fix for na.data = "listwise" when using standard cor()

hfgolino · Mar 16, 2024 · 1c3430b · 1c3430b
1 parent fbd39bf
commit 1c3430b
Show file tree

Hide file tree

Showing 4 changed files with 44 additions and 16 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -1,7 +1,7 @@
 Package: EGAnet
 Title: Exploratory Graph Analysis – a Framework for Estimating the Number of Dimensions in Multivariate Data using Network Psychometrics
-Version: 2.0.5
-Date: 2024-03-14
+Version: 2.0.6
+Date: 2024-03-16
 Authors@R: c(person("Hudson", "Golino", email = "hfg9s@virginia.edu", role = c("aut", "cre"), comment = c(ORCID = "0000-0002-1601-1447")),
 	     person("Alexander", "Christensen", email = "alexpaulchristensen@gmail.com", role = "aut", comment = c(ORCID = "0000-0002-9798-7037")),
 	     person("Robert", "Moulder", email = "rgm4fd@virginia.edu", role = "ctb", comment = c(ORCID = "0000-0001-7504-9560")),
@@ -11,15 +11,15 @@ Authors@R: c(person("Hudson", "Golino", email = "hfg9s@virginia.edu", role = c("
 Maintainer: Hudson Golino <hfg9s@virginia.edu>
 Description: Implements the Exploratory Graph Analysis (EGA) framework for dimensionality
              and psychometric assessment. EGA estimates the number of dimensions in
-	     psychological data using network estimation methods and community detection
+	     	 psychological data using network estimation methods and community detection
              algorithms. A bootstrap method is provided to assess the stability of dimensions
-	     and items. Fit is evaluated using the Entropy Fit family of indices. Unique 
+	     	 and items. Fit is evaluated using the Entropy Fit family of indices. Unique 
              Variable Analysis evaluates the extent to which items are locally dependent (or
              redundant). Network loadings provide similar information to factor loadings and
-	     can be used to compute network scores. A bootstrap and permutation approach are
+	     	 can be used to compute network scores. A bootstrap and permutation approach are
              available to assess configural and metric invariance. Hierarchical structures
              can be detected using Hierarchical EGA. Time series and intensive longitudinal 
-	     data can be analyzed using Dynamic EGA, supporting individual, group, and 
+	     	 data can be analyzed using Dynamic EGA, supporting individual, group, and 
              population level assessments.
 Depends: R (>= 3.5.0)
 License: GPL (>= 3.0)

diff --git a/NEWS b/NEWS
@@ -1,5 +1,10 @@
 WEBSITE: https://r-ega.net
 
+Changes in version 2.0.6
+
+o FIX: bug when using `na.data = "listwise"` in standard `cor()` function
+
+
 Changes in version 2.0.5
 
 o FIX: 'stroke' parameter in `hierEGA` that broke with {ggplot2} update

diff --git a/R/auto.correlate.R b/R/auto.correlate.R
@@ -97,7 +97,7 @@
 #' @export
 #'
 # Automatic correlations ----
-# Updated 07.09.2023
+# Updated 16.03.2024
 auto.correlate <- function(
     data, # Matrix or data frame
     corr = c("kendall", "pearson", "spearman"), # allow changes to standard correlations
@@ -140,8 +140,11 @@ auto.correlate <- function(
 
     # Obtain correlation matrix
     correlation_matrix <- cor(
-      x = data, use = na.data,
-      method = corr
+      x = data, use = swiftelse(
+        na.data == "pairwise",
+        "pairwise.complete.obs",
+        "complete.obs"
+      ), method = corr
     )
 
   }else{ # Proceed with determination of categorical correlations
@@ -191,7 +194,11 @@ auto.correlate <- function(
           continuous_variables, continuous_variables # ensure proper indexing
         ] <- cor(
           x = data[,continuous_variables],
-          use = na.data, method = corr
+          use = swiftelse(
+            na.data == "pairwise",
+            "pairwise.complete.obs",
+            "complete.obs"
+          ), method = corr
         )
 
       }
@@ -222,8 +229,11 @@ auto.correlate <- function(
 
       # Compute Pearson's correlations
       correlation_matrix <- cor(
-        x = data, use = na.data,
-        method = corr
+        x = data, use = swiftelse(
+          na.data == "pairwise",
+          "pairwise.complete.obs",
+          "complete.obs"
+        ), method = corr
       )
 
     }
@@ -357,7 +367,7 @@ auto.correlate_errors <- function(data, ordinal.categories, forcePD, verbose, ..
 #' Uses two-step approximation from {polycor}'s \code{polyserial}
 #'
 #' @noRd
-# Updated 03.07.2023
+# Updated 16.03.2024
 polyserial.vector <- function(
     categorical_variable, continuous_variables,
     na.data = c("pairwise", "listwise")
@@ -388,7 +398,14 @@ polyserial.vector <- function(
     sqrt((categorical_cases - 1) / categorical_cases) *
       sd(categorical_variable, na.rm = TRUE) *
       # Correlations with scaled continuous variables
-      cor(categorical_variable, scale(continuous_variables), use = na.data) /
+      cor(
+        categorical_variable, scale(continuous_variables),
+        use = swiftelse(
+          na.data == "pairwise",
+          "pairwise.complete.obs",
+          "complete.obs"
+        )
+      ) /
       # Compute sum of thresholds
       sum(dnorm(obtain_thresholds(categorical_variable)), na.rm = TRUE)
   )

diff --git a/R/helpers.R b/R/helpers.R
@@ -1448,7 +1448,7 @@ needs_usable <- function(ellipse)
 #' @noRd
 # Obtain data, sample size, correlation matrix ----
 # Generic function to get the usual needed inputs
-# Updated 04.09.2023
+# Updated 16.03.2024
 obtain_sample_correlations <- function(data, n, corr, na.data, verbose, ...)
 {
 
@@ -1501,7 +1501,13 @@ obtain_sample_correlations <- function(data, n, corr, na.data, verbose, ...)
       )
 
     }else{
-      correlation_matrix <- cor(data, use = na.data, method = corr)
+      correlation_matrix <- cor(
+        data, use = swiftelse(
+          na.data == "pairwise",
+          "pairwise.complete.obs",
+          "complete.obs"
+        ), method = corr
+      )
     }
 
   }