Avaliação de erros de mensuração em métodos de estimação sob a abordagem de cadastro duplo

Abstract

The Dual Frame approach involves a sample survey where two A and B, are used to provide coverage for a single target population. This approach has been used successfully in the literature in situations where a single cadastre can not provide complete coverage of this population. This work plan aims to study strategies to mitigate the effects of the measurement error, using estimators for population totals based on known literatures. These estimates were evaluated on the effect of one and two registers. In the Double Register, where only the A register is contaminated with measurement errors and B is free, it is noticed that when the contamination increases Bootstrap has difficulty correcting the estimates, regardless of the distribution used. For the tests the distributions Uniforme (0,1), Gamma (2,1) and Weibull (2,1) were used and the estimators used were Horvitz Thompson, Hartley and Bankier.