Evaluation of maximum likelihood procedures in ranked set sampling under imperfect ranking

AbstractAbstractRanked set sampling (RSS) is a cost-efficient alternative to simple random (SRS) when ranking small sets of population units much simpler than effectively measuring them. Maximum likelihood estimation (MLE) in RSS usually based on the restrictive assumption perfect ranking, assuming that sample lead order statistics. However, this procedure not flexible and often produces highly biased estimators errors are present. To remedy these constraints, work we suggest evaluate MLE procedures for RSS: (i) assumption; (ii) SRS assumptions; (iii) weighted estimator derived from (ii); (iv) choice or by testing null hypothesis ranking. We carried out extensive simulations investigate efficiency each strategy, which allowed us conclude although performed best under (ii-iv) preferable imperfect have also evaluated parametric bootstrap algorithm obtain confidence intervals suggested estimators. Two applications using data age determination fishes anthropometric measures athletes illustrate corroborate our findings.Keywords: Asymptotic inferenceMonte-Carlo simulationOrder statisticsParametric bootstrapPerfect testWeighted