Add validation to ensure es_type parameter is only used with effectsize = "boot"

R/estimate_contrasts.R

@@ -71,8 +71,9 @@
 #' size will be computed.
 #' @param es_type Specifies the type of effect-size measure to estimate when
 #' using `effectsize = "boot"`. One of `"unstandardized"`, `"cohens.d"`,
-#' `"hedges.g"`, `"cohens.d.sigma"`, `"r"`, or `"akp.robust.d"`. See`
-#' effect.type` argument of [bootES::bootES] for details.
+#' `"hedges.g"`, `"cohens.d.sigma"`, `"r"`, or `"akp.robust.d"`. See `effect.type`
+#' argument of [`bootES::bootES()`] for details. If not specified, defaults to
+#' `"cohens.d"`.
 #' @param iterations The number of bootstrap resamples to perform.
 #' @inheritParams estimate_means
 #'
@@ -301,7 +302,7 @@ estimate_contrasts.default <- function(
   keep_iterations = FALSE,
   effectsize = NULL,
   iterations = 200,
-  es_type = "cohens.d",
+  es_type = NULL,
   backend = NULL,
   verbose = TRUE,
   ...
@@ -317,6 +318,12 @@ estimate_contrasts.default <- function(
   # validate input
   estimate <- .validate_estimate_arg(estimate)
   comparison <- .check_for_inequality_comparison(comparison)
+  # Validate es_type usage
+  if (is.null(effectsize) && !is.null(es_type)) {
+    insight::format_error(
+      "`es_type` can only be used when `effectsize` is specified. Currently `effectsize = NULL`."
+    )
+  }
 
   if (backend == "emmeans") {
     # Emmeans ----------------------------------------------------------------

R/estimate_contrasts_effectsize.R

@@ -1,15 +1,23 @@
-.estimate_contrasts_effectsize <- function(model,
-                                           estimated,
-                                           contrasts_results,
-                                           effectsize,
-                                           bootstraps,
-                                           bootES_type,
-                                           backend) {
+.estimate_contrasts_effectsize <- function(
+  model,
+  estimated,
+  contrasts_results,
+  effectsize,
+  bootstraps,
+  bootES_type,
+  backend
+) {
   # Add standardized effect size
   insight::validate_argument(effectsize, c("none", "emmeans", "marginal", "boot"))
 
   if (effectsize == "emmeans" && backend != "emmeans") {
-    insight::format_error("`effectsize = \"emmeans\"` only possible with `backend = \"emmeans\"`")
+    insight::format_error(
+      "`effectsize = \"emmeans\"` only possible with `backend = \"emmeans\"`"
+    )
+  }
+
+  if (!is.null(bootES_type) && effectsize != "boot") {
+    insight::format_error("`es_type` can only be used when `effectsize = \"boot\"`.")
   }
 
   # Check if the model includes any random effects. Effect size calculations in
@@ -36,7 +44,8 @@
     ))
   }
 
-  switch(effectsize,
+  switch(
+    effectsize,
     emmeans = {
       eff <- emmeans::eff_size(
         estimated,
@@ -54,10 +63,16 @@
       # d_adj <- contrasts$t * contrasts$SE / sigma(model) * sqrt(1 - R2)
       # New: d_adj <- difference * (1- R2)/ sigma
       R2 <- summary(model)$r.squared
-      d_adj <- contrasts_results$Difference * (1 - R2) / insight::get_sigma(model, verbose = FALSE)
+      d_adj <- contrasts_results$Difference *
+        (1 - R2) /
+        insight::get_sigma(model, verbose = FALSE)
       contrasts_results <- cbind(contrasts_results, marginal_d = d_adj)
     },
     boot = {
+      # set default
+      if (is.null(bootES_type)) {
+        bootES_type <- "cohens.d"
+      }
       insight::check_if_installed("bootES")
       dat <- insight::get_data(model)
       resp <- insight::find_response(model)
@@ -84,7 +99,11 @@
 
       eff <- do.call(rbind, es.lists)
       eff <- eff[1:3]
-      names(eff) <- c(bootES_type, paste0(bootES_type, "_CI_low"), paste0(bootES_type, "_CI_high"))
+      names(eff) <- c(
+        bootES_type,
+        paste0(bootES_type, "_CI_low"),
+        paste0(bootES_type, "_CI_high")
+      )
 
       contrasts_results <- cbind(contrasts_results, eff)
     }

R/estimate_means.R

@@ -306,7 +306,7 @@
 #'
 #' The output from `equivalence_test()` returns a column `SGPV`, the "second
 #' generation p-value", which is equivalent to the `p (Equivalence)` column when
-#' using the `equivalence` argument. It is basically representiv the ROPE coverage
+#' using the `equivalence` argument. It is basically representative of the ROPE coverage
 #' from the confidence interval of the estimate (i.e. the proportion of the
 #' confidence intervals that lies within the region of practical equivalence).
 #'
@@ -414,16 +414,18 @@
 #' estimate_means(model, by = "Sepal.Width", length = 3)
 #' }
 #' @export
-estimate_means <- function(model,
-                           by = "auto",
-                           predict = NULL,
-                           ci = 0.95,
-                           estimate = NULL,
-                           transform = NULL,
-                           keep_iterations = FALSE,
-                           backend = NULL,
-                           verbose = TRUE,
-                           ...) {
+estimate_means <- function(
+  model,
+  by = "auto",
+  predict = NULL,
+  ci = 0.95,
+  estimate = NULL,
+  transform = NULL,
+  keep_iterations = FALSE,
+  backend = NULL,
+  verbose = TRUE,
+  ...
+) {
   # Process argument ---------------------------------------------------------
   # --------------------------------------------------------------------------
 
@@ -469,12 +471,16 @@ estimate_means <- function(model,
   info <- attributes(estimated)
 
   # Table formatting
-  attr(means, "table_title") <- c(switch(estimate,
-    specific = "Model-based Predictions",
-    typical = "Estimated Marginal Means",
-    average = "Average Predictions",
-    population = "Average Counterfactual Predictions"
-  ), "blue")
+  attr(means, "table_title") <- c(
+    switch(
+      estimate,
+      specific = "Model-based Predictions",
+      typical = "Estimated Marginal Means",
+      average = "Average Predictions",
+      population = "Average Counterfactual Predictions"
+    ),
+    "blue"
+  )
 
   attr(means, "table_footer") <- .table_footer(
     means,

inst/WORDLIST

@@ -1,45 +1,39 @@
-’s
 ANCOVAs
-Arel
-ATE
 ATEs
 ATT
 ATU
+Analysing
+Arel
 Axelsson
-BH
 BLUPs
 Bundock
-cdot
-Chatton
 CJ
-coefficients’
-diag
-Dickerman
-Dom
+Chatton
 DS
+Dickerman
 EFC
 EMM
 EMMs
 EUROFAMCARE
 Fitzsimons
 GAMM
+GLM's
 GLMM
 GLMMs
-GLM's
 GMM
 GMMs
 Greifer
 Heiss
-Hernan
 Hernán
 ICC's
 ICCs
 IPW
 Intersectional
 Intersectionality
+JOSS
+LM
 Leckie
 Lenth
-LM
 MAIHDA
 Mattan
 McCabe
@@ -49,154 +43,113 @@ Merlo
 Mize
 Mmmh
 Modelisation
-Modelling
-Møller
 Montiel
 Mulinari
+Møller
 Neyman
-Olea
+Nonresponse
 ORCID
+Olea
 Owww
-PCVs
 PCV
+PCVs
 Plagborg
-QoL
 Psychometrika
+QoL
 RCTs
 README
-Rohrer
 RStudio
 RTs
 Ratcliff
-Reproducibility
-Rescaling
+Rohrer
+SSM
+SV
+SVARs
 Setosa
 Shachar
 Spiller
-SSM
 Subramanian
-SV
-SVARs
 Versicolor
 Virginica
-Visualisation
 Visualising
 Wagenmakers
 Weisberg
 Wemrell
 Wiernik
 al
 analysing
-bc
 behaviour
-blogpost
-bocode
-bonferroni
 bootES
 brglm
 brms
-caregiving
+cdot
 ception
 codecov
-computable
+coefficients’
 conditionspeed
-counterfactuals
 confounder
-confounders
 coxme
 d'être
-d’être
 dat
 datagrid
 datawizard
+diag
 doi
 dpar
+d’être
 easystats
-edu
-effectsize
 emmeans
-emtrends
 et
 exchangeability
 favour
-fdr
-fmwww
 foR
 generalizability
 geoms
 ggplot
 github
 glmmTMB
 glms
-grano
 grey
-hochberg
-holm
-hommel
-http
 https
-individuals'
-interpretable
+individuals’
 intersectional
 intersectionality
 intra
-io
 jmr
 joss
 labelled
-lifecycle
 lme
-lme4
 loess
 marginaleffects
-marginalizations
 mattansb
 modelisation
 modelled
 modelling
-natively
 nd
 nnet
 ol
 onwards
 partialled
-patilindrajeets
-pscl
 pre
-quartiles
+pscl
 rOpenSci
-Remotes
 raison
-recoding
-recodings
-repec
-reproducibility
-rescaled
-rescales
-residualize
 residualized
+residualizes
 rootSolve
 rstanarm
-salis
 spinoff
 ssmph
-strengejacke
-summarised
 summarises
-terciles
-transint
-tukey
-ultron
-uninformativeness
+tinyplot
 unidimensional
+uninformativeness
 unstandardizing
 usecases
 versicolor
 virginica
 visualisation
 visualise
-visualising
 walkthrough
-Nonresponse
-tinyplot
 ’
+’s