Inequality in the Military: Fact or Fiction?
指出，先前研究中发现的军队晋升中的种族歧视证据实际上是统计假象，源于未控制数据中的删失偏差，少数族裔士兵自愿延长服役导致了平均晋升时间的差异。
With the exception of Hauser' s (1978) cautionary note, Butler's original findings have been accepted as valid. We show that the apparent evidence of racial discrimination is a statistical artifact caused by the investigator's failure to control for censoring bias in the data. In the case of military promotions, this censoring has the perverse effect of indicating racial discrimination when the differential is, in fact, due to voluntary decisions by minority soldiers to remain in the service longer than nonminority soldiers when faced with identical promotion opportunities. The use of observed differences in mean time to promotion to show racial bias illustrates a problem with relying on statistical measures that fail to control for the use of a nonrandom sample. In his comment on Butler's article, Hauser questioned the validity of using information on mean time to promotion to infer the presence of racism. In particular, Hauser noted that