Statistical analysis was conducted with SPSS 13.0 for Windows, with the level of significance set at 5.0% (p-value < 0.05). Frequencies, means, standard deviations (SD), and percentages were computed for descriptive purposes. We used the Kolmogorov-Smirnov (KS) test to determine the distribution of DMFT scores. Since non-normal distribution was found, non-parametric tests (Wilcoxon rank sum test and Kruskal-Wallis H test) were used. Pearson's chi-square test was used to analyze the differences in the distribution of socio-demographic and oral health behavior variables (independent variables) between the case group (DMFT > 0) and the control group (DMFT = 0), and crude odds ratios (OR’s) and 95% confidence intervals (95% CI) were estimated. Unordered categorical variable (school districts) is defined as dummy viable with the method of reference cell coding. For the lowest prevalence rate of caries experience in our survey, Liwan was chosen as the reference cell, which is compared with all other districts. Variables that were significantly associated with DMFT scores were selected in the final model. Multivariate logistic regression analysis (Forward LR procedure) was conducted to investigate which independent variables were significant for explaining the dental condition