Merge pull request #1278 from cmu-delphi/update-codebook-wave-12

Update qsf tools to work with experimental wave 12

Author: krivard

facebook/qsf-tools/generate-codebook.R

@@ -326,7 +326,7 @@ process_qsf <- function(path_to_qsf,
 
   if (any(is.na(qdf$description))) {
     nonlabelled_items <- qdf$variable[is.na(qdf$description)]
-    stop(sprintf("items %s do not have a short name assigned",
+    stop(sprintf("items %s do not have a description provided",
                  paste(nonlabelled_items, collapse=", "))
     )
   }
@@ -396,7 +396,14 @@ add_static_fields <- function(codebook,
                               path_to_static_fields="./static/static_microdata_fields.csv") {
   static_fields <- get_static_fields(wave, path_to_static_fields)
 
-  return(bind_rows(codebook, static_fields))
+  codebook <- bind_rows(codebook, static_fields) %>% 
+    filter(!(variable == "module" & wave < 11), # module field is only available for wave >= 11
+           !(variable %in% c("wave", "UserLanguage", "fips") & wave < 4), # wave, UserLangauge, and fips fields are only available for wave >= 4
+           !(variable == "w12_treatment" & wave == 12.5), # experimental arm field is only available for wave == 12.5
+           variable != "Random_Number"
+    )
+
+  return(codebook)
 }
 
 #' Load dataframe of non-Qualtrics data fields
@@ -417,8 +424,7 @@ get_static_fields <- function(wave,
                                              response_option_randomization = col_character()
                             )) %>%
     mutate(wave = wave) %>% 
-    select(wave, everything()) %>% 
-    filter(!(variable == "module" & wave < 11))
+    select(wave, everything())
 
   return(static_fields)
 }

facebook/qsf-tools/qsf-utils.R

@@ -46,20 +46,21 @@ get_block_item_map <- function(qsf) {
 
 
 #' Get wave number from qsf filename
+#' 
+#' Wave number as provided in the qsf name should be an integer or a float with
+#' one decimal place.
 #'
 #' @param path_to_qsf
 #'
-#' @return integer wave number
+#' @return (mostly) integer wave number
 get_wave <- function(path_to_qsf) {
-  qsf_name_pattern <- "(.*Wave_)([0-9]*)([.]qsf)$"
+  qsf_name_pattern <- "(.*Wave_)([0-9]*([.][0-9])?)([.]qsf)$"
   if (!grepl(qsf_name_pattern, path_to_qsf)) {
     stop("qsf filename should be of the format 'Survey_of_COVID-Like_Illness_-_Wave_XX.qsf'")
   }
 
-  wave <- as.integer(sub(
-    "(.*Wave_)([0-9]*)([.]qsf)$",
-    "\\2",
-    path_to_qsf)
+  wave <- as.numeric(
+    sub(qsf_name_pattern, "\\2", path_to_qsf)
   ) 
 
   return(wave)

facebook/qsf-tools/static/item_replacement_map.csv

@@ -35,3 +35,11 @@ V11a,V11
 V12a,V12
 C7a,C7
 B10c,B10a
+V15c,V15a
+V1alt,V1
+B13a,B13
+P1,E1
+P2,E1
+P3,E4
+P5,E2
+P6,E3

facebook/qsf-tools/static/item_shortquestion_map.csv

@@ -126,4 +126,13 @@ V12a,"tried get vaccine"
 V15a,"vaccine access barriers vaccinated"
 V15b,"vaccine access barriers unvaccinated"
 V16,"when try vaccinated"
-V3a,"vaccine accepting"
+V3a,"vaccine accepting"
+V15c,"vaccine access barriers vaccinated"
+V1alt,"know anyone vaccinated"
+B13a,"COVID ever"
+P1,"have children <18"
+P2,"oldest child age"
+P3,"vaccinate oldest child"
+P4,"oldest child school type"
+P5,"oldest child school remote type"
+P6,"oldest child school safety measures"

facebook/qsf-tools/static/static_microdata_fields.csv

@@ -6,3 +6,4 @@ UserLanguage,NA,"survey language",NA,NA,NA,NA
 fips,NA,"county FIPS code",NA,NA,NA,NA
 weight,NA,"Facebook respondent weight",NA,NA,NA,NA
 module,NA,"question set respondent was assigned to",NA,NA,NA,NA
+w12_treatment,NA,"experimental treatment respondent was assigned to",NA,NA,NA,NA