Goodness-of-fit tests in ranked set sampling

In literature, many authors have studied goodness-of-fit (GOF) tests based on ranked set sampling (RSS). In these studies, many different distribution function estimators have been suggested. In this thesis, empirical distribution function (EDF) estimators based on sampling designs which are level-0, level-1 and level-2 in RSS are proposed and GOF tests based on EDF are studied. Also, efficiencies of these EDF estimators are investigated with respect to EDF estimator of simple random sampling (SRS) under perfect and imperfect ranking for finite population. Moreover, powers of different EDF based GOF test statistics for the sampling designs are examined under perfect ranking for finite population. Besides the sampling designs, partially rank-ordered set (PROS) is used in RSS procedure. By using different simulation algorithms, powers, critical values for different GOF tests and efficiencies of the EDF estimators are obtained. Based on these efficiency and power values, in general, it is observed that RSS has higher performance than SRS. These results are presented in tables and illustrated in figures.