Objective: To describe the sampling design, operative strategy, and data management used in the Encuesta Nacional de Evaluaci6n del Desempeño (Mexican National Performance Assessment Survey).
Material and methods: A probabilistic, multistage, stratified sample was used to make inferences for each state of the country with representativeness of the urban and rural localities of Mexico. The survey was conducted between November 2002 and April 2003. The study population consisted of individuals in their households.
Results: Forty-thousand households were visited all over the country, and 38,746 interviews were obtained,with a non-response rate considerably lower than expected.
Conclusions: The methods used to carry out the National Performance Assessment Survey allowed calculation of estimates with measurable accuracy in terms of error and confidence, making the results inferable to the national population.