Gambling and the trait of addiction in a sample of New Zealand university students
New Zealand Journal of Psychology, Jun 2003 by Clarke, Dave
Continued from page 2.
Participants were recruited according to the guidelines of the Massey University Human Ethics Committee which approved the research. Volunteers were treated in accordance with the "New Zealand Psychological Society Code of Ethics" (New Zealand Psychological Society, 1986). After the second lecture, information sheets were distributed to the introductory psychology class by the author, before asking for volunteers to complete the questionnaire in their own time, and to return it anonymously to the School of Psychology. The information sheet included details about the purposes of the survey, respondents' rights as participants, the handling and confidentiality of the anonymous data, and the contact details at the Student Health and Counseling Services if they were distressed by the questionnaire. It clearly specified that all responses were confidential and that no identifying details such as name or university identification number would be sought. The information sheet also told them that they could withdraw from the study at any time until returning the questionnaire. After that time, it was not possible to identify an individual's questionnaire for retrieval. The questionnaire took about 30 minutes to complete.
Prior to inferential analyses, the data for each scale were examined for assumptions of normality. Gambling frequency was moderately skewed (1.23), so logarithmic transformations were computed to approximate normal distributions for the frequency data (Tabachnik & Fidell, 1989). For all other scales, except the SOGS-R, the distributions of data were within normal ranges and raw data were used in the computations.
Prevalence of gambling
The prevalence of gambling in the past 12 months (recent gambling) among the present sample was compared with that of a recent similar representative national sample (Amey, 2001). Only respondents who gambled on one or more activities were included in the comparisons. Table 1 shows the percentages of the university participants' and the national sample's recent gambling for each activity, and the prevalence of recent gambling for the group 15 to 24 years of age, compared with the same age group in the national sample. The most popular, recent gambling activities were Lotto and instant scratch tickets (73% each), lotteries/raffles (57%), and gaming machines (50%). The prevalence rates for these activities among the national participants (Amey, 2001) were 75%, 48%, 67% and 34%, respectively. For the present total sample, the chi-squares of differences between the observed rates and the rates expected from the representative national sample were statistically significant for casinos (30.25), card games (20.00), scratch tickets (13.02), and gaming machines (7.83), [chi]^sup 2^(1, N= 165), ps
Chi-squared tests comparing problem to non-problem gamblers, males to females and the four main ethnic groups to one another on the dichotomous variables of amount gambled, parents' gambling, and continuous gambling, were computed. The proportions of women to men were not significantly different on amount spent ($10 or less, vs. >$10), [chi]^sup 2^ (1, N = 165) = 0.58, perception of parents' excessive gambling (yes, no), [chi]^sup 2^ (1, N = 165) = 1.61, or continuous gambling, [chi]^sup 2^(1, N= 165) = 0.67, ps > .05. AMANOVA showed that there were no significant differences for gender on gambling frequency, number of activities, depression or Addiction, Wilks' lambda= .94, F(4, 157) = 2.42,p > .05.
Using post hoc tests of significance of differences between means with Bonferroni corrections for multiple comparisons, a MANOVA showed that the 13 Maori gambled on significantly more activities (M= 8.92, SD = 2.99) than the European/Pakeha (M = 6.64, SD = 2.30), Pacific Island (M= 6.73, SD = 2.65) or Asian (M= 5.46, SD = 2.82) groups, F (3, 151) = 4.78, p
Gambling in Casinos
More casino gamblers (76%) than non-casino gamblers (25%) spent $10 or more on any one bet, [chi]^sup 2^(1, N= 160) = 24.87, p
Problem versus Non-problem Gamblers
Gamblers were classified into groups according to their scores on the SOGS-R (Ladouceur et al., 1997): problem gamblers (scores 3 or greater) versus non-problem gamblers (scores of less than 3). Problem gamblers were compared to non-problem gamblers on gambling frequency, number of gambling activities, amount gambled, parents' gambling, continuous gambling, depression, and the trait of addiction.
According to the criterion of scores of three or greater on the SOGS-R, 26 of the 165 gamblers (16%) were classified as problem gamblers. Six of these (4%) met the criterion for probable pathological gamblers. There were proportionately equivalent numbers of male (6 of 34) and female (20 of 131) problem gamblers, [chi]^sup 2^ (1, N = 165) = 0.12, p > .05. Similarly, there were no significant differences in proportions of problem gamblers for age, [chi]^sup 2^ (3, ,N = 165) = 3.31, or ethnicity, [chi]^sup 2^ (3, N = 165) = 1.83,ps > .05.
In answer to the question if they thought that either of their parents gambled excessively, 36 of the participants circled yes and 114 circled no. A significantly greater proportion of problem gamblers thought that their parents gambled excessively (44%) compared with non-problem gamblers (20%), [chi]^sup 2^ (1, N= 150) = 6.58, p = .01. Similarly, a significantly greater proportion of problem gamblers (73%) spent more than $10 per bet in the last 12 months than non-problem gamblers (7%), [chi]^sup 2^ (1, N= 161) = 21.74, p
Further, proportionately more problem gamblers (64%) engaged in the continuous gambling activities of scratch tickets, gaming machines, track betting, card games, and casinos, than non-problem gamblers (24%), [chi]^sup 2^ (1, N = 110) = 12.83, p
For the 101 gamblers with complete sets of data, logistic regression (Walsh & Ollenburger, 2001) was computed to ascertain the unique contribution of each of the relevant variables related to the SOGS-R classification of problem and non-problem gamblers, while controlling for the other variables in the model (Kerlinger & Lee, 2000). Depression was omitted from the analysis because of its collinearity with Addiction. The results appear in Table 3. Amount gambled, continuous gambling, parents' gambling and Addiction were significant predictors in discriminating between problem and non-problem gamblers. The Wald statistic is used to test for the significance of each predictor in the model. The Hosmer-Lemeshow goodness-of-fit test was not significant, [chi]^sup 2^ (8, N = 101) = 5.69, p = .68, indicating that there was an adequate fit to a perfect model (Tabachnick & Fidell, 1996). Overall, 87% of the gamblers were correctly classified.
The odds ratios of 5.21 and 5.35 for continuous gambling and parents' gambling, respectively, indicate that students who engaged in continuous gambling or thought that their parents gambled excessively, were more than five times as likely to be problem gamblers as those who gambled non-continuously or who did not think that their parents gambled excessively. Participants who gambled more than $10 were more than four times (4.79) at risk of being problem gamblers than those who gambled $ 10 or less. For Addiction, with every one point increase in score (out of 32), there was a 17% increase in odds of being classified as a problem gambler.
The university students had gambling preferences similar to the representative national participants, except for gambling in casinos, card games, scratch tickets and gaming machines. More than twice as many gambled on four or more activities as the national sample, perhaps because of the greater choice and ready availability of gambling activities in the Auckland region (Abbott & Volberg, 2000). As in the earlier study (Clarke & Rossen, 2000), many of them were underage when they gambled illegally in casinos.