easystats · Copilot · Aug 31, 2025 · Aug 31, 2025
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -1,7 +1,7 @@
 Package: report
 Type: Package
 Title: Automated Reporting of Results and Statistical Models
-Version: 0.6.1.2
+Version: 0.6.1.3
 Authors@R:
     c(person(given = "Dominique",
              family = "Makowski",

diff --git a/NEWS.md b/NEWS.md
@@ -2,6 +2,7 @@
 
 Bug fixes
 
+* `report()`: fix failure when reporting Kruskal-Wallis tests with degenerate cases (one observation per group) (#454)
 * Fixed issue with missing effect size for the Intercept term in type 3 anova tables (#451)
 
 # report 0.6.1

diff --git a/R/report_htest_kruskal.R b/R/report_htest_kruskal.R
@@ -10,7 +10,29 @@
 
 .report_effectsize_kruskal <- function(x, table, dot_args, rules = "funder2019") {
   args <- c(list(x), dot_args)
-  table <- do.call(effectsize::effectsize, args)
+
+  # Try to calculate effect size with confidence intervals
+  table <- tryCatch(
+    {
+      do.call(effectsize::effectsize, args)
+    },
+    error = function(e) {
+      # If CI calculation fails (e.g., degenerate cases), fallback to no CI
+      if (grepl("confidence intervals", e$message, ignore.case = TRUE) ||
+        grepl("differing number of rows", e$message)) {
+        args_no_ci <- c(list(x), dot_args, list(ci = NULL))
+        do.call(effectsize::effectsize, args_no_ci)
+      } else if (grepl("Unable to retrieve data", e$message)) {
+        # Data retrieval failed - this happens with certain htest object forms
+        # For now, we'll let this error propagate as the user should use the formula interface
+        # or provide data manually as suggested in the main report.htest function
+        stop(e)
+      } else {
+        stop(e)
+      }
+    }
+  )
+
   ci <- attributes(table)$ci
   estimate <- names(table)[1]
 
@@ -22,13 +44,23 @@
   rules <- .text_effectsize(attr(attr(interpretation, "rules"), "rule_name"))
 
   main <- paste0("Epsilon squared (rank) = ", insight::format_value(table$rank_epsilon_squared))
-  statistics <- paste0(
-    main,
-    ", ",
-    insight::format_ci(table$CI_low, table$CI_high, ci)
-  )
 
-  table <- table[names(table)[-2]]
+  # Handle cases where CI calculation failed and CI columns are missing
+  if (all(c("CI_low", "CI_high") %in% names(table))) {
+    statistics <- paste0(
+      main,
+      ", ",
+      insight::format_ci(table$CI_low, table$CI_high, ci)
+    )
+  } else {
+    # No CI available - report only the effect size
+    statistics <- main
+  }
+
+  # Remove CI_lower column if present (original code logic)
+  if (ncol(table) > 1) {
+    table <- table[names(table)[-2]]
+  }
 
   list(
     table = table, statistics = statistics, interpretation = interpretation,

diff --git a/tests/testthat/_snaps/windows/report.lmer.new.md b/tests/testthat/_snaps/windows/report.lmer.new.md
@@ -0,0 +1,45 @@
+# report-lmer
+
+    Code
+      report(m1)
+    Output
+      We fitted a linear mixed model (estimated using REML and nloptwrap optimizer)
+      to predict Reaction with Days (formula: Reaction ~ Days). The model included
+      Days as random effects (formula: ~1 + Days | Subject). The model's total
+      explanatory power is substantial (conditional R2 = 0.80) and the part related
+      to the fixed effects alone (marginal R2) is of 0.28. The model's intercept,
+      corresponding to Days = 0, is at 251.41 (95% CI [237.94, 264.87], t(174) =
+      36.84, p < .001). Within this model:
+
+        - The effect of Days is statistically significant and positive (beta = 10.47,
+      95% CI [7.42, 13.52], t(174) = 6.77, p < .001; Std. beta = 0.54, 95% CI [0.38,
+      0.69])
+
+      Standardized parameters were obtained by fitting the model on a standardized
+      version of the dataset. 95% Confidence Intervals (CIs) and p-values were
+      computed using a Wald t-distribution approximation.
+
+---
+
+    Code
+      report(m2)
+    Message
+      boundary (singular) fit: see help('isSingular')
+      boundary (singular) fit: see help('isSingular')
+    Output
+      We fitted a linear mixed model (estimated using REML and nloptwrap optimizer)
+      to predict Reaction with Days (formula: Reaction ~ Days). The model included
+      mysubgrp as random effects (formula: list(~1 | mysubgrp:mygrp, ~1 | mygrp, ~1 |
+      Subject)). The model's total explanatory power is substantial (conditional R2 =
+      0.71) and the part related to the fixed effects alone (marginal R2) is of 0.28.
+      The model's intercept, corresponding to Days = 0, is at 252.09 (95% CI [232.50,
+      271.69], t(174) = 25.39, p < .001). Within this model:
+
+        - The effect of Days is statistically significant and positive (beta = 10.35,
+      95% CI [8.78, 11.93], t(174) = 12.95, p < .001; Std. beta = 0.53, 95% CI [0.45,
+      0.61])
+
+      Standardized parameters were obtained by fitting the model on a standardized
+      version of the dataset. 95% Confidence Intervals (CIs) and p-values were
+      computed using a Wald t-distribution approximation.
+
diff --git a/tests/testthat/test-report.htest-kruskal.R b/tests/testthat/test-report.htest-kruskal.R
@@ -7,3 +7,23 @@ test_that("report.htest-kruskal report", {
     report(x, verbose = FALSE)
   )
 })
+
+test_that("report.htest-kruskal degenerate case (one observation per group)", {
+  # Test case from issue #454: one observation per group causes CI calculation to fail
+  n <- 10
+  set.seed(123)
+  df <- data.frame(a = as.factor(1:n), b = rnorm(n))
+  # Use the formula interface to allow effectsize to retrieve data
+  test <- kruskal.test(b ~ a, data = df)
+
+  # Should not fail and should complete quickly - provide data manually
+  expect_no_error({
+    result <- report(test, data = df, verbose = FALSE)
+  })
+
+  # Result should contain effect size but no CI due to degenerate case
+  result <- report(test, data = df, verbose = FALSE)
+  result_text <- as.character(result)
+  expect_true(grepl("Epsilon squared \\(rank\\) = 1\\.00", result_text))
+  expect_false(grepl("95% CI", result_text))  # No CI should be present
+})