Stings in the tails: Detecting and dealing with censored data
Variables often show evidence of clustering at extreme values and of graininess, that is, of a limited number of distinct values. Scores on two subscales of a quality-of-life measure, traditionally analyzed with OLS regression or ANOVA models, provide examples. Ignoring or failing to detect such features of the data will result in poor estimates of effect size. Copyright 2005 by StataCorp LP.