check_heterogeneity_bias() -> check_group_variation()

DESCRIPTION

@@ -183,7 +183,7 @@ Suggests:
     panelr,
     pbkrtest,
     PCDimension,
-    performance (>= 0.12.0),
+    performance (>= 0.14.0),
     plm,
     PMCMRplus,
     poorman,

R/1_model_parameters.R

@@ -129,7 +129,7 @@
 #'
 #' - **pseudo** (*for 2-level (G)LMMs only*): In this (post-hoc) method, the
 #' response and the predictor are standardized based on the level of prediction
-#' (levels are detected with [performance::check_heterogeneity_bias()]): Predictors
+#' (levels are detected with [performance::check_group_variation()]): Predictors
 #' are standardized based on their SD at level of prediction (see also
 #' [datawizard::demean()]); The outcome (in linear LMMs) is standardized based
 #' on a fitted random-intercept-model, where `sqrt(random-intercept-variance)`
@@ -185,7 +185,7 @@
 #'
 #' Compared to fixed effects (or single-level) models, determining appropriate
 #' df for Wald-based inference in mixed models is more difficult.
 #' See [the R GLMM FAQ](https://bbolker.github.io/mixedmodels-misc/glmmFAQ.html#what-are-the-p-values-listed-by-summaryglmerfit-etc.-are-they-reliable)
 #' for a discussion.
 #'
 #' Several approximate methods for computing df are available, but you should

R/standardize_info.R

@@ -415,7 +415,8 @@ standardize_info.default <- function(model,
 
   f <- if (two_sd) 2 else 1
 
-  within_vars <- unclass(performance::check_heterogeneity_bias(model))
+  gv <- performance::check_group_variation(model)
+  within_vars <- gv[gv$Variation %in% c("both", "within"), "Variable"]
   id <- insight::get_random(model)[[1]]
   w <- insight::get_weights(model, remove_na = TRUE)
 

R/standardize_parameters.R

@@ -64,7 +64,7 @@
 #' delta.
 #' - **pseudo** (*for 2-level (G)LMMs only*): In this (post-hoc) method, the
 #' response and the predictor are standardized based on the level of prediction
-#' (levels are detected with [`performance::check_heterogeneity_bias()`]): Predictors
+#' (levels are detected with [`performance::check_group_variation()`]): Predictors
 #' are standardized based on their SD at level of prediction (see also
 #' [`datawizard::demean()`]); The outcome (in linear LMMs) is standardized based
 #' on a fitted random-intercept-model, where `sqrt(random-intercept-variance)`
@@ -125,7 +125,7 @@
 #' @family standardize
 #' @family effect size indices
 #'
 #' @seealso See also [package vignette](https://easystats.github.io/parameters/articles/standardize_parameters_effsize.html).
 #'
 #' @examples
 #' model <- lm(len ~ supp * dose, data = ToothGrowth)
@@ -260,7 +260,7 @@
     colnames(pars)[colnames(pars) == "Coefficient"] <- gsub(" ", "_", coefficient_name, fixed = TRUE)
   }
 
   i <- colnames(pars) %in% c("Coefficient", "Median", "Mean", "MAP", "Odds_Ratio", "Risk_Ratio", "IRR", "Prevalence_Ratio")
   colnames(pars)[i] <- paste0("Std_", colnames(pars)[i])
 
   ## SE attribute?

tests/testthat/test-standardize_parameters.R

@@ -329,6 +329,7 @@ test_that("standardize_parameters (Bayes)", {
 # Pseudo - GLMM --------------------------------
 test_that("standardize_parameters (Pseudo - GLMM)", {
   skip_if_not_installed("datawizard", minimum_version = "0.12.0")
+  skip_if_not_installed("performance", minimum_version = "0.14.0")
   skip_on_cran()
   skip_if_not_installed("lme4")
   set.seed(1)

vignettes/demean.Rmd

@@ -85,11 +85,11 @@
 are captured in *one* estimate.
 
 You can check if your model may suffer from heterogeneity bias using the
-`check_heterogeneity_bias()` function:
+`check_group_variation()` function:
 
 ```{r}
 library(performance)
-check_heterogeneity_bias(qol_cancer, select = c("phq4", "education"), by = "ID")
+check_group_variation(qol_cancer, select = c("phq4", "education"), by = "ID")
 ```
 
 # Adressing heterogeneity bias: the Fixed Effects Regression (FE) approach
@@ -258,12 +258,12 @@
 [@bell_fixed_2019]:
 
 ```{r echo=FALSE}
 f <- "y<sub>it</sub> = &beta;<sub>0</sub> + &beta;<sub>1W</sub> (x<sub>it</sub> - &#x035E;x<sub>i</sub>) + &beta;<sub>2B</sub> &#x035E;x<sub>i</sub> + &beta;<sub>3</sub> z<sub>i</sub> + &upsilon;<sub>i0</sub> + &upsilon;<sub>i1</sub> (x<sub>it</sub> - &#x035E;x<sub>i</sub>) + &epsilon;<sub>it</sub>"
 knitr::asis_output(f)
 ```
 
 ```{r echo=FALSE}
 f <- "<ul><li>x<sub>it</sub> - &#x035E;x<sub>i</sub> is the de-meaned predictor, <em>phq4_within</em></li><li>&#x035E;x<sub>i</sub> is the group-meaned predictor, <em>phq4_between</em></li><li>&beta;<sub>1W</sub> is the coefficient for phq4_within (within-subject)</li><li>&beta;<sub>2B</sub> is the coefficient for phq4_between (bewteen-subject)</li><li>&beta;<sub>3</sub> is the coefficient for time-constant predictors, such as `hospital` or `education` (bewteen-subject)</li></ul>"
 knitr::asis_output(f)
 ```