Stein's Paradox is Impossible in the Nonanticipative Context
研究了在序列估计中，若不能使用后续数据估计先前参数，则对每个参数使用可容许估计量在组合问题中也是可容许的，这与斯坦因悖论相反。
Abstract Suppose several parameters are to be estimated in sequence. Independent data relating to them are available, but data later in the sequence cannot be used to estimate earlier parameters. Then the use of an admissible estimator for each individual parameter is an admissible procedure in the combined problem using summed loss. If the estimator were not required to be nonanticipative, the use might be inadmissible as in Stein's famous inadmissibility result.