两组计量资料比较时的标准化问题

两组计量资料比较时，如果存在混杂因素且内部构成不同，便可能影响两组平均水平比较的可比性，若直接用基于原始资料的平均水平进行两组间的比较有可能得出错误结论。正确的处理方法应当是：如果混杂因素在3个以下，可以应用标准化及分层标准化的方法；如果混杂因素较多，就不再适合应用标准化方法，而是考虑多因素分析。本文通过三个实例，针对医学实践中计量资料比较可能出现的问题，阐述了标准化法的思想及其应用，以便医务工作者能够根据数据本身的特点选择合适的统计学处理方法，获得两组计量资料平均水平比较的正确结论。

standardization in Comparison between Two sets of Measurement Data

When two sets of measurement data are compared and there exist confounders which are unbalanced between them, the confounders may violate the comparability. Comparing raw arithmetic mean directly may lead to wrong conclusions. A proper strategy is introduced here. If the number of confounders is less than three, we can use standardization. Otherwise, the standardization is not practicable, while statistical methods such as multiple regression analysis, co-variance analysis should be applied. In this article, three examples are discussed to show how to perform standardization in order to improve the comparability during comparison between two sets of measurement data. It is supposed to guide medical researchers to choose proper statistical methods to obtain correct conclusions when dealing with comparison among measurement data.