Table 1 Current peer-reviewed research on regression-based model estimates of covariates and game type specific risks between 2009 and 2018

From: The Gambling Consumption Mediation Model (GCMM): A Multiple Mediation Approach to Estimate the Association of Particular Game Types with Problem Gambling

References Sample N Pivotal method Predictors of problem gambling Main results
#1
Welte et al. (2009)
Random telephone survey in the USA among adolescents (14–21 years old) 1535 last year gamblers Simultaneous negative binomial regression of all game types Participation in 15 game types (both tested: dichotomous or ordinal level), covariates (gender, age, socio-economic status), and interaction terms of forms with the three covariates On a dichotomous participation level, the most important predictors were: card games, casino gambling, other games, and games of skill; on an ordinal participation level the most important predictors were casino gambling, office pools, and charity, other games, lottery, card games, and games of skill; significant interaction effects between gender and some game types
#2
LaPlante et al. (2011)
Secondary data analysis of the 2007 British Gambling Prevalence Survey; at least 16 years old 8968 last year gamblers A series of logistic regressions for each game type Participation in 15 game types (dichotomous level) as single predictors and also adjusted for the total number of game types a person played Adjusting for the number of game types involved in strongly reduced the impact of each game type on problem gambling; the only significant predictor left after adjustment was participation in virtual gaming machines; private betting and betting on horses even became significant protective factors
#3
Afifi et al. (2010)
Secondary data analysis of the Canadian Community Health Survey 2002; only women, at least 15 years old 10,056 women A series of logistic regressions for each game type Participation in 15 game types (ordinal level) as single predictors and also adjusted for several covariates (age, income, education, marital status, life stress, social support, and negative coping) The general trend noted in the results was that the probability of problem gambling increased with greater frequency of gambling in any game type; after adjusting for covariates, regular gambling involvement in video lottery terminals inside or outside casinos, other casino games, bingo and instant win tickets posed highest risks for problem gambling
#4
Brosowski et al. (2012)
Active customers of an online gambling provider; predictors extracted from actual gambling behavior across 7 months 27,653 A series of logistic regressions for each game type Participation in 6 game types (dichotomous level) adjusted for the number of game types played Adjusting for the number of game types only left live action betting and poker as significant risk factors for simultaneously transgressing two established offline-thresholds of probable at-risk gambling; each additional game type increased the transgression risks
#5
Haß et al. (2012)
Secondary data analysis of three German gambling surveys 2007, 2009, 2011; at least 16 years old 15,165 last year gamblers A series of bivariate and simultaneous logistic regressions for all game types Participation in 14 game types (dichotomous level) adjusted for (1) participation in all other game types, (2) total number of game types played, (3) covariates (age, gender, education, ethnicity, employment status) After adjusting for covariates, number of game types involved in, and participation in all other game types simultaneously, lotto became a significant protective factor; highest risks were still posed by electronic gambling machines, Internet casino games, keno, casino table games (not slots), and other sport betting (private provider)
#6
LaPlante et al. (2013)
Survey of casino patrons at two Las Vegas resort casinos 676 last year gamblers A series of logistic regressions for each game type Participation in 10 game types (dichotomous level) adjusted for (1) the frequency of gambling across all game types during the last year, (2) the total number of game types played during the current visit), and (3) both Frequency of play during the past year, rather than game played or number of games played during the current visit, consistently predicted a history of gambling problems
#7
LaPlante et al. (2014)
Active customers of an online gambling provider who responded to an online survey to assess problematic gambling behavior; predictors extracted from actual gambling behavior 1440 A series of logistic regressions for each game type Participation in 16 game types (dichotomous level) as single predictors and also adjusted for (1) the total number of game types played and (2) the number of days a person placed a bet, (3) and both Adjusting for the number of game types only left live action betting as a significant risk factor; after adjusting for the number of active gambling days the strongest predictors were phone-based casino software, simulated sports games, backgammon, downloadable casino software, games played in between bets, browser-based casino software, fortune games, and live action betting; after adjusting for both indicators of involvement poker even became a significant protective factor; both indicators were slightly correlated and the risk of problem gambling increased almost linearly with ascending levels of involvement
#8
Afifi et al. (2014)
Secondary data analysis of the Canadian Community Health Survey 2002; at least 15 years old 18,913 last year gamblers A series of bivariate and simultaneous logistic regressions for all game types Participation in 13 game types (ordinal level; partially applied dichotomously) (1) adjusted for participation in other types, (2) adjusted for other types and the total number of game types played, and (3) interactions of involvement with age and gender After adjustment for participation in other game types and the number of game types, regular gambling increased the risks of almost all game types; irregular gambling in some game types posed a protective effect, for instance in instant win or lottery tickets; regular involvement in electronic gaming machines inside casinos, other casino games, and electronic gaming machines outside casinos posed the largest risks for problem gambling; the number of game types involved in did not interact significantly with age or gender
#9
Scalese et al. (2016)
Secondary data analysis of an Italian population survey from 2010–2011; 15–64 years old 5292 individuals who completed the Problem Gambling Severity Index Ordinal logistic regressions in two separated samples: one-game-users and multi-game users Participation in 5 game types (dichotomous level) adjusted for age, gender, employment status, and educational level Being involved in more than one game type was associated with levels of problem gambling; after adjustment for demographic variables sport betting and slots increased risks of problem gambling in both samples—one-game and multi-game players; poker and other card games increased risks only for multi-game players; lotteries were protective in both samples, scratch cards only for multi-game players
#10
Yeung and Wraith (2017)
Secondary data analysis of an Australian population survey from 2008; at least 18 years old 11,177 last year gamblers A series of multinomial logistic and negative binomial regressions Participation in 11 game types (ordinal level) adjusted for (1) covariates (gender, age, education, residency, language, marital status, employment status), (2) the total number of games played in the past year, (3) total frequency of participation in other gambling activities, (4) interaction effects of age and number of game types or total frequency Negative binomial regression and logistic regression showed similar results; no interaction effect occured between age and both measures of gambling involvement; after adjustment of covariates and total number of game types or total gambling frequency, large significant risks remained for electronic gambling machines and casino table games
#11
Baggio et al. (2017)
Sample of French adolescents 9910 A series of increasingly complex Quasi-Poisson regressions Participation in Internet gambling (dichotomous) was adjusted for (1) demographic covariates (age, gender, and parental occupational status), (2) number of game types, (3) sum of gambling days, and (4) both measures Online gambling remained a significant risk factor after adjusting for covariates, number of game types, and number of gambling days; after including all variables simultaneously the association collapsed and only measures of breadth and depth remained significant predictors
#12
Nelson et al. (2018)
Online survey among residents of Massachusetts; at least 18 years old 274 last year gamblers A series of bivariate and simultaneous logistic regressions for each game type; finally a hierarchical regression Participation in 19 game types (dichotomous level) was successively adjusted for (1) patterns of game play (4 factor scores, previously extracted from a principal component analysis of gambling frequency measures), (2) overall depth of involvement (maximum frequency, daily hours, amount gambled, amount lost), and (3) number of game types All bivariate associations of game type participation and at least one DSM-IV criterion of problem gambling disappeared after adjusting for different measures of gambling involvement; number of game types had the strongest impact on effect reductions; component measures of gambling patterns and overall involvement were strongly interrelated
#13
Castrén et al. (2018)
Secondary data analysis of a Finnish population survey from 2015; at least 18 years old 3555 Stepwise negative binomial regressions for game types after controlling for demographics and overall involvement (number of game types was dropped due to collinearity) from a final simultaneous model Participation in gambling mode (online; offline) and 12 game types (ordinal level) adjusted for (1) covariates (gender, age), (2) overall expenditure by net income, and (3) overall gambling frequency After adjusting for demographics and the risk increasing impacts of online gambling, overall frequency and spent percent of income, ascending levels of gambling frequency in scratch games, betting games, slot machines, non-poker games on Finland’s Slot Machine Association online casino, and non-monopoly gambling (mainly online and strongly associated with the number of game types) remained significant risk factors for more gambling related harms
#14
Cavalera et al. (2018)
Online survey among Italian adults; 18–94 years old 4773 Stepwise and simultaneous multinomial logistic regressions Participating in online games, strategic, non-strategic, and a combination of both were included with (1) demographic covariates (gender, age, marital status, education, parental gambling problems) and (2) number of game types After simultaneously adjusting for significant predictors of the stepwise regression, number of game types, strategic games (poker, betting), and the combined use of strategic and non strategic games (lottery, slot machines, pull-tabs, bingo), remained significant risk factors for at risk gambling and problem gambling