Got it, let's tackle this problem step by step. First, I'll provide a reproducible dataset that matches your variable structure, then walk you through two reliable ways to add the correlation stats (with p-value prioritized) to your plot.

Reproducible Dataset #

First, let's create simulated data you can test with:

library(tidyverse)
set.seed(123) # Ensures results are consistent every time you run the code

sim_data <- tibble(
  Year = rep(2010:2020, each = 30),
  Position = rep(c("Manager", "Staff", "Intern"), times = 11*10),
  Score = case_when(
    Position == "Manager" ~ rnorm(330, mean = 70 + 0.8*(Year-2010), sd = 5),
    Position == "Staff" ~ rnorm(330, mean = 65 + 0.5*(Year-2010), sd = 6),
    Position == "Intern" ~ rnorm(330, mean = 60 + 0.3*(Year-2010), sd = 7)
  )
)

Method 1: Manual Calculation + Custom Annotations #

This method gives you full control over where to place your labels. We'll first compute the Pearson correlation stats for each Position group, then add them to the plot.

Step 1: Compute Grouped Correlation Stats #

cor_stats <- sim_data %>%
  group_by(Position) %>%
  summarize(
    cor_test = list(cor.test(Score, Year, method = "pearson")),
    r = round(cor_test[[1]]$estimate, 3),
    p_val = round(cor_test[[1]]$p.value, 4),
    # Format label to prioritize p-value, include r for context
    label = str_glue("p = {p_val}\nr = {r}")
  )

Step 2: Plot with Custom Labels #

We'll place labels on the right side of the plot (adjust x/y values to fit your actual data):

ggplot(sim_data, aes(x = Score, y = Year, color = Position)) +
  geom_smooth(method = "lm", se = FALSE) +
  # Add precomputed stats as text annotations
  geom_text(
    data = cor_stats,
    aes(x = quantile(sim_data$Score, 0.95), y = median(sim_data$Year), label = label),
    hjust = 0, vjust = 0.5, size = 4, show.legend = FALSE
  ) +
  theme_minimal() +
  labs(title = "Score vs. Year by Position", x = "Performance Score", y = "Year")

Method 2: Automated Labeling with ggpmisc #

For a more streamlined workflow, use the ggpmisc package to auto-generate and place the stats directly on your linear smooth lines.

Step 1: Install & Load the Package #

install.packages("ggpmisc")
library(ggpmisc)

Step 2: Plot with Auto-Generated Stats #

We'll configure the label to show p-value first, followed by r:

ggplot(sim_data, aes(x = Score, y = Year, color = Position)) +
  geom_smooth(method = "lm", se = FALSE) +
  # Add correlation stats to the plot
  stat_poly_eq(
    aes(label = paste(..p.value.label.., ..rr.label.., sep = "~~~")),
    formula = y ~ x,
    parse = TRUE,
    size = 4,
    label.x = "right" # Place labels on the right edge
  ) +
  theme_minimal() +
  labs(title = "Score vs. Year by Position", x = "Performance Score", y = "Year")

Quick Notes: #

• The ~~~ in the label string creates a line break between p-value and r.
• Adjust label.x/label.y in stat_poly_eq to move labels to your preferred position.
• If you only want to display p-value, simplify the label to ..p.value.label...

内容的提问来源于stack exchange，提问作者Evan