Datos agregados para corregir los sesgos de no respuesta y de cobertura en encuestas.

In the last decades the effect of nonresponse and coverage bias in surveys have questioned the ability of inferring the results to the population. An extended procedure used to correct nonresponse and coverage problems is the use of weights to balance the sample of respondents. However auxiliary information available for respondents and nonrespondents is required to compute weights. In this paper statistical simulations are used to test the potential of aggregate data to correct nonresponse bias. This research compares individual data adjustments to the use of auxiliary aggregate data. The results show the use of aggregate data can improve survey representativity if three requirements are met: 1) the dependent variable is grouped, 2) the dependent and auxiliary variables are correlated and 3) the auxiliary variable is correlated with response propensities. [ABSTRACT FROM AUTHOR]