diff --git a/analysis/generate_paper.sh b/analysis/generate_paper.sh
index 4b830592..51a55870 100755
--- a/analysis/generate_paper.sh
+++ b/analysis/generate_paper.sh
@@ -59,7 +59,7 @@ main() {
   # RQ1
   #
   Rscript analysis/paper/performance_distribution.R "${TMP_DIR}/decomposition_scores.csv" "${PAPER_REPOSITORY}/figures/rq1-performance-distribution.pdf"
-  Rscript analysis/paper/statistical_analysis_untangling_tool.R "${TMP_DIR}/decomposition_scores.csv" "${PAPER_REPOSITORY}/data/rq1.txt"
+  Rscript src/r/main/statistical_analysis_untangling_tool.R "${TMP_DIR}/decomposition_scores.csv" "${PAPER_REPOSITORY}/data/"
   Rscript analysis/paper/compare_models.R "${TMP_DIR}/decomposition_scores.csv" "${PAPER_REPOSITORY}/tables/model-comparison.tex" > "${PAPER_REPOSITORY}"/data/model-comparison.txt
 
   #
diff --git a/analysis/paper/statistical_analysis_untangling_tool.R b/analysis/paper/statistical_analysis_untangling_tool.R
deleted file mode 100755
index ccc510e0..00000000
--- a/analysis/paper/statistical_analysis_untangling_tool.R
+++ /dev/null
@@ -1,58 +0,0 @@
-#!/usr/bin/env Rscript
-#
-# Generates statistical analysis for the untangling tools with p-value, Cohen's d, R^2, and model residuals normality test results.
-#
-# Arguments:
-# - 1: The `decomposition_scores.csv` file generated by `decompose.sh`.
-# - 1: The path to the file where the results will be saved.
-#
-# Output:
-# The results are saved as text data. The output file contains
-# the output of the summary() and cohen.d() functions.
-
-library(librarian)
-library(tidyverse)
-library(car)
-library(ggpubr)
-library(lme4)
-library(effsize)
-library(lmerTest)
-library(flexplot)
-library(rsq)
-shelf(broom)
-
-args = commandArgs(trailingOnly=TRUE)
-
-if (length(args)!=2) {
-  stop("Please provide an input file and output file.", call.=FALSE)
-}
-inputFile = args[1]
-outputFile = args[2]
-
-data <- read.csv(inputFile, header = FALSE, col.names = c('Project', 'BugID', 'SmartCommit', 'Flexeme', 'FileUntangling'))
-data <- subset(data, select = -c(FileUntangling))
-data$BugID <- as_factor(data$BugID)
-
-# Convert to long format
-data_long = pivot_longer(data, cols = 3:4, names_to = 'Tool', values_to = 'Performance')
-
-# The summary can be interpreted as follows:
-
-# Intercept row shows whether the baseline treatment (whichever is first) is significantly different from 0.
-# The second row, containing the other treatment, shows whether the other treatment is significantly
-# different from the intercept.
-
-model <- lm(Performance ~ Tool, data=data_long)
-# Residuals
-# It is recommended to look at the residuals to check for normality rather than apply a statistical test.
-visualize(model, "residuals")
-
-sink(outputFile)
-summary(model)
-#estimates(model_simple)
-
-tidy(model)
-
-cohen.d(data_long$Performance[data_long$Tool == "SmartCommit"], data_long$Performance[data_long$Tool == "Flexeme"])
-sink()
-
diff --git a/src/r/main/statistical_analysis_untangling_tool.R b/src/r/main/statistical_analysis_untangling_tool.R
new file mode 100755
index 00000000..a7ab7d39
--- /dev/null
+++ b/src/r/main/statistical_analysis_untangling_tool.R
@@ -0,0 +1,64 @@
+#!/usr/bin/env Rscript
+#
+# Run and export the statistical analysis for the untangling tools with p-value, Cohen's d, R^2, and model residuals normality test results.
+# The statistical analysis results are saved in two files:
+# - `outputDir/untangling_tool_performance_statistical_analysis.txt` contains the summary of the model, ANOVA, Shapiro-Wilk normality test, and Cohen's d.
+# - `outputDir/untangling_tool_performance_residuals.pdf` contains the residual plots of the model.
+#
+# Arguments:
+# - 1: The untangling performance file `decomposition_scores.csv` containing the performance of the untangling tools for a dataset.
+# - 1: The path to the directory where the analysis results will be saved.
+#
+# Output:
+# The results are saved as text data. The output file contains
+# the output of the summary() and cohen.d() functions.
+
+args = commandArgs(trailingOnly=TRUE)
+
+if (length(args)!=2) {
+  stop("Please provide the untangling performance file and the path where to store the results. Example: 'statistical_analysis_untangling_tool.R tool_performance.csv analysis/'", call.=FALSE)
+}
+untanglingPerformanceFile = args[1]
+outputDir= args[2]
+
+residuals_plot_file = paste(outputDir, "untangling_tool_performance_residuals.pdf", sep="/")
+untangling_tool_performance_statistical_analysis_file = paste(outputDir, "untangling_tool_performance_statistical_analysis.txt", sep="/")
+
+library(librarian)
+library(tidyverse)
+library(car)
+library(ggpubr)
+library(lme4)
+library(effsize)
+library(lmerTest)
+library(flexplot)
+library(rsq)
+shelf(broom)
+
+data <- read.csv(untanglingPerformanceFile, header = FALSE, col.names = c('Project', 'BugID', 'SmartCommit', 'Flexeme', 'FileUntangling'))
+data <- subset(data, select = -c(FileUntangling))
+data$BugID <- as_factor(data$BugID)
+
+# Convert to long format and select only SmartCommit and Flexeme to compare.
+data_long = pivot_longer(data, cols = c('SmartCommit', 'Flexeme'), names_to = 'Tool', values_to = 'Performance')
+
+# The summary can be interpreted as follows:
+# Intercept row shows whether the baseline treatment (whichever is first) is significantly different from 0.
+# The second row, containing the other treatment, shows whether the other treatment is significantly
+# different from the intercept.
+model <- lm(Performance ~ Tool, data=data_long)
+
+# Residuals
+# It is recommended to look at the residuals to check for normality rather than apply a statistical test.
+pdf(residuals_plot_file)
+visualize(model, "residuals")
+dev.off()
+
+# Analysis results
+sink(untangling_tool_performance_statistical_analysis_file)
+summary(model)
+anova(model)
+shapiro.test(residuals(model))
+cohen.d(data_long$Performance[data_long$Tool == "SmartCommit"], data_long$Performance[data_long$Tool == "Flexeme"])
+sink()
+