Representativity evaluation of sample micro-records in population censuses based on IPUMS database

Le résumé fourni par la source

Many demographers use samples of micro-records from population censuses for direct statistical analysis. If the sample lacks representativity, the resulting analysis may lead to invalid conclusions. It is necessary to evaluate the sample’s representativity before conducting any analytical analysis. Data comparison and sampling variance are the two main methods of assessing the representativity of samples. Studies show that the selection of sampling method should be on the basis of the needs of the researcher; under stratified double sampling, the linear estimator and ratio estimator estimating the numbers of men and women can be combined into a synthetic estimator; using actual data, the calculated sampling variance is relatively small, and the confidence intervals include parameter values, so the representativity of the sub-samples is high. A complete theoretical framework for evaluating the sample’s representativity of the census micro-records is established. It helps demographers master the method of evaluating sample representativity and draw correct conclusions from demographic analysis. The proposed sample representativity evaluation methods are not only suitable for the micro-records of population census but also for those of agricultural and economic censuses.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.