Merge pull request #1294 from cmu-delphi/bodge-E1

Fix E1 naming in survey microdata

Author: krivard

facebook/delphiFacebook/R/responses.R

@@ -170,6 +170,7 @@ load_response_one <- function(input_filename, params, contingency_run) {
   input_data <- bodge_v4_translation(input_data, wave)
   input_data <- bodge_C6_C8(input_data, wave)
   input_data <- bodge_B13(input_data, wave)
+  input_data <- bodge_E1(input_data, wave)
 
   input_data <- code_symptoms(input_data, wave)
   input_data <- code_hh_size(input_data, wave)
@@ -486,6 +487,34 @@ bodge_B13 <- function(input_data, wave) {
   return(input_data)
 }
 
+#' Fix E1_* names in Wave 11 data after ~June 16, 2021.
+#' 
+#' Items E1_1 through E1_4 are part of a matrix. Qualtrics, for unknown reasons,
+#' switched to naming these E1_4 through E1_7 in June. Convert back to the
+#' intended names.
+#'
+#' @param input_data data frame of responses, before subsetting to select
+#'   variables
+#' @param wave integer indicating survey version
+#'   
+#' @return corrected data frame
+#' @importFrom dplyr rename
+bodge_E1 <- function(input_data, wave) {
+  E14_present <- all(c("E1_1", "E1_2", "E1_3", "E1_4") %in% names(input_data))
+  E47_present <- all(c("E1_4", "E1_5", "E1_6", "E1_7") %in% names(input_data))
+  assert(!(E14_present && E47_present), "fields E1_1-E1_4 should not be present at the same time as fields E1_4-E1_7")
+  
+  if ( E47_present ) {
+    input_data <- rename(input_data,
+                         E1_1 = "E1_4",
+                         E1_2 = "E1_5",
+                         E1_3 = "E1_6",
+                         E1_4 = "E1_7"
+    )
+  }
+  return(input_data)
+}
+
 #' Process module assignment column.
 #' 
 #' Rename `module` and recode to A/B/`NA`. Note: module assignment column name

facebook/delphiFacebook/man/bodge_E1.Rd

@@ -349,3 +349,43 @@ test_that("B13 bodge works correctly", {
                input)
 })
 
+test_that("E1 bodge works correctly", {
+  input <- tibble(
+    E1_1 = c(1, 2, 3, 1),
+    E1_2 = c(1, 2, 3, 2),
+    E1_3 = c(1, 2, 3, 3),
+    E1_4 = c(1, 2, 3, 4)
+  )
+  expect_equal(bodge_E1(input, wave = NA), input)
+  
+  input_4_to_7 <- tibble(
+    E1_4 = c(1, 2, 3, 1),
+    E1_5 = c(1, 2, 3, 2),
+    E1_6 = c(1, 2, 3, 3),
+    E1_7 = c(1, 2, 3, 4)
+  )
+
+  expect_equal(bodge_E1(input_4_to_7, wave = NA), input)
+  
+  input_mixed <- tibble(
+    E1_4 = c(1, 2, 3, 1),
+    E1_2 = c(1, 2, 3, 2),
+    E1_3 = c(1, 2, 3, 3),
+    E1_7 = c(1, 2, 3, 4)
+  )
+  expect_equal(bodge_E1(input_mixed, wave = NA), input_mixed)
+  
+  input_both <- tibble(
+    E1_1 = c(1, 2, 3, 1),
+    E1_2 = c(1, 2, 3, 2),
+    E1_3 = c(1, 2, 3, 3),
+    E1_4 = c(1, 2, 3, 1),
+    E1_5 = c(1, 2, 3, 2),
+    E1_6 = c(1, 2, 3, 3),
+    E1_7 = c(1, 2, 3, 4)
+  )
+  expect_error(bodge_E1(input_both, wave = NA),
+               "fields E1_1-E1_4 should not be present at the same time as fields E1_4-E1_7"
+  )
+})
+