Robust Measures of Location

Abstract

Prior to the year 1960, there was a very poor understanding of how to study nonnormality and its effects on conventional hypothesis testing methods Limited studies seemed to suggest that nonnormality is not an issue when the goal is to make inferences about population means. But starting in 1960, a series of crises made it clear that finding improved methods is critical to those conducting applied research. These crises can be roughly classified into three groups, with a myriad of details in each.