堆叠条形图：如何添加样本量数值与统计检验结果？

要给每个颜色的堆叠条形顶部标注样本量，需先单独统计每组样本量，再用geom_text()将标签定位到条形顶部：

# 加载依赖包
library(ggplot2)
library(dplyr)
library(ggsci)

# 计算各颜色组的样本量
sample_sizes <- my_data %>% 
  count(flower_color, name = "sample_size")

# 绘制带样本量的堆叠条形图
table(my_data$flower_color, my_data$flower_length) %>%
  as.data.frame() %>%
  filter(Freq > 0) %>%
  ggplot(aes(x = Var1, y = Freq, fill = Var2)) +
  geom_bar(position = "fill", stat = "identity") + 
  # 添加样本量标签，position_fill(vjust=1.05)确保标签在条形上方
  geom_text(data = sample_sizes, 
            aes(x = flower_color, y = 1, label = sample_size, fill = NULL),
            position = position_fill(vjust = 1.05), size = 4) +
  scale_y_continuous(labels = scales::percent_format()) +
  theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(),
        panel.background = element_blank(), axis.line = element_line(colour = "black")) +
  scale_fill_jco()  # 替代原代码中的set_palette，需加载ggsci包

你的数据是分类变量的列联表（花颜色与花长度的交叉分布），适合用**卡方检验（Chi-squared test）**验证组间分布差异；若单元格期望频数小于5则改用Fisher精确检验，你的数据集样本量足够，卡方检验适用。

方法1：添加全局卡方检验p值 #

用ggpubr包的stat_compare_means()直接在图中添加全局检验结果：

library(ggpubr)

table(my_data$flower_color, my_data$flower_length) %>%
  as.data.frame() %>%
  filter(Freq > 0) %>%
  ggplot(aes(x = Var1, y = Freq, fill = Var2)) +
  geom_bar(position = "fill", stat = "identity") + 
  geom_text(data = sample_sizes, 
            aes(x = flower_color, y = 1, label = sample_size, fill = NULL),
            position = position_fill(vjust = 1.05), size = 4) +
  scale_y_continuous(labels = scales::percent_format()) +
  theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(),
        panel.background = element_blank(), axis.line = element_line(colour = "black")) +
  scale_fill_jco() +
  # 添加全局卡方检验p值，label.y控制垂直位置
  stat_compare_means(method = "chisq", label = "p.format", label.y = 1.1)

方法2：添加组间两两比较的显著性标记 #

如果需要对比特定组（如蓝色与红色、蓝色与粉色）的分布差异，用pairwise.prop.test做事后检验，再用ggsignif的geom_signif()添加标记：

library(ggsignif)

# 执行两两比例检验，用bonferroni调整p值
pairwise_result <- pairwise.prop.test(table(my_data$flower_color, my_data$flower_length), 
                                      p.adjust.method = "bonferroni")

# 整理比较结果
compare_df <- data.frame(
  group1 = c("blue", "blue"),
  group2 = c("red", "pink"),
  p_val = pairwise_result$p.value[lower.tri(pairwise_result$p.value)],
  y_pos = c(1.05, 1.1)
) %>%
  mutate(signif = case_when(
    p_val < 0.001 ~ "***",
    p_val < 0.01 ~ "**",
    p_val < 0.05 ~ "*",
    TRUE ~ "ns"
  ))

# 绘图添加显著性标记
table(my_data$flower_color, my_data$flower_length) %>%
  as.data.frame() %>%
  filter(Freq > 0) %>%
  ggplot(aes(x = Var1, y = Freq, fill = Var2)) +
  geom_bar(position = "fill", stat = "identity") + 
  geom_text(data = sample_sizes, 
            aes(x = flower_color, y = 1, label = sample_size, fill = NULL),
            position = position_fill(vjust = 1.05), size = 4) +
  scale_y_continuous(labels = scales::percent_format()) +
  theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(),
        panel.background = element_blank(), axis.line = element_line(colour = "black")) +
  scale_fill_jco() +
  geom_signif(
    data = compare_df,
    aes(xmin = group1, xmax = group2, annotations = signif, y_position = y_pos),
    tip_length = 0.01,
    manual = TRUE
  )

内容的提问来源于stack exchange，提问作者Edoardo Scali