2. Access to the Proposition
2.3 Gambling and the Availability Hypothesis
2.3.1 Studying the Availability Hypothesis
Large-scale interviews or surveys in Canada (Cox et al., 2005), New Zealand (Abbott
& Volberg, 1996), Norway (Lund, 2008), and Sweden (Abbott, Volberg & Ronnberg, 2004) have been used to study the availability hypothesis. Despite gambling being available and legalised in NZ for decades, the construction of casinos in the late 1980s saw a rise in per-capita expenditure on gambling, doubling from 1988 to 1990 (Abbott & Volberg, 1996). In addition to increased availability of gambling, Abbott and Volberg (1996) noted changes in the public awareness of problem gambling, with a rise in the number of people agreeing that gambling could be a problem for some people (up from 66% to 71%) and that such people should be able to access help services if they wished to give up gambling (from 86%-91%).
In a national survey in New Zealand which included 3,933 adults aged 18 and above, Abbott and Volberg (1996) found that problem gamblers were more likely to regularly participate (once per week or more) in a wider variety of forms of gambling than
non-problem gamblers. In addition to this, they also spent more money and displayed a preference for continuous forms of gambling compared to non-problem gamblers.
In Canada, the 1990s saw a spread in the availability of different forms of legalised gambling including casinos and electronic gaming machines (EGMs). Cox et al. (2005) interviewed 34,770 Canadian’s aged 15 and over using the CPGI to compare problems across 10 provinces of the country. They found that the highest rates of gambling problems were seen in provinces with highest concentration of EGMs in addition to permanent casinos. Four out of five provinces that had both EGMs and casinos produced the four highest prevalence figures for problem gambling. However, this study only looked at macro-relationships and did not consider the amount of time spent on EGMs or other variables which may lead to at risk behaviour such as the concurrent consumption of alcohol.
Delfabbro (2008) examined the effect of a 15% reduction in the number of EGMs in South Australia. Data were obtained from 594 EGM venues. The removal of around 2000 EGMs in 2005 had very little impact upon levels of EGM expenditure. About 50% of participants reported difficulty gaining access to an EGM and about a quarter felt that this reduced their urge to gamble. Although machines might have been more difficult to find, there was little indication that the frequency, time and effort devoted to gambling decreased.
However the absence of effects may reflect the magnitude of the reduction in the number of machines or the one year time frame of the study.
In the US, it has been agued that the legalisation of gambling was due in part to the financial troubles many states were experiencing in the 1970s (Volberg, 1994). Gambling has been introduced in many states in the US in a more staggered fashion beginning with state lotteries, followed by scratchies (1970s), followed by card rooms (1980s), then by the late 1980s, riverboat casinos, EGMs, and Indian Reservation casinos. Volberg (1994) conducted a prevalence study (similar to the Canadian one described above) to look at issues of the availability of different forms of gambling in different states relative to the number of problem pathological gamblers and also the health services offered within each state.
Relationships between availability and problem were found. For states with greater access to gambling, there was a greater incidence of problem gambling. Volberg found that eastern states (including Massachusetts and New Jersey) and California had much higher prevalence rates of pathological gambling (2.3%, 1.4% and 1.2% respectively) compared with Iowa (0.1%). Volberg (1994) argued that this difference in prevalence rates was a function of both availability and time. That is, legalised gambling was more common in the states with a higher prevalence of problem gamblers and legalised gambling had been available for over 20 years in these states, compared with less than 10 years for Iowa.
Increases in the availability of gambling have occurred in much of the western world including North America, Europe, New Zealand and Australia. Despite a longer history of gambling in Australia (O'Hara, 1988), increases in availability of legal gambling in Australia have also occurred. Gambling in Australia previously consisted primarily of lotteries and racing, and that increases in the availability of gambling occurred in part because of technological developments (Productivity Commission, 1999). Gambling became more accessible with the availability of local TAB and EGM outlets in suburbia while electronic gambling devices allowed for faster play, and casinos allowed higher bets. As there have been increases in the number of people reporting problems resulting from their gambling, researchers have suggested that an epidemiological or public health approach to gambling may be appropriate (Chipper, Govoni & Roerecke, 2006).
Lund (2008) described the association between increases in ‘consumption’ of gambling and the proportion of heavy gamblers as conforming to single distribution theory.
Ledermann originally proposed this in relation to alcohol consumption in France during World War II, where decreases in the availability of alcohol led to decreases in alcohol-related problems. Rose (1985; 1982) further developed this idea and applied the concept to both sick individuals e.g. those with high blood pressure and also to individuals who drank too much alcohol causing health problems. Ledermann observed that the distribution of alcohol consumption could be described by a lognormal curve that was positively skewed (skewed to the right tail) (Chipper et al., 2006). The distribution curve has since been described as uni-modal, positively skewed and approaching normal when log transformed (Chipper et al., 2006; Grun & McKeigue, 2000; Lund, 2008). Regarding a Ledermann curve, the number of people who have a problem (the tail of the distribution) depends upon the average level of consumption or use in the relevant population.
If consumption within the population has a skewed distribution, then changes in the mean level of consumption on a population level can result in significant changes to the heavy consumers but only slight changes in the rest of the distribution (Chipper et al., 2006;
Lund, 2008). Grun and McKeigue (2000) argue that if the single distribution theory applies to gambling, there should be differences between communities in the frequency of excessive gamblers or gambling that correlate with measures of central tendency for those particular communities.
One way to determine if the single distribution theory applies is to examine how changes in governmental policy such as the introduction of a National Lottery, influences gambler behaviour (Grun & McKeigue, 2000). Grun and McKeigue (2000) examined expenditure on gambling using data from the Family Expenditure Survey before and after 1994, when the National Lottery was introduced in the UK. They found that the mean gambling expenditure rose from £1.45 a week (equivalent to 0.5% of household income) prior to the introduction of the lottery in 1993-1994 to £3.81 a week (equivalent to 1.5% of household income) in 1995-1996 and that this was almost exclusively due to expenditure on the UK lottery. Furthermore, the number of households gambling also increased from 40% to 75% after the introduction of the lottery. To examine potential problem behaviours, Grun and McKeigue (2000) then looked at the number of households spending more than £20 a week which increased from 0.8% of households to 3.2% of households after the introduction of the lottery) or more than 10% of their household income (almost a 4-fold increase). These findings provide support that the single distribution theory can be applied to gambling because there is a close relationship between the average expenditure level on gambling in the community and the proportion of gambling that may be regarded as excessive.
Furthermore, this study also found that increases in gambling problems exceeded the increases in the averages amounts of gambling. A strength of this paper was to define excessive gambling as based upon household finance, as any financial hardship resulting from problem gambling is likely to affect more than just the individual concerned (Grun &
McKeigue, 2000). However, a limitation in this paper was the reliance of using the Family Expenditure Survey as an index of gambling problems rather than a more specific gambling tool such as the CPGI or SOGS which are well validated tools and include items in addition to financial ones in relation to problem gambling.
Another study in Canada by Room, Turner and Ialomiteanu (1999) used telephone surveys to compare respondents in Niagara Falls before and after the opening of a Casino in 1996 and compared them with responses from control groups of participants living in the wider state on Ontario. This study included questions about individual’s gambling behaviour, their gambling problems reported using the short SOGS, and responses about other people’s perceptions of their gambling and whether any close friends or relatives had gambling problems. Room et al. (1999) found that although gambling in a casino increased in Ontario as a whole (11%) after the opening of the casino, that there was an even greater increase in the percentage of Niagara residents who had gambled in a casino after the survey (43%), and the average reported spending at casinos also significantly increased from $2.30 in a month to
$11.10. Increases in the short SOGS also increased in the Niagara residents, with the
proportion scoring 2 or more rising from 2.5% to 4.4%, and the proportion scoring 3 or more rose from 0.7% to 2.3%, with the authors using a cut-off score of 2 to indicate a problem gambler. However, as only the short measure of the SOGS was used no more information about problem gambling status could be ascertained from this study. In contrast, the problem gambling status of residents of Ontario (the control group) remained stable after the opening of the casino in Niagara. This paper demonstrated both increases in the amounts and
frequency of visits to casinos following opening in Niagara and also an increase in problems arising from gambling providing support for the single distribution theory. Whether such increases in money spent or the number of people reporting problems with their gambling post-one year is similar or not is not known from this study.
Chipper et al. (2006) tested the applicability of the single distribution theory using archival survey data from 3,554 participants in Ontario, Canada. Their study sought to
determine if the population conformed to the features proposed by the single distribution theory, to determine the best measure of problem gambling that was associated with problem gambling and to determine the relationship between gambling ‘consumption’ and the risk of having a gambling-related problem. They found support for the theory with consumption of gambling and the amount of money being spent on gambling being highly skewed, and being
‘normal’ when log transformed. Multivariate logistic regression was then used to show that an individual’s level of consumption of gambling was important as the odds of having problems associated with gambling (measured by CPGI) were greater for the fourth and highest quartile. Regarding the best measure of problem gambling, similar to the study by Grun and McKeigue (2000), it was found that the percentage of household income was the best predictor of gambling problems. This may be because this measure incorporates both a measure of the actual amount of money spent on gambling and the resources available for the individual to gamble (Chipper et al., 2006).
Lund (2008) has also found some support for the single distribution theory in relation to gambling by comparing three data sets of surveys (including a national survey of gambling in 2002, a school based survey in 2004 and a postal national survey in 2005). Lund (2008) found that the statistical assumptions of a Ledermann curve were satisfied (uni-modal, skewed right hand tail) in relation to gambling in Norway. In this study, there was a positive correlation between the mean gambling frequency and the proportion of frequent gamblers in each of the three surveys and Lund (2008) argues that this has implication for government policy regarding changes to gambling regulations. Lund argued that it is important to consider changes at the community level not just changes that target particular vulnerable groups who may develop gambling problems. To this end, Lund (2008) suggests that restricting availability (opening/closing hours, number of available games, restrictions on maximum bet) may be worthwhile policies for governments to take in relation to gambling.
However, other authors (see Chipper et al., 2006) have suggested different approaches such as recommending ‘safe levels of consumption’ defined by percentage of house-hold income, in a similar way that governments help to prevent alcohol abuse and binging by proposing a safe level on the number of drinks that males and females can consume within a session.
Despite this, the relationship between increases in availability and problem gambling prevalence is not a simple linear one (Abbott, 2004). For example, in Australia the
Productivity Commission’s report found that with a number of different availability measures (e.g. EGM’s per 1000 adults, EGM expenditure per adult and total continuous expenditure per adult) that there was a higher prevalence of problem gambling in Australian states where there was increased accessibility and expenditure (Productivity Commission, 1999), the report also showed that while the states with the highest availability (more EGMs and expenditure on gambling) had higher rates of problems associated with gambling, that increased availability within these states was not associated with increases in problem gambling prevalence (Abbott, 2001; Productivity Commission, 1999). Abbott, Williams and Volberg (1999) looked at problem gamblers after a period of seven years and suggested that problems related to gambling within a community may level out over time because of greater public awareness of potential problems that may be caused by gambling. There is likely to be more services and industry controls after the introduction of new gambling forms and that there is increased knowledge within the community of warning signs of problem gambling through advertising and public information campaigns. However, more research is needed addressing changes in problems related to gambling over time, especially in different
jurisdictions, to determine how quickly these factors have an influence on problem gambling prevalence, and their relative influence compared with one another (Abbott, 2004). Abbott
(2006) concluded that the effects of exposure are complex and require considerations of the role of the individual and environmental risk and protective factors. Indeed, some
researchers have likened the use of technologies such the internet to that of fads such as Citizen Band radios (Grohol, 1999).