多因子水平在两个分数之间的Cohen d

问题描述

我有9个不同的因子水平（等级），两个得分（得分1和得分2）不同。

通过使用以下代码，我可以毫无问题地获得这两个分数之间的效果大小：

cohens_d(df$score1,df$score2)

我想要做的是针对每个因子水平计算这两个得分之间的效应大小。以下代码不起作用：

df %>%
group_by(class) %>%
cohens_d(score1,score2)

到目前为止，我已经尝试了lsr，effsize和effectize软件包，但是它们都不起作用。

编辑2：这是我得到的错误：

Error in .deal_with_cohens_d_arguments(x,y,data) : object 'score1' not found

看起来像我的代码的前两行，该行应该用来选择数据并按因子级别进行分组。

df %>%
group_by(class) %>%

谢谢。

这是dput（head（df，20））的结果：

structure(list(`Filter 1` = c(1L,1L,1L),ID = 301:320,Grade = structure(c(1L,.Label = c("class1","class2","class3","class4","class5","class6","class7","class8","class9"),class = "factor"),Class = structure(c(9L,9L,9L),.Label = c("8/D","8/G","8/A","8/B","7/E","7/G","6/H","6/D","5/G","5/H","5CK","5AA","adults","7/D"),birthdate = c(2005,2005,2005),Q1 = c(1L,0L,Q2 = c(0,1,0),Q3 = c(1,1),Q4 = c(1,Q5 = c(0,Q6 = c(1,Q7 = c(1,Q8 = c(1,Q9 = c(0,Q10 = c(1,Q11 = c(1,Q12 = c(1,Q13 = c(0,Q14 = c(1,Q15 = c(0,Q16 = c(1,Q17 = c(1,Q18 = c(0,Q19 = c(1,Q20 = c(0,Q21 = c(1,Q22 = c(0,Q23 = c(1,Q24 = c(1,Q25 = c(1,Q26 = c(0,Q27 = c(1,Q28 = c(1,Q29 = c(1,Q30 = c(0,Q31 = c(0,Q32 = c(1,Q33 = c(1,Q34 = c(1,Q35 = c(1,Q36 = c(0,Q37 = c(0,Q38 = c(1L,0L),Q39 = c(1,Q40 = c(1,Q41 = c(0,Q42 = c(1,Q43 = c(0,Q44 = c(0,Q45 = c(1,Q46 = c(0,Q47 = c(1,Q48 = c(0,Q49 = c(0,Q50 = c(0,totalscore = c(30,16,19,15,24,17,32,36,20,42,12,18,40,25,6,9,21,15),transformed = c(-0.489961597251375,-1.6416086159321,-1.29700091350809,-1.78769329956576,-1.02498741152925,-1.52396770612663,-0.347537009605224,0.0378608125108563,-1.19639199386294,0.72280668820074,-2.01768420296626,-1.40146915348254,0.471079599763142,-0.94308469425977,-2.93312341654017,-2.57351548813345,-1.13020640446587,-1.78769329956576),score2= c(8,5,4,2,8,10,7,11,5
    ),score1= c(8,3,13,4)),row.names = c(NA,-20L),class = c("grouped_df","tbl_df","tbl","data.frame"),groups = structure(list(Grade = structure(1L,.rows = list(1:20)),-1L),class = c("tbl_df",.drop = TRUE))

解决方法

这是一种方法。规则很简单，如果要在dplyr管道中计算某些内容，请使用mutate或summarise，请勿直接通过管道传输到要使用的函数。

library(dplyr)
library(apa)

df %>%
  group_by(Class) %>%
  mutate(effect = cohens_d(score1,score2))

,

如果此功能适用于整个数据帧：

cohens_d(df$score1,df$score2)

然后考虑将by跨因子级别运行以返回cohens_d结果的命名列表，其中列表中的项目数等于因子级别数：

cohens_d_list <- by(df,df$grade,function(sub) 
                       cohens_d(sub$score1,sub$score2)
                 )

cohens_d_list$class1
cohens_d_list$class2
cohens_d_list$class3
...