Merge branch 'main' of https://github.com/easystats/effectsize

DESCRIPTION

@@ -1,7 +1,7 @@
 Type: Package
 Package: effectsize
 Title: Indices of Effect Size
-Version: 0.8.5
+Version: 0.8.6
 Authors@R: 
     c(person(given = "Mattan S.",
              family = "Ben-Shachar",
@@ -66,16 +66,16 @@ BugReports: https://github.com/easystats/effectsize/issues/
 Depends:
     R (>= 3.6)
 Imports:
-    bayestestR (>= 0.13.0),
-    insight (>= 0.19.3.2),
-    parameters (>= 0.20.2),
-    performance (>= 0.10.2),
-    datawizard (>= 0.6.5),
+    bayestestR (>= 0.13.1),
+    insight (>= 0.19.5),
+    parameters (>= 0.21.1),
+    performance (>= 0.10.5),
+    datawizard (>= 0.8.0),
     stats,
     utils
 Suggests:
-    correlation (>= 0.8.3),
-    see (>= 0.7.4),
+    correlation (>= 0.8.4),
+    see (>= 0.8.0),
     afex,
     BayesFactor,
     boot,
@@ -107,4 +107,3 @@ Config/Needs/website:
     rstudio/bslib,
     r-lib/pkgdown,
     easystats/easystatstemplate
-Remotes: easystats/insight

NEWS.md

@@ -1,3 +1,16 @@
+# effectsize 0.8.6
+
+This is a minor update to bring `effectsize` in-line with the formula methods 
+in `t.test()` and `wilcox.test()` in `R>=4.4.0`.
+
+## Breaking Changes
+
+- `cohens_d()`, `hedges_g()`, `p_superiority()`, `wmw_odds()`, `means_ratio()` and `rank_biserial()` no longer support setting `paired = TRUE` when using the formula method.
+
+## Bug fixes
+
+- `eta_squared(<gam>)` returns (approximate) effect sizes for smooths.
+
 # effectsize 0.8.5
 
 ## New features

R/cohens_d.R

@@ -206,6 +206,7 @@ glass_delta <- function(x, y = NULL, data = NULL,
   out <- .get_data_2_samples(x, y, data, paired = paired, verbose = verbose, ...)
   x <- out[["x"]]
   y <- out[["y"]]
+  paired <- out[["paired"]]
 
   if (is.null(y)) {
     if (type == "delta") {
@@ -217,6 +218,10 @@ glass_delta <- function(x, y = NULL, data = NULL,
 
   # Compute index
   if (paired) {
+    if (type == "delta") {
+      insight::format_error("This effect size is only applicable for two independent samples.")
+    }
+
     d <- mean(x - y)
     n <- length(x)
     s <- stats::sd(x - y)

R/common_language.R

@@ -123,6 +123,7 @@ p_superiority <- function(x, y = NULL, data = NULL,
   )
   x <- data[["x"]]
   y <- data[["y"]]
+  paired <- data[["paired"]]
 
   if (parametric) {
     d <- cohens_d(
@@ -162,14 +163,15 @@ cohens_u1 <- function(x, y = NULL, data = NULL,
     return(effectsize(x, type = "u1", ci = ci, verbose = verbose, ...))
   }
 
-
   data <- .get_data_2_samples(x, y, data,
     allow_ordered = !parametric,
     verbose = verbose, ...
   )
   x <- data[["x"]]
   y <- data[["y"]]
-  if (is.null(y)) insight::format_error("cohens_u3 only applicable to two sample case.")
+  if (is.null(y) || isTRUE(match.call()$paired) || isTRUE(data[["paired"]])) {
+    insight::format_error("This effect size is only applicable for two independent samples.")
+  }
 
   if (!parametric) {
     insight::format_error("Cohen's U1 only available for parametric estimation.")
@@ -202,14 +204,15 @@ cohens_u2 <- function(x, y = NULL, data = NULL,
     return(effectsize(x, type = "u2", ci = ci, verbose = verbose, ...))
   }
 
-
   data <- .get_data_2_samples(x, y, data,
     allow_ordered = !parametric,
     verbose = verbose, ...
   )
   x <- data[["x"]]
   y <- data[["y"]]
-  if (is.null(y)) insight::format_error("cohens_u3 only applicable to two sample case.")
+  if (is.null(y) || isTRUE(match.call()$paired) || isTRUE(data[["paired"]])) {
+    insight::format_error("This effect size is only applicable for two independent samples.")
+  }
 
   if (parametric) {
     d <- cohens_d(
@@ -253,7 +256,9 @@ cohens_u3 <- function(x, y = NULL, data = NULL,
   )
   x <- data[["x"]]
   y <- data[["y"]]
-  if (is.null(y)) insight::format_error("cohens_u3 only applicable to two sample case.")
+  if (is.null(y) || isTRUE(match.call()$paired) || isTRUE(data[["paired"]])) {
+    insight::format_error("This effect size is only applicable for two independent samples.")
+  }
 
   if (parametric) {
     d <- cohens_d(
@@ -289,14 +294,15 @@ p_overlap <- function(x, y = NULL, data = NULL,
     return(effectsize(x, type = "overlap", ci = ci, verbose = verbose, ...))
   }
 
-
   data <- .get_data_2_samples(x, y, data,
     allow_ordered = !parametric,
     verbose = verbose, ...
   )
   x <- data[["x"]]
   y <- data[["y"]]
-  if (is.null(y)) insight::format_error("Overlap only applicable to two sample case.")
+  if (is.null(y) || isTRUE(match.call()$paired) || isTRUE(data[["paired"]])) {
+    insight::format_error("This effect size is only applicable for two independent samples.")
+  }
 
   if (parametric) {
     d <- cohens_d(

R/eta_squared-methods.R

@@ -354,13 +354,16 @@
   model <- stats::anova(model)
 
   p.table <- as.data.frame(model$pTerms.table)
+  p.table$Component <- "conditional"
   s.table <- as.data.frame(model$s.table)
+  s.table$Component <- "smooth_terms"
+  colnames(s.table)[colnames(s.table) == "Ref.df"] <- "df"
   s.table[setdiff(colnames(p.table), colnames(s.table))] <- NA
   p.table[setdiff(colnames(s.table), colnames(p.table))] <- NA
   tab <- rbind(p.table, s.table)
-  colnames(tab)[colnames(tab) == "F"] <- "F-value"
   colnames(tab)[colnames(tab) == "df"] <- "npar"
   tab$df_error <- model$residual.df
+  # tab$df_error <- Inf
 
   out <-
     .anova_es.anova(
@@ -371,6 +374,8 @@
       ci = ci, alternative = alternative,
       verbose = verbose
     )
+  out$Component <- tab$Component
+  out <- datawizard::data_relocate(out, select = "Component", before = 1)
 
   attr(out, "anova_type") <- 3
   attr(out, "approximate") <- TRUE

R/means_ratio.R

@@ -76,6 +76,7 @@ means_ratio <- function(x, y = NULL, data = NULL,
   )
   x <- out[["x"]]
   y <- out[["y"]]
+  paired <- out[["paired"]]
 
   if (is.null(y)) {
     insight::format_error("Only one sample provided. y or data must be provided.")

R/pooled.R

@@ -28,6 +28,9 @@ sd_pooled <- function(x, y = NULL, data = NULL, verbose = TRUE, ...) {
   data <- .get_data_2_samples(x, y, data, verbose = verbose, ...)
   x <- data[["x"]]
   y <- data[["y"]]
+  if (is.null(y) || isTRUE(match.call()$paired) || isTRUE(data[["paired"]])) {
+    insight::format_error("This effect size is only applicable for two independent samples.")
+  }
 
   V <- cov_pooled(
     data.frame(x = x),
@@ -46,6 +49,9 @@ mad_pooled <- function(x, y = NULL, data = NULL,
   data <- .get_data_2_samples(x, y, data, verbose = verbose, ...)
   x <- data[["x"]]
   y <- data[["y"]]
+  if (is.null(y) || isTRUE(match.call()$paired) || isTRUE(data[["paired"]])) {
+    insight::format_error("This effect size is only applicable for two independent samples.")
+  }
 
   n1 <- length(x)
   n2 <- length(y)

R/rank_diff.R

@@ -62,28 +62,34 @@
 #' # Same as:
 #' # rank_biserial("mpg", "am", data = mtcars)
 #' # rank_biserial(mtcars$mpg[mtcars$am=="0"], mtcars$mpg[mtcars$am=="1"])
+#' # cliffs_delta(mpg ~ am, data = mtcars)
 #'
 #' # More options:
 #' rank_biserial(mpg ~ am, data = mtcars, mu = -5)
 #' print(rb, append_CLES = TRUE)
 #'
 #'
 #' # One Sample ----------
-#' rank_biserial(wt ~ 1, data = mtcars, mu = 3)
+#' # from help("wilcox.test")
+#' x <- c(1.83,  0.50,  1.62,  2.48, 1.68, 1.88, 1.55, 3.06, 1.30)
+#' y <- c(0.878, 0.647, 0.598, 2.05, 1.06, 1.29, 1.06, 3.14, 1.29)
+#' depression <- data.frame(first = x, second = y, change = y - x)
+#'
+#' rank_biserial(change ~ 1, data = depression)
+#'
 #' # same as:
-#' # rank_biserial("wt", data = mtcars, mu = 3)
-#' # rank_biserial(mtcars$wt, mu = 3)
+#' # rank_biserial("change", data = depression)
+#' # rank_biserial(mtcars$wt)
+#'
+#' # More options:
+#' rank_biserial(change ~ 1, data = depression, mu = -0.5)
 #'
 #'
 #' # Paired Samples ----------
-#' dat <- data.frame(
-#'   Cond1 = c(1.83, 0.5, 1.62, 2.48, 1.68, 1.88, 1.55, 3.06, 1.3),
-#'   Cond2 = c(0.878, 0.647, 0.598, 2.05, 1.06, 1.29, 1.06, 3.14, 1.29)
-#' )
-#' (rb <- rank_biserial(Pair(Cond1, Cond2) ~ 1, data = dat, paired = TRUE))
+#' (rb <- rank_biserial(Pair(first, second) ~ 1, data = depression))
 #'
 #' # same as:
-#' # rank_biserial(dat$Cond1, dat$Cond2, paired = TRUE)
+#' # rank_biserial(depression$first, depression$second, paired = TRUE)
 #'
 #' interpret_rank_biserial(0.78)
 #' interpret(rb, rules = "funder2019")
@@ -127,8 +133,9 @@ rank_biserial <- function(x, y = NULL, data = NULL,
     allow_ordered = TRUE,
     verbose = verbose, ...
   )
-  x <- out$x
-  y <- out$y
+  x <- out[["x"]]
+  y <- out[["y"]]
+  paired <- out[["paired"]]
 
   if (is.null(y)) {
     y <- 0
@@ -208,7 +215,7 @@ cliffs_delta <- function(x, y = NULL, data = NULL,
   )
   x <- data$x
   y <- data$y
-  if (is.null(y) || isTRUE(eval.parent(cl$paired))) {
+  if (is.null(y) || isTRUE(match.call()$paired) || isTRUE(data[["paired"]])) {
     insight::format_error("This effect size is only applicable for two independent samples.")
   }
 

R/utils_validate_input_data.R

@@ -3,11 +3,18 @@
                                 paired = FALSE, allow_ordered = FALSE,
                                 verbose = TRUE, ...) {
   if (inherits(x, "formula")) {
+    if (isTRUE(paired)) {
+      # This is to be consistent with R>=4.4.0
+      insight::format_error("cannot use 'paired = TRUE' in formula method.")
+    }
+
     # Validate:
     if (length(x) != 3L) {
       insight::format_error(
         "Formula must have one of the following forms:",
-        "\n\ty ~ group,\n\ty ~ 1,\n\tPair(x,y) ~ 1"
+        "          y ~ group   (independant samples)",
+        "          y ~ 1       (one sample)",
+        "  Pair(x,y) ~ 1       (paired samples)"
       )
     }
 
@@ -49,6 +56,7 @@
   } else if (inherits(x, "Pair")) {
     x <- x[, 1] - x[, 2]
     y <- NULL
+    paired <- TRUE
   }
 
 
@@ -90,7 +98,7 @@
   }
 
 
-  list(x = x, y = y)
+  list(x = x, y = y, paired = paired)
 }
 
 

inst/WORDLIST

@@ -83,6 +83,7 @@ Koo
 Kruschke
 LMM
 Labelled
+labelled
 Lajeunesse
 Lakens
 Landis

tests/testthat/test-eta_squared.R

@@ -660,6 +660,7 @@ test_that("ets_squared | gam", {
   b <- mgcv::gam(y ~ x0 + s(x1) + s(x2) + t2(x1, x2) + s(x3), data = dat)
 
   expect_error(out <- eta_squared(b), regexp = NA)
+  expect_warning(eta_squared(b), regexp = NA)
   expect_output(print(out), "Type III")
 })
 

tests/testthat/test-rom.R

@@ -73,7 +73,14 @@ test_that("means_ratio paired - adjusted", {
   expect_error(means_ratio(extra ~ group, data = sleep), "negative")
 
   sleep$y <- sleep$extra + 4
-  x <- means_ratio(y ~ group,
+  sleep_wide <- datawizard::data_to_wide(sleep,
+    id_cols = "ID",
+    values_from = "y",
+    names_from = "group",
+    names_prefix = "extra_"
+  )
+
+  x <- means_ratio(sleep_wide[["extra_1"]], sleep_wide[["extra_2"]],
     data = sleep,
     adjust = TRUE, paired = TRUE
   )
@@ -86,7 +93,13 @@ test_that("means_ratio paired - adjusted", {
 test_that("means_ratio paired - not adjusted", {
   data(sleep)
   sleep$y <- sleep$extra + 4
-  x <- means_ratio(y ~ group,
+  sleep_wide <- datawizard::data_to_wide(sleep,
+    id_cols = "ID",
+    values_from = "y",
+    names_from = "group",
+    names_prefix = "extra_"
+  )
+  x <- means_ratio(sleep_wide[["extra_1"]], sleep_wide[["extra_2"]],
     data = sleep,
     adjust = FALSE, paired = TRUE
   )