论文部分内容阅读
目的：两大样本比较问题，传统上对同方差正态情形可采用双样本ｔ检验法，对非正态情形可采用非参数秩检验法；但对两小样本比较问题，若采用以上方法其可靠性较差。本研究旨在建立一种两小样本比较的统计处理方法。方法：假定（１）样本来自同方差正态总体（２）参数具有正态先验（３）在平方损失下，采用了经验Ｂａｙｅｓ方法。结果：建立了参数的Ｂａｙｅｓ估计及Ｂａｙｅｓ置信区间，并给出了两小样本比较的Ｂａｙｅｓ检验法则。结论：利用实例进一步说明了对两小样本问题经验Ｂａｙｅｓ方法较双样本ｔ检验法具有更高可靠性
OBJECTIVE: To compare the two samples, the traditional two-sample t-test can be used for the normal case of the same variance and the non-parametric rank test for the non-normal case. However, for the comparison of two small samples, if the above method is adopted Poor reliability. This study aims to establish a statistical comparison of two small samples. Methods: Assume that (1) the samples come from the same normal population (2) the parameters have normal a priori. (3) Empirical Bayes method is used in the square loss. Results: Bayesian estimation and Bayes confidence interval of parameters were established and the Bayes test rule of two small samples was given. CONCLUSION: The example shows that the empirical Bayes method is more reliable than the two-sample t-test for the two-sample problem