Comparación entre dos estrategias bietápicas de muestreo para la estimación de un total en diferentes escenarios simulados

Abstract

This work is about the comparison between two strategies of two stages sampling through estimated coe cients of variation (ECV) produced by each strategy. The rst strategy is a sampling proportional to a variable of auxiliary information, in addition to a simple random sampling named PS-SRS; the second strategy is a strati ed random sampling named STSI-SRS. Both strategies use the Horvitz - Thompson estimator. This comparison was made by using a variable of simulated answer in several scenarios, where it changed the correlation coe cient between the variable of interest and the variable of auxiliary information (total of people per geographical area of the database of Houses, Homes and People - VIHOPE of DANE) and it changed too the rate between the intercept of linear model which links them and the average of the simulated variable, in order to determine the performance of each strategy in each simulated scenario and decide which one has the best performance, according to the estimated coe cient of variation (ECV) produced by each strategy.