This way, the range of activities online and in the Venue is calculated as in previous studies [41]. These composite count variables use the total number of activities participants have been involved in at least once in the past 30 days. Online activities 메이저사이트were held from one to seven times, and venue activities were held from zero to seven times.
Statistical analyzes
The data were analyzed using R [59]. The relationship between the frequency of participation in each activity was examined using Rho of Spearman. Relationship between activity frequency, PGSI score and K6 score were also examined using Spearman’s Rho. To facilitate the interpretation of these correlations, median PGSI and K6 and the 25th-75th percentiles are reported for each level of activity frequency. The method 메이저사이트was used for the irregular distribution of PGSI and K6 scores and the ordered nature of the activity frequency variables. For these analyzes, we corrected the multiple comparisons using the Bonferroni method. The unique contribution of demographic details potentially associated with gambling activities on each online or Venue to the scores of PGSI and K6 was examined using quasi-Poisson regression.
Quasi-Poisson regression was used because the distribution of PGSI and K6 was extremely positive and kleptocratic, and early studies showed that these variables were overdistributed (e.g., the variance was greater than the average). We derived estimates of the variance explained by each regression using a recently developed variance-based R2 calculation method [60]. These R2v values were calculated using the rsq package of R [61]. We report an adjusted R2v value for the number of predictors in each model (e.g., adj.R2v). In addition, whether there are multiple collinearities between predictor variables was examined using Variance Inflation Factors (VIF). VIF evaluated a model that included individual activity frequencies and nonclassified demographic variables (e.g., age) ranging from 1.35 to 3.96 (M = 2.66), below the usual cutoff of 5 or 10.
The highest VIF variables were the attendance frequency of poker (3.96) at the Venue, e-sports (3.87) at the Venue, e-sports (3.48) online, poker (3.46) at the online, casino card/table games (3.46) at the venue and casino card/table games (3.28) at the online. The other VIFs were all < 3.00. We also examined the VIF of the model, including the range of involvement online and in the Venue. The VIF of the Online (13.58) and Venue (17.34) involvement was excluded from the regression model because it exceeded the recommended cutoff. In addition to the main regression analysis, a series of exploratory quasi-Poisson regressions were performed for each activity pair (e.g., online EGM and Venue EGM). These analyses include the frequency of gambling in each activity pair, demographic variables, the breadth of online gambling, and the breadth of involvement in venue-based gambling. The results of these analyzes are summarized in the main text, and the complete table is provided in the supplementary information.
The related VIF scores of these analyzes are shown in Additional File 1: Table S1 and the results of each regression are shown in Additional File 1: Table S2-S15. Demographic characteristics of the samples are shown in Table 1. Most participants were men, Europeans, married or de facto companionship, with the highest educational level after secondary education, working full or part-time, were born in Australia and were confirmed to not speak any language other than English at home. The participants ranged in age from 18 to 85. The PGSI score was strongly distorted (M = 3.91, SD = 5.56, Mdn = 1.00, Skew = 1.73, Kurtosis = 2.62, Min = 0, Max = 27) and similar to the Kessler 6 score (M = 5.64, SD = 6.14, Max = 3.00, Skew = 1.08, Kurtosis = 0.33, Min = 0, Max = 27).
