单因素ANOVA后，如何在箱线图展示与TukeyHSD一致的p值？

我已经完成了单因素ANOVA和Tukey事后检验，但在把Tukey检验的p值添加到箱线图时，发现图上显示的p值和我的检验结果对不上。下面是我用到的包、分析代码、Tukey检验结果，还有绘图的代码，想请教怎么调整才能让箱线图上的p值和Tukey的结果一致？

用到的R包与ANOVA、Tukey检验代码 #

library(ggplot2) 
library(ggpubr) 
library(rstatix) 

CR.anova <- aov(CR ~ Group, data = df)
summary(CR.anova)
tukey_hsd(CR.anova)

Tukey检验结果 #

term  group1    group2    estimate conf.low conf.high p.adj    p.adj.signif
1 Group K02       K12       -1.90    -3.14    -0.660   0.00105  **
2 Group K02       BASELINE   0.153   -1.06     1.37    0.987    ns
3 Group K02       PLACEBO    0.239   -1.00     1.48    0.955    ns
4 Group K12       BASELINE   2.05     0.838    3.27    0.000286 ***
5 Group K12       PLACEBO    2.14     0.898    3.38    0.000215 ***
6 Group BASELINE  PLACEBO    0.0861  -1.13     1.30    0.998    ns

箱线图绘制代码 #

ggboxplot(
  Metabolite.data, 
  x= "Group", 
  y= "CRP",
  color = "black", 
  title = "CRP", 
  ylab = "Normalized value", 
  rug = TRUE, 
  legend = "right", 
  font.label = list(size = 40, face = "bold"), 
  linetype = "solid", 
  size = 1.1, 
  palette = c("#cc3300","#E7B800","#99ccff", "#003366"),
  fill= "Group", 
  order = c("BASELINE", "PLACEBO", "K02","K12"), 
  bxp.errorbar= TRUE, 
  alpha= 0.6
) + 
  border(color = "black", size= 2)+ 
  geom_jitter(aes(fill= Group),shape = 21, size = 4,color = "black", stroke= 1, position = position_jitter()) + 
  theme(plot.title = element_text(hjust=0.5 , face = "bold", size = 20 ),
        axis.text = element_text(face = "bold"))+ 
  theme(axis.title.x=element_blank())+ 
  theme(axis.title.y = element_text(face = "bold", size= 16 ))+ 
  theme(legend.position = "none")+ 
  stat_boxplot(geom ="errorbar",size= 1.1, width= 0.4)+ 
  stat_compare_means(comparisons = CRP_comparisons)

解决方法：手动导入Tukey检验结果进行标注 #

问题核心在于stat_compare_means()默认不会调用你已经计算好的Tukey检验p值，它会自动执行其他事后检验（比如配对t检验或Wilcoxon检验），所以才会出现结果不一致的情况。我们可以用stat_pvalue_manual()直接导入Tukey的结果来标注，步骤如下：

1. 整理Tukey检验结果为标注所需的格式

# 提取并保存Tukey检验结果
tukey_results <- tukey_hsd(CR.anova)

# 创建包含对比组、p值、显著性标记的标注数据框，y值需要根据你的数据范围调整（避免标注重叠）
tukey_p_df <- data.frame(
  group1 = tukey_results$group1,
  group2 = tukey_results$group2,
  p = tukey_results$p.adj,
  p.signif = tukey_results$p.adj.signif,
  y = c(15, 16, 17, 18, 19, 20) # 这里的数值请根据你的箱线图y轴最大值调整
)

2. 修改绘图代码，替换stat_compare_means()为stat_pvalue_manual()

ggboxplot(
  Metabolite.data, 
  x= "Group", 
  y= "CRP",
  color = "black", 
  title = "CRP", 
  ylab = "Normalized value", 
  rug = TRUE, 
  legend = "right", 
  font.label = list(size = 40, face = "bold"), 
  linetype = "solid", 
  size = 1.1, 
  palette = c("#cc3300","#E7B800","#99ccff", "#003366"),
  fill= "Group", 
  order = c("BASELINE", "PLACEBO", "K02","K12"), 
  bxp.errorbar= TRUE, 
  alpha= 0.6
) + 
  border(color = "black", size= 2)+ 
  geom_jitter(aes(fill= Group),shape = 21, size = 4,color = "black", stroke= 1, position = position_jitter()) + 
  theme(plot.title = element_text(hjust=0.5 , face = "bold", size = 20 ),
        axis.text = element_text(face = "bold"))+ 
  theme(axis.title.x=element_blank())+ 
  theme(axis.title.y = element_text(face = "bold", size= 16 ))+ 
  theme(legend.position = "none")+ 
  stat_boxplot(geom ="errorbar",size= 1.1, width= 0.4)+
  # 手动添加Tukey检验的p值标注
  stat_pvalue_manual(
    tukey_p_df,
    x = "group1", xend = "group2", y = "y",
    label = "p = {round(p, 6)} ({p.signif})", # 可以自定义显示格式，只显示标记就写"{p.signif}"
    tip.length = 0.01,
    size = 4
  )

额外提示 #

• 如果只需要显示显著性标记（比如***、**、ns），把label参数改成"{p.signif}"即可，图表会更简洁。
• y列的数值一定要根据你的实际数据调整，确保标注线不会和箱线、散点重叠，如果有重叠可以适当增大数值。

内容的提问来源于stack exchange，提问作者suad alshammari