain, Becona, 2009; 3.5 seconds in South Australia, 2.1 seconds in Victoria, and 3 seconds in Tasmania and Queensland; Australian DFHCSIA, 2012). Blaszczynski, Sharpe, and Walker (2001) found no significant difference in money or time spent between individuals who played EGMs with 3.5 second versus 5 second game speeds in a study conducted with 210 EGM players in clubs and hotels in New South Wales. This finding was attributed to the fact that only 12% of players normally played at a wager cycle faster than 5 seconds. Delfabbro, Falzon, & Ingram (2005) found that 3.5 second games (relative to slower games) did produce an increase in number of games played, but not total time spent playing.

Results of a study conducted by Ladouceur & Sevigny (2006) indicated that 5 second game speeds caused gamblers to play significantly more games and spend more money compared to 15 second game speed EGMs. Similarly, a 30% reduction in game speed was reported to be an important factor in a 14% reduction in expenditure and time spent by gamblers (particularly higher risk gamblers) in a study of VLTs in Nova Scotia (Corporate Research Associates, 2006).29 The introduction of auto-play EGMs (machines that play automatically on insertion of money and pressings of an ‘AutoPlay’ button), is relevant to the issue of game speed play, but there has been no research on their impact. Nonetheless, the presumption of greater harm is reflected in the fact that AutoPlay has been banned in Victoria, South Australia, and Western Australia (Caraniche Pty Ltd., 2005). Near Misses Most EGMs are programmed to display ‘near misses’ (or ‘near wins’) more frequently than they would occur by chance occurrence despite some legal efforts to ban this feature (Harrigan, 2008).30 The importance of ‘near misses’ in EGM gambling to perpetuate play has been

29

30

The actual time speeds used were not reported.

In Ontario, for example, the Alcohol and Gaming Commission of Ontario (AGCO), in their Minimum Technical Standards for Electronic Gaming Equipment (Dec 2007) states that regarding ‘Near Misses”, that “after the selection of game outcome, the gaming equipment must not make a variable secondary decision which affects the result shown to the player. For instance, the Random Number Generator (RNG) chooses an outcome that the game will lose. The game must not substitute a particular type of loss to show to the player. Lotteries have similar features, as the most common lottery format worldwide involves selecting between 5 and 7 numbers between 1

54

recognized for some time (Reid, 1986; Griffiths, 1991; Harrigan, 2009). For example, Strickland and Grote (1967) found that placing frequent winning symbols early in the slot machine’s 3 symbol sequences increased persistence in laboratory slot machine play compared to when they were placed late in the sequence. Skaer (1985) similarly found that his subjects preferred to play the slot machine that was perceived to come close to paying off more frequently. This same finding was obtained by MacLin, Dixon, Daugherty, & Small (2007). Both Kassinove and Schare (2001) and Cote, Caron, Aubert, Desrochers, and Ladouceur (2003) have confirmed that people who come close to a ‘near win’ persist significantly longer in subsequent laboratory EGM play. Wohl and Enzle (2003) found that wagers following a near loss were significantly higher in subsequent games of laboratory-based, computerized roulette play.31 Clark et al. (2009) has further documented that near misses enhance motivation to gamble. Giroux & Ladouceur (2006) found that more gamblers chose to continue their gambling session of VLTs showing near wins. However, this preference did not translate into more money spent on that VLT.

Number of Play Lines In recent years there has been a tendency to increase the number of play lines available on EGMs, which has the effect of increasing the overall rate of operant reinforcement (Haw, 2008a; 2009; Harrigan et al., 2011). Not surprisingly, then, in a series of observational studies, Williamson & Walker (2000) and Walker (2001) have found that gamblers have a preference for playing large number of lines (up to 20), with a minimum bet per line. This is known as the ‘mini-max’ strategy (Livingstone et al., 2008). In a laboratory investigation, Delfabbro, Falzon and Ingram (2005) also found that 3 line betting produced an increase in the number of games played and time spent compared to single line betting. Very few jurisdictions impose constraints on play lines. An exception is Australia, where no more than 50 lines are permissible in Queensland and no more than 30 in Tasmania (Australian DFHCSIA, 2012). Bill Acceptors Note or bill acceptors were introduced EGMs beginning in the late 1980s and early 1990s primarily to decrease the need to handle large amounts of change (both for the gambler and the venue). However, an added consequence was a significant increase in EGM revenues.

It is thought that these increased revenues may have disproportionately impacted problem gamblers, as note acceptors decrease the need to visit the cashier on a regular basis (which might act as a social or psychological constraint) and decrease the need for breaks (Productivity

and 60. As a result, many players have correctly selected a significant portion of the winning numbers (and receive small prizes), and will believe they were close to winning the jackpot when they were not. 31

A related phenomenon likely with the same effects, is ‘nudging’, where an EGM reel appears to come to a halt with a winning sequence, and then a second or two later, nudges over to a different outcome.

55

Commission, 1999). Note acceptors are now common on EGMs, but are prohibited in some jurisdictions (e.g., Norway; South Australia; within hotels and clubs in Tasmania and the Australian Northern Territory (Australian DFHCSIA, 2012)). Other jurisdictions limit the size of bill the note acceptor will take (e.g., $100 note acceptors are banned in the Australian Capital Territory and Victoria (Australian DFHCSIA, 2012)). There is limited formal research on the impact of bill acceptors. In 2006 the Norwegian government banned them on EGMs, following which a 16% drop occurred in the number of calls to the national problem gambling helpline and a 24% reduction in the number of people seeking help for gambling problems (Engebø & Gyllstrøm, 2008; Gotestam & Johansson, 2009). In adolescents (age 13 -19), there was 20% reduction in both slot machine use and the proportion of problem gamblers (Hansen & Rosow, 2009). Limiting the size of the note, on the other hand, has not proven very effective. Blaszczynski, Sharpe, and Walker (2001) (also reported in Sharpe et al., 2005) found that limiting bill size to a $20 AU maximum had no significant effect on time or money spent gambling (even though more problem gamblers than recreational gamblers used large note acceptors). They attributed this to the fact that all venues have facilities to change larger denominations to smaller notes.

This same restriction was reported to be ineffective in a study of Queensland EGM players (Brodie, Honeyfield, & Whitehead, 2003), a significant portion of whom indicated the maximum note amount should be lower than $20 or that bill acceptors should be eliminated altogether. Stake and Prize Size There is a presumption in many European countries that EGMs allowing larger bets and offering larger jackpots may be more problematic. This is reflected in the fact that it is common to have different ‘classes’ of EGMs, with greater restrictions on the numbers and placement of machines that allow larger stakes and prizes (Meyer, Hayer, & Griffiths, 2009). In Belgium, wins at casino EGMs are allowed to reach €70 per hour, whereas arcade EGMs are limited to €25/hr and bingo machines €12.5/hr. Also, the maximum jackpot is €10,000 for casino EGMs, €2,000 for arcade EGMs and €500 for bingo machines. In Germany, EGMs in arcades, restaurants and bars are required to have lower stakes and jackpots than in casinos. In Great Britain, EGMs are categorized according to the maximum stake and prize available, with the maximum stake ranging from 10p (D category machines) to £100 (B2 category machines) and the maximum prize ranging from £8 (D) to £4000 (B1).

Machines with the highest stake and prizes are limited to casinos and betting shops/tracks, with lower stake/prize machines being permitted in pubs, and arcades. In Hungary, category 2 machines in pubs and restaurants have a maximum stake of €0.8 and a win not more than 25 times the stake, and category 1 machines in gaming arcades and casinos permit a win of 200 times the stake. In Iceland, the maximum prize is €110 in shops and food outlets without a liquor license, but €1100 in bars and restaurants with a liquor license. In Lithuania, machines with unlimited payouts can only be operated in casinos, but machines with limited payouts can be operated in gambling machine halls. In the Netherlands, the loss on EGMs outside casinos is limited to €40 per hour but is unlimited for EGMs inside casinos. In Poland, EGMs outside casinos and arcades are limited to €.07 per bet with the maximum prize being €15. In Australia, there is a $5 bet limit in Tasmania, Victoria, Queensland

56

(hotels and clubs), and a $10 limit in other states/territories (Australian DFHCSIA, 2012). Certain states have no win limit (Australian Capital Territory, Northern Territory, South Australia, Tasmania, Victoria, Western Australia), whereas others have limited it to $10,000 for non-linked machines (South Australia, Queensland, New South Wales). Relatively little formal research has been conducted on this issue. In Australia, the top revenuegenerating EGMs tend to have a higher average bet size, which is due to having a larger number of play lines to bet on (Livingstone & Woolley, 2008). In laboratory settings, Weatherly and Brandt (2004) found that students tended to bet more when given $1 and each bet was worth $0.01, compared to students who were given $10 and each bet was worth $0.10, suggesting that the perceived magnitude of bets and losses influences duration of play. In contrast, Delfabbro, Falzon, and Ingram (2005) found no influence of bet size (1 versus 3 credits) on time spent gambling or number of plays. However, it is perhaps worth noting that the lower rates of problem gambling found in many European countries (Williams, Volberg, & Stevens, 2012) is associated with significantly lower bet and win sizes on most European EGMs. For example, in 1998 the maximum bet in the U.K.

was only £0.50, whereas it was $10 in Australia, $100 in Canada, and $500 in the U.S (Caraniche Pty Ltd., 2005). Although higher stakes EGMs have been introduced in most European countries in recent years, lower stake EGMs are still common (Meyer, Hayer, & Griffiths, 2009).32 The introduction of a policy to limit EGM bets to a maximum of $10 in the Australian Capital Territory did not result in behavioural change for either recreational or problem gamblers, presumably because $10 was a higher bet than the large majority of EGM players typically made (McMillen & Pitt, 2005). However, Blaszczynski , Sharpe, & Walker (2001) (also reported in Sharpe et al., 2005) found that reducing maximum bet from $10 to $1 resulted in significantly decreased EGM expenditures and time spent playing. Furthermore, arguing from primarily a theoretical perspective, both Livingstone & Woolley (2008) and the Australian Productivity Commission (2010) concluded that reducing the maximum bet size to $1 would significantly reduce the harm from EGMs, both because of the lower losses that would be incurred and because problem gamblers are much more likely to bet more than $1 per spin. Payback Percentage Payback (or payout) percentages on EGMs vary from 60% to 101%33, with many countries mandating a minimum payback percentage. In Poland, no payback percentage is mandated by law (Dzik, 2009).

In the Netherlands, a payout ratio of at least 60% is required in EGMs outside 32

There are several other possible reasons for the lower current rates of problem gambling in European countries (e.g., longer exposure to legal gambling). 33

In Nevada, it is not uncommon to have one or two machines in a venue paying more than 100% so as to draw in customers (who end up playing the lower payback machines because the machine with 100+% is in constant use).

57

of casinos and 80% for casino EGMs. In the U.K. payout rates start at 70%. In Spain the minimum payback rate is 75% (Becona, 2009). In Nevada it is 75% and 78% in New Jersey. In Belgium the minimum payback percentage is 84% (Druine, 2009). In Australia the payback percentage ranges from 85% to 90% (Australian DFHCSIA, 2012). In Canada it varies between 85% and 92% depending on the province and whether it is a video lottery terminal (in a bar) or a slot machine (in a casino). Somewhat surprisingly, empirical research has failed to find a strong association between payback percentage and player behaviour. For example, in a laboratory simulation, Weatherly and Brandt (2004) found that percentage payback rate (75%, 83%, and 95%) did not influence EGM gambling behaviour over a 15 minute session. Similar findings have been obtained by Gillis, McDonald, & Weatherly (2008) and Weatherly, Thompson, Hodny, & Meier (2009). A contrary finding was obtained by Haw (2008b), where participants did show a behavioural preference for machines with higher payback percentages.

An overall review of this literature by Parke (2010) found no empirical support that payback percentage was related to the development of problem gambling.34 From a theoretical perspective, higher payback percentages will produce higher reward frequency, and heightened prospects for an ‘early big win’, both of which will normally strengthen the behaviour. However, this is offset by the fact that higher payback percentages will also be associated with less financial loss and harm. Thus, there is certainly a payback range that should theoretically have a closer relationship with problem gambling: i.e., one that offers a high enough reward frequency but also ensures significant financial losses and ‘chasing’ (these conditions will also maximize profit for the operator). Thus, the fact that almost all EGMs have a payback rate between 75% and 92% is probably no coincidence. Interactive Features The increasingly interactive nature of EGMs almost certainly promotes the illusion of control (Griffiths, 1993; Langer, 1975). However, there has been very little empirical research that investigates the magnitude of this effect on EGM play. Loba, Stewart, Klein & Blackburn (2001) found that gamblers did not believe that the presence or absence of a ‘stop reel’ function (button) would alter gambling behaviour.

Despite this perception, two studies that examined actual behaviour found the ‘stop reel’ function significantly increased the length of gambling sessions (Ladouceur & Sevigny, 2005) and money gambled (Corporate Research Associates, 2006). In the latter study, higher risk gamblers were most influenced.

34

These general findings are consistent with the fact that consumer patronage tends to be highest for forms of gambling with the lowest payback percentages (i.e., lotteries, instant win tickets), and lowest for forms of gambling with the highest payback percentages (casino table games, sports betting).

58

Pop-Up Messages Ladouceur & Sevigny (2003) investigated the effectiveness of two different pop-up messages on 30 EGM players. The study found a significant reduction in the number of bets made by players who saw a message about randomness and by players who simply saw the word ‘break’ compared with a group not exposed to pop-up messages. They hypothesized that pop-ups interrupted cognitive processes that tended to facilitate narrowed attention and ‘loss of reality’. Schellinck & Schrans (2002) found that an EGM pop-up message after 60-minutes of continuous play (and 30 again minutes thereafter), telling players how long they had played and asking if they wished to continue, resulted in a small but significant reduction in session length and expenditure among higher-risk players. In the second phase of this study, Schrans, Grace and Schellinck (2004) found that a pop-up message after every 30 minutes produced no significant improvement over the 60 minute message. However, habituation to these messages was noted as a potential problem to the maintenance of this effect. Relevant to this finding is another laboratory study involving 92 university students which showed that participants recalled translucent, dynamic messages scrolling across the EGM screens to a significantly greater extent than government-mandated static messages (Monaghan & Blaszczynski, 2007).

Cloutier, Ladouceur, and Sevigny (2006) found that pop-up messages regarding erroneous beliefs produced a significant decrease in these targeted beliefs at post-test. Jardin & Edelgard (2009) found that college students given occasional messages that correctly described the random nature of a computerized game spent less virtual money gambling, played fewer trials, and were more likely to quit when they still had money remaining. An electronic banner message indicating that VLT outcomes were random produced a significant two week decrease in hours playing VLTs (from 8.59 hours to 6.91 hours) in 54 Canadian VLT players, with this impact being greater for problem gamblers (Gallagher, Nicki, Otteson, & Elliot, 2011). However, this behavioural impact was not maintained two weeks after the banner ended, and only 64.3% of participants participated in both the post-banner and follow-up assessments. Similarly, the animated video produced by the Ontario Problem Gambling Research Centre and the Ontario Lottery and Gaming Corporation was able to reduce the frequency of exceeding self-set financial limits when measured immediately after the viewing. Although this impact may constitute “proof of principle” evidence that behaviour change can occur, it was not sustained over a period of 30 days (Wohl et al., 2010).

Researchers are subsequently assessing modifications to the intervention and to the gambling environment (e.g., removing ATMs from the slots floor) as strategies to sustain the behavioural impact. There is at least one study that has not found any impact of pop-up messages. Wynne & Stinchfield (2004) found that pop-up reminders indicating how much time the person has played did not influence the amount of money spent on VLTs.

59

On-Screen Clocks Playing longer than planned is a risk practice that increased the likelihood of developing impaired control over gambling. Accordingly, researchers have explored the value of introducing clocks into play. Schellinck and Schrans (2002) found that an on-screen clock was associated with improvements in keeping track of time and staying within desired time limits, but had no effect in reducing session length or expenditure. An on-screen clock also had no effect on gambling behaviour in a study by Wynne and Stinchfield (2004). EGM gamblers in a Quebec study reported that having a clock was not a helpful tool in promoting responsible gambling (Ladouceur & Sevigny, 2009). Mandatory Cash-Out A recent naturalistic study of EGM play found a logarithmic distribution of time spent playing EGMs, with a median gambling duration of 13 minutes, an average of 24 minutes and a maximum of 525 minutes (Townshend & Stansfield, 2007). Only 10% of players play for one hour or more, which suggests that time limitations or mandatory cash-outs may have some theoretical utility. However, the only formal research on this issue is a study by Schellinck and Schrans (2002) which found that mandatory cash-out after 145 minutes of continuous play did not alter average EGM session length or expenditure. To date, shorter cash-out periods have not been investigated.

The Netherlands has a mandatory cash-out when €40 or more has been won on EGMs located in restaurants. Privacy There is conflicting evidence about the effects of privacy. Video lottery terminal players reported that placing VLTs in isolated areas of bars and taverns likely contributed to impaired control in a study by Ladouceur, Jacque, Sevigny, and Cantinotti (2005). However, a laboratory test using these same people did not find that a secluded setting increased their gambling. Similarly, Lucas & Roehl (2002) found that video poker machines located centrally in a Las Vegas casino had significantly higher revenue compared to machines in all other locations, which they attributed to central machines being more easily accessed and having higher traffic volume. On the other hand, a laboratory study by Mishra, Morgan, Lalumiere, & Williams (2010) found that having someone directly observe their VLT play resulted in a very significant reduction in the amount of time played among 180 male gamblers. The effect was similar for male and female observers, and occurred for both problem and non-problem gamblers. A laboratory study by Rockloff & Greer (2011) had a similar finding in that participants made smaller bet sizes and had greater final payouts when their EGM play was being observed compared to when it was not.

60

Money versus Credits Although it is believed that the use of credit and debit cards increases spending relative to actual cash, there is not much evidence on this topic with respect to EGMs. EGM gamblers in a Quebec study reported that having a cash display was a helpful tool in promoting responsible gambling (Ladouceur & Sevigny, 2009). Similarly, Loba et al. (2002) found that pathological gamblers reported that a money won/lost display was superior to a credit display in helping to stop playing. However, there is a lack of behavioural evidence to support these beliefs. For example, money (versus credit) counters on VLTs were not found to influence gambling behaviour in a study by Wynne and Stinchfield (2004). In Canada, although all EGMs outside of casinos (i.e., ‘VLTs’) pay winnings via a credit slip rather than cash, there tends to be very little difference in their revenue generation compared to cash-paying slot machines inside gambling venues (Canadian Gambling Digest, 2004; Williams, Belanger, & Arthur, 2011). In the Australian Capital Territory, policies to restrict EGM cash payments to winnings less than $1,000 simply caused gamblers to cash out their winnings before the $1,000 limit was reached and then resume play (McMillen & Pitt, 2005).

Lights and Sounds Although there is considerable speculation and observational commentary about the effects of lights and sounds on EGM behaviour (e.g., Griffiths, 1993; Griffiths & Parke, 2003), there is relatively little empirical research. There is some evidence that alterations of light and sound are related to subjective enjoyment of EGM gambling (e.g., Loba et al., 2001). In one of the few empirical studies, Delfabbro, Falzon, and Ingram (2005) found that EGMs with 35% lower illumination significantly increased either the number of plays or time spent playing. However, the presence or absence of sound did not influence gambling behaviour. Further investigation is needed to determine whether such features have differential effect on problem gambler subtypes such as those posited by Blaszczynski and Nower (2002). The ‘emotionally vulnerable’ sub-type, for example, gambles to cope with negative affect and may be drawn to more calming environments and games, whereas the ‘impulsive anti-social’ sub-type may seek more stimulating environments. Some jurisdictions (e.g., Netherlands, Spain) prohibit sounds and flashing lights when a slot machine is not occupied (Gourdriaan, de Bruin, & Koeter, 2009; Becona, 2009). Seating Seating is commonly provided for EGM players in many countries, but is less common in the low stake low prize machines in Europe.

Although there is no research on the impact of having seating, it seems fairly obvious that the duration of EGM play would be shortened without it.

61

Summary of EGM Parameter Modifications In summary, the above research on EGM features has identified several measures that appear to have some potential to reduce harm. Such features include: eliminating early big wins (perhaps by decreasing maximum win size), slower speed of play, reducing the frequency of near misses, reducing the number of betting lines, eliminating bill acceptors, reducing the interactive nature of EGMs, presenting dynamic pop-up messages, reducing maximum bet size, and perhaps, reducing the availability of seating. There is insufficient evidence to support the utility of: varying payback rates, establishing a maximum win size, requiring mandatory cashouts, introducing on-screen clocks, substituting monetary for credit displays, providing less privacy, and manipulating ambient light and sound. There are two important caveats about this research. The first is that almost all of these studies have been conducted on people with prior EGM experience. Accordingly, the effectiveness of EGM parameter modifications as primary prevention tools, while plausible, is less certain. Second, the magnitude of the effects tends to be small.

The reality is that any automated device employing a variable ratio schedule (or more properly, random ratio schedule) with significant reinforcers and an event frequency of 5 seconds will tend to produce very strong behavioural patterns that are resistant to modification or extinction (Ferster & Skinner, 1957).35 Thus, EGMs will likely always be ‘high-risk’ devices with a strong association to problem gambling. In many ways, efforts to mitigate the harm of EGMs is reminiscent of the ultimately unsuccessful efforts to mitigate the harm of tobacco consumption by putting filters on cigarettes or promoting low-tar varieties of tobacco.

Player Pre-Commitment “Pre-commitment” refers to a strategy whereby pre-set limits on time, frequency, or money spent gambling are registered prior to the start of play. Pre-commitment is believed to be a useful harm minimization strategy because it allows players to make rational decisions about gambling involvement prior to actually engaging in gambling and obliges them to retain these limits despite subsequent temptations that arise during play (Parke, Rigbye, & Parke, 2008). Research indicates it is fairly common for regular gamblers (including problem gamblers) to have budgetary limits in mind prior to gambling but to exceed these limits as play progresses (McDonnell-Phillips, 2006; Focal Research Consultants, 2010). 35

Indeed, there are many who would argue that operant conditioning is the main theoretical framework with which to understand EGM play (Delfabbro, Falzon, & Ingram, 2005; Delfabbro & Winefield, 1999a,1999b; Dickerson, 1979; Dickerson, 1993; Dickerson, Cunningham, Legg-England, & Hinchy, 1991; Dickerson, Hinchy, England, et al., 1992; Dixon & Schreiber, 2002; Knapp, 1976; Petry & Roll, 2001). However, while principles of learning (both operant and classical) are likely of fundamental importance, it is clear that they are insufficient on their own. While pigeons pecking keys for food under a variable ratio schedule will all develop very persistent behaviour, only a minority of people who play EGMs find them appealing and continue playing them. Secondly, the evolutionary purpose of operant conditioning is to shape behaviour so as to optimize returns. Excessive EGM play (i.e., problem gambling) is not an adaptive response to the environmental contingencies these machines offer.

62

Online gambling is very well suited for pre-commitment, as unlike most types of land-based gambling, all activity is electronically recorded and linked to an identifiable individual. Consequently, pre-commitment has been offered for several years on a small portion of the ~2,400 existing online gambling sites (Griffiths, 2012; Williams, Wood & Parke, 2012ab). The pre-commitment parameters available to online gamblers vary depending on the site. Most common are deposit limits (e.g., daily, weekly, monthly); bet size limits; loss limits (e.g., weekly, monthly, yearly); short-term exclusion from certain game types (e.g., 6 months); and shortterm account suspension (e.g., 6 months). For most sites, deposit and loss limits tend to be revocable, but account suspension irrevocable (Williams, 2010). Some of the voluntary online gambling regulatory bodies (e.g., e-Commerce and Online Gaming Regulation and Assurance (eCOGRA )) require some form of pre-commitment for site accreditation. Currently, however, there is no system in operation which links all Internet gambling sites within a jurisdiction, or across jurisdictions, with an integrated self-exclusion or pre-commitment regime (Williams, 2010). Accordingly, all player-determined parameters and limits can be ignored simply by changing sites. For land-based gambling, pre-commitment is most often implemented using a ‘smart card’.

A smart card is any pocket-sized plastic card with embedded integrated circuits providing some limited memory and/or microprocessor capabilities when interacting with external card-reading devices. Smart cards have often been used for Player Loyalty/Reward cards and/or as debit cards for cashless gambling. Sometimes these same cards are programmed to allow precommitment and sometimes the pre-commitment card is a different card altogether. In 2002 the Crown Casino, in Melbourne, Victoria was the first casino in the world to introduce player pre-commitment monetary limits on a smart card for EGM play. Since 2002 smart cards have been introduced throughout Victoria, as well as Queensland, New South Wales, Nova Scotia, New Zealand, Sweden, Denmark, and Norway (Regis Controls, 2009; Williams, 2010). Precommitment is for EGMs only in Victoria, Queensland, and New South Wales; for VLTs only in Nova Scotia; for online lotteries only in New Zealand; for online poker and sports betting, as well as land-based bingo, lotteries, and sports betting in Sweden; and for all forms of online and land-based gambling in Norway.

There has been some research on the effectiveness of pre-commitment:  Schrans, Grace and Schellinck (2004) found that a feature allowing players to set a time limit on their VLT play was only effective in influencing one of the six behaviours being targeted for improvement.  Anecdotally, problems have been expressed about the effectiveness of pre-commitment when non-card EGMs are also readily available (as is the case in New South Wales (NSW)) (Nisbet, 2005). There is also a concern that because of the significant amount initially put on the card (e.g., $200 in NSW) gamblers may increase spending, either due to more money being readily available or because they require less embarrassing interactions with cashiers and other venue staff (Parke, Rigbye, & Parke, 2008). That being said, people who have opted to use cards for pre-commitment and have agreed to report on their use tend to indicate that the card helped them better manage their spending (Nisbet, 2005).

63







Focal Research Consultants (2007) tracked VLT play for a 6 month period in a region of Nova Scotia that only had player-card activated machines available (~51 EGMs in 9 locations played by 1,824 players). Roughly 71% of regular players (i.e., those playing once a month or more) opted to try one of the responsible gambling (RG) features included on the cards (i.e., spending limit, time limit, 2 day exclusion, or playing history reports). Roughly 65% of these people continued to use one or more RG features in subsequent sessions. A subsample of these RG adopters (n = 122) had a baseline period of non-RG use that allowed for a pre-post comparison. These individuals were found to have a significant decrease in per session expenditure ($47 to $40), an increase in play length (82 min to 98 min), and no change in frequency of play per month (9.3 to 9.3). Examination of individuals with high risk characteristics found no decrease in expenditure for high frequency players (18+ times in 6 months). People with Canadian Problem Gambling Index (CPGI) scores of 5 or higher showed a tendency toward decreased per session expenditure, but this outcome was offset by a tendency toward increased frequency of play.

It was also found that approximately 37% of players swapped cards and/or obtained them from venue staff to circumvent the system, with this being particularly true for people with higher CPGI scores (Bernhard et al., 2006; Omnifacts Bristol Research, 2007). Follow-up research by Focal Research Consultants (2010) on the roll-out of the province-wide pre-commitment found that although most gamblers reported that they needed pre-commitment (endorsed by 98% of problem gamblers), only a minority planned to use it (as adoption was voluntary). This turned out to be an accurate prediction, as only 2,740 players were voluntarily enrolled in the “My-Play” pre-commitment system by December 2010 (roughly 2.8% of VLT players). The program was subsequently made mandatory in April 2012. Results from the 2009 Worldsmart South Australian (SA) trial showed that fewer than 1% of loyalty card holders had voluntarily activated the pre-commitment options on their card (Productivity Commission, 2010). For those who did, the most popular options were limits on: daily spending (59% of cards); breaks in play (19%); weekly spending (14%); monthly spending (12%); daily duration of play (10%); monthly duration of play (8%); fortnight spending (6%); weekly duration of play (4%); fortnight duration of play (4%); and spending history (3%). Utilization of pre-commitment features tended to increase with time.

Among the 94 individuals with 3 months of baseline data, there was a 25% reduction in daily turnover subsequent to utilization of pre-commitment. However, it is possible these individuals continued to gamble and simply decreased use of their voluntary card. There were also 600 instances where players exceeded their pre-established limits (in the SA system, when a limit has been met it can be overridden by venue staff). It is not known what percentage of pre-commitment breaches this 600 figure represents (Productivity Commission, 2010). Results from the 2009 Queensland Maxgaming’s Simplay system showed that only 15% of Simplay cardholders used pre-commitment, although this percentage may be higher for atrisk groups (Productivity Commission, 2010). The most popular option (13%) was a daily spending limit. No one set playing time limits. There was tentative evidence that expenditure may have decreased subsequent to limit setting. However, here again, this assumes that players did not continue playing using cash (i.e., once a limit was exceeded the only consequence was the inability to use the card for cashless gambling). Since the system

64

 

has been expanded throughout Queensland approximately 14,000 people have opted to use this voluntary system. Of these, 9% have set spending limits (with this percentage increasing over time). Results from the Odyssey trial in Queensland were similar, with only 5% of players opting into cashless gambling (that would also allow pre-commitment), and 28% of these individuals opting to set a daily spending limit. Sixty percent of users reported that pre-commitment was useful, with this percentage being higher for high-risk players. There was also a 40% reduction in spending by players who set limits compared to a 3% reduction in players who did not set limits. However, this reduction might be offset by increased cash-based play, which continued to be available (Productivity Commission, 2010). Because of the mandatory nature of the Swedish and Norwegian systems, a large percentage of the gambling population has obtained smart cards (in 2008 this was roughly 1.3 million Swedes and 1.9 million Norwegians) (Responsible Gambling Council, 2009). The Internet Poker Committee (2008) (cited in Responsible Gambling Council, 2009) surveyed approximately 3,000 participants of the Swedish online poker pre-commitment system.

Participants reported that a) monetary limits were more useful than time limits; b) 1/3 of players set monetary limits that were excessively high and 40% set time limits that were excessively high (essentially disabling the pre-commitment system); c) for individuals who hit their limits, 37% went to another online site and 32% simply changed their limits; d) 5% of players barred themselves for a period of time, with one week being the most common length (25% of people who barred themselves also began playing online poker at other sites); e) 26% used the risk assessment option and 52% of these individuals found it useful.

In summary, there does not exist an abundance of research on the effectiveness of precommitment, but that which does exist indicates that it holds promise as a harm-minimization technique (Griffiths, 2012; Parke, Rigbye, & Parke, 2008; Productivity Commission, 2010; Williams, 2010). Part of the problem in evaluating pre-commitment arises from the many different ways it can be implemented (i.e., mandatory or voluntary; exceedable vs. nonexceedable limits; revocable vs. non-revocable limits; short vs. long duration of limits; presence on all or just some EGMs; etc.). Similar to research on other problem gambling prevention initiatives, the ‘devil is in the details’ and the actual effectiveness of a technique is usually very much dependent on how it is applied. Drawing on the lessons of problem gambling prevention research, it is reasonable to surmise that the degree to which pre-commitment is voluntary36, revocable, exceedable, of short duration, limited to some (but not all) EGMs or some Internet sites, limited to some (but not all) forms of gambling, and does not use biometric ID or some other form of identification that deters identity swapping, is the degree to which the utility of the technique is primarily limited to non-problem gamblers (which may or may not translate into a decreased future incidence of problem gambling). In contrast, the degree to which all