问题可能是你的df$width变量。尝试设置它们，使每个类别的总和为1：

set.seed(1) # Make the result reproducible

df <- data.frame(
  group = rep(c("A", "B", "C"), each = 3),
  category = rep(c("X", "Y", "Z"), 3),
  value = rnorm(9)
)

df$width = ifelse(df$category=="X",.5,.25) # This is what I changed

ggplot(df, aes(x = group, y = value, fill = category)) +
  geom_col(width = df$width,
           position = position_dodge2(padding = c(0.5,0.5,0.5))) + # Also edited your padding variables
  scale_fill_manual(values = c("red", "green", "blue"))

结果：

在position_dodge2()函数中使用preserve参数，这是您想要的吗？

ggplot(df, aes(x = group, y = value, fill = factor(category))) +
  geom_col(width = df$width,position = position_dodge2(padding = c(0.5,0,0),
                                                       preserve = 'single')) +
  scale_fill_manual(values = c("red", "green", "blue"))

根据OP的注解更新，尝试删除宽度参数。

ggplot(df, aes(x = group, y = value, fill = factor(category))) +
  geom_col(position = position_dodge2(padding = c(0.5,0,0),
                                                       preserve = 'single')) +
  scale_fill_manual(values = c("red", "green", "blue"))