This paper describes how the complex sample design aspects of stratification, clustering, unequal selection probabilities and calibration adjustments of the sample weights affecting the National Household Sample Survey (PNAD) of IBGE can be incorporated into the analysis of its data. For this purpose, it includes: a brief but comprehensive description of this survey's sample design; indication of how its data can be used for estimating totals; and also a description of the essential methods required to fit parametric models taking into account the complex sample survey design aspects mentioned. It also presents some estimates for household and personal characteristics obtained from PNAD/1998, for which standard errors and design effects are computed as an illustration. It concludes with an indication of the precautions users must take when analysing data from PNAD in their survey practice.
Design effect; Estimation; Survey data analysis; Structured data; Household survey; Parametric models
