Comparison of software programs for data analysis of complex surveys

OBJECTIVE: To compare specific software programs for data analysis of complex surveys regarding the following characteristics: ease of application, computer efficiency and accuracy of the results. METHODS: Secondary data from the Pesquisa Nacional sobre Demografia e Saúde (National survey on demography and health) (1996) with a target population of women aged 15 to 49 years old were used. This was a probabilistic subsampling drawn in two stages, then stratified, with the probability proportional to size in the first stage. The northern and mid-western regions of the country were selected for the study. The parameters of interest were mean for the age variable, and the proportion for five other qualitative variables. The software programs used were Epi Info, Stata and WesVarPC. RESULTS: The programs have two common options for the files import: the dBASE and text type files. The number of steps previous to the execution of the analyses were twenty- one for Epi Info, eleven for Stata and nine for WesVarPC. Efficiency was high for all them, that is, less that three seconds. The standard errors estimated using Epi Info and Stata were the same, with approximation up to the third decimal; those for WesVarPC were generally higher. CONCLUSIONS: Epi Info is the most limited software program regarding the analyses currently performed; however it is easy to use and free. Stata and WesVarPC are far more complete, however the disadvantage is their cost. The choice of the software program will depend mainly on the user's specific needs.