方差分析基本概念
- 事前未得到数据时,根据实验设计可以确定的ANOVA类型:
- Between-Subjects ANOVA
- One-way ANOVA
- Factorial ANOVA is a general term applied when examining multiple independent variables.
- Within-Subjects ANOVA or Repeated Measures ANOVA. 需要考虑球形检验的问题。
- Mixed-Model ANOVA: A mixed model ANOVA, sometimes called a within-between ANOVA
- ANCOVA: An analysis of covariance (ANCOVA) is appropriate when examining for differences in a continuous dependent variable between groups, while controlling for the effect of additional author justinvariables.
- Multivariate analysis of variance (MANOVA) is an extension on the ANOVA, and is appropriate when examining for differences in multiple continuous level variables between groups.
- 数据采集后,同一种ANOVA类型,需要检验数据的平衡情况,即每一个因子的每个水平的个例数目是否一致,一致则平衡,不一致不平衡。当数据不一致的时候,至少有三种方法来计算平方和(sum of squares,SS),一般称为Type I, II and III 平方和。
Steps
- 使用软件:JPSP
- 导入SPSS或csv格式的数据
- 主要需要关注和报告的为group的BF01值。注意,BF01和BF10是等价的!BF01是支持H0的概率比H1的比率,BF10则是H1比H0。用BF01还是BF10可以在JPSP中进行设定。
- 贝叶斯因子为 BF01 = 113.769, 说明在(假定没有效应的)零假设下出现当前数据的可能性是在(假定存在效应的)备择假设下可能性的 113.769 倍。根据 Jeffreys (1961)提出的分类标准, 这是极强地支持了零假设, 即组间没有差异。
References
- 统计知识 | 如何报告贝叶斯因子
https://mp.weixin.qq.com/s/Kv2kyWNGpXxwsclEYG16cw
- 胡传鹏, 孔祥祯, Wagenmakers, E.-J., Ly, A., 彭凯平. (2018). 贝叶斯因子及其在JASP中的实现. 心理科学进展, 26(6), 951-965. doi:10.3724/SP.J.1042.2018.00951
- Anova – Type I/II/III SS explained https://mcfromnz.wordpress.com/2011/03/02/anova-type-iiiiii-ss-explained/
- Anova http://www.statisticshowto.com/probability-and-statistics/hypothesis-testing/anova/