Sıralı küme örneklemesinde sıralama hata modelleri, maliyet ve en uygun küme büyüklüğü

Ranked Set Sampling (RSS) is a sampling method commonly used in recent years. RSS is developed as an alternative to Simple Random Sampling (SRS) in order to estimate population parameters more efficiently where the measurement of sampling units is difficult or costly but the units are easier to rank. There are several factors that make this method useful especially for studies in medicine, agriculture, forestry and ecology. The most important of these factors are the set size and the relative costs of some operations such as sampling, measurement and ranking. Ranking of the units in the set is made on the basis of the visual judgment of the researcher or a concomitant variable which has a strong correlation with the variable of interest. These ranking methods are defined as ranking error models. In this thesis, the widely used cost and ranking error models in RSS literature are investigated. Also, it is aimed to explore the effect of ranking error models on the mean estimator based on RSS and some of its modified methods for different distribution, set and cycle size in infinite population. Besides, it is aimed to examine whether RSS is cost effiective with respect to SRS in terms of mean squared error of the mean estimator considering ranking error models and the N-KPST cost model in infinite population and if so, to determine the optimal set size for RSS. Monte Carlo simulation studies are conducted for these purposes. Additionally, the study is supported by real life data.