使用group_by后调用shapiro.test报错：unused argument (ratio_log)求助

The error you're seeing comes down to two key issues in your code:

1. You didn't save the ratio_log column to your dataframe after using mutate(). The mutate() function returns a new dataframe, but you didn't assign it back to df—so the column doesn't exist when you try to reference it in your pipe chain.
2. shapiro.test() isn't a dplyr verb, so you can't pass a column name directly to it after grouping. You need to use a dplyr function that applies the test to each group individually.

Here are a few corrected, working approaches:

Option 1: Use group_modify for tidy test results #

This method applies shapiro.test() to each group and returns the results in a clean, structured dataframe:

library(dplyr)

# Create dataframe and add calculated columns in one pipe
df <- data.frame(
  Type = c("Bark", "Redwood", "Oak"),
  size = c(10,15,13),
  width = c(3,4,5)
) %>%
  mutate(Ratio = size/width, ratio_log = log10(Ratio))

# Run Shapiro-Wilk test per group
df %>%
  group_by(Type) %>%
  group_modify(~ as.data.frame(shapiro.test(.$ratio_log)))

Option 2: Capture full test objects with summarize #

If you want to keep the complete test objects for further inspection, use summarize() with a list, then unnest the results:

library(dplyr)
library(tidyr)

df %>%
  group_by(Type) %>%
  summarize(shapiro_result = list(shapiro.test(ratio_log))) %>%
  unnest_wider(shapiro_result)

Important Note #

Your sample dataset only has one observation per group, which is too small for a Shapiro-Wilk test (it requires at least 3 observations to produce meaningful results). Make sure your actual dataset has enough data points per group before running this test.

内容的提问来源于stack exchange，提问作者Brandon Jablon