This is an article on probability sampling written to describe the methods used to calculate and calibrate sample weights of the Brazilian Oral Health Survey (SBBrasil 2003) and identify sample structural variables. It presents an overview of the efforts made to retrieve information from the records used to select the sample units in the sampled municipalities and describes the main problems observed with the survey's database, that acted as constraints tocal culating the natural sample weights and identifying primary sampling units and selection strata. It compares weighted sample estimates with previously published unweighted sample descriptive statistics, concluding that the oral health of the Brazilian population was better than previously disclosed.
Oral Health; Sampling Studies; Health Status; Oral Health
