組內相關係數點估計分析

Abstract

組內相關係數的運用領域橫跨自然科學與社會科學，反應了總變異可以被組間變異解釋的程度。由於應用領域廣泛，因此組內相關係數估計量數量眾多，本文由過去文獻挑選出五項組內相關係數估計量進行計算評比。因此探討在母體相關係數、組內樣本數與組數所組成的各種情境下，加入常態分配與非常態分配情境模擬，計算此五項估計量在各種不同的情境下的期望值偏誤、偏誤比例與平均平方誤差，並對其進行評比。

Intraclass correlation coefficients are used extensively to measure the reliability or degree of resemblance among group members in multilevel research. The study compares the behavior of several intraclass correlation estimators in terms of bias, bias ratio, and mean square error. Monte Carlo simulation results revealed that the mutual dominance relationship among the indices. The numerical investigations provide guidelines to help researchers choose between the competing measures.