期刊名称：The Proceedings of Zagreb Faculty of Economics and Business
印刷版ISSN：1333-8900
出版年度：2016
卷号：14
期号：2
页码：19-44
语种：English
出版社：Faculty of Economics and Business, University of Zagreb
摘要：In some cases, a sampling approach must be used instead of a census even if the target population is small. In such a case, the drawn sample will also be small and every sample unit can significantly influence the results. This paper aims to address the aforementioned issue from the perspective of different sampling procedures. The observed small population consists of 57 Croatian hospitals. The averages of total due obligations for medicines of Croatian hospitals in 2014 are estimated using three different sampling procedures and compared to the real population parameter value. The analysis has shown that the minimum recommended overall sampling rate, which would result in satisfactory precision and accuracy levels of parameter estimates, for simple random sampling and stratified random sampling is 70%. The two-stage cluster sampling procedure, on the other hand, did not achieve a satisfactory precision and accuracy level at any of the observed sample sizes because of a high level of heterogeneity between clusters.
