ABSTRACT
Exposure to schistosomiasis cercariae can be measured in terms of frequency of bathing, washing clothes, washing animals, cleaning canals, etc. As randomized studies are neither practical nor ethical, evidence must be based on observational studies with uncontrollable patterns of activities. For individuals who are already at high risk because of one type of exposure, adding another risk may be less relevant than for individuals where the first risk factor is not present. The observed differences between the two villages could be explained, for instance, through different crops being cultivated. Also, if in one village a larger proportion of the population is engaged in farming, this could lead, on the average, to a lesser degree of water contact per person. The proposed marginal-likelihood procedure provides a solution for a problem which frequently arises in the analysis of questionnaire data.
