Una revisión de los distintos estimadores robustos para muestreo en poblaciones finitas
Fecha
2016-11
Título de la revista
ISSN de la revista
Título del volumen
Editor
Resumen
Se presentaron estimadores clásicos y robustos para la estimación de parámetros de pobla-ciones finitas a partir de muestras seleccionadas en forma probabilística. Los primeros po-seen el inconveniente de ser sensibles ante la aparición de valores atípicos. Una solución surge a partir del empleo de estimadores denominados robustos, los cuales son menos sen-sibles ante la existencia de outliers. Se presenta un conjunto de funciones existentes en el programa R que permite el cálculo de los estimadores y de sus correspondientes estimacio-nes de variancia.
En estudios futuros se planea la evaluación de los estimadores clásicos y robustos a partir de simulaciones considerando diversos diseños muestrales y datos contaminados con distin-tos números de observaciones atípicas.
One of the objectives of sample surveys is the estimation of parameters of variables of inter-est. One solution is the use of the classical estimators that have good distributional proper-ties such as, for example, unbiasedness. The problem arise when in a survey, values from some variables are far from the common data. Given this situation, the classical estimators present difficulties that are translated in a poor performance with respect to precision measures. We present a review of robust estimators, which are less sensitive to the appear-ance of outliers. Furthermore, we show variance estimators for the proposed estimators, as well as the available codes in the statistical program R.
One of the objectives of sample surveys is the estimation of parameters of variables of inter-est. One solution is the use of the classical estimators that have good distributional proper-ties such as, for example, unbiasedness. The problem arise when in a survey, values from some variables are far from the common data. Given this situation, the classical estimators present difficulties that are translated in a poor performance with respect to precision measures. We present a review of robust estimators, which are less sensitive to the appear-ance of outliers. Furthermore, we show variance estimators for the proposed estimators, as well as the available codes in the statistical program R.
Descripción
Palabras clave
muestreo en poblaciones finitas, estimador de Horvitz-Thompson, estima-dor de Hájek, estimadores robustos, sampling from finite populations, Horvitz-Thompson estimator, Hájek estimator, robust estimators