Demographics
The 408 males in the sample had significantly higher PGSI scores (median = 1, mean rank = 335.98) compared to the 231 females (median = 0, mean rank = 291.78), Mann–Whitney U = 40,605.5, Z = −3.11, p = 0.002. Younger respondents were significantly more likely to have higher PGSI scores compared to older respondents, Spearman’s rho = −0.31, p < 0.001. No significant difference was found in PGSI scores between those who were born in Australia and elsewhere, Mann–Whitney U = 30,004.5, Z = −.155, p = 0.122.
Significant differences were found between the different marital statuses, Kruskal–Wallis χ(3) = 9.83, p = 0.020. Post-hoc tests revealed that the 133 never married respondents had significantly higher PGSI scores compared to the 307 married respondents (median = 1, mean rank = 245.88 vs median = 0, mean rank = 209.50 respectively) and compared to the 126 respondents living with their partner/de facto (median = 1, mean rank = 139.71 vs median = 0, mean rank = 119.75 respectively), Mann–Whitney U = 17,040, Z = −2.95, p = 0.003 and Mann–Whitney U = 7087, Z = −2.25, p = 0.024 respectively. The 73 widowed/divorced/separated respondents had a median PGSI of 1 and did not differ significantly from any other group.
Significant differences in problem gambling severity were found between different household types, Kruskal–Wallis χ(5) = 31.47, p < 0.001. The 38 respondents in one parent families with children (median = 5) had significantly higher PGSI scores compared to the 215 respondents living as a couple with children (median = 5, mean rank = 147.91 vs median = 0, mean rank = 123.30 respectively) and the 186 respondents living as a couple with no children (median = 5, mean rank = 148.32 vs median = 0, mean rank = 105.18 respectively), Mann–Whitney U = 3290.5, Z = −2.03, p = 0.043 and Mann–Whitney U = 2173, Z = −4.16, p < 0.001 respectively. Furthermore, those living as a couple with no children had significantly lower PGSI scores compared to the 117 people living singly (median = 0, mean rank = 138.40 vs median = 1, mean rank = 173.62 respectively) and the 70 in group households (median = 0, mean rank = 116.86 vs median = 2, mean rank = 159.42 respectively), Mann–Whitney U = 8,351, Z = −3.75, p < 0.001 and Mann–Whitney U = 4,345.5, Z = −4.48, p < 0.001 respectively. Finally, a significant difference was observed between those living as a couple with children compared to those living as a couple with no children. While these groups had the same medians (0), a non-parametric test demonstrated that the couple with children group (mean rank = 216.87) had significantly higher PGSI scores compared to those living as a couple with no children (mean rank = 182.66), Mann–Whitney U = 16,583.5, Z = −3.25, p = 0.001. The 13 respondents living in an ‘other’ arrangement did not differ significantly from any of the other groups.
Significant differences were found between those with different levels of education, Kruskal–Wallis χ(4) = 9.84, p = 0.043. Specifically, the 93 respondents who did not complete high school (median = 0) had a significantly lower level of problem gambling severity compared to the 133 with year 12 as their highest educational qualification (median = 0, mean rank = 103.44 vs median = 1, mean rank = 120.54 respectively, Mann–Whitney U = 5248.5, Z = −2.07, p = 0.039), to the 146 respondents with an undergraduate qualification (median = 0, mean rank = 106.99 vs median = 1, mean rank = 128.28 respectively, Mann–Whitney U = 5579.5, Z = −2.48, p = 0.013) and to the 85 respondents with a postgraduate qualification (median = 0, mean rank = 82.77 vs median = 1, mean rank = 96.86 respectively, Mann–Whitney U = 3326.5, Z = −1.97, p = 0.049). Further, those with an undergraduate qualification (median = 1, mean rank = 175.97) had significantly higher PGSI scores compared to the 182 respondents with a trade or technical certificate or diploma (median = 0, mean rank = 155.30, Mann–Whitney U = 11,612.6, Z = −2.10, p = 0.036). No other group differences were significant.
Significant differences were found between those of differing work status, Kruskal–Wallis χ(8) = 24.53, p = 0.002. The 310 full-time respondents had significantly higher PGSI scores compared to the 54 self-employed respondents (median = 1, mean rank = 188.20 vs median = 0, mean rank = 149.81 respectively, Mann–Whitney U = 6604.5, Z = −2.60, p = 0.009), the 75 retired respondents (median = 1, mean rank = 203.13 vs median = 0, mean rank = 151.14 respectively, Mann–Whitney U = 8485.5, Z = −3.86, p < 0.001) and the 24 respondents on sick or disability pensions (median = 1, mean rank = 170.30 vs median = 0, mean rank = 131.33 respectively, Mann–Whitney U = 2852.0, Z = −2.00, p = 0.046). Further, the 21 full-time students had significantly higher PGSI scores compared to the self-employed (median = 1, mean rank = 46.43 vs median = 0, mean rank = 34.72 respectively, Mann–Whitney U = 390, Z = −2.25, p = 0.025), the retired (median = 1, mean rank = 62.38 vs median = 0, mean rank = 44.61 respectively Mann–Whitney U = 496, Z = −2.91, p = 0.004) and sick or disability pensioners (median = 1, mean rank = 27.14 vs median = 0, mean rank = 19.38 respectively, Mann–Whitney U = 165, Z = −2.11, p = 0.035). Finally, the 94 part-time workers had a median of 0, as did the retired respondents, although the post hoc tests revealed a significant difference in ranked scores, with part-time workers having significantly higher scores (mean rank = 91.84 vs mean rank = 76.43 respectively), Mann–Whitney U = 2882.0, Z = −2.28, p = 0.023. The 25 unemployed respondents and 26 on full-time home duties did not differ significantly from any of the other groups and no other differences were significant. No significant relationship was found between household income and PGSI score (Spearman’s rho = 0.013, p = 0.757).