Past-year Gambling Research Articles

Comparing ‘fair’ machine learning models for detecting at-risk online gamblers

ABSTRACT Researchers have worked to develop machine learning models that detect at-risk online gamblers, enabling personalized harm prevention tools. However, existing research has not evaluated these models’ potential to reinforce or amplify sociodemographic biases leading to treatment disparity, a recognized issue in the machine learning field. We sought to develop and compare three examples of potentially fair models using online gambling data. In two large samples of transaction data from a provincially owned Canadian gambling website (N1 = 9,145, N2 = 10,716), we developed three machine learning models based on competing concepts of fairness: fairness via unawareness, classification parity, and outcome calibration. We hypothesized that significant relationships existed between reporting a high risk of past-year gambling problems (the dependent variable) and participants’ age and sex. Further, we hypothesized that the three ‘fair’ models would show differing levels of classification performance both in aggregate and within sociodemographic groups. Significant age and sex effects were found, refuting the fairness via unawareness modeling strategy. Superiority across all performance metrics was not present for either of the remaining models. For the fairest practices in any jurisdiction, classification parity and outcome calibration models should be tested in situ, and incorporate the perspectives and preferences of end users who will be affected.

Demographic characteristics, gambling engagement, mental health, and associations with harmful gambling risk among UK Armed Forces serving personnel

IntroductionHarmful gambling negatively impacts individuals, families and communities. Growing international evidence indicates that the Armed Forces (AF) community may be at a comparatively higher risk of experiencing harm from gambling...

Psychometric Validation of the Compulsive Internet Use Scale in Spanish Adolescents.

Despite problematic internet use being especially high among adolescents, there are no screening instruments in Spain specifically for adolescents that would facilitate early detection of this problem. The main goal of this study was to validate the Compulsive Internet Use Scale (CIUS) in the adolescent Spanish population as well as to analyze the discriminative capacity of CIUS based on sociodemographic characteristics, grade point average, and other addictive behaviors. Data were obtained from the ESTUDES, a representative survey of addictive behaviors of Spanish adolescents. The sample consisted of a total of 34,308 adolescents aged between 14 and 18 years old (Mage = 15.70; SD = 1.19; 51.7% females). Results indicated that the CIUS fit a unidimensional structure, exhibited measurement invariance with respect to sex and age, and demonstrated excellent reliability (ω = .94). Past-month tobacco, alcohol, and cannabis use, as well as past-year gambling and gaming, were related to CIUS scores. A standardized screening instrument that provides valid, reliable information on young people's use of the Internet in Spain is a critical requirement for successful early detection and intervention in this population.

Establishing the temporal stability of machine learning models that detect online gambling-related harms

Artificial Intelligence (AI) models can detect at-risk online gamblers by analyzing patterns in their betting behaviour, but their performance over time has not been assessed. Linking online gamblers’ self-reported gambling problems to their transactional gambling data, we investigated the temporal stability of online gambling behaviours at two timepoints (2019–2022). We then assessed the impacts of shifting gambling behaviours on the performance of two previously-validated AI harm detection models.Adult users of a Canadian online gambling website (n2019 = 9,145, n2022 = 11,258) completed the Problem Gambling Severity Index (PGSI), a validated questionnaire examining past-year gambling problems. Population Stability Index assessed the temporal stability of 10 previously-validated indicators of problematic online gambling. Two AI models were then revalidated using overall and threshold-dependent performance metrics.All measured indicators of problematic online gambling behaviour shifted significantly from 2019 to 2022 (all p < 0.001). Existing AI models showed significant changes in Area Under the Precision-Recall Curve (PGSI 5+ ΔAUPRC = +2.87%; t(20401) = 2.83, p = 0.004; PGSI 8+ ΔAUPRC = +7.06%; t(20401) = 7.21, p < 0.001). Adjusting these models’ decision thresholds realigned their classification performance with originally-validated levels. These results support the use of AI for online gambling harm detection.

Intersection of gambling with smoking and alcohol use in Great Britain: a cross-sectional survey in October 2022

ObjectivesGambling is associated with cigarette smoking and alcohol consumption. We explored the intersection of gambling across all risk levels of harm with smoking and alcohol use among adults in Great...

To screen, or not to screen: An experimental comparison of two methods for correlating video game loot box expenditure and problem gambling severity

Loot boxes are gambling-like products found in video games that players can buy with real-world money to obtain random rewards. A positive correlation between loot box spending and problem gambling severity has been well-replicated. Some researchers recently argued that this observed positive correlation may be due to participants incorrectly interpreting problem gambling questions as applying to their loot box expenditure because they see loot box purchasing as a form of ‘gambling.’ We experimentally tested this alternative explanation for the observed positive correlation (N = 2027) by manipulating whether all participants were given the problem gambling scale, as the previous literature generally had done (the ‘non-screening’ approach; n = 1005), or by ‘screening’ participants (n = 1022) by only giving the problem gambling scale to those reporting recent gambling expenditure. Through the latter screening process, we clarified and calibrated what ‘gambling’ means by providing an exhaustive list of activities that should be accounted for and specifically instructed participants that loot box purchasing is not to be considered a form of ‘gambling.’ Results showed positive correlations between loot box spending and problem gambling across both experimental conditions. In addition, a predicted positive correlation emerged between binary past-year gambling participation and loot box expenditure in the screening group. These experimental results confirm that the association between loot box spending and problem gambling severity is likely not due to participants misinterpreting problem gambling questions as being relevant to their loot box spending. However, problem gambling severity was inflated in the non-screening group, meaning that future research on gambling-like products should include gambling participation screening questions; better define what ‘gambling’ means; potentially exclude non-gamblers from analysis; and, importantly, provide explicit instructions on whether certain activities should not be considered a form of ‘gambling.’

Loot boxes, gambling-related risk factors, and mental health in Mainland China: A large-scale survey

Loot boxes can be bought with real-world money to obtain random content inside video games. Loot boxes are viewed by many as gambling-like and are prevalently implemented globally. Previous Western and international studies have consistently found loot box spending to be positively correlated with problem gambling. Previous Western studies presented mixed results as to the correlations between loot box purchasing and gambling-related risk factors, mental wellbeing, and psychological distress. A large-scale survey of adult video game players from the People’s Republic of China (PRC) (N = 2601) was conducted through Tencent Survey. The positive correlations between loot box spending and problem gambling, and between loot box spending and problem videogaming, were successfully replicated. However, other potential risk factors (i.e., impulsivity/impulsiveness; binary past-year gambling participation status; and sensation-seeking tendencies) either did not positively correlate with loot box spending or only did so weakly. Contrary to expectations, high impulsivity was negatively associated with loot box engagement. The Risky Loot Box Index (RLI) most strongly positively correlated with, and was the best predictor in multiple linear regression models for, loot box spending. The RLI may be effective at measuring loot box harms cross-culturally. A surprising weak positive correlation was found between loot box engagement and PRC players’ mental wellbeing, and high psychological distress unexpectedly negatively predicted loot box purchasing. Although gambling-like, the risk and protective factors of loot boxes are seemingly different, meaning they should rightfully be treated as novel products. Cross-cultural research can contribute to a better understanding of loot box harms.

Longitudinal joint trajectories of gambling disorder and hypomentalization: A latent class growth analysis among young adults

IntroductionOnly a few studies have investigated the relationship between risk for gambling disorder (GD) and mentalization difficulties. However, these results on positive correlations were all from cross-sectional studies; therefore, it would be important to investigate the longitudinal relationship between the constructs. AimsIt was aimed to identify subgroups of gamblers with different joint growth trajectories of risk for GD and hypomentalization (i.e., uncertainty of understanding own and others' mental states) and to compare the identified latent classes along impulsivity, sensation seeking and depression. MethodsData from waves 1–3 of the Budapest Longitudinal Study were used. The final sample included 219 young adult gamblers who completed the study questionnaire and reported past-year gambling in all three waves (Females: N = 93 [42.29%]; Mean age: 28.36 years [SD = 4.29]). Standardized questionnaires were administered to measure the main constructs. ResultsA four-class model was retained based on latent class growth analysis. Class 1 had consistently moderately high risk for GD and hypomentalization, and Class 4 presented consistently low risk for GD and hypomentalization. Classes 2–3 were characterized by low risk for GD and increasing hypomentalization. Class 1 had significantly higher sensation seeking and depression compared with the other subgroups in multiple cases. Class 4 showed significantly lower impulsivity and sensation seeking than the other classes. ConclusionsElevated hypomentalization was presented among individuals with risk for GD. However, there is little empirical evidence to support that the two constructs have convergently changing trajectories. A cautious interpretation of the results is recommended due to low symptom presence.

Gambling behavior and criminal convictions: a register-based investigation

Background Gambling severity and criminal convictions are known to be correlated, but there is a lack of understanding of the associations between the type and frequency of offending, game types, gambling motives and gambling intensity. This study aims to fill this gap by examining systematically whether conviction prevalence varies with gambling behavior. Method The study combined register-based variables on criminal convictions from the past 45 years to a gambling-related population survey (n = 3993). Descriptive data are provided on the prevalence of criminal convictions among different past-year gambling behavior subgroups. Logistic regression models were built to explore the associations between having convictions on one’s record and gambling behavior traits more in depth. Results The results of this study indicate that the prevalence of having criminal convictions not only varies among different demographic groups and well-being correlates, but also among different groups of past-year gamblers based on played game types, gambling versatility, gambling severity and gambling motives. Having a history of convictions was predicted by male gender, older age group, receiving basic social allowance, at-risk gambling and risky alcohol use. Escape as a gambling motive and playing EGMs proved to be significant predictors of having multiple criminal convictions. Conclusions Gamblers differ in conviction prevalence as a function of gambling game types, versatility and gambling motives. These findings contribute to better understanding the dialectic interplay between gambling and criminal behavior and may inform the development of targeted interventions.

Loot box purchasing is associated with gambling and problem gambling when controlling for known psychological risk factors of gambling

Background A robust literature has found that loot box purchasing is associated with gambling and problem gambling. However, it remains unclear whether this association is merely an artifact of known psychological risk factors for gambling. The present study thus examined associations of loot box purchasing with gambling and problem gambling while controlling for potential psychological confounders. Methods Current gamers, recruited from five Canadian universities (N = 1189) and Academic Prolific and Reddit (N = 499), reported on loot box engagement, gambling engagement, and psychological characteristics. In each sample, binomial logistic regression analyses were used to examine the association between past-year loot box purchasing and likelihood of past-year gambling. Hierarchical linear regression analyses were used to examine the associations of each of past-year loot box purchasing and risky loot box engagement with problem gambling severity. Results In both samples, having purchased loot boxes in the past year was significantly associated with increased likelihood of having gambled in the past year and greater problem gambling severity. In the student sample, greater risky loot box engagement was significantly associated with increased problem gambling severity. Conclusions Consistent with previous research, there exist associations between loot box purchasing and gambling. Results suggest that these associations are robust to known psychological risk factors for gambling, reducing plausibility of the notion that the association between loot box purchasing and gambling exists only due to shared psychological vulnerabilities. Loot box purchasing represents an important marker of risk for gambling and problem gambling among people who play video games.

Opening Pandora’s Loot Box: Weak Links Between Gambling and Loot Box Expenditure in China, and Player Opinions on Probability Disclosures and Pity-Timers

Loot boxes are quasi-gambling virtual products in video games that provide randomised rewards of varying value. Previous studies in Western contexts have identified a positive correlation between loot box purchasing and problem gambling severity. A preregistered survey of People’s Republic of China (PRC) video game players (N = 879) failed to replicate this correlation. We observed statistically significant but weak positive correlations between loot box expenditure and past-year gambling participation, and between loot box expenditure and impulsiveness. This study cannot prove that loot boxes are not disproportionately purchased by people with problem gambling symptomatology in the PRC or that PRC players are not potentially at risk of loot box-related harms. Instead, the evidence suggests that the relationship between loot boxes and gambling might be weaker in the PRC than in Western countries. We identified multiple unique factors about the PRC that might be affecting this relationship. For example, the lotteries are the only legally permitted form of gambling. More gamified electronic gambling products are unavailable. The limited availability of gambling meant that a low level of gambling participation (n = 87) was observed, which is a limitation of this study. Additionally, the PRC is presently the only country to legally require loot box probability disclosures as a consumer protection measure. Most loot box purchasers (84.6%) reported seeing loot box probability disclosures, but only 19.3% of this group reported consequently spending less money. Most loot box purchasers (86.9%) thought that pity-timers, which increase the winning probabilities of obtaining rarer rewards, are appropriate for implementation. Future loot box research should give greater consideration to differing cultural contexts and novel consumer protection measures.

Compulsory School Achievement and Future Gambling Expenditure: A Finnish Population-Based Study

Background: Gambling is associated with many conditions that can compromise young people’s health and wellbeing, such as substance use and poor school achievement. Conversely, low school achievement can be linked to lower socio-economic position. Thus, the aim of this study is to examine whether compulsory school achievement is linked with gambling participation and gambling expenditure (GE) later in youth and whether GE is linked with lower socio-economic position. Methods: The Finnish Gambling Harms survey data (n = 7186) were used. The data were collected in three regions during spring 2017. Participants aged 18–29 years old were selected from the data. Past-year GE was examined using two measures: weekly gambling expenditure (WGE, in €) and relative gambling expenditure (RGE, in %). Logistic regression and log-linear regression models for past-year gambling, WGE and RGE were created. Results: Persons who had no more than a mediocre grade point average (GPA) had a 25% higher WGE and 30% higher RGE in 2016 than those who had an outstanding GPA in the compulsory school. Compared with persons with an outstanding GPA, those with a satisfactory to very good GPA spent 13% more on gambling, and their RGE was 17% higher. Additionally, those with lower socio-economic status (SES) had a higher WGE and RGE compared with higher SES. Conclusions: Even after controlling for other crucial background characteristics, early life success, in the form of compulsory school outcomes, seems to correlate with gambling expenditures later in youth. This suggests that the gambling behaviour can be linked to the cognitive ability of an individual. Our findings also imply that gambling could be more heavily concentrated on individuals that are already more socially disadvantaged. However, it is worth noting that individual factors such as traumas, antisocial personality, anxiety and depression are all associated with gambling and poor academic achievement. Overall, this suggests that various educational tools at a younger age can be effective in preventing gambling-related problems in later life.

Associations Between Addictive Behaviors, Individual Characteristics, and the Use of Gambling Services Within the World of Gaming: Cross-sectional Survey Study.

BackgroundGambling within the world of gaming is an emerging phenomenon that may share common conceptual characteristics with traditional forms of gambling. The current literature suggests a higher degree of problematic behaviors in this gambling pattern, but studies are few, prompting for further research regarding individual characteristics and comorbid conditions associated with this activity.ObjectiveThe aim of the study is to investigate correlations between the use of gambling services within the world of gaming and individual characteristics and addictive behaviors including problem gambling.MethodsA cross-sectional web survey was distributed to an existing panel of online respondents in Sweden. A total of 2001 respondents were included. Chi-square and Mann-Whitney U tests, followed by a logistic regression, were used in order to determine independent variables associated with gambling in the context of gaming.ResultsA total of 2.9% (58/1984) of respondents reported past-year gambling within gaming. Significant associations were found with male sex, younger age, history of treatment-seeking for alcohol problems, and higher Gaming Addiction Scale scores.ConclusionsThe demonstrated findings strengthen previously found associations between gambling in gaming and younger age, male sex, and problematic gaming behaviors. Additionally, the association with a history of treatment needs for alcohol problems adds to the previous impression of increased problem severity and comorbidity in within-gaming gamblers.

Relations Among Gambling Behavior, Associated Problems, Game Type, and Risk Factors in a Rural, African American, Adolescent Sample.

Problem gambling is a rising concern among adolescent populations; youth gamble more frequently than adults, and those who gamble are more susceptible than adults to maladaptive outcomes. Research shows that gambling problems are exacerbated among minorities, despite lower rates of gambling frequency. Minority youth are at especially high risk for problematic gambling outcomes, though they have not been widely studied. The objectives of this study were to (a) investigate gambling frequency and rates of associated problems among rural, African American youth, (b) examine risk factors associated with gambling problems, and (c) explore preferences for game type. Hypotheses were tested with survey data from 270 African American youth from rural communities in Georgia, ages 14-17. Past-year gambling prevalence was 38% (48% of males and 28% of females), and 30% of those who gambled (11% of the total sample) reported at least one problem behavior associated with gambling.Confirmatory factor analysis established a distinction between games of skill versus luck. Gambling problems were associated with skill games, and youth played skill games more than luck games. Substance use and anger scores predicted gambling frequency, and gambling frequency predicted gambling problems. Depression scores provided no predictive utility. Poverty status was negatively associated with skill gambling, and there was no association between poverty status and luck gambling. Males gambled more frequently, had more gambling problems, and were more likely to engage in skill gambling relative to females.

Associations between gambling severity and criminal convictions: implications for the welfare state

This study addresses criminal convictions, social disadvantage and problem gambling as an interwoven set of problems. It makes use of data from a population-based gambling survey (n = 7,186) conducted in three Finnish regions. The survey data are combined with national registers to examine associations between sociodemographic factors and gambling severity, comparing persons with and without a criminal record. Gambling behavior included past-year (2016) gambling severity and perceived life-time problem gambling. Social disadvantage was assessed using sociodemographic factors such as education, employment status, level of income and receipt of basic social assistance. Logistic regression analysis showed that both past-year problem or pathological gambling (OR: 2.725) and perceived life-time gambling problems (OR: 2.363) were associated with having a conviction, compared to recreational gambling. Low education, unemployment, low income and receipt of basic social assistance were associated with receiving a conviction. When gender, age and sociodemographic factors were controlled for, odds ratios for both past-year gambling problems (OR: 1.223) and perceived life-time gambling problems (OR: 1.586) did not remain statistically significant. The current study suggests that preventive efforts against problem gambling and interventions in criminal justice systems should be expanded to incorporate the aim of reducing social disadvantage.

Gambling Behavior Among Hong Kong College and University Students

This study examined gambling behavior and correlates of pathological gambling among college and university students in Hong Kong. A survey questionnaire was administered to 510 Chinese students (302 men, 208 women) recruited from twelve tertiary institutions. The standardized questionnaire included questions on socio-demographic background, preferred lifetime and past-year gambling forms, attitudes towards gambling, perceived life satisfaction, social influence, intention to seek help, and a gambling screen to assess problematic gambling. The response rate is 86%. Results indicate the prevalence rate of lifetime and past-year gambling are 79.6% and 41.8% respectively with male domination. Many (60%) started gambling before 18 years. The estimate of lifetime vulnerability to pathological gambling is 14.7%. Pathological gambling is associated with the male gender, Internet gambling, monthly gambling expenditure, gambling attitude, betting on a great variety of games, and life dissatisfaction. Survey results have implications for campus awareness programs and future research.

Lottery-Purchasing Adolescents: Gambling Perceptions, Problems, and Characteristics.

Gambling on the lottery is a prevalent behavior, and lottery products are increasingly available in online and electronic formats. As lottery-purchasing is prevalent in adolescents, this study systematically examined relationships between lottery-purchasing and problem-gambling severity and gambling perceptions and attitudes, as well as differences in the relationships between problem-gambling severity and measures of health/functioning and gambling characteristics among lottery-purchasing groups. Participants were 1517 Connecticut high-school adolescents with past-year gambling. Chi-square and logistic regression models were conducted, and interactions between problem-gambling severity and lottery-purchasing status on multiple outcomes were determined. Lottery-purchasing, compared to non-lottery-purchasing, adolescents had greater problem-gambling severity and reported more permissive gambling attitudes and greater parental approval of gambling. Significant between-group differences were observed, with at-risk/problem-gambling more strongly associated with having friends and adult gambling partners among non-lottery-purchasing adolescents, and machine and online gambling, and gambling alone more strongly associated among lottery-purchasing adolescents. Greater problem-gambling severity, permissive gambling attitudes, and parental approval of gambling in lottery-purchasing adolescents suggest that parenting contexts are important considerations in prevention efforts targeting problem gambling in youths. Between-group differences in associations between problem-gambling severity and gambling types and partners may identify high-risk groups associated with more solitary gambling behaviors for whom targeted interventions may be adapted.

Online Survey of Gambling Participation and Problem Gambling in Chiba City: Problem Gambling Rates May be Strongly Influenced by the Administration Mode.

Prevalence studies on gambling participation and problem gambling in Japan have been limited. To obtain data showing the current status of gambling in Japan, we conducted an online survey on gambling participation and problem gambling among residents in Chiba City. The online data collection was conducted through an Internet research firm. Questionnaires were consisted of personal demographics, past-year gambling participation and severity of gambling risks. The Problem Gambling Severity Index (PGSI) was used as a survey instrument. Males (51.5%) were significantly more likely than females (25.5%) to have gambled in the past year. Problem gamblers (PGSI score ≧ 8) were 7.2% among males and 1.2% among females. Problem gambling was strongly correlated with frequent gambling, preference for Pachinko/Slot and smoking habit. The gambling participation rates were roughly lower than those reported in many overseas jurisdictions, whereas the problem gambling rates were considerably higher than those reported in the same jurisdictions. According to the total consumption model, it would be reasonable to consider that gambling participation rates show a positive correlation with problem gambling rates. The high levels of problem gambling may be due to the administration mode using online sampling, by which answers admitting unapprovable behaviors like problem gambling tend to increase. This suggests that the previous studies using conventional face-to-face or telephone methods may have underestimated problem gambling rates.

Adolescent Land-Based and Internet Gambling: Australian and International Prevalence Rates and Measurement Issues

This review summarizes recent studies of adolescent gambling in Australia and overseas. Its purpose is to examine variations in participation in gambling activities, including land-based and Internet gambling, and in measurement approaches to studying gambling-related problems. Between 15 and 80% of young people report past-year gambling participation, typically involving scratch cards, lotteries, card games and sports betting. Around 5% report past-year Internet gambling. Estimates of problem gambling among adolescents vary, with Australian figures generally between 1 and 5%, and studies in the USA, Canada, UK, Italy and Finland generally reporting rates of 1 to 7%. Simulated gambling involvement (e.g., gambling-like apps, social casino games) appears more prevalent but its relationship to monetary gambling and problem gambling risk requires further research. Youth gambling and exposure to gambling-like activities via digital technologies is a global phenomenon that warrants continuing research. Research is required at the conceptual level, for example, to identify relevant harms and map links to other risk behaviours, and at the methodological level to identify optimal measurement approaches. This includes considering respondents’ level of involvement in reported gambling (active vs passive), recall issues and clearly distinguishing monetary from simulated gambling. Understanding youth gambling is particularly relevant as new digital technology-based gambling activities and promotions become more prevalent and available to young people with the means to access them.

The longitudinal effect of early-life sensation seeking on gambling and gambling problems among Puerto Rican young adults.

Sensation seeking has been proposed as a risk factor for gambling and gambling problems; however, existing evidence for a relationship between sensation seeking and gambling behaviors is inconclusive and data are lacking for emerging adults and racial and ethnic minorities. In this longitudinal study, we explored the association between developmental trajectories of sensation seeking in childhood and adolescence and gambling and gambling problems in early adulthood in individuals of Puerto Rican origin. Gambling data were collected during 2014-2018 from a subsample of participants in the Boricua Youth Study who were recruited in the South Bronx of New York City and in San Juan and Caguas, Puerto Rico. Sensation seeking was measured using a 10-item instrument modified from the scale created by Russo et al. for use in children as young as 5 years old. Developmental trajectories of age-adjusted sensation seeking were created using growth mixture models. Gambling and gambling problems were assessed based on the Canadian Adolescent Gambling Inventory (CAGI) Version 1.09. Data were analyzed using descriptive methods and multivariable logistic regression. Individuals in the high sensation-seeking class had lower adjusted odds of past-year gambling (OR = .36; 95% confidence interval [.14, .92]) than did those in the normative sensation-seeking class, whereas no differences were observed for individuals in the low and accelerated classes. No relationship was found between sensation seeking and past-year gambling problems. Given the severe consequences of early initiation of gambling and gambling problems, other early life risk factors and alternative hypotheses for the elevated prevalence of gambling problems in young adults and racial and ethnic minority populations should be explored. (PsycINFO Database Record (c) 2020 APA, all rights reserved).