Conclusion

Table 7 shows the relative risk ratios of falling into different frequency categories from the multinominal logistic regression. There are several interesting points. First, reading after work is not significant in predicting who falls into the former punter category, but reading is inversely related to being an occasional, frequent, or constant punter. Second, higher income increases the probability of being a former or constant punter, but has no effect on being an occasional or frequent punter. Third, being religious increases the probability of being a former or occasional punter, but has no statistical effect on being a frequent, or constant, punter. This result implies that for frequent and constant punters, superstition is less influential in playing the illegal lottery. This is consistent with casual observation that frequent participants in the lotteries prefer to calculate the probability of winning numbers rather than looking for clues hidden in the hint sheets, and that playing the lottery is a normal consumption for some punters. Fourth, playing social gambling games and buying official lottery tickets are the most significant factors increasing the relative risk ratios of falling into each of the frequency categories from former to constant. Finally, when it comes to explaining the characteristics of those who constantly participate in the illegal lottery, the most important predictors are playing card games, participating in the legal lottery, and having a relatively high income.