Clean alter data

Goal

Clean alter data and remove faulty alter data.

Set up

#packages
library(tidyverse)

#load
load(file = "data/data-processed/liss_merged/lisscdn_afterrel_V4_240624.Rdata")

Data prep

#create check data
#first create a long format
alter_long <- liss_long %>% 
  select(nomem_encr,
         survey_wave,
         starts_with("alter_id"),
         starts_with("rel_alter"),
         starts_with("gender_alter")) %>% 
  rename(gender_alter.1 = gender_alter1,
         gender_alter.2 = gender_alter2,
         gender_alter.3 = gender_alter3,
         gender_alter.4 = gender_alter4,
         gender_alter.5 = gender_alter5,
         rel_alter.1 = rel_alter1,
         rel_alter.2 = rel_alter2,
         rel_alter.3 = rel_alter3,
         rel_alter.4 = rel_alter4,
         rel_alter.5 = rel_alter5,
         alter_id.1 = alter_id_1,
         alter_id.2 = alter_id_2,
         alter_id.3 = alter_id_3,
         alter_id.4 = alter_id_4,
         alter_id.5 = alter_id_5) %>% 
  pivot_longer(cols = c(starts_with("alter_id"),
                        starts_with("gender"),
                        starts_with("rel")),
               names_to = c("measure", "alter"),
               values_to = "value",
               names_pattern = "(.+)\\.(.+)") %>% 
  pivot_wider(names_from = measure,
              values_from = value) 

#create dyad id and remove na's on alter_id
check_data <- alter_long %>% 
  filter(!is.na(alter_id)) %>% 
  mutate(dyad_id = paste0(nomem_encr, alter_id)) %>% 
  select(1,7,4,2,3,5,6) %>% 
  arrange(dyad_id, survey_wave)

Alter Checks

Gender check

Check wether an alter switched gender. Although it is possible, it is still highly unlikely.

#create mean of gender over time. If not 0 or 1, then we have a problem.
check_data_gender <- check_data %>%
  arrange(nomem_encr, dyad_id, survey_wave) %>%
  group_by(dyad_id) %>%
  mutate(mean_gender = mean(gender_alter)) %>%
  ungroup()

#we lose 461 respondents due to gender transitions
gender_fault_ids <- check_data_gender %>% 
  filter(mean_gender != 2 & mean_gender != 1) %>%
  select(nomem_encr) %>%
  distinct()

Relationship check

Check wheter alters have weird and impposible relationship transformations

#Use paste0 to create a new variable which contain unique transition combinations
#then we can actually filter out the impossible combinations.
check_data <- check_data %>%
  group_by(dyad_id) %>%
  select(nomem_encr, dyad_id, survey_wave, rel_alter) %>%
  filter(!is.na(rel_alter)) %>% 
  mutate(rel_check = paste0(rel_alter, lag(rel_alter))) %>% #create unique variable
  ungroup()

#set impossible combinations. See codebook for the meaning of these. 
impossible_combinations <- c(12,13,14,15,16,17,18,110,
                             21,23,24,25,26,27,28,29,210,
                             31,32,34,35,36,37,38,39,310,
                             41,42,43,45,46,47,48,49,410,
                             51,52,53,54,56,57,58,59,510,
                             62,63,64,65,
                             72,73,74,75,
                             82,83,84,85,
                             92,93,94,95,
                             102,103,104,105)

#we lose 1000 ego over impossible combinations
#select the rows with impossible combinations
faulty_dyads_ids <- check_data %>%
  filter(rel_check %in% impossible_combinations) %>%
  select(nomem_encr) %>% 
  distinct() %>% 
  pull()

Clean data

Filter networks withimpossible gender or relationship transformations.

#filter out the networks with impossible combinations
liss_long <- liss_long %>%
  filter(!nomem_encr %in% faulty_dyads_ids) %>%  #remove dyads_ids
  filter(!nomem_encr %in% gender_fault_ids)  #remove gender_ids

#filter out from wide file
liss_wide <- liss_wide %>%
  filter(!nomem_encr %in% faulty_dyads_ids) %>%  #remove dyads_ids
  filter(!nomem_encr %in% gender_fault_ids)  #remove gender_ids

Data export

#clean global environment.
#only keep the wide and long files. 
rm(list=ls()[! ls() %in% c("liss_wide","liss_long")])

#save data
save.image(file = "data/data-processed/liss_merged/lisscdn_cleaned_V4_240624.Rdata")

---

Copyright © 2024 Jeroense Thijmen