如何在R中制作符合APA格式的相关系数表

Perfect, let's build that APA-compliant correlation table in R exactly how you need it. I'll walk you through a step-by-step solution with code that's easy to adapt to your data:

Step 1: Calculate Correlations & P-Values #

First, we'll use the psych package to compute both correlation coefficients and their corresponding p-values (it's more straightforward than base R for this task). We'll use sample data from mtcars here—just replace it with your own dataset.

# Load required package
library(psych)

# Replace this with your actual dataset
my_data <- mtcars[, c("mpg", "disp", "hp", "wt")]

# Compute Pearson correlations and p-values (use method = "spearman" for rank correlations)
cor_results <- corr.test(my_data, method = "pearson")
cor_matrix <- cor_results$r  # Correlation coefficients
p_matrix <- cor_results$p    # P-values for significance

Step 2: Keep Only Lower (or Upper) Triangle #

We'll zero out the upper triangle and diagonal (since APA tables typically only show one half of the correlation matrix to avoid redundancy). If you want the upper triangle instead, just swap upper.tri() with lower.tri().

# Keep only lower triangle (set upper triangle + diagonal to NA)
lower_tri_cor <- cor_matrix
lower_tri_cor[upper.tri(lower_tri_cor, diag = TRUE)] <- NA

# For upper triangle instead:
# upper_tri_cor <- cor_matrix
# upper_tri_cor[lower.tri(upper_tri_cor, diag = TRUE)] <- NA

Step 3: Format Coefficients (2 Decimals, No Leading Zero) #

We'll create a helper function to format the correlation values: it rounds to 2 decimals and removes the leading zero (e.g., 0.25 becomes .25, -0.05 becomes -.05).

# Helper function to format correlation values
format_cor <- function(x) {
  if (is.na(x)) return("")  # Leave empty cells for unused triangle
  # Round to 2 decimals
  formatted <- sprintf("%.2f", x)
  # Remove leading zero for positive values, adjust for negatives
  formatted <- gsub("^0\\.", "\\.", formatted)
  formatted <- gsub("^-0\\.", "-\\.", formatted)
  return(formatted)
}

# Apply formatting to our triangle matrix
formatted_cors <- matrix(sapply(lower_tri_cor, format_cor), 
                         nrow = nrow(lower_tri_cor))
# Keep original column/row names
colnames(formatted_cors) <- colnames(cor_matrix)
rownames(formatted_cors) <- rownames(cor_matrix)

Step 4: Add Significance Stars #

Next, we'll map p-values to APA-style significance stars:

• * for p < 0.05
• ** for p < 0.01
• *** for p < 0.001

# Helper function to add stars based on p-value
add_stars <- function(cor_val, p_val) {
  if (is.na(p_val)) return(cor_val)  # Skip empty cells
  stars <- ""
  if (p_val < 0.001) stars <- "***"
  else if (p_val < 0.01) stars <- "**"
  else if (p_val < 0.05) stars <- "*"
  return(paste0(cor_val, stars))
}

# Combine formatted correlations with significance stars
final_table <- matrix(mapply(add_stars, formatted_cors, p_matrix), 
                      nrow = nrow(formatted_cors))
colnames(final_table) <- colnames(formatted_cors)
rownames(final_table) <- rownames(formatted_cors)

Step 5: Output as Markdown Table #

Finally, convert the matrix to a clean markdown table (great for reports, papers, or copy-pasting into documents) using the knitr package:

# Load knitr for markdown table output
library(knitr)

# Generate markdown table with caption
kable(final_table, align = "c", 
      caption = "Pearson Correlations (Lower Triangle, APA Format)")

Example Output #

For the mtcars sample data, the output will look like this:

Quick Adjustments #

• To use upper triangle instead of lower: Swap the triangle logic in Step 2
• For Spearman/Kendall correlations: Change method = "spearman" or method = "kendall" in Step 1
• To adjust significance thresholds: Modify the p-value cutoffs in the add_stars function

内容的提问来源于stack exchange，提问作者Ehsan88