The NextGen Harmonised Data Gateway
  • Home
  • Variable Database
    • Variable Database
    • Make it Green!
    • Make it Digital!
    • Make it Equal!
    • Make it Healthy!
    • Make it Strong!
  • Harmonisation
    • Harmonisation Toolbox
    • Make it Green!
    • Make it Digital!
    • Make it Equal!
    • Make it Healthy!
    • Make it Strong!
  • Data
  • Training
  • Contact
    • User Support
    • Team

On this page

  • Set-up
  • 1. Prepare Data: EB 2002
  • 2. Prepare Data: ESS2016
  • 3. Harmonisation
  • Summary

Climate Change Worries

This page guides you through the harmonisation process of the Climate Change Worries item.

The item was administered in ESS 2016 and EB 2002. Before we start harmonising, please download these data sets from the respective repositories. You can find links to the survey repositories on the Make It Green! data documentation page, in the section devoted to Climate Change Worries. If you click on the single year, the link takes you to the specific data set. We are working with the .sav format. So please make sure to download the datasets in this format.

Set-up

Please download the statistical software R (R Core Team, 2024) and RStudio (RStudio Team, 2023). Then start a project in RStudio by clicking File > New Project and set up a new directory or assign your project to an existing directory (i.e., folder). Within the project directory, create a folder called “data”. Now save the downloaded ESS and EB survey data into this “data” folder. Lastly, please download the following packages using the install.packages() function if they have not been before.

install.packages()
packages <- c( "tidyverse", "readxl", "xlsx", "haven", "labelled", "tidyr", "knitr", "kableExtra", "dplyr", "openxlsx")

installed <- packages %in% rownames(installed.packages())
if (any(!installed)) install.packages(packages[!installed])

Every package installed in R must also be loaded. This is what the library-function is for.

library(tidyverse)
library(readxl)
library(openxlsx) # If you have problems installing or loading the xlsx-package, use the openxlsx. 
library(haven)
library(labelled)
library(tidyr)
library(knitr)
library(kableExtra)
library(dplyr)

An important step during the harmonisation process is to prepare the source data sets in a consistent and systematic way. For that, the functions “codebook” (create_codebook_short()) and ”crosswalk table” (create_cwt()) by Kołczyńska (2022) are used. They show the correspondence between source data sets and harmonised data regarding variable names and values. These functions are adopted into our needs, and the R file containing them can be downloaded here . Please save this file in the R project folder and import it into the R project via source("kolczynska2022_i4ng.R").

How-to

If you can not download the R-file and if it is opened in a new tab, please open a new file in your R-Studio project under File -> New File -> R Script and copy the text from the kolczynska2022-file into this new file. Make sure that you save this new file in your project folder under the name “kolczynska2022_i4ng.R”.

source("kolczynska2022_i4ng.R")

Now we can name our target variable as “g_cc_worr”. The name is a combination of “g” representing the main topic “green”, the concept “Climate Change” and the variable label “Climate Change Worries”.

target_var_hrms <- "g_cc_worr"

Before we can harmonise any data, we need to prepare the source data sets:

1. Prepare Data: EB 2002

Six steps must be carried out in the following coding section:

1.1 Import data

eb <- haven::read_sav("data/ZA3692_v1-0-1.sav", user_na=TRUE)
Check the source item

Use this output to compare it with your output. Check especially if the labels match.

attr(eb$v568, "label")
[1] "Q39 ENVIRONM PROBL: CLIMATE CHANGE"
eb %>% count(v568)
# A tibble: 5 × 2
  v568                       n
  <dbl+lbl>              <int>
1 1 [Very worried]        6115
2 2 [Fairly worried]      7534
3 3 [Not very worried]    2705
4 4 [Not at all worried]   461
5 5 (NA) [DK]              226

1.2. Create codebook

Not just the substantive variables but also protocol variables need to be selected from this data set. Protocol variables selected for this harmonisation are the source data’s country, weight, and ID variables. In this step, we create a codebook that provides an overview of the selected variables and their naming in the harmonised data.

codebook_eb <- create_codebook_short(eb) %>%
  mutate(survey = "EB",
         target_var = ifelse(var_source == "isocntry", "country_source", NA),
         target_var = ifelse(var_source == "v11", "w2_poststr", target_var),
         target_var = ifelse(grepl("RESPONDENT ID|identi", var_label_source), "caseID_source", target_var),
         target_var = ifelse(grepl("Q39 ENVIRONM PROBL: CLIMATE CHANGE", var_label_source), 
                             target_var_hrms, 
                             target_var),
         target_var = gsub(" ", "_", target_var),
         target_var = tolower(target_var))

codebook_eb includes the columns called: var_source, var_label_source, survey, target_var. The variable names which will be used in the harmonised data are listed under target_var.

var_source var_label_source survey target_var
isocntry NATION - ALL SAMPLES ISO 3166 (CROSSTABULATION VARIABLE) EB country_source
v11 WEIGHT SPECIAL GERMANY EB w2_poststr
v568 Q39 ENVIRONM PROBL: CLIMATE CHANGE EB g_cc_worr
v788 ORIGINAL RESPONDENT ID EB caseid_source

 

1.3. Retrieve names of the chosen source variables

variables <- codebook_eb %>% drop_na(target_var) %>% pull(var_source)

The retrieved variables are: isocntry, v11, v568, v788.

 

1.4. Select a subset of the survey data containing only the chosen variables

eb_small <- eb %>% select(all_of(variables)) %>% mutate(caseid = row_number())

You will see that the following data set is a subset of the EB 2002 with the chosen variables for the harmonised data. The first six rows of the dataset are as follows:

isocntry v11 v568 v788 caseid
BE 0.9370 2 1000001 1
BE 1.0399 2 1001003 2
BE 1.0399 2 1001003 3
BE 1.0071 2 1001004 4
BE 1.1645 1 1001005 5
BE 1.1133 3 1001006 6

 

1.5. Create a crosswalk table for the chosen variables

A crosswalk table provides a connection between the source variable values and the target variable values. First, it provides the variables, their labels, and the possible values. Then, the information is linked with the codebook information of the newly specified target variables. Lastly, target values are defined in a variable called target_value.

Here, weight and ID variables are excluded from the crosswalk table. Original codes for the country variable are preserved. For the substantive variable, the response scales of the EB and ESS items are reversed. Considering the big picture of Make it Green! items, we decided to adjust the EB response scale. Lastly, NAs are assigned to missing values.

eb_cwt <- create_cwt(eb_small %>% select(all_of(variables))) %>% filter(!grepl("WEIGHT|ID", var_label_source)) %>% 
  left_join(codebook_eb, by=join_by(var_source, var_label_source)) %>% 
  mutate(value_code = as.character(value_code),
         target_value = case_when(target_var == target_var_hrms & !grepl("know|efusal|answer|NA|applicable| DK|Can't choose|Na", value_label_source) 
                                  ~ plyr::mapvalues(value_code, c("1", "2", "3", "4"), # original one
                                                       c("4", "3", "2", "1")), # reversed one
                                  target_var == target_var_hrms & grepl("know|efusal|answer|NA|applicable| DK|Can't choose|Na", value_label_source) ~ NA,
                                  TRUE ~ as.character(value_code))
         )
Important

There are decisions about the substantive variable made by taking into account both EB and ESS data. The utility of creating a crosswalk table is that the results of these decisions can be seen explicitly by comparing the value_code and target_value columns:

  1. Definitions of missing values differ across EB and ESS. Missing values are not harmonised but recoded into the single missing value “NA”.

  2. Response scales of the EB and the ESS items point in different directions. The EB response scale is adjusted at this step.

Let’s look at the correspondence between g_cc_worr and v568:

eb_cwt %>% filter(var_source == "v568") %>% 
  kable(row.names = FALSE) %>%
  kable_styling(full_width = FALSE, bootstrap_options = c("striped", "hover"))
var_source var_label_source npos value_n value_label_source value_code survey target_var target_value
v568 Q39 ENVIRONM PROBL: CLIMATE CHANGE 1 [1] Very worried: 6115 [1] Very worried 1 EB g_cc_worr 4
v568 Q39 ENVIRONM PROBL: CLIMATE CHANGE 2 [2] Fairly worried: 7534 [2] Fairly worried 2 EB g_cc_worr 3
v568 Q39 ENVIRONM PROBL: CLIMATE CHANGE 3 [3] Not very worried: 2705 [3] Not very worried 3 EB g_cc_worr 2
v568 Q39 ENVIRONM PROBL: CLIMATE CHANGE 4 [4] Not at all worried: 461 [4] Not at all worried 4 EB g_cc_worr 1
v568 Q39 ENVIRONM PROBL: CLIMATE CHANGE 5 [5] DK: 226 [5] DK 5 EB g_cc_worr NA
Optional

It might be useful to export the crosswalk table as a spreadsheet since it serves as a reference document that visualises the connection between the source and harmonised data set.

write.xlsx(eb_cwt, paste("data/",  target_var_hrms, "_cwt_eb.xlsx", sep = ''))

 

1.6. Recode survey data using the crosswalk table

To be able to merge survey data with the crosswalk table information, we first need to convert the wide data format to a long data format.

eb_small_long <- eb_small %>% 
  mutate(across(everything(), as.character)) %>% 
  pivot_longer(cols = c(v568, isocntry),
               names_to = "var_source", values_to = "value_source")

This is how the first six rows of the eb_small_long should look like:

v11 v788 caseid var_source value_source
0.937 1000001 1 v568 2
0.937 1000001 1 isocntry BE
1.0399 1001003 2 v568 2
1.0399 1001003 2 isocntry BE
1.0399 1001003 3 v568 2
1.0399 1001003 3 isocntry BE

Now we use left_join() to recode source values according to the crosswalk table, eb_cwt.

eb_hrmsd <- eb_small_long %>%
  mutate(value_source = as.character(value_source)) %>% 
  left_join(eb_cwt, by = c("value_source" = "value_code", "var_source")) %>% 
  select(-c(npos, value_n)) %>% 
  mutate(target_value = as.character(target_value)) %>% 
  pivot_wider(
    id_cols =c( v11, v788, caseid, survey),    # Columns to keep as unique identifiers*
    names_from = target_var,      # Column that will become the new column names
    values_from = target_value         # Column containing the values for those new columns
  ) %>% pivot_longer(
    cols = all_of(target_var_hrms), 
    names_to = "target_var", values_to = "target_value") %>% 
  rename(caseID_source = v788)

Now, EB data is ready for harmonisation. This is how the first six rows of the eb_hrmsd should look like:

v11 caseID_source caseid survey country_source target_var target_value
0.937 1000001 1 EB BE g_cc_worr 3
1.0399 1001003 2 EB BE g_cc_worr 3
1.0399 1001003 3 EB BE g_cc_worr 3
1.0071 1001004 4 EB BE g_cc_worr 3
1.1645 1001005 5 EB BE g_cc_worr 4
1.1133 1001006 6 EB BE g_cc_worr 2
Optional

At this step, the prepared data can be exported as an Rdata file and saved for later re-use.

saveRDS(eb_hrmsd, paste("data/",  target_var_hrms, "_eb.rds", sep = ''))

 

2. Prepare Data: ESS2016

We will follow the same steps used for the EB 2002 item. As expected, ESS and EB data have differences, and they are taken into account in implementing these steps.

 

2.1. Import data

ess <- haven::read_sav("data/ESS8e02_3.sav", user_na=TRUE)
Check the source item

Use this output to compare it with your output. Check especially if the labels match.

attr(ess$wrclmch, "label")
[1] "How worried about climate change"
ess %>% count(wrclmch)
# A tibble: 9 × 2
  wrclmch                     n
  <dbl+lbl>               <int>
1 1 [Not at all worried]   2590
2 2 [Not very worried]     8493
3 3 [Somewhat worried]    19729
4 4 [Very worried]         9649
5 5 [Extremely worried]    2193
6 6 (NA) [Not applicable]  1317
7 7 (NA) [Refusal]           29
8 8 (NA) [Don't know]       364
9 9 (NA) [No answer]         23

 

2.2. Create codebook

Protocol variables selected for this harmonisation are the source data’s country, weight, and ID variables. Note that there are two weight variables: design and post-stratification weights. Both of them are included.

codebook_ess <- create_codebook_short(ess) %>%
  mutate(survey = "ESS",
         target_var = ifelse(var_source == "cntry", "country_source", NA),
         target_var = ifelse(var_source == "dweight", "w1_design", target_var),
        target_var = ifelse(var_source == "pspwght", "w2_poststr", target_var),
         target_var = ifelse(grepl("RESPONDENT ID|identi", var_label_source), "caseID_source", target_var),
         target_var = ifelse(grepl("worried about climate change", var_label_source), 
                             target_var_hrms, 
                             target_var),
         target_var = gsub(" ", "_", target_var),
         target_var = tolower(target_var))

The variable names which will be used in the harmonised data are listed under target_var. The same names are used except for the weight variables because there are two weight variables.

var_source var_label_source survey target_var
idno Respondent's identification number ESS caseid_source
cntry Country ESS country_source
dweight Design weight ESS w1_design
pspwght Post-stratification weight including design weight ESS w2_poststr
wrclmch How worried about climate change ESS g_cc_worr

 

2.3. Retrieve names of the chosen source variables

variables <- codebook_ess %>% drop_na(target_var) %>% pull(var_source)

The retrieved variables are: idno, cntry, dweight, pspwght, wrclmch.

 

2.4. Select a subset of the survey data containing only the chosen variables

ess_small <- ess %>% select(all_of(variables)) %>% mutate(caseid = row_number())

You will see that the following data set is a subset of the ESS 2016 with the chosen variables for the harmonised data. The first six rows of the dataset are as follows:

idno cntry dweight pspwght wrclmch caseid
1 AT 0.6116771 1.1784955 4 1
2 AT 1.2233541 0.8994715 4 2
4 AT 0.3890577 0.3157533 4 3
6 AT 0.6425940 0.4724674 3 4
10 AT 3.4324021 2.2467055 4 5
11 AT 1.1805793 1.0113790 3 6

 

2.5. Create a crosswalk table for the chosen variables

Similar to what was implemented in step 1.5, weight and ID variables are excluded from the crosswalk table. Only country and the substantive variable are used for creating the crosswalk table. Original codes for the country variable are preserved. For the substantive variable, “NA” is assigned to missing values; the other values are kept identical.

ess_cwt <- create_cwt(ess_small %>% select(cntry, wrclmch)) %>%
  left_join(codebook_ess, by = join_by(var_source, var_label_source)) %>%
  mutate(target_value = NA,
         target_value = ifelse(!grepl("know|efusal|answer|NA|applicable|Can't choose|]DK|Na", value_label_source), 
                               value_code, 
                               target_value),
         value_code = as.character(value_code)
  )
Important

The ESS item has multiple definitions for missing values, and the EB item has only one. Because of this difference in missing values, they are recoded into the single missing value “NA”.

Let’s look at the correspondence between g_cc_worr and wrclmch:

ess_cwt %>% filter(var_source == "wrclmch") %>%
  kable(row.names = FALSE) %>%
  kable_styling(full_width = FALSE, bootstrap_options = c("striped", "hover"))
var_source var_label_source npos value_n value_label_source value_code survey target_var target_value
wrclmch How worried about climate change 1 [1] Not at all worried: 2590 [1] Not at all worried 1 ESS g_cc_worr 1
wrclmch How worried about climate change 2 [2] Not very worried: 8493 [2] Not very worried 2 ESS g_cc_worr 2
wrclmch How worried about climate change 3 [3] Somewhat worried: 19729 [3] Somewhat worried 3 ESS g_cc_worr 3
wrclmch How worried about climate change 4 [4] Very worried: 9649 [4] Very worried 4 ESS g_cc_worr 4
wrclmch How worried about climate change 5 [5] Extremely worried: 2193 [5] Extremely worried 5 ESS g_cc_worr 5
wrclmch How worried about climate change 6 [6] Not applicable: 1317 [6] Not applicable 6 ESS g_cc_worr NA
wrclmch How worried about climate change 7 [7] Refusal: 29 [7] Refusal 7 ESS g_cc_worr NA
wrclmch How worried about climate change 8 [8] Don't know: 364 [8] Don't know 8 ESS g_cc_worr NA
wrclmch How worried about climate change 9 [9] No answer: 23 [9] No answer 9 ESS g_cc_worr NA
Optional

It might be useful to export the crosswalk table as a spreadsheet since it serves as a reference document that visualises the connection between the source and harmonised data set.

write.xlsx(ess_cwt, paste("data/",  target_var_hrms, "_cwt_ess.xlsx", sep = ''))

 

2.6. Recode survey data using the crosswalk table

We first convert the wide data format to a long data format, and then merge survey data with the crosswalk table information.

ess_small_long <- ess_small %>%
  mutate(across(everything(), as.character)) %>%
  pivot_longer(cols = c(wrclmch, cntry),
               names_to = "var_source", values_to = "value_source")

This is how the first six rows of the ess_small_long should look like:

idno dweight pspwght caseid var_source value_source
1 0.611677050590515 1.17849552631378 1 wrclmch 4
1 0.611677050590515 1.17849552631378 1 cntry AT
2 1.22335410118103 0.899471521377563 2 wrclmch 4
2 1.22335410118103 0.899471521377563 2 cntry AT
4 0.389057725667953 0.31575334072113 3 wrclmch 4
4 0.389057725667953 0.31575334072113 3 cntry AT

Now we use left_join() to recode source values according to the crosswalk table, ess_cwt.

ess_hrmsd <- ess_small_long %>% 
  left_join(ess_cwt, by = c("value_source" = "value_code", "var_source")) %>% 
  select(-c(npos, value_n)) %>% 
  mutate(target_value = as.character(target_value)) %>% 
  pivot_wider(
    id_cols = c(idno, dweight, pspwght, caseid, survey),  # Columns to keep as unique identifiers
    names_from = target_var,   # Column that will become the new column names
    values_from = target_value # Column containing the values for those new columns
  ) %>% pivot_longer(
    cols = all_of(target_var_hrms), 
    names_to = "target_var", values_to = "target_value") %>% 
  rename(caseID_source = idno)

ESS data is ready for harmonisation. This is how the first six rows of the ess_hrmsd should look like:

head(ess_hrmsd) %>%
  kable(row.names = FALSE) %>%
  kable_styling(full_width = FALSE, bootstrap_options = c("striped", "hover"))
caseID_source dweight pspwght caseid survey country_source target_var target_value
1 0.611677050590515 1.17849552631378 1 ESS AT g_cc_worr 4
2 1.22335410118103 0.899471521377563 2 ESS AT g_cc_worr 4
4 0.389057725667953 0.31575334072113 3 ESS AT g_cc_worr 4
6 0.64259397983551 0.472467392683029 4 ESS AT g_cc_worr 3
10 3.43240213394165 2.24670553207397 5 ESS AT g_cc_worr 4
11 1.18057930469513 1.01137900352478 6 ESS AT g_cc_worr 3
Optional

At this step, the prepared data can be exported as an Rdata file and saved for later re-use.

saveRDS(ess_hrmsd, paste("data/",  target_var_hrms, "_ess.rds", sep = ''))

 

3. Harmonisation

Linear stretch (de Jonge et al., 2014) or identity linking is used to harmonise items. Kolen & Brennan (2014) and Albano (2016) provide extensive information about identity linking as well as other methodological approaches. Interested users are encouraged to consult these sources for further details.

With the following steps, we will create the harmonised Climate Change Worries data.

 

3.1. Standardise country names & apply labels

There are functions specifically written to standardise country names and apply labels. Please download the R file containing these functions here . Then save this file in the R project folder and import it into the R project via source("func_recoded_country.R").

How-to

If you cannot download the R-file, please open a new file in your R-Studio project under File -> New File -> R Script and copy the text from the func_recoded_country-file into this new file. Make sure that you save this new file in your project folder under the name “func_recoded_country.R”.

source("func_recoded_country.R")

eb_1 <- eb_hrmsd %>% 
  mutate(year = "2002") %>%
 recoded_country("country_source") %>% 
  country_new_labels("country_new") %>%
  func_country_sample("country_source") %>%
  country_source_recoded_labels("country_source_recoded")


ess_1 <- ess_hrmsd %>% 
  mutate(year = "2016") %>%
   recoded_country("country_source") %>% 
  country_new_labels("country_new") %>%
  func_country_sample("country_source") %>%
  country_source_recoded_labels("country_source_recoded")

# in ESS GB refers to UK: in all ESS waves GB_GBN needs to be recoded to GB_UK for country_new and country_recoded
ess_1$country_new[ess_1$country_new == "GB_GBN"] <- "GB_UK"
ess_1$country_source_recoded[ess_1$country_source_recoded == "GB_GBN"] <- "GB_UK"
Check

You can run the following code to check the country variable:

table(eb_hrmsd$country_source, eb_1$country_new)
table(ess_hrmsd$country_source, ess_1$country_new)

 

3.2. Weights

For details on decisions and procedures for including weight variables, please see the Make It Green! data documentation page, section Weight Variables.

eb_fnl <- eb_1  %>% mutate(w1_design = v11,
                      w2_poststr = v11, 
                      w3_combined = as.numeric(w2_poststr))

ess_fnl <- ess_1 %>% mutate(w1_design =dweight,
                       w2_poststr = pspwght, 
                       w3_combined = as.numeric(w2_poststr))

  3.3. Import additional information about the source data sets

This file, including additional information, can be downloaded here . Please save it into the “data” folder you generated previously.

add_data <- read_excel("data/from_doi_links_green.xlsx") %>% select(-studyno_source)

 

3.4. Combine all data

commons <- c("caseID_source", "caseid", "survey", 
             "target_var","target_value","year", "country_new", "country_source_recoded",
             "w1_design", "w2_poststr", "w3_combined") 


all_comb <- bind_rows(ess_fnl[, commons],
                      eb_fnl[, commons]) %>% 
  left_join(add_data, by = c("target_var", "survey","year")) %>% 
  mutate(
    
    caseID_I4NG = paste(survey, year, country_new, caseid, sep = ''),

    #country_year
    country_year = paste(country_new, year, sep = '_'),
    
    # caseID = NA,
    caseID = row_number(),

    # code the lengths of response scales in each project
    npoints = case_when( survey == "ESS" ~ 5, survey == "EB" ~ 4, TRUE ~ NA_real_)
    )  %>% 
  rename(caseid_sw = caseid, country_I4NG = country_new) 
Check

Let’s check the data set:

head(all_comb) %>%
  kable(row.names = FALSE) %>%
  kable_styling(full_width = FALSE, bootstrap_options = c("striped", "hover"))
caseID_source caseid_sw survey target_var target_value year country_I4NG country_source_recoded w1_design w2_poststr w3_combined title_source wave doi_source caseID_I4NG country_year caseID npoints
1 1 ESS g_cc_worr 4 2016 AT AT 0.611677050590515 1.17849552631378 1.1784955 Welfare & climate change attitudes 8 https://doi.org/10.21338/ess8e02_3 ESS2016AT1 AT_2016 1 5
2 2 ESS g_cc_worr 4 2016 AT AT 1.22335410118103 0.899471521377563 0.8994715 Welfare & climate change attitudes 8 https://doi.org/10.21338/ess8e02_3 ESS2016AT2 AT_2016 2 5
4 3 ESS g_cc_worr 4 2016 AT AT 0.389057725667953 0.31575334072113 0.3157533 Welfare & climate change attitudes 8 https://doi.org/10.21338/ess8e02_3 ESS2016AT3 AT_2016 3 5
6 4 ESS g_cc_worr 3 2016 AT AT 0.64259397983551 0.472467392683029 0.4724674 Welfare & climate change attitudes 8 https://doi.org/10.21338/ess8e02_3 ESS2016AT4 AT_2016 4 5
10 5 ESS g_cc_worr 4 2016 AT AT 3.43240213394165 2.24670553207397 2.2467055 Welfare & climate change attitudes 8 https://doi.org/10.21338/ess8e02_3 ESS2016AT5 AT_2016 5 5
11 6 ESS g_cc_worr 3 2016 AT AT 1.18057930469513 1.01137900352478 1.0113790 Welfare & climate change attitudes 8 https://doi.org/10.21338/ess8e02_3 ESS2016AT6 AT_2016 6 5
distinct(all_comb, target_var, survey, year)
# A tibble: 2 × 3
  target_var survey year 
  <chr>      <chr>  <chr>
1 g_cc_worr  ESS    2016 
2 g_cc_worr  EB     2002 

 

3.5. Find equivalents of the response categories

transformation <- function(x, x_min, x_max, y_min, y_max){
  
  # x: source scale
  # y: target scale
  
  y = ((x - x_min)/(x_max - x_min))*(y_max-y_min) + y_min
  
  return(y)
}

# source scale
scale_source <- c(1:4)

# Transform the 4-point source scale to the 5-point target scale
scale_target <- transformation(x = scale_source, x_min = 1, x_max = 4, y_min = 1, y_max = 5)

Equivalents of 1, 2, 3, 4 on the target scale are 1, 2.33, 3.67, 5. In the following step, EB item scores are recoded according to the correspondence between the source scale and the target scale.

 

3.6. Apply transformation

all_hrmsd <- all_comb %>%
  # apply transformation
  mutate(value_hrmsd = case_when(!(npoints == 5) ~ plyr::mapvalues(target_value,
                                                             scale_source, 
                                                             scale_target),
                           TRUE ~ target_value)) %>% 
  # style and format
  rename(var_hrmsd = target_var) %>% 
  select(-var_hrmsd, -value_hrmsd, var_hrmsd, value_hrmsd, -caseID, caseID, 
         -ends_with("_source"), ends_with("_source"), -target_value, -npoints) %>% 
  pivot_wider(
    id_cols = c(doi_source, title_source, caseID_source, caseid_sw, survey,  country_I4NG, country_source_recoded, year, country_year, 
                w1_design, w2_poststr, caseID_I4NG),    
    names_from = var_hrmsd,      # Column that will become the new column names
    values_from = value_hrmsd # Column containing the values for those new columns
    ) 

Now the harmonised Climate Change Worries data are ready. This is how the first six rows of the all_hrmsd should look like:

doi_source title_source caseID_source caseid_sw survey country_I4NG country_source_recoded year country_year w1_design w2_poststr caseID_I4NG g_cc_worr
https://doi.org/10.21338/ess8e02_3 Welfare & climate change attitudes 1 1 ESS AT AT 2016 AT_2016 0.611677050590515 1.17849552631378 ESS2016AT1 4
https://doi.org/10.21338/ess8e02_3 Welfare & climate change attitudes 2 2 ESS AT AT 2016 AT_2016 1.22335410118103 0.899471521377563 ESS2016AT2 4
https://doi.org/10.21338/ess8e02_3 Welfare & climate change attitudes 4 3 ESS AT AT 2016 AT_2016 0.389057725667953 0.31575334072113 ESS2016AT3 4
https://doi.org/10.21338/ess8e02_3 Welfare & climate change attitudes 6 4 ESS AT AT 2016 AT_2016 0.64259397983551 0.472467392683029 ESS2016AT4 3
https://doi.org/10.21338/ess8e02_3 Welfare & climate change attitudes 10 5 ESS AT AT 2016 AT_2016 3.43240213394165 2.24670553207397 ESS2016AT5 4
https://doi.org/10.21338/ess8e02_3 Welfare & climate change attitudes 11 6 ESS AT AT 2016 AT_2016 1.18057930469513 1.01137900352478 ESS2016AT6 3

 

3.7. Export data

# Rdata
saveRDS(all_hrmsd, paste("data/",  target_var_hrms, "_hrmsd.rds", sep = ''))

# SPSS file
write_sav(all_hrmsd, paste("data/",  target_var_hrms, "_hrmsd.sav", sep = ''))

# STATA file
write_dta(all_hrmsd, paste("data/",  target_var_hrms, "_hrmsd.dta", sep = ''))

# CSV file
write.csv(all_hrmsd, paste("data/",  target_var_hrms, "_hrmsd.csv", sep = ''), row.names = F, quote = T)

 

Summary

Climate Change Worries was administered in ESS 2016 and EB 2002. Our script, first, prepares these two data sets for harmonisation so that they are consistent in format via codebooks and crosswalk tables. Then, the response categories are harmonised using the identity linking/linear stretch approach. After this step, the item is ready for subsequent analysis!

To go back to the Make it Green! Toolbox page, please click here!

Back to top

References

Albano, A. D. (2016). equate: An R package for observed-score linking and equating. Journal of Statistical Software, 74(8), 1–36. https://doi.org/10.18637/jss.v074.i08
de Jonge, T., Veenhoven, R., & Arends, L. (2014). Homogenizing responses to different survey questions on the same topic: Proposal of a scale homogenization method using a reference distribution. Social Indicators Research, 117, 375–300. https://doi.org/10.1007/s11205-013-0335-6
Kołczyńska, M. (2022). Combining multiple survey sources: A reproducible workflow and toolbox for survey data harmonization. Methodological Innovations, 15(1), 62–72. https://doi.org/10.1177/20597991221077923
Kolen, M. J., & Brennan, R. L. (2014). Test equating, scaling, and linking. Springer. https://doi.org/10.1007/978-1-4939-0317-7
R Core Team. (2024). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/
RStudio Team. (2023). RStudio: Integrated development environment for R. https://www.rstudio.com/

© EU Funded Project 101131118 — Infra4NextGen

 

Subject to Terms of Use | Privacy notice | Cookie policy | Built with R and Quarto