Problemy jakości danych statystycznych w przypadku badania cech rzadkich

摘要

The authors presented estimation of parameters for rare characteristics as well as methods of the sample selection to the rare and hardly perceptible populations, which may provide an alternative or complement to research aimed at the common characteristics. The basic sampling schemes for rare characteristics, including the inverse sampling, capture-recapture sampling, location sampling, as well as line-intercept sampling and link-tracing design were described. Also the multiple frame concept and the trials of increased coverage are presented in the article. Difficulties arising from the use of conventional sampling schemes were exemplified with a simulation based on data from the Agricultural Census and household budget surveys. The existence of alternatives for such cases, along with the advantages and disadvantages of those solutions are pointed, too.%W artykule przedstawiono szacowanie parametrów dla cech rzadkich oraz metody wyboru próby dla populacji rzadkich i trudno uchwytnych, które mogą stanowić alternatywę lub uzupełnienie badań nakierowanych na cechy powszechne. Zaprezentowano podstawowe schematy losowania dla cech rzadkich, w tym losowania: odwrotnego (inverse sampling), wielokrotnego połowu (losowanie typu pojmanie-uwolnienie, capture-recapture sampling), lokacyjnego (location sampling) oraz schematy linia-przecięcie (line-intercept sampling) i śledzenia łączy (link-tracing design).