如何在同一画布重叠绘制两个emmeans对象的估计边际均值图

现有两个emmeans输出表如下：

emm_1
$emmeans
 Order rate   SE  df asymp.LCL asymp.UCL
 1    19.3 1.51 Inf      16.5      22.5
 2    26.0 2.33 Inf      21.8      31.0

emm_2
$emmeans
 Order rate   SE  df asymp.LCL asymp.UCL
 1    25.6 1.62 Inf      22.6      28.9
 2    18.8 2.34 Inf      14.8      24.0

最初尝试的绘图代码如下，运行后仅显示第二张emmeans的图：

plot(emm_1,col="steelblue4") + theme_bw() +
  labs(title = "Choice1", 
       x = "Estimated marginal mean",
       y = "Order") + theme(plot.title = element_text(hjust = 0.5)) +
  scale_x_continuous(breaks = seq(0, 33,5), limits =c(0,33))
par(new=TRUE)
plot(emm_2,col="green") + theme_bw() +
  labs(title = "Choice2", 
       x = "Estimated marginal mean",
       y = "Order") + theme(plot.title = element_text(hjust = 0.5)) +
  scale_x_continuous(breaks = seq(0, 33,5), limits =c(0,33))

问题原因 #

par(new=TRUE)是R基础绘图系统的参数，对你当前使用的ggplot2绘图系统完全无效，每次运行ggplot的+语法生成的绘图对象，只有最后打印的那个会显示在画布上，之前的会被直接覆盖。

实现方案 #

该需求完全可以实现，有两种常见实现方式，按需选择：

方案1：两组数据放在同一坐标轴内绘制 #

先把两个emmeans结果提取出来，添加分组标识后合并，再统一绘图，最终两组结果会展示在同一个坐标系下，用颜色区分组别：

# 加载需要的包
library(emmeans)
library(ggplot2)
library(dplyr)

# 提取emmeans数据并添加分组标识
emm1_df <- as.data.frame(emm_1$emmeans) %>% mutate(group = "Choice1")
emm2_df <- as.data.frame(emm_2$emmeans) %>% mutate(group = "Choice2")
all_emm <- bind_rows(emm1_df, emm2_df)
all_emm$Order <- factor(all_emm$Order) # 将Order转成因子方便绘图

# 统一绘图
ggplot(all_emm, aes(x = rate, y = Order, color = group)) +
  geom_point(position = position_dodge(width = 0.3), size = 2) +
  geom_errorbarh(aes(xmin = asymp.LCL, xmax = asymp.UCL), 
                 height = 0.2, position = position_dodge(width = 0.3)) +
  theme_bw() +
  labs(title = "两组估计边际均值对比", 
       x = "Estimated marginal mean",
       y = "Order",
       color = "组别") + 
  theme(plot.title = element_text(hjust = 0.5)) +
  scale_x_continuous(breaks = seq(0, 33,5), limits =c(0,33)) +
  scale_color_manual(values = c("Choice1" = "steelblue4", "Choice2" = "green"))

方案2：分面展示两组结果 #

如果希望保留两个独立的子图，排布在同一画布中，可以用patchwork包拼接两个ggplot对象：

# 加载patchwork包
library(patchwork)

# 分别生成两个绘图对象
p1 <- plot(emm_1,col="steelblue4") + theme_bw() +
  labs(title = "Choice1", 
       x = "Estimated marginal mean",
       y = "Order") + 
  theme(plot.title = element_text(hjust = 0.5)) +
  scale_x_continuous(breaks = seq(0, 33,5), limits =c(0,33))

p2 <- plot(emm_2,col="green") + theme_bw() +
  labs(title = "Choice2", 
       x = "Estimated marginal mean",
       y = "Order") + 
  theme(plot.title = element_text(hjust = 0.5)) +
  scale_x_continuous(breaks = seq(0, 33,5), limits =c(0,33))

# 横向拼接两个图（如果要纵向拼就换成p1/p2）
p1 + p2

内容的提问来源于stack exchange，提问作者Camila Jones