123 derive language status (#130)

* update derive_language_status and the test for it

* fix typo

Author: Sara-Jade-O

R/derive_vars.R

@@ -21,7 +21,7 @@ derive_vars <- function(data) {
 
 #' @title Derive language status
 #'
-#' @description Derive the status of each programmming language as "access" (access only), "knowledge" (knowledge only), "both" or "neither".
+#' @description Derive the status of each programmming language as "use" (use only), "knowledge" (knowledge only), "both" or "neither".
 #'
 #' @param data tidied CARS wave 3 data (data.frame).
 #'
@@ -30,21 +30,21 @@ derive_vars <- function(data) {
 
 derive_language_status <- function(data) {
 
-  lang_list <- colnames(data)[grepl("access_", colnames(data))]
+  lang_list <- colnames(data)[grepl("use_", colnames(data))]
 
   lang_list <- lang_list[!grepl("other", lang_list)]
 
-  lang_list <- gsub("access_", "", lang_list)
+  lang_list <- gsub("use_", "", lang_list)
 
   new_vars <- sapply(lang_list, function(lang) {
-    access_col <- data[[paste0("access_", lang)]]
+    use_col <- data[[paste0("use_", lang)]]
 
     knowledge_col <- data[[paste0("knowledge_", lang)]]
 
-    dplyr::case_when(access_col == "Yes" & knowledge_col == "Yes" ~ "Both",
-                     access_col == "Yes" & knowledge_col != "Yes" ~ "Access Only",
-                     access_col != "Yes" & knowledge_col == "Yes" ~ "Knowledge Only",
-                     access_col != "Yes" & knowledge_col != "Yes" ~ "Neither")
+    dplyr::case_when(use_col == "Yes" & knowledge_col == "Yes" ~ "Both",
+                     use_col == "Yes" & knowledge_col != "Yes" ~ "Use Only",
+                     use_col != "Yes" & knowledge_col == "Yes" ~ "Knowledge Only",
+                     use_col != "Yes" & knowledge_col != "Yes" ~ "Neither")
   })
 
   colnames(new_vars) <- paste0("status_", lang_list)

R/frequency-tables.R

@@ -186,7 +186,7 @@ summarise_rap_opinions <- function(data, config, question, prop = TRUE, sample =
 
 #' @title Summarise programming language status
 #'
-#' @description calculate counts of responents reporting access to, knowledge of, or both for each programming language.
+#' @description calculate counts of respondents reporting knowledge of, use of, or both for each programming language.
 #'
 #' @param data full CARS dataset after pre-processing
 #'
@@ -203,9 +203,11 @@ summarise_language_status <- function(data) {
                  "status_python",
                  "status_SPSS",
                  "status_stata",
-                 "status_matlab")
+                 "status_matlab",
+                 "status_dax",
+                 "status_spark")
 
-  levels <- c("Access Only", "Both", "Knowledge Only")
+  levels <- c("Use Only", "Both", "Knowledge Only")
 
   labels <- c("R",
               "SQL",
@@ -214,9 +216,13 @@ summarise_language_status <- function(data) {
               "Python",
               "SPSS",
               "Stata",
-              "Matlab")
+              "Matlab",
+              "DAX",
+              "Spark")
 
-  frequencies <- calculate_freqs(data, questions, levels, labels)
+  data[] <- lapply(data, factor, levels = levels)
+
+  frequencies <- calculate_freqs(data, questions, labels, prop = TRUE, sample = FALSE)
 
   return(frequencies)
 

man/derive_language_status.Rd

@@ -3,21 +3,25 @@
 # Input data should never include missing data.
 dummy_data <- data.frame(
   knowledge_R = c("Yes", "Yes"),
-  access_R = c("No", "No"),
+  use_R = c("No", "No"),
   knowledge_SQL = c("Yes", "Yes"),
-  access_SQL = c("Don't know", "No"),
+  use_SQL = c("Don't know", "No"),
   knowledge_SAS = c("Yes", "Yes"),
-  access_SAS = c("Yes", "Yes"),
+  use_SAS = c("Yes", "Yes"),
   knowledge_VBA = c("No", "No"),
-  access_VBA = c("Yes", "Yes"),
+  use_VBA = c("Yes", "Yes"),
   knowledge_python = c("No", "No"),
-  access_python = c("Don't know", "Don't know"),
+  use_python = c("Don't know", "Don't know"),
   knowledge_SPSS = c("No", "No"),
-  access_SPSS = c("No", "Yes"),
+  use_SPSS = c("No", "Yes"),
   knowledge_stata = c("Don't know", "No"),
-  access_stata = c("Yes", "No"),
+  use_stata = c("Yes", "No"),
   knowledge_matlab = c("Yes", "No"),
-  access_matlab = c("Yes", "No")
+  use_matlab = c("Yes", "No"),
+  knowledge_dax = c("Yes", "No"),
+  use_dax = c("No", "No"),
+  use_spark = c("Don't know", "Don't know"),
+  knowledge_spark = c("Don't know", "Don't know")
 )
 
 test_that("derive_language_status output is as expected", {
@@ -29,11 +33,13 @@ test_that("derive_language_status output is as expected", {
     status_R = c("Knowledge Only", "Knowledge Only"),
     status_SQL = c("Knowledge Only", "Knowledge Only"),
     status_SAS = c("Both", "Both"),
-    status_VBA = c("Access Only", "Access Only"),
+    status_VBA = c("Use Only", "Use Only"),
     status_python = c("Neither", "Neither"),
-    status_SPSS = c("Neither", "Access Only"),
-    status_stata = c("Access Only", "Neither"),
-    status_matlab = c("Both", "Neither")
+    status_SPSS = c("Neither", "Use Only"),
+    status_stata = c("Use Only", "Neither"),
+    status_matlab = c("Both", "Neither"),
+    status_dax = c("Knowledge Only", "Neither"),
+    status_spark = c("Neither", "Neither")
   )
 
   expect_equal(got, expected)