如何在R中自动化实现嵌套变量的t-test并提取统计结果

原代码错误说明 #

• 循环维度设置错误：你需要遍历的是Age和Location两个分组变量，每个检验需要同时用到同一分组下的Control和Treatment两组数据，不需要遍历Treatment
• col_t_welch传参错误：该函数接收两个数值向量x（对照组数值）和y（处理组数值），不支持公式+subset的传参方式

修正后的for循环实现 #

library(matrixTests)
set.seed(123)
df1 <- data.frame(matrix(ncol = 4, nrow = 36))
x <- c("Age","Location","Treatment","Value")
colnames(df1) <- x
df1$Age <- as.factor(rep(c(1,2,3), each = 12))
df1$Location <- as.factor(rep(c("Central","North"), each = 6))
df1$Treatment <- as.factor(rep(c("Control","Treatment"), each = 3))
df1$Value <- round(rnorm(36,200,25),0)

# 初始化结果框，和你期望的d2结构一致
d2 <- expand.grid(Age = factor(c(1,2,3)),
                  Location = factor(c("Central","North")), 
                  mean_diff = NA,
                  SE_diff = NA,
                  pvalue = NA)

# 遍历每个Age+Location组合
for (i in seq_len(nrow(d2))) {
  # 提取当前分组的子数据
  sub_df <- df1[df1$Age == d2$Age[i] & df1$Location == d2$Location[i], ]
  # 分别提取对照组和处理组的Value
  control_val <- sub_df$Value[sub_df$Treatment == "Control"]
  treat_val <- sub_df$Value[sub_df$Treatment == "Treatment"]
  # 执行welch t检验
  t_res <- col_t_welch(x = control_val, y = treat_val)
  # 写入结果
  d2$mean_diff[i] <- t_res$mean.diff
  d2$SE_diff[i] <- t_res$stderr
  d2$pvalue[i] <- t_res$pvalue
}

# 查看结果
d2

更简洁的非循环实现（可选） #

如果你不想写for循环，可以用dplyr的分组运算直接得到结果，代码更简洁易读：

library(dplyr)
library(matrixTests)

d2 <- df1 %>%
  group_by(Age, Location) %>%
  summarise(
    t_res = list(col_t_welch(x = Value[Treatment == "Control"], y = Value[Treatment == "Treatment"])),
    mean_diff = t_res[[1]]$mean.diff,
    SE_diff = t_res[[1]]$stderr,
    pvalue = t_res[[1]]$pvalue,
    .groups = "drop"
  ) %>%
  select(-t_res)

两种方法得到的结果和你示例中手动计算的d2完全一致。

内容的提问来源于stack exchange，提问作者tassones