Adoption, Adaptation or Exposure? Novel Digital Gambling Activities and Links with Gambling Problems

Gambling problems are much more common amongst people who use novel gambling products, including skin gambling, esports betting and fantasy sports betting. The exposure and adaptation effects suggest that, like a novel pathogen, new products produce gambling problems until adaptation can counteract vulnerabilities. The purpose of this review is to posit, based on recent data, that an adoption effect provides a more parsimonious account for why novel gambling products are associated with problems. People who are more vulnerable to a gambling problem are more likely to adopt novel gambling forms, consistent with their tendency to gamble on many forms. The high prevalence of gambling problems cannot be accounted for by the exposure effect alone, unless one assumes, implausibly, that these novel forms are dramatically more toxic than established products. As newer products diffuse in their acceptance throughout the population, the proportion of people with a gambling problem who play such games goes down. The evolution of gambling products can be described by a product lifecycle model where gambling problems are common amongst early adopters, but proportionately decrease amongst users as a product grows in popularity, reaches maturity and finally stagnates and is replaced. The adoption effect is important because it indicates the need for interventions when new gambling products are introduced. It cautions against assuming that interventions are working just because the prevalence of gambling problems declines.


Introduction
Several novel forms of digital gambling have become available in recent years. Esports betting enables people to bet on professional video game competitions. Skin gambling involves the use of skins acquired in video games to bet on competitive events and games of chance. Fantasy sports betting allows participants to compete for prize pools by assembling virtual teams of professional sports players whose performance is based on their real-world play. These three novel products currently attract relatively small markets. For example, amongst Australian adults, past-year prevalence is 0.6% for esports betting, 0.6% for fantasy sports betting and 0.3% for skin gambling [1••]. These activities attract mainly young adult males [2][3][4][5][6][7]. Most studies have found positive associations between participation in these activities and higher problem gambling severity, although causal relationships remain unclear [4,[8][9][10].
In marketing terms, these novel products are in an early stage of their product lifecycle, and, by definition, current users are early adopters [11]. Derived from Rogers' [12] seminal diffusion of innovations theory, which explains the adoption of innovations over time throughout a population or social system, early adopters are individuals who take up an innovative product, practice or technology before the majority of the population [12]. Research across multiple product categories has found that early adopters tend to differ from later adopters on a range of characteristics. These include greater prior involvement with similar products, higher technical/digital skills, higher risk-taking/novelty-seeking tendencies and peer networks who use and recommend the innovation [13•]. Theoretically, it follows that early adopters of novel gambling products are likely to be highly involved gamblers, who have prior experience of digital games and online gambling and who have social networks who also engage in these activities. These characteristics suggest that gambling problems are likely to be elevated amongst these early adopters, since higher gambling intensity and involvement, online gambling and peer gambling influences are associated with more harmful levels of gambling [14][15][16][17]. Accordingly, in the early adoption stage, participation in these novel activities may be skewed towards vulnerable gamblers.
This concentration of vulnerability amongst early adopters has parallels in epidemiology. Following exposure to a toxin or virus, individuals who are most susceptible to the disease are the first in a population to become ill [18]. Epidemiological (epi) curves typically indicate an initial sharp increase in case numbers as the most vulnerable people become ill, followed by a slowing rate of a new disease that reflects greater resistance amongst less vulnerable people [19]. Gambling research has drawn on epidemiological models to point to an exposure effect and an adaptation effect on the epi curve of gambling problems over time [20, 21, 22••, 23, 24••]. Based on patterns observed in epidemiological and innovation models, we suggest that a third effect may also be present, which we call the adoption effect. This paper aims to explain this effect, discuss how it may be reflected in rates of gambling problems amongst users of novel compared to more mature gambling products and encourage new lines of inquiry in this area. We argue that the adoption effect is parsimonious in explaining exposure and change over time within the one framework to explain how novel gambling products produce a spike in gambling problems that later decrease over time as gambling products mature.

Exposure and Adaptation Effects
The exposure effect posits that the introduction of harmful forms of gambling into a population brings exposure to a public health toxin that will increase gambling problems [20,21]. Developing a gambling problem is contingent on engaging in gambling, which in turn relies on its availability. Accordingly, a widely held assumption is that increased gambling availability will result in increased gambling problems [22••]. A meta-analysis of 119 studies and two reviews found that this assumption held true during the 1970s-1990s in markets where gambling was newly legalised [23, 24••, 25]. These analyses observed that the rapid expansion of commercial gambling, particularly the more harmful forms of electronic gaming machines (EGMs) and casino gambling, was accompanied by an increased prevalence of gambling problems.
However, subsequent studies in more mature markets have found that the exposure effect does not remain linear and may break down at a certain level, leading to questions over its durability across time points and locations [20, 22 ••]. Researchers have observed that the prevalence of gambling problems does not remain directly correlated with indicators of exposure to gambling in a population, such as the number of venues or duration of exposure [20, 22••, 24••, 25, 26, 27•]. For example, using a Regional Index of Gambling Exposure, a comparison of several US states found a curvilinear relationship between exposure and problem gambling [20,21]. Like many epidemics, a flattening of the curve has been observed where the prevalence of gambling problems stabilises or even declines after some time, despite continued exposure to gambling in the population [20, 28••, 29]. One explanation for this phenomenon is the adaptation effect. This posits that, over time, individuals in an exposed population adapt by building resistance and immunity, leading to a decline in gambling problems [20, 27•]. This adaptation may be due to several factors, including "social learning, waning of novelty effects, increases in harmful consequences [which may prompt people to moderate their gambling], developed interventions, and new interests that preclude engaging in the initially harmful activity" [20]. Increased public health measures may also improve protection for individuals, such as public health messaging, self-regulatory tools and treatment options [30].
It is now widely accepted in the gambling literature that both exposure and adaptation occur. During initial exposure to new forms of harmful gambling, the population prevalence of gambling problems increases, followed by a plateauing or decline as adaptation is said to occur. Naturally, this represents a broad pattern, and some variations in specific settings are expected depending on factors such as community vulnerability, the mix of gambling products and degree of market saturation [20, 31••].

Adoption Effect
Notwithstanding the possibilities of exposure and adaptation effects, we propose a third possible explanation for the flattening of the curve for gambling problems. We refer to this as the adoption effect. Supported by innovation and epidemiological models, this effect may occur because early adopters of gambling products are likely to be more susceptible to gambling problems, while later adopters have greater resistance. As the diffusion of a new gambling product in the population occurs in later stages of its product lifecycle, the greater resistance of later adopters to gambling problems might help to explain the flattening of the curve. Individual adaptation and public health measures may not fully explain the stabilising or declining prevalence of gambling problems over time. In fact, these factors may not be sufficient or even necessary explanations. Instead, the mainstreaming of gambling products in the maturity stage of their lifecycle might dilute the problem gambling rate found amongst early adopters as more resistant people start to engage in the activity and make up a larger proportion of users. Figure 1 depicts a hypothetical example of the adoption effect for a gambling product as it progresses through the lifecycle stages of market development, growth, maturity and stagnation [11]. During this progression, the proportion of consumers with a moderate to severe gambling problem (the vertical bars in Fig. 1, to be read with the reference scale on the right-hand side) is highest in the market development stage, when the consumer base is comprised of the more vulnerable early adopters. This proportion then decreases in later stages, as less vulnerable later adopters start engaging with the product and dilute the prevalence of gambling problems in the customer base. Thus, for instance, initially, 80% of people playing the novel product have pre-existing moderate or high-risk gambling problems because they are early adopters. At the final period, however, less than 20% of people using the product have MR or PG status problems. That is, the lower percentage of PG/MR gamblers using the "no-longer-novel" product (80% vs ~ 20%) is principally a consequence of the introduction of new players with lessor or no problems as the product matures. The lines shown, in contrast, reference the left-hand scale of the figure and show how the absolute number of people in each gambling-risk category who are using the product changes over time. The number of PG/MRs barely changes over time, whereas the NP/LR gamblers start adopting the product later.
In summary, we propose that, in addition to the potential exposure and adaptation effects, the adoption effect may also be a plausible contributor to the epi curves observed for gambling problems. Below, we present data from two of our Australian surveys that provide preliminary support that the adoption effect deserves consideration in gambling research.

Sample 1: Nationally Representative Telephone Survey of Australian Adults
This 2019 survey of Australian adults (N = 15,000) used a random digit dialling sample provided by a national database of mobile phone prefixes. The representativeness of the sample was further improved by weighting to key population demographics. The detailed methodology is explained in the research report [1••]. All respondents were asked whether they had gambled on each of 13 activities in the past 12 months. The Problem Gambling Severity Index was administered, with scores categorised as 0 = non-problem gambling, 1-2 = low-risk gambling, 3-7 = moderate-risk gambling and 8-27 = problem gambling [33]. Table 1 shows the proportion of respondents who had gambled on each of eight gambling forms who were in the problem gambling/moderate-risk (PG/MR) categories. These eight forms include five mature gambling products that are most often linked to gambling problems, as well as the three novel forms [14][15][16]34]. Gambling on all forms was significantly associated with PG/MR. However, Fig. 1 Hypothetical example of the adoption effect for a gambling product compared to the five established gambling activities that are in later stages of their product lifecycle, considerably higher proportions of gamblers on the three novel gambling products were in the PG/MR categories: skin gambling (43.5%), esports betting (46.2%) and fantasy sports betting (50.0%).
While based on small numbers of gamblers on the three novel gambling activities, Table 1 supports our expectation that products in the early adoption stage attract a customer base that is skewed towards those with higher problem gambling severity. In contrast, products in more mature stages of development, such as EGMs, casino games and race and sports betting, attract lower proportions of PG/MRs, consistent with greater resistance amongst later adopters who comprise the majority of their customers.

Sample 2: National Online Survey of Australian Gamblers
This survey recruited a non-probability sample of Australian adults who reported gambling at least once during 2019 (N = 5,019). Participants were recruited through a panel aggregator, Qualtrics (n = 4,060) and an in-house panel of respondents to our previous gambling studies (n = 959). Using a non-probability sample yielded much larger numbers of gamblers on each product than could be achieved in our national telephone survey. Questions on gambling participation and the PGSI were the same as in the national telephone survey. The research report contains full details of the methods [1 ••]. Table 2, drawn from [1••], shows the proportion in the PG/MR categories amongst respondents who had gambled on each of the eight gambling activities in 2019. While gambling on all forms was significantly associated with PG/ MR, gamblers on the three novel gambling products were predominantly in the PG/MR categories: skin gambling (81.5%), esports betting (70.7%) and fantasy sports betting (70.6%). These proportions were considerably higher than for the other more established forms of gambling.
The data in Table 2, based on much larger subsamples of gamblers, demonstrate the same pattern evident in Table 1that the customer base for each novel gambling form has considerably higher PG/MR rates compared to those who gamble on more mature products. If gambling is the "toxin" that comprises exposure and thereafter adaptation, it is less likely that there should be such dramatic differences by product during the exposure cycle. Instead, we suggest that adoption is a simpler explanation. Novel products attract adoption by the most at-risk players in the population, and only later adopters dilute the proportion of players who are vulnerable over time.

Discussion
Our finding that early adopters of the three novel gambling products in Australia have substantially higher rates of gambling problems, compared to users of more mature gambling products, suggests that an adoption effect is possible. It is not known whether all novel gambling forms or innovations disproportionately attract an early market with a high prevalence of PG/MR, or whether the adoption effect applies most to those products with features known to increase the risk of problem gambling. A recent meta-analysis of 104 studies found that the strongest risk factors for problem gambling  are online gambling and continuous-play gambling products [32]. These features are present in the three novel forms examined.

Evidence for the Adoption Effect
The existence of an adoption effect may not completely displace alternative explanations such as the exposure and adaptation effects. Consistent with the exposure effect, the introduction of the three novel forms of gambling has been accompanied by a high prevalence of gambling problems amongst a newly exposed customer group. However, this elevated prevalence cannot be explained by the exposure effect alone, and in fact, adoption may be an alternative explanation. If gambling problems are purely a function of exposure to a harmful gambling product, the relative prevalence of PG/MR amongst a product's users indicates that the three novel forms must be incredibly more dangerous than the other forms. It is difficult to accept, for example, that novel forms such as fantasy sports betting are vastly more toxic than EGMs, which are known to be associated with excessive expenditure and present a substantial risk of dependence and harm to individual users [33][34][35][36]42].

Evidence for the Adaptation Effect
Our findings also appear to be at least consistent with an adaptation effect too since the lower PG/MR rates amongst the customer base for more mature products may indicate that individuals and populations have adapted to exposure to these products. At the population level, this adaptation may occur for the novel products in the future if public health interventions improve and provide better protection for users. In short, we cannot entirely dismiss that adaptation, and not just adoption, may influence the association between product use and gambling problems. At present in Australia, esports betting and fantasy sports betting are provided by licensed operators who offer selfregulatory tools, including limit-setting and self-exclusion, although many customers use unlicensed operators who may provide no consumer protection [1••]. The high prevalence of gambling problems amongst users in our two samples suggests these tools, where available, currently have limited effectiveness. Skin gambling is largely unregulated and currently offers no gambling harm minimisation measures, but these may improve with time as the market grows and attracts regulatory attention.
At the individual level, some people may also adapt in the future by curtailing their engagement in these novel forms as their novelty and honeymoon effects wane, or if individuals moderate their gambling on these products to reduce the harmful consequences they have experienced. Given the young average age of early adopters on the three novel gambling products, some individuals may age out of risky gambling behaviour [2][3][4][5][6][7]. This maturing out of addictions as young people transition to adult roles and responsibilities and experience psychological maturation has been observed for alcohol and other drug addictions [37][38][39]. Longitudinal studies have also observed patterns consistent with this natural maturation effect in relation to problem gambling [40,41]. Percentages are those who are classified as moderate risk or problem gamblers. If a percentage is, say, 35.1%, then the remaining 64.9% in that cell are classified as non-problem or low-risk gamblers. Ns in the "form" column refer to the number who participated in that form in the previous 12 months PG/MR problem or moderate risk gambler based on the PGSI

Which Effect Predominates?
While our analyses do not directly contest either the exposure or adaptation effects, we do however suggest that an adoption effect may be in play. In addition, we suggest that the adoption effect may be a simpler explanation that describes the pattern of effects observed in the data. That is, it is not necessary to propose two mechanisms, exposure followed by adaptation, to arrive at a theory of how novel gambling products produce a spike in gambling problems that later trails off over time. Instead, simple early adoption by vulnerable populations can explain both the early spikes in gambling-related problems in a product's lifecycle and a dilution over time in the proportion of people with gambling problems as less vulnerable people, being lateradopters, start using the product. Occam's razor suggests that the adoption effect is the most compact and thus compelling representation of what we have observed. The very high rates of PG/MR amongst current users of novel products point to a small customer base which is comprised of many individuals with little resistance to gambling problems. As well as possible adaptation to exposure over time, this customer base is also likely to expand to include the more resistant later adopters, which would dilute the current PG/MR rates. This effect may also partially explain the lower rates of gambling problems we found for the more mature products, whose customer base is comprised mainly of later adopters. Theoretically, this effect is supported by product innovation and epidemiological models. In short, the adoption effect might provide an alternative or additional explanation for the shape of epi curves for gambling problems.
Naturally, rigorous and extensive research is needed to test the hypothesis that the adoption effect at least partially, or even fully, explains the flattening of the curve of gambling problems over time as gambling products mature. For example, prospective longitudinal studies could examine trends in individual and population gambling problems as new products are introduced and mature. Retrospective longitudinal analysis might also be conducted where studies have tracked individuals over the early adoption and maturity stages of harmful gambling products. If the adoption effect holds true, we expect these analyses to find high rates of pre-existing PG/MR problems amongst new users when novel products are introduced, but lower rates of pre-existing problems amongst later-arriving new users when these products are in more mature stages of their lifecycle. This would show that most individuals do not adapt, but that products only appear safer over time by virtue of dilution. Alternatively, declines in PG/MR amongst early adopters who continue to gamble on these same products as they mature would instead support exposure and adaptation effects.

Conclusion
An adoption effect in gambling would have important implications for policy and practice. It would draw attention to the heightened vulnerability to gambling problems of early adopters, and to their increased potential risk of gambling harm from novel products that add to their pre-existing burdens. This increased risk emphasises the dangers of introducing novel products that have features that are strongly linked to problem gambling, including continuous-play products and online access [32], and that are likely to interact with psychological vulnerability to exacerbate gambling problems amongst early adopters. This indicates the need for consumer protection measures to accompany the introduction of these products instead of lagging behind. At present, vulnerable early adopters form a large part of the market at a time when these products have the least consumer protections. These early adopters also present a logical target group for interventions that encourage help-seeking, given the high concentration of gambling problems in this group. Further, if the adoption effect explains the flattening of the curve, this would indicate that individuals and populations do not adapt to the extent previously thought. This would caution against assuming that prevention and intervention measures are working just because the prevalence of gambling problems declines. In any case, the long-term reduction in the percentage of people experiencing harm from a product comes at a considerable cost. A dilution of poison is still poison. The high prevalence of gambling problems amongst early adopters, as found in our samples, indicates that the pool of early users and their families are likely to suffer serious harm before recovery might occur. Communities also bear costs from gambling harm while waiting for products to "appear" less harmful (even if this may never occur), for example, in the health, welfare and justice systems. Since continued technological re-invention is part of the business model of gambling, the pattern of heightened vulnerability at product introduction and innovation warrants the early provision of preventive public health measures to help ensure that novel products do not amplify gambling problems amongst early adopters who generally appear to have more pre-existing problems.
If the adoption effect is not supported in further research, the three novel forms discussed in this paper appear to be extremely toxic. It is possible that some new gambling products, or innovations to existing products, represent a more virulent gambling mutation that sharply increases the incidence of gambling problems. If this is the case, policy should focus on removing the availability of these products. However, we suggest that the adoption effect is a more plausible explanation for the extraordinarily high rates of gambling problems found amongst early adopters. We should therefore expect to see a flattening of the curve as these products mature-not because of adaptation, but because the pool of early adopters, many of whom have moderate and severe gambling problems, becomes diluted by much larger numbers of more resistant later adopters.
We fully acknowledge that our ideas in this paper are tentative, although they appear to have a good theoretical basis based on epidemiological and public health frameworks. Our purpose is to prompt further inquiry rather than provide definitive answers. We welcome comments that might refine or refute these ideas and analyses that can test our hypothesised adoption effect.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.