Retrospective Bias in the Displaced Worker Surveys
美国劳工统计局被解雇工人调查因受访者记忆随时间消退，低估了过去五年被解雇工人数量约三分之一，且记忆损失在不同人口群体中分布不均，影响分析可靠性。
The Displaced Worker Survey of the U.S. Bureau of Labor Statistics understates the number of workers displaced over the preceding five years covered by the survey by approximately one-third. The bias results from the fact that respondents' memory of displacement erodes over time. Memory loss is not randomly distributed across demographic groups, thereby making analyses based on these survey data suspect. The note reports memory-loss adjusted estimates of the number of displaced workers between 1979 and 1989 based on Displaced Worker Surveys for 1984, 1986, 1988, and 1990.