Introducción al diseño de una muestra maestra para la provincia de Santa Fe
No Thumbnail Available
Date
2022-04-25
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Description
Las oficinas de estadística utilizan entre sus operativos a las encuestas a hogares para
realizar mediciones de indicadores específicos de las problemáticas públicas, con el fin
de favorecer la toma de decisiones políticas para mejorar la calidad de vida humana
en general. Para ello se valen de operativos que, en su mayoría, se apoyan en
muestras que son seleccionadas a través de procedimientos probabilísticos, los cuales aseguran la representatividad de la muestra respecto a la población. Para la selección de una muestra por procedimientos aleatorios, se requiere un marco de muestreo que está compuesto por el conjunto de materiales que sirven para identificar, localizar y acceder a cada uno de los elementos de la población. Una opción para la construcción
de un marco es la de seleccionar una muestra grande en una primera fase inicial que
posteriormente sirva de marco para seleccionar las muestras. El IPEC tiene como
objetivo la realización de una Muestra Maestra en la provincia de Santa Fe, la cual
estará basada en el diseño de la MMUVRA desarrollada por el INDEC. El objetivo del
presente trabajo es comenzar con el estudio de los distintos componentes que forman
parte del diseño muestral a partir del cual se construirá la Muestra Maestra para ir
determinando aquellos aspectos que brinden estimaciones más precisas. Se consideró
el diseño muestral que se emplea en la localidad más grande de la provincia, Rosario,
teniendo como objetivo evaluar la precisión que se obtiene en la estimación de
parámetros relacionados con el ámbito laboral que forman parte de la Encuesta
Permanente de Hogares. Se consideraron diseños con distintos métodos de selección
y distintos tamaños de unidades de muestreo de primera etapa basados en las
desagregaciones geográficas censales que considera el INDEC en los Censos de
Población. Las unidades de mayor tamaño formadas por la unión de 2 o 3 radios
censales tuvieron los mejores comportamientos en términos de precisión, mientras que
las desagregaciones menores, radios y segmentos, tuvieron un desempeño pobre. El
diseño que considera probabilidades de inclusión distintas para las unidades de
primera etapa brinda mejores resultados respecto a la precisión comparado a un
diseño con igual probabilidad como el muestreo simple al azar
National Statistical Offices use household surveys among their statistical tools to measure specific indicators of public problems, to take political decision-making to improve the quality of human life in general. To do this, they use operations that rely on samples that are selected through probabilistic procedures, which ensure the representativeness of the sample with respect to the population. For the selection of a sample by random procedures, a sampling frame is required that is made up of the set of materials that serve to identify, locate, and access each of the elements of the population. One option for the construction of a sampling frame is to select a large sample in a first initial phase that later serves as a frame to select the samples. The objective of IPEC is to carry out a Master Sample in the province of Santa Fe, which will be based on the design of the MMUVRA developed by INDEC. The objective of this work is to begin with the study of the different components that are part of the sample design from which the Master Sample will be built to determine those aspects that provide more precise estimates. The sample design used in the largest town in the province, Rosario, was considered, with the objective of evaluating the precision obtained in the estimation of parameters related to the work environment, which are part of the Permanent Household Survey. Designs with diverse selection methods and different sizes of first-stage sampling units were considered based on the geographical census disaggregation considered by INDEC in the Population Censuses. The largest units formed by the union of 2 or 3 census radios had the best behaviour in terms of precision, while the smaller disaggregation, radios and segmentos, had an inferior performance. The design which considers different inclusion probabilities for the primary stage units provides better results regarding precision compared to a design with equal probability such as simple random sampling.
National Statistical Offices use household surveys among their statistical tools to measure specific indicators of public problems, to take political decision-making to improve the quality of human life in general. To do this, they use operations that rely on samples that are selected through probabilistic procedures, which ensure the representativeness of the sample with respect to the population. For the selection of a sample by random procedures, a sampling frame is required that is made up of the set of materials that serve to identify, locate, and access each of the elements of the population. One option for the construction of a sampling frame is to select a large sample in a first initial phase that later serves as a frame to select the samples. The objective of IPEC is to carry out a Master Sample in the province of Santa Fe, which will be based on the design of the MMUVRA developed by INDEC. The objective of this work is to begin with the study of the different components that are part of the sample design from which the Master Sample will be built to determine those aspects that provide more precise estimates. The sample design used in the largest town in the province, Rosario, was considered, with the objective of evaluating the precision obtained in the estimation of parameters related to the work environment, which are part of the Permanent Household Survey. Designs with diverse selection methods and different sizes of first-stage sampling units were considered based on the geographical census disaggregation considered by INDEC in the Population Censuses. The largest units formed by the union of 2 or 3 census radios had the best behaviour in terms of precision, while the smaller disaggregation, radios and segmentos, had an inferior performance. The design which considers different inclusion probabilities for the primary stage units provides better results regarding precision compared to a design with equal probability such as simple random sampling.
Keywords
Muestra maestra, Diseño muestral complejo, Efecto de diseño, Master sample, Complex sampling design, Design efect