RANKING INDIVIDUALS VERSUS GROUPS

摘要: Ranking groups (schools, regions, counties) according to the average score of their constituent parts – say, ranking schools by academic achievements students is a common yardstick in evaluation and cornerstone any planning process. In this paper we show that under certain circumstances groups, unlike individuals, can be affected examiner. The fact reversal possible does not necessarily imply test itself deficient, it merely reveals are non-homogeneous therefore, meaningless. An investigation conditions which such manipulation occur leads us suggest new statistical indicators as warning signals help one recognize situations if when they arise. It turns out these related for finding whether distribution dominates another.