r-"stat_signif()"中的计算失败:数据本质上是常量

Question

我正在尝试绘制方框图，我想查看组比较，但由于某些原因，它没有显示

library(ggbeeswarm)
library(ggpubr)
library(ggplot2)
ggplot(data = data2, aes(wcc_group, 2750.3)) +
geom_boxplot() +
geom_quasirandom(aes(colour = as.factor(X28_day_mortality) , shape = sex, fill = 
as.factor(X28_day_mortality)), groupOnX = FALSE) +
labs(x = "WCC Group", y = "Apolipoprotein A-I", title = "Expression of Apolipoprotein A-I in WCC 
Groups", fill = "Mortality", shape = "Sex", colour = "Mortality") +  stat_compare_means(method = 
"t.test", comparisons = list(c("Low", "Medium"), c("Low", "High"), c("Medium", "High") ), aes(label = 
..p.signif..)) +
scale_colour_manual(values =c("red", "black")) +
scale_fill_manual(values =c("red", "black")) +
scale_shape_manual(values =c(24,22))

在此处输入图像描述

为什么这个错误不断出现？

Answer 1

当您在示例中指定y值时，它是2750.3，这是有原因的吗？

这个奇异的y值就是为每个组绘制黑线的原因。它不是方框图，因为只有一个值——不能有Q1、中位数、Q3等。如果你让y值是该变量(载脂蛋白a-I？(的值，那么方框图将能够显示分数的分布。

由于类似的原因，t检验不起作用并产生误差——由于y的唯一值为2750.3，因此没有计算t检验值的方差。

ggplot(data = data2, aes(wcc_group, [enter associated protein var here])) +
geom_boxplot() +
geom_quasirandom(aes(colour = as.factor(X28_day_mortality) , shape = sex, fill = 
as.factor(X28_day_mortality)), groupOnX = FALSE) +
labs(x = "WCC Group", y = "Apolipoprotein A-I", title = "Expression of Apolipoprotein A-I in WCC 
Groups", fill = "Mortality", shape = "Sex", colour = "Mortality") +  stat_compare_means(method = 
"t.test", comparisons = list(c("Low", "Medium"), c("Low", "High"), c("Medium", "High") ), aes(label = 
..p.signif..)) +
scale_colour_manual(values =c("red", "black")) +
scale_fill_manual(values =c("red", "black")) +
scale_shape_manual(values =c(24,22))