| 高飞,刘媛媛,李长平,胡良平.如何正确运用t检验——t检验的基本概念与前提条件[J].四川精神卫生杂志,2020,33(3):211-216.Gao Fei,Liu Yuanyuan,Li Changping,Hu Liangping,How to use t test correctly——the basic concepts and preconditions of t test[J].SICHUAN MENTAL HEALTH,2020,33(3):211-216 |
| 如何正确运用t检验——t检验的基本概念与前提条件 |
| How to use t test correctly——the basic concepts and preconditions of t test |
| 投稿时间：2020-05-26 |
| DOI：10.11886/scjsws20200526003 |
| 中文关键词: 小样本 t分布 t检验 假设检验 参数检验 区间估计 |
| 英文关键词:Small sample t distribution t test Hypothesis testing Parameter test Interval estimation |
| 基金项目:国家自然科学基金项目（项目名称：贝叶斯生存分析方法在肝细胞癌肝移植患者预后预测中的应用研究，项目编号：81803333） |
|
| 摘要点击次数: |
| 全文下载次数: |
| 中文摘要: |
| 本文旨在全面介绍t检验的基本概念与前提条件（涉及实验设计和t分布两个方面）。t分布和t检验的适用场合包括：①估计符合某些条件时定量指标的参考值范围；②直线回归分析中个体因变量Y值的容许区间计算；③估计定量指标总体均数的置信区间；④线性相关与回归分析（包括多重线性回归分析）中某些参数的假设检验；⑤定量资料均值的假设检验，此时，严格地说，t检验仅适用于以下三种实验设计类型，即单组设计、配对设计和成组设计。在实际运用中，应正确辨析定量资料所取自的实验设计类型并检查资料的参数检验条件；若属于其他情形（参见前述提及的多种场合），一般都应有相应的统计学理论为依据（即可以证明问题中涉及的统计量服从t分布），否则，就可能属于滥用或误用。 |
| 英文摘要: |
| The purpose of this paper was to comprehensively introduce the concept of t test and the prerequisites for using (involving experimental design and t-distribution). The applicable occasions of t distribution and t test included: ① Estimate the reference range of quantitative indicators when certain conditions were met. ② Calculation of the tolerance interval of individual dependent variable Y in linear regression analysis. ③ Confidence interval estimation for the population mean of quantitative indicators. ④ Hypothesis testing of certain parameters in linear correlation and regression analysis (including multiple linear regression analysis). ⑤ Hypothesis testing of the mean of quantitative data, at this time, strictly speaking, the t test was only applicable to three experimental designs, single group design, paired design and group design. In practical application, the type of experimental design and parameter test conditions for quantitative data should be correctly distinguished and checked. If it belonged to other situations (please see the aforementioned multiple occasions), it should generally be based on the corresponding statistical theory (that is, it can be proved that the statistics involved in the problem follows the t distribution), otherwise it might be an abuse or misuse. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |