Dental researchers are interested in measures of association between explanatory variables and various disease indicators, where both can be measured at the tooth or surface level. The usual chi-square tests for event rate comparisons are inadequate to analyze such data since teeth within the same individual are correlated with one another. This situation may become further complicated when stratification adjustment is of interest, particularly when the stratification factor is also at the tooth or surface level, such as tooth type or presence of previous disease for the tooth. Consequently, statistical methods are needed to control for the correlation of sites in the mouth. A survey sampling approach is suggested; with this method, the sample is managed as a one stage cluster sample of patients to obtain the design base covariance matrix for the cell counts in a contingency table. The covariance matrix for the cell counts can be obtained from statistical software used to analyze data from large sample surveys. Approximate estimates for variance for various measures of association can then be obtained by Taylor series methods. Event rate comparisons through the odds ratio, relative risk, risk difference as well as other measures of association can be made with adjustments for the intra-patient correlation. Methods for summary measures of association across strata are considered as well.