Institutional Trust Data Exploration

knitr::opts_chunk$set(echo = TRUE, 
                      message = F, 
                      warning = F)
library(tidyverse)
library(here)
library(RColorBrewer)
library(janitor)
library(psych)
library(ggtext)
library(knitr)
library(kableExtra)
library(forcats)
library(gtools)
library(ggrepel)
library(DT)
library(papeR)
library(compareGroups)
library(ggcorrplot)

cb_colors = brewer.pal(n = 8, name = "Dark2")
custom_theme = theme_minimal() + 
      theme(strip.text = element_text(size = 16),
        axis.text=element_text(size=16),
        axis.title=element_text(size=16,face="bold"),
        panel.spacing = unit(1, "lines"), 
        legend.title = element_text(size=16,face="bold"), 
        legend.text = element_text(size=16), 
        plot.title = element_text(size = 20, face = "bold"), 
        plot.title.position = "plot", 
        plot.subtitle = element_text(size = 16), 
        plot.caption = element_text(size = 16))

df_full = read.csv("Share_Your_Voice_2023_01_11_19_08_50.csv")

df_retarget = df_full %>% 
  filter(pilot_version == 'retarget_pilot_regular_ads')

1. Summary Statistics Tables - Trust

Question: On a scale from 1 to 5, how much would you say you trust the follwing people or organizations to give you good information about the vaccine?

Takeaway:

• The institutional trust variables exhibit significant differences (range: 2.459-3.862)
• Community-based trust is lower

df_retarget = 
  df_retarget %>%
  mutate(
    vax_status_retarget = case_when(
      num_dose == "0" ~ 0,
      num_dose %in% c("1", "2", "3") ~ 1
    ),
    vax_status_retarget_4 = case_when(
      num_dose == "0" ~ 0,
      num_dose == "1" ~ 1,
      num_dose == "2" ~ 2,
      num_dose == "3" ~ 3
    )
  )

df_retarget$vax_status_retarget = factor(df_retarget$vax_status_retarget, levels = c(0, 1))
df_retarget$vax_status_retarget_4 = factor(df_retarget$vax_status_retarget_4, levels = c(0,1,2,3))

df_retarget$country_answer[!df_retarget$country_answer %in% c("South Africa","Kenya","Nigeria","Ghana")] = NA
df_retarget$country_answer = factor(df_retarget$country_answer, levels = c("South Africa","Kenya","Nigeria","Ghana"))
df_retarget = df_retarget %>% filter(!is.na(country_answer))

## clean data
df_retarget$doctor[!df_retarget$doctor %in% c("1","2","3","4","5")] = NA
df_retarget$doctor = strtoi(df_retarget$doctor)
df_retarget$family[!df_retarget$family %in% c("1","2","3","4","5")] = NA
df_retarget$family = strtoi(df_retarget$family)
df_retarget$friends[!df_retarget$friends %in% c("1","2","3","4","5")] = NA
df_retarget$friends = strtoi(df_retarget$friends)
df_retarget$minister_religious_leader[!df_retarget$minister_religious_leader %in% c("1","2","3","4","5")] = NA
df_retarget$minister_religious_leader = strtoi(df_retarget$minister_religious_leader)
df_retarget$community_leader[!df_retarget$community_leader %in% c("1","2","3","4","5")] = NA
df_retarget$community_leader = strtoi(df_retarget$community_leader)

df_retarget$clinic_worker[!df_retarget$clinic_worker %in% c("1","2","3","4","5")] = NA
df_retarget$clinic_worker = strtoi(df_retarget$clinic_worker)
df_retarget$community_based_org[!df_retarget$community_based_org %in% c("1","2","3","4","5")] = NA
df_retarget$community_based_org = strtoi(df_retarget$community_based_org)
df_retarget$health_department[!df_retarget$health_department %in% c("1","2","3","4","5")] = NA
df_retarget$health_department = strtoi(df_retarget$health_department)
df_retarget$WHO[!df_retarget$WHO %in% c("1","2","3","4","5")] = NA
df_retarget$WHO = strtoi(df_retarget$WHO)
df_retarget$african_scientist[!df_retarget$african_scientist %in% c("1","2","3","4","5")] = NA
df_retarget$african_scientist = strtoi(df_retarget$african_scientist)
df_retarget$international_scientist[!df_retarget$international_scientist %in% c("1","2","3","4","5")] = NA
df_retarget$international_scientist = strtoi(df_retarget$international_scientist)
df_retarget$international_ngo[!df_retarget$international_ngo %in% c("1","2","3","4","5")] = NA
df_retarget$international_ngo = strtoi(df_retarget$international_ngo)
df_retarget$local_news_media[!df_retarget$local_news_media %in% c("1","2","3","4","5")] = NA
df_retarget$local_news_media = strtoi(df_retarget$local_news_media)
df_retarget$international_news_media[!df_retarget$international_news_media %in% c("1","2","3","4","5")] = NA
df_retarget$international_news_media = strtoi(df_retarget$international_news_media)

Summary Statistics

Note: Sort by Mean

df_retarget %>%
  #select_if(is.numeric) %>%
  select(doctor, family, friends, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo, `Local News/Media` = local_news_media, `International News/Media` = international_news_media) %>%
  describe(quant = c(.25,.75) ) %>%
  select(n, mean, sd, se, min, min, first_quartile = Q0.25, median, third_quartile = Q0.75, max) %>%
  clean_names(case = "title") %>%
  rename(SD = Sd, SE = Se) %>%
  arrange(desc(Mean)) %>%
  kable(caption = "Summary Statistics for 'Who do they trust'", digits = 3)%>%
  kable_styling(bootstrap_options = c("striped", "hover")) %>%
  row_spec(c(1,2,3,4,5,6,8,9,12,13), bold = T, color = "black", background = "lightgrey")

Summary Statistics for ‘Who do they trust’

	N	Mean	SD	SE	Min	First Quartile	Median	Third Quartile	Max
WHO	1521	3.862	1.492	0.038	1	3	5	5	5
doctor	1602	3.721	1.443	0.036	1	3	4	5	5
Department of Health	1526	3.635	1.504	0.039	1	2	4	5	5
Local Health Clinic Worker	1544	3.498	1.384	0.035	1	2	4	5	5
International Scientists	1509	3.318	1.486	0.038	1	2	4	5	5
African scientists	1517	3.178	1.457	0.037	1	2	3	5	5
family	1580	3.175	1.482	0.037	1	2	3	5	5
International NGOs	1500	3.096	1.394	0.036	1	2	3	4	5
Community-Based Organizations	1531	2.937	1.370	0.035	1	2	3	4	5
International News/Media	1493	2.903	1.388	0.036	1	2	3	4	5
Local News/Media	1496	2.721	1.381	0.036	1	1	3	4	5
Ministers/Religious Leaders	1561	2.689	1.442	0.036	1	1	3	4	5
Community Leaders	1553	2.459	1.336	0.034	1	1	2	3	5
friends	1568	2.416	1.300	0.033	1	1	2	3	5

Summary Statistics by Country

Note: Sort by South Africa-Mean

df_retarget %>%
  select(country_answer, doctor, family, friends, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo, `Local News/Media` = local_news_media, `International News/Media` = international_news_media) %>% 
  split(.$country_answer) %>% 
  map(describe) %>% 
  do.call(rbind, .) %>%
  data.frame() %>%
  select(n, mean , se) %>%
  rownames_to_column() %>%
  separate(rowname, c("country_answer", "variable"), sep = "\\.") %>%
  filter(variable != "country_answer*") %>%
  pivot_longer(cols = c("n", "mean", "se")) %>%
  mutate(column = paste(country_answer, name, sep = "_")) %>%
  select(column, variable, value) %>%
  pivot_wider(names_from = column) %>%
  arrange(desc(`South Africa_mean`)) %>%
  kable(digits = 3,caption = "Summary Statistics for Trust Variables by Country", 
        col.names = c("", "N", "Mean", "SE", "N", "Mean", "SE", "N", "Mean", "SE", "N", "Mean", "SE")) %>%
  kable_styling(bootstrap_options = c("striped", "hover")) %>%
  add_header_above(c(" " = 1, "South Africa" = 3, "Kenya" = 3, "Nigeria" = 3, "Ghana" = 3)) %>%
  row_spec(c(1,2,3,4,6,7,8,9,12,13), bold = T, color = "black", background = "lightgrey")

Summary Statistics for Trust Variables by Country

	South Africa			Kenya			Nigeria			Ghana
	N	Mean	SE	N	Mean	SE	N	Mean	SE	N	Mean	SE
doctor	435	3.667	0.073	408	3.652	0.075	334	3.793	0.075	425	3.786	0.066
WHO	409	3.592	0.081	400	3.902	0.072	319	4.163	0.073	393	3.858	0.075
Local Health Clinic Worker	417	3.585	0.070	402	3.557	0.068	324	3.373	0.076	401	3.449	0.069
Department of Health	411	3.545	0.078	401	3.626	0.073	320	3.747	0.082	394	3.647	0.075
family	429	3.086	0.074	405	3.264	0.071	331	3.387	0.080	415	3.010	0.073
African scientists	409	3.042	0.078	399	3.043	0.070	317	3.334	0.079	392	3.332	0.071
International Scientists	407	2.943	0.076	399	3.328	0.072	314	3.640	0.078	389	3.440	0.075
Community-Based Organizations	414	2.819	0.069	401	3.182	0.068	320	2.878	0.073	396	2.861	0.068
International NGOs	404	2.703	0.070	398	3.234	0.070	313	3.383	0.072	385	3.132	0.071
International News/Media	399	2.536	0.068	396	3.008	0.069	312	3.106	0.076	386	3.010	0.071
Local News/Media	401	2.476	0.070	397	2.844	0.069	312	2.753	0.076	386	2.821	0.070
Ministers/Religious Leaders	421	2.373	0.069	403	2.826	0.070	328	2.963	0.077	409	2.658	0.074
Community Leaders	418	2.165	0.064	402	2.724	0.064	326	2.442	0.074	407	2.514	0.068
friends	423	2.156	0.061	404	2.522	0.064	329	2.581	0.069	412	2.447	0.066

Distribution tables

doctor

df_retarget$doctor = as.factor(df_retarget$doctor)
doctor_df = df_retarget %>% filter(!is.na(doctor))
doctor_plot <- doctor_df %>% count(doctor) %>% mutate(percentage = n / nrow(doctor_df))
ggplot(doctor_plot, aes(x = fct_inorder(doctor), y = n, fill = doctor)) +
  geom_bar(stat = "identity") +
  geom_text(aes(label = paste0(n, " (", round(percentage * 100,2), "%)")), vjust = -0.25) +
  ggtitle(paste0("Trust in 'doctor' (n = ", sum(doctor_plot$n), ")")) +
  theme(plot.title = element_text(face = "bold"), legend.position = "none") +
  labs(x = "trust score", y = "count")

WHO

df_retarget$WHO = as.factor(df_retarget$WHO)
WHO_df = df_retarget %>% filter(!is.na(WHO))
WHO_plot <- WHO_df %>% count(WHO) %>% mutate(percentage = n / nrow(WHO_df))
ggplot(WHO_plot, aes(x = fct_inorder(WHO), y = n, fill = WHO)) +
  geom_bar(stat = "identity") +
  geom_text(aes(label = paste0(n, " (", round(percentage * 100,2), "%)")), vjust = -0.25) +
  ggtitle(paste0("Trust in 'WHO' (n = ", sum(WHO_plot$n), ")")) +
  theme(plot.title = element_text(face = "bold"), legend.position = "none") +
  labs(x = "trust score", y = "count")

Ministers/Religious Leaders

df_retarget$minister_religious_leader = as.factor(df_retarget$minister_religious_leader)
minister_religious_leader_df = df_retarget %>% filter(!is.na(minister_religious_leader))
minister_religious_leaderO_plot <- minister_religious_leader_df %>% count(minister_religious_leader) %>% mutate(percentage = n / nrow(minister_religious_leader_df))
ggplot(minister_religious_leaderO_plot, aes(x = fct_inorder(minister_religious_leader), y = n, fill = minister_religious_leader)) +
  geom_bar(stat = "identity") +
  geom_text(aes(label = paste0(n, " (", round(percentage * 100,2), "%)")), vjust = -0.25) +
  ggtitle(paste0("Trust in 'Ministers/Religious Leaders' (n = ", sum(minister_religious_leaderO_plot$n), ")")) +
  theme(plot.title = element_text(face = "bold"), legend.position = "none") +
  labs(x = "trust score", y = "count")

Community Leaders

df_retarget$community_leader = as.factor(df_retarget$community_leader)
graph_df = df_retarget %>% filter(!is.na(community_leader))
graph_plot <- graph_df %>% count(community_leader) %>% mutate(percentage = n / nrow(graph_df))
ggplot(graph_plot, aes(x = fct_inorder(community_leader), y = n, fill = community_leader)) +
  geom_bar(stat = "identity") +
  geom_text(aes(label = paste0(n, " (", round(percentage * 100,2), "%)")), vjust = -0.25) +
  ggtitle(paste0("Trust in 'Community Leaders' (n = ", sum(graph_plot$n), ")")) +
  theme(plot.title = element_text(face = "bold"), legend.position = "none") +
  labs(x = "trust score", y = "count")

Local Health Clinic Worker

df_retarget$clinic_worker = as.factor(df_retarget$clinic_worker)
graph_df = df_retarget %>% filter(!is.na(clinic_worker))
graph_plot <- graph_df %>% count(clinic_worker) %>% mutate(percentage = n / nrow(graph_df))
ggplot(graph_plot, aes(x = fct_inorder(clinic_worker), y = n, fill = clinic_worker)) +
  geom_bar(stat = "identity") +
  geom_text(aes(label = paste0(n, " (", round(percentage * 100,2), "%)")), vjust = -0.25) +
  ggtitle(paste0("Trust in 'Local Health Clinic Worker' (n = ", sum(graph_plot$n), ")")) +
  theme(plot.title = element_text(face = "bold"), legend.position = "none") +
  labs(x = "trust score", y = "count")

Community-Based Organizations

df_retarget$community_based_org = as.factor(df_retarget$community_based_org)
graph_df = df_retarget %>% filter(!is.na(community_based_org))
graph_plot <- graph_df %>% count(community_based_org) %>% mutate(percentage = n / nrow(graph_df))
ggplot(graph_plot, aes(x = fct_inorder(community_based_org), y = n, fill = community_based_org)) +
  geom_bar(stat = "identity") +
  geom_text(aes(label = paste0(n, " (", round(percentage * 100,2), "%)")), vjust = -0.25) +
  ggtitle(paste0("Trust in 'Community-Based Organizations' (n = ", sum(graph_plot$n), ")")) +
  theme(plot.title = element_text(face = "bold"), legend.position = "none") +
  labs(x = "trust score", y = "count")

Department of Health

df_retarget$health_department = as.factor(df_retarget$health_department)
graph_df = df_retarget %>% filter(!is.na(health_department))
graph_plot <- graph_df %>% count(health_department) %>% mutate(percentage = n / nrow(graph_df))
ggplot(graph_plot, aes(x = fct_inorder(health_department), y = n, fill = health_department)) +
  geom_bar(stat = "identity") +
  geom_text(aes(label = paste0(n, " (", round(percentage * 100,2), "%)")), vjust = -0.25) +
  ggtitle(paste0("Trust in 'Department of Health' (n = ", sum(graph_plot$n), ")")) +
  theme(plot.title = element_text(face = "bold"), legend.position = "none") +
  labs(x = "trust score", y = "count")

African scientists

df_retarget$african_scientist = as.factor(df_retarget$african_scientist)
graph_df = df_retarget %>% filter(!is.na(african_scientist))
graph_plot <- graph_df %>% count(african_scientist) %>% mutate(percentage = n / nrow(graph_df))
ggplot(graph_plot, aes(x = fct_inorder(african_scientist), y = n, fill = african_scientist)) +
  geom_bar(stat = "identity") +
  geom_text(aes(label = paste0(n, " (", round(percentage * 100,2), "%)")), vjust = -0.25) +
  ggtitle(paste0("Trust in 'African scientists' (n = ", sum(graph_plot$n), ")")) +
  theme(plot.title = element_text(face = "bold"), legend.position = "none") +
  labs(x = "trust score", y = "count")

International Scientists

df_retarget$international_scientist = as.factor(df_retarget$international_scientist)
graph_df = df_retarget %>% filter(!is.na(international_scientist))
graph_plot <- graph_df %>% count(international_scientist) %>% mutate(percentage = n / nrow(graph_df))
ggplot(graph_plot, aes(x = fct_inorder(international_scientist), y = n, fill = international_scientist)) +
  geom_bar(stat = "identity") +
  geom_text(aes(label = paste0(n, " (", round(percentage * 100,2), "%)")), vjust = -0.25) +
  ggtitle(paste0("Trust in 'International Scientists' (n = ", sum(graph_plot$n), ")")) +
  theme(plot.title = element_text(face = "bold"), legend.position = "none") +
  labs(x = "trust score", y = "count")

International NGOs

df_retarget$international_ngo = as.factor(df_retarget$international_ngo)
graph_df = df_retarget %>% filter(!is.na(international_ngo))
graph_plot <- graph_df %>% count(international_ngo) %>% mutate(percentage = n / nrow(graph_df))
ggplot(graph_plot, aes(x = fct_inorder(international_ngo), y = n, fill = international_ngo)) +
  geom_bar(stat = "identity") +
  geom_text(aes(label = paste0(n, " (", round(percentage * 100,2), "%)")), vjust = -0.25) +
  ggtitle(paste0("Trust in 'International NGOs' (n = ", sum(graph_plot$n), ")")) +
  theme(plot.title = element_text(face = "bold"), legend.position = "none") +
  labs(x = "trust score", y = "count")

2. Trust Questions

## clean data
df_retarget$doctor[!df_retarget$doctor %in% c("1","2","3","4","5")] = NA
df_retarget$doctor = strtoi(df_retarget$doctor)
df_retarget$family[!df_retarget$family %in% c("1","2","3","4","5")] = NA
df_retarget$family = strtoi(df_retarget$family)
df_retarget$friends[!df_retarget$friends %in% c("1","2","3","4","5")] = NA
df_retarget$friends = strtoi(df_retarget$friends)
df_retarget$minister_religious_leader[!df_retarget$minister_religious_leader %in% c("1","2","3","4","5")] = NA
df_retarget$minister_religious_leader = strtoi(df_retarget$minister_religious_leader)
df_retarget$community_leader[!df_retarget$community_leader %in% c("1","2","3","4","5")] = NA
df_retarget$community_leader = strtoi(df_retarget$community_leader)

df_retarget$clinic_worker[!df_retarget$clinic_worker %in% c("1","2","3","4","5")] = NA
df_retarget$clinic_worker = strtoi(df_retarget$clinic_worker)
df_retarget$community_based_org[!df_retarget$community_based_org %in% c("1","2","3","4","5")] = NA
df_retarget$community_based_org = strtoi(df_retarget$community_based_org)
df_retarget$health_department[!df_retarget$health_department %in% c("1","2","3","4","5")] = NA
df_retarget$health_department = strtoi(df_retarget$health_department)
df_retarget$WHO[!df_retarget$WHO %in% c("1","2","3","4","5")] = NA
df_retarget$WHO = strtoi(df_retarget$WHO)
df_retarget$african_scientist[!df_retarget$african_scientist %in% c("1","2","3","4","5")] = NA
df_retarget$african_scientist = strtoi(df_retarget$african_scientist)
df_retarget$international_scientist[!df_retarget$international_scientist %in% c("1","2","3","4","5")] = NA
df_retarget$international_scientist = strtoi(df_retarget$international_scientist)
df_retarget$international_ngo[!df_retarget$international_ngo %in% c("1","2","3","4","5")] = NA
df_retarget$international_ngo = strtoi(df_retarget$international_ngo)
df_retarget$local_news_media[!df_retarget$local_news_media %in% c("1","2","3","4","5")] = NA
df_retarget$local_news_media = strtoi(df_retarget$local_news_media)
df_retarget$international_news_media[!df_retarget$international_news_media %in% c("1","2","3","4","5")] = NA
df_retarget$international_news_media = strtoi(df_retarget$international_news_media)

Takeaway:

• All institutional trust variables are positively correlated

All Trust Variables

df_trust_corr = df_retarget %>%
  select(doctor, family, friends, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo, `Local News/Media` = local_news_media, `International News/Media` = international_news_media)

ggcorrplot(cor(df_trust_corr, use = "pairwise.complete.obs"), type = "lower", lab = TRUE, lab_size = 5/.pt, tl.cex = 7, colors= c(cb_colors[1], "white", cb_colors[2]), hc.order = TRUE) + 
  labs(y = "", x = "", title = "Correlation Matrix: \nTrust Variables", 
       caption = paste("Number of Observations:", scales::comma(nrow(df_trust_corr))))+
  theme(axis.text.x = element_text(angle = 45, hjust=1))

Institutional Trust Variables Only

df_trust_corr_inst = df_retarget %>%
  select(doctor, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo)

ggcorrplot(cor(df_trust_corr_inst, use = "pairwise.complete.obs"), type = "lower", lab = TRUE, lab_size = 5/.pt, tl.cex = 7, colors= c(cb_colors[1], "white", cb_colors[2]), hc.order = TRUE) + 
  labs(y = "", x = "", title = "Correlation Matrix: \nInstitutional Trust Variables", 
       caption = paste("Number of Observations:", scales::comma(nrow(df_trust_corr_inst))))+
  theme(axis.text.x = element_text(angle = 45, hjust=1))

Institutional Trust Variables vs Non-Institutional Trust Variables

df_trust_corr = df_retarget %>%
  select(doctor, family, friends, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo, `Local News/Media` = local_news_media, `International News/Media` = international_news_media)

df_trust_corr_int = data.frame(cor(df_trust_corr, use = "pairwise.complete.obs"))
df_trust_corr_int = df_trust_corr_int %>% 
  select(friends, family, `Local.News.Media`, `International.News.Media`)
df_trust_corr_int = df_trust_corr_int[!(row.names(df_trust_corr_int) %in% c("friends", "family", "Local News/Media", "International News/Media")),]

ggcorrplot(df_trust_corr_int, lab = TRUE, lab_size = 5/.pt, tl.cex = 7, colors= c(cb_colors[1], "white", cb_colors[2])) + 
  labs(y = "", x = "", title = "Correlation Matrix: \nInstitutional Trust Variables vs Non-Institutional Trust Variables", 
       caption = paste("Number of Observations:", scales::comma(nrow(df_trust_corr_int))))+
  theme(axis.text.x = element_text(angle = 45, hjust=1))

#data.frame(cor(df_trust_corr, use = "pairwise.complete.obs"))

3. Sorting Questions

Questions: Tell me more about what YOU believe by telling me how much you agree or disagree with the statements you’re about to read.

• 1 = I Disagree Strongly
• 2 = I Disagree
• 3 = I’m Unsure
• 4 = I Agree
• 5 = I Agree Strongly

Takeaway:

• Most of the sorting question variables and institutional trust variables are weakly positively correlated
• afraid needle is slightly negatively correlated with trust variables

## clean data
df_retarget$covid_is_a_problem[!df_retarget$covid_is_a_problem %in% c("1","2","3","4","5")] = NA
df_retarget$covid_is_a_problem = strtoi(df_retarget$covid_is_a_problem)
df_retarget$covid_is_problem_my_country[!df_retarget$covid_is_problem_my_country %in% c("1","2","3","4","5")] = NA
df_retarget$covid_is_problem_my_country = strtoi(df_retarget$covid_is_problem_my_country)
df_retarget$vax_is_safe[!df_retarget$vax_is_safe %in% c("1","2","3","4","5")] = NA
df_retarget$vax_is_safe = strtoi(df_retarget$vax_is_safe)
df_retarget$vax_prevent_sick_death[!df_retarget$vax_prevent_sick_death %in% c("1","2","3","4","5")] = NA
df_retarget$vax_prevent_sick_death = strtoi(df_retarget$vax_prevent_sick_death)
df_retarget$need_vax_for_protection[!df_retarget$need_vax_for_protection %in% c("1","2","3","4","5")] = NA
df_retarget$need_vax_for_protection = strtoi(df_retarget$need_vax_for_protection)
df_retarget$vax_safer_covid[!df_retarget$vax_safer_covid %in% c("1","2","3","4","5")] = NA
df_retarget$vax_safer_covid = strtoi(df_retarget$vax_safer_covid)
df_retarget$developer_want_to_help[!df_retarget$developer_want_to_help %in% c("1","2","3","4","5")] = NA
df_retarget$developer_want_to_help = strtoi(df_retarget$developer_want_to_help)
df_retarget$health_worker_want_to_help[!df_retarget$health_worker_want_to_help %in% c("1","2","3","4","5")] = NA
df_retarget$health_worker_want_to_help = strtoi(df_retarget$health_worker_want_to_help)
df_retarget$gov_want_to_help[!df_retarget$gov_want_to_help %in% c("1","2","3","4","5")] = NA
df_retarget$gov_want_to_help = strtoi(df_retarget$gov_want_to_help)
df_retarget$important_to_protect_myself[!df_retarget$important_to_protect_myself %in% c("1","2","3","4","5")] = NA
df_retarget$important_to_protect_myself = strtoi(df_retarget$important_to_protect_myself)
df_retarget$important_to_protect_other[!df_retarget$important_to_protect_other %in% c("1","2","3","4","5")] = NA
df_retarget$important_to_protect_other = strtoi(df_retarget$important_to_protect_other)
df_retarget$vax_moral[!df_retarget$vax_moral %in% c("1","2","3","4","5")] = NA
df_retarget$vax_moral = strtoi(df_retarget$vax_moral)
df_retarget$worry_short_term_side_effect[!df_retarget$worry_short_term_side_effect %in% c("1","2","3","4","5")] = NA
df_retarget$worry_short_term_side_effect = strtoi(df_retarget$worry_short_term_side_effect)
df_retarget$worry_long_term_side_effect[!df_retarget$worry_long_term_side_effect %in% c("1","2","3","4","5")] = NA
df_retarget$worry_long_term_side_effect = strtoi(df_retarget$worry_long_term_side_effect)
df_retarget$covid_is_real[!df_retarget$covid_is_real %in% c("1","2","3","4","5")] = NA
df_retarget$covid_is_real = strtoi(df_retarget$covid_is_real)
df_retarget$probably_exposed[!df_retarget$probably_exposed %in% c("1","2","3","4","5")] = NA
df_retarget$probably_exposed = strtoi(df_retarget$probably_exposed)
df_retarget$afraid_needle[!df_retarget$afraid_needle %in% c("1","2","3","4","5")] = NA
df_retarget$afraid_needle = strtoi(df_retarget$afraid_needle)
df_retarget$deeply_religious[!df_retarget$deeply_religious %in% c("1","2","3","4","5")] = NA
df_retarget$deeply_religious = strtoi(df_retarget$deeply_religious)
df_retarget$important_moral[!df_retarget$important_moral %in% c("1","2","3","4","5")] = NA
df_retarget$important_moral = strtoi(df_retarget$important_moral)
df_retarget$fit_in_group_importance[!df_retarget$fit_in_group_importance %in% c("1","2","3","4","5")] = NA
df_retarget$fit_in_group_importance = strtoi(df_retarget$fit_in_group_importance)
df_retarget$responsible_importance[!df_retarget$responsible_importance %in% c("1","2","3","4","5")] = NA
df_retarget$responsible_importance = strtoi(df_retarget$responsible_importance)

All Trust Variables vs Sorting Questions

df_sort_corr = df_retarget %>%
  select(doctor, family, friends, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo, `Local News/Media` = local_news_media, `International News/Media` = international_news_media,
         covid_is_a_problem, covid_is_problem_my_country,vax_is_safe,vax_prevent_sick_death, need_vax_for_protection,vax_safer_covid,developer_want_to_help,health_worker_want_to_help,gov_want_to_help,important_to_protect_myself,important_to_protect_other,vax_moral,worry_short_term_side_effect,worry_long_term_side_effect,covid_is_real,probably_exposed,afraid_needle,deeply_religious,important_moral,fit_in_group_importance,responsible_importance)

df_sort_corr = data.frame(cor(df_sort_corr, use = "pairwise.complete.obs"))
df_sort_corr = df_sort_corr %>% 
  select(friends, family, `Local.News.Media`, `International.News.Media`, doctor, `Ministers.Religious.Leaders`, `Community.Leaders`, `Local.Health.Clinic.Worker`, `Community.Based.Organizations`, `Department.of.Health`, `African.scientists`, `International.Scientists`, `International.NGOs`, WHO)
df_sort_corr = df_sort_corr[!(row.names(df_sort_corr) %in% c("friends", "family", "Local News/Media", "International News/Media", "doctor", "Ministers/Religious Leaders", "Community Leaders", "Local Health Clinic Worker", "Community-Based Organizations", "Department of Health", "African scientists", "International Scientists", "International NGOs", "WHO")),]

ggcorrplot(df_sort_corr, lab = TRUE, lab_size = 5/.pt, tl.cex = 7, colors= c(cb_colors[1], "white", cb_colors[2])) + 
  labs(y = "", x = "", title = "Correlation Matrix: \nTrust Variables vs Sorting Questions", 
       caption = paste("Number of Observations:", scales::comma(nrow(df_sort_corr))))+
  theme(axis.text.x = element_text(angle = 45, hjust=1))

Variable Descriptions:

• covid_is_a_problem: People are still dying from COVID
• covid_is_problem_my_country: COVID is a problem in [my country]
• vax_is_safe: I think the COVID vaccines are safe
• vax_prevent_sick_death: COVID vaccines help prevent serious sickness and death
• need_vax_for_protection: You need a vaccine for protection from sickness (not because you are sick)
• vax_safer_covid: Getting the vaccine is much safer than getting COVID
• developer_want_to_help: I think the people who developed the vaccine wanted to help people
• health_worker_want_to_help: I think my local healthcare workers want me to be healthy and well
• gov_want_to_help: I think my government’s department of health workers want me to be healthy and well
• important_to_protect_myself: It is important to me that I protect myself from the effects of COVID
• important_to_protect_other: It is important to me that I protect others from the effects of COVID
• vax_moral: Getting vaccinated is a moral issue
• worry_short_term_side_effect: I worry about short-term side effects of the COVID vaccine
• worry_long_term_side_effect: I worry about long-term side effects of the COVID vaccine
• covid_is_real: COVID has killed millions of people worldwide
• probably_exposed: I will probably be exposed to someone with COVID over the next year
• afraid_needle: I am very afraid of needles
• deeply_religious: I am a deeply religious person
• important_moral: It is important to me that I be a moral person
• fit_in_group_importance: It is important to me to feel like I “fit in” in with my group
• responsible_importance: It is important to me to feel like I am a responsible member of my community

Institutional Trust Variables vs Sorting Questions

df_sort_corr = df_retarget %>%
  select(doctor, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo,
         covid_is_a_problem, covid_is_problem_my_country,vax_is_safe,vax_prevent_sick_death, need_vax_for_protection,vax_safer_covid,developer_want_to_help,health_worker_want_to_help,gov_want_to_help,important_to_protect_myself,important_to_protect_other,vax_moral,worry_short_term_side_effect,worry_long_term_side_effect,covid_is_real,probably_exposed,afraid_needle,deeply_religious,important_moral,fit_in_group_importance,responsible_importance)

df_sort_corr = data.frame(cor(df_sort_corr, use = "pairwise.complete.obs"))
df_sort_corr = df_sort_corr %>% 
  select(doctor, `Ministers.Religious.Leaders`, `Community.Leaders`, `Local.Health.Clinic.Worker`, `Community.Based.Organizations`, `Department.of.Health`, `African.scientists`, `International.Scientists`, `International.NGOs`, WHO)
df_sort_corr = df_sort_corr[!(row.names(df_sort_corr) %in% c("doctor", "Ministers/Religious Leaders", "Community Leaders", "Local Health Clinic Worker", "Community-Based Organizations", "Department of Health", "African scientists", "International Scientists", "International NGOs", "WHO")),]

ggcorrplot(df_sort_corr, lab = TRUE, lab_size = 5/.pt, tl.cex = 7, colors= c(cb_colors[1], "white", cb_colors[2])) + 
  labs(y = "", x = "", title = "Correlation Matrix: \nInstitutional Trust Variables vs Sorting Questions", 
       caption = paste("Number of Observations:", scales::comma(nrow(df_sort_corr))))+
  theme(axis.text.x = element_text(angle = 45, hjust=1))

Variable Descriptions:

• covid_is_a_problem: People are still dying from COVID
• covid_is_problem_my_country: COVID is a problem in [my country]
• vax_is_safe: I think the COVID vaccines are safe
• vax_prevent_sick_death: COVID vaccines help prevent serious sickness and death
• need_vax_for_protection: You need a vaccine for protection from sickness (not because you are sick)
• vax_safer_covid: Getting the vaccine is much safer than getting COVID
• developer_want_to_help: I think the people who developed the vaccine wanted to help people
• health_worker_want_to_help: I think my local healthcare workers want me to be healthy and well
• gov_want_to_help: I think my government’s department of health workers want me to be healthy and well
• important_to_protect_myself: It is important to me that I protect myself from the effects of COVID
• important_to_protect_other: It is important to me that I protect others from the effects of COVID
• vax_moral: Getting vaccinated is a moral issue
• worry_short_term_side_effect: I worry about short-term side effects of the COVID vaccine
• worry_long_term_side_effect: I worry about long-term side effects of the COVID vaccine
• covid_is_real: COVID has killed millions of people worldwide
• probably_exposed: I will probably be exposed to someone with COVID over the next year
• afraid_needle: I am very afraid of needles
• deeply_religious: I am a deeply religious person
• important_moral: It is important to me that I be a moral person
• fit_in_group_importance: It is important to me to feel like I “fit in” in with my group
• responsible_importance: It is important to me to feel like I am a responsible member of my community

4. Vaccination Status/Attitudes

Takeaway:

• Vaccinated people have higher institutional trust level than that of unvaccinated people
• Higher institutional trust level associated with higher intention to have vaccine in the future
• Whether received other vaccines or not is not that correlated with institutional trust level

Vaccination Status

Note: Sort by Unvaccinated-Mean

df_retarget %>%
  select(vax_status_retarget, doctor, family, friends, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo, `Local News/Media` = local_news_media, `International News/Media` = international_news_media) %>% 
  split(.$vax_status_retarget) %>% 
  map(describe) %>% 
  do.call(rbind, .) %>%
  data.frame() %>%
  select(n, mean , se) %>%
  rownames_to_column() %>%
  separate(rowname, c("vax_status_retarget", "variable"), sep = "\\.") %>%
  filter(variable != "vax_status_retarget*") %>%
  pivot_longer(cols = c("n", "mean", "se")) %>%
  mutate(column = paste(vax_status_retarget, name, sep = "_")) %>%
  select(column, variable, value) %>%
  pivot_wider(names_from = column) %>%
  arrange(desc(`0_mean`)) %>%
  kable(digits = 3,caption = "Summary Statistics for Trust Variables by Vaccination Status", 
        col.names = c("", "N", "Mean", "SE", "N", "Mean", "SE")) %>%
  kable_styling(bootstrap_options = c("striped", "hover"))%>%
  add_header_above(c(" " = 1, "Unvaccinated" = 3, "Vaccinated" = 3)) %>%
  row_spec(c(1,2,3,5,6,7,8,9,11,13), bold = T, color = "black", background = "lightgrey")

Summary Statistics for Trust Variables by Vaccination Status

	Unvaccinated			Vaccinated
	N	Mean	SE	N	Mean	SE
WHO	292	3.767	0.085	795	4.169	0.046
doctor	291	3.629	0.079	795	4.020	0.045
Department of Health	292	3.295	0.088	795	3.980	0.047
family	292	3.192	0.085	795	3.360	0.051
International Scientists	292	3.188	0.082	795	3.628	0.049
Local Health Clinic Worker	292	3.134	0.076	795	3.845	0.044
African scientists	292	3.075	0.078	795	3.459	0.049
International NGOs	291	2.993	0.078	795	3.316	0.048
Community-Based Organizations	292	2.760	0.071	795	3.157	0.047
International News/Media	292	2.723	0.075	795	3.116	0.049
Ministers/Religious Leaders	292	2.705	0.084	795	2.808	0.050
Local News/Media	292	2.455	0.073	795	2.923	0.048
Community Leaders	292	2.260	0.074	795	2.605	0.046
friends	292	2.226	0.068	795	2.600	0.046

Vaccination Dose

Note: Sort by Unvaccinated-Mean

df_retarget %>%
  select(vax_status_retarget_4, doctor, family, friends, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo, `Local News/Media` = local_news_media, `International News/Media` = international_news_media) %>% 
  split(.$vax_status_retarget_4) %>% 
  map(describe) %>% 
  do.call(rbind, .) %>%
  data.frame() %>%
  select(n, mean , se) %>%
  rownames_to_column() %>%
  separate(rowname, c("vax_status_retarget_4", "variable"), sep = "\\.") %>%
  filter(variable != "vax_status_retarget_4*") %>%
  pivot_longer(cols = c("n", "mean", "se")) %>%
  mutate(column = paste(vax_status_retarget_4, name, sep = "_")) %>%
  select(column, variable, value) %>%
  pivot_wider(names_from = column) %>%
  arrange(desc(`0_mean`)) %>%
  kable(digits = 3,caption = "Summary Statistics for Trust Variables by Vaccination Dose", 
        col.names = c("", "N", "Mean", "SE", "N", "Mean", "SE", "N", "Mean", "SE", "N", "Mean", "SE")) %>%
  kable_styling(bootstrap_options = c("striped", "hover")) %>%
  add_header_above(c(" " = 1, "Unvaccinated" = 3, "1 dose" = 3, "2 doses" = 3, "3 doses" = 3)) %>%
  row_spec(c(1,2,3,5,6,7,8,9,11,13), bold = T, color = "black", background = "lightgrey")

Summary Statistics for Trust Variables by Vaccination Dose

	Unvaccinated			1 dose			2 doses			3 doses
	N	Mean	SE	N	Mean	SE	N	Mean	SE	N	Mean	SE
WHO	292	3.767	0.085	231	4.091	0.088	376	4.269	0.063	188	4.064	0.102
doctor	291	3.629	0.079	231	4.026	0.079	376	4.008	0.066	188	4.037	0.095
Department of Health	292	3.295	0.088	231	3.913	0.089	376	4.045	0.065	188	3.931	0.103
family	292	3.192	0.085	231	3.281	0.094	376	3.410	0.072	188	3.356	0.110
International Scientists	292	3.188	0.082	231	3.662	0.090	376	3.662	0.069	188	3.516	0.108
Local Health Clinic Worker	292	3.134	0.076	231	3.684	0.084	376	3.949	0.059	188	3.835	0.096
African scientists	292	3.075	0.078	231	3.424	0.094	376	3.503	0.069	188	3.415	0.102
International NGOs	291	2.993	0.078	231	3.234	0.089	376	3.465	0.065	188	3.117	0.106
Community-Based Organizations	292	2.760	0.071	231	3.039	0.086	376	3.279	0.068	188	3.059	0.100
International News/Media	292	2.723	0.075	231	3.017	0.090	376	3.184	0.071	188	3.101	0.101
Ministers/Religious Leaders	292	2.705	0.084	231	2.740	0.091	376	2.819	0.074	188	2.867	0.102
Local News/Media	292	2.455	0.073	231	2.779	0.091	376	3.029	0.069	188	2.888	0.102
Community Leaders	292	2.260	0.074	231	2.481	0.083	376	2.646	0.068	188	2.676	0.099
friends	292	2.226	0.068	231	2.502	0.082	376	2.598	0.066	188	2.723	0.102

Intention to get vaccine in future

Question: On a scale from 1 -5, How likely are you to get a COVID vaccine/booster next year? (1 = very unlikely, 2 = somewhat unlikely, 3 = unsure, 4 = probably will, 5 = definitely will)

df_retarget$vax_next_year = strtoi(df_retarget$vax_next_year)
df_retarget$vax_next_year[!df_retarget$vax_next_year %in% c("1","2","3","4","5")] = NA

df_future_corr = df_retarget %>%
  select(doctor, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo,
         vax_next_year)

df_future_corr = data.frame(cor(df_future_corr, use = "pairwise.complete.obs"))
df_future_corr = df_future_corr %>% 
  select(vax_next_year)
#df_future_corr = df_future_corr[!(row.names(df_future_corr) %in% c("vax_next_year")),]

ggcorrplot(df_future_corr, type = "lower", lab = TRUE, lab_size = 5/.pt, tl.cex = 7, colors= c(cb_colors[1], "white", cb_colors[2])) + 
  labs(y = "", x = "", title = "Correlation Matrix: \nInstitutional Trust Variables vs Vax Future", 
       caption = paste("Number of Observations:", scales::comma(nrow(df_future_corr))))+
  theme(axis.text.x = element_text(angle = 45, hjust=1))

Received any other vaccines

Question: Have you ever received any other vaccines? Options: yes; no

df_retarget$receive_other_vax[df_retarget$receive_other_vax == "yes"] = 1
df_retarget$receive_other_vax[df_retarget$receive_other_vax == "no"] = 0
df_retarget$receive_other_vax[!df_retarget$receive_other_vax %in% c("0","1")] = NA
df_retarget$receive_other_vax = strtoi(df_retarget$receive_other_vax, base = 0L)

df_future_corr = df_retarget %>%
  select(doctor, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo,
         receive_other_vax)

df_future_corr = data.frame(cor(df_future_corr, use = "pairwise.complete.obs"))
df_future_corr = df_future_corr %>% 
  select(receive_other_vax)
#df_future_corr = df_future_corr[!(row.names(df_future_corr) %in% c("vax_next_year")),]

ggcorrplot(df_future_corr, type = "lower", lab = TRUE, lab_size = 5/.pt, tl.cex = 7, colors= c(cb_colors[1], "white", cb_colors[2])) + 
  labs(y = "", x = "", title = "Correlation Matrix: \nInstitutional Trust Variables vs Received any other vaccines", 
       caption = paste("Number of Observations:", scales::comma(nrow(df_future_corr))))+
  theme(axis.text.x = element_text(angle = 45, hjust=1))

5. Demographics

Takeaway:

• In general, demographics are not that correlated with institutional trust level
• White or caucasian associated with lower institutional trust level
• People live in more urban area associated with lower institutional trust level
• Higher educational level associated with higher institutional trust level
• Hinduism and Judaism have highest trust level in Ministers/Religious Leaders
• People choose “Other religion” have lower institutional trust level in general

Continuous Demographic Variables

Starred variables have a corresponding variable, with the suffix _num (ie education_num) that provides a numeric ordering of the variables. This order is listed in parenthesis.

Variable	Description
gender*	male (0) or female (1)
age	integer 18 to 99
education*	< high school (1), high school (2), some college (3), 2-year degree (4), 4-year degree (5), graduate degree (6), other, prefer not to say
location*	rural(1), suburban (2), urban(3), other, prefer not to say
religiosity*	not very religious (1), somewhat religious (2), very religious (3), other, prefer not to say
white	1 if the participant is a white or caucasian, 0 if not
ethnicity	asian or indian, black or african, coloured, white or caucasian, other, prefer not to say

## clean demographic data
numeric_cols <- function(df){
  out <- df %>%
    mutate(
      age = as.numeric(ifelse(demog_age %in% paste(18:99), demog_age, "")),
      gender_num = case_when(
        gender == "female" ~ 1,
        gender == "male" ~ 0
        ),
      education_num = case_when(
        demog_education == "< High school" ~ 1,
        demog_education == "High school" ~ 2,
        demog_education == "Some college" ~ 3,
        demog_education == "2-year degree" ~ 4,
        demog_education == "4-year degree" ~ 5,
        demog_education == "Graduate degree" ~ 6,
      ),
      religiosity_num = case_when(
        religiosity == "Not very religious" ~ 1,
        religiosity == "Somewhat religious" ~ 2,
        religiosity == "Very religious" ~ 3,
      ),
      location_num = case_when(
        location == "Rural" ~ 1,
        location == "Suburban" ~ 2,
        location == "Urban" ~ 3,
      ),
      location =factor(location, levels = c("Urban", "Suburban", "Rural", "Prefer not to say")),
      white = ifelse(ethnicity == "White or Caucasian", 1, 0),
      ethnicity = factor(ethnicity, levels=c("Asian or Indian", 
                                                "Black or African", 
                                                "coloured", 
                                                "White or Caucasian", 
                                                "Other",
                                                "Prefer not to say")
    ))
    return(out)
}

df_retarget_num = numeric_cols(df_retarget)

df_demog_corr = df_retarget_num %>%
  select(doctor, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo,
         age, education = education_num, location=location_num, religiosity = religiosity_num, gender = gender_num, white)

df_demog_corr = data.frame(cor(df_demog_corr, use = "pairwise.complete.obs"))
df_demog_corr = df_demog_corr %>% 
  select(-c(doctor, `Ministers.Religious.Leaders`, `Community.Leaders`, `Local.Health.Clinic.Worker`, `Community.Based.Organizations`, `Department.of.Health`, `African.scientists`, `International.Scientists`, `International.NGOs`, WHO))
df_demog_corr = df_demog_corr[(row.names(df_demog_corr) %in% c("doctor", "Ministers/Religious Leaders", "Community Leaders", "Local Health Clinic Worker", "Community-Based Organizations", "Department of Health", "African scientists", "International Scientists", "International NGOs", "WHO")),]

ggcorrplot(df_demog_corr, lab = TRUE, lab_size = 5/.pt, tl.cex = 7, colors= c(cb_colors[1], "white", cb_colors[2])) + 
  labs(y = "", x = "", title = "Correlation Matrix: \nInstitutional Trust Variables vs Continuous Demographics", 
       caption = paste("Number of Observations:", scales::comma(nrow(df_future_corr))))+
  theme(axis.text.x = element_text(angle = 45, hjust=1))

Other Demographic Variables

1. Ethnicity

df_retarget$ethnicity[!df_retarget$ethnicity %in% c("Black or African","White or Caucasian", "Asian or Indian", "Other", "Prefer not to say")] = NA
df_retarget$ethnicity = factor(df_retarget$ethnicity, levels=c("Black or African","White or Caucasian", "Asian or Indian", "Other", "Prefer not to say"))
ethnicity_df = df_retarget %>% 
  filter(!is.na(ethnicity)) %>% 
  select(doctor, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo,
         ethnicity) %>%
  group_by(ethnicity) %>%
  drop_na() %>%
  dplyr::summarise(
    doctor = mean(doctor),
    `Ministers/Religious Leaders` = mean(`Ministers/Religious Leaders`),
    `Community Leaders` = mean(`Community Leaders`),
    `Local Health Clinic Worker` = mean(`Local Health Clinic Worker`),
    `Community-Based Organizations` = mean(`Community-Based Organizations`),
    `Department of Health` = mean(`Department of Health`),
    `African scientists` = mean(`African scientists`),
    `International Scientists` = mean(`International Scientists`),
    `International NGOs` = mean(`International NGOs`),
    WHO = mean(WHO)
  )

ethnicity_df %>%
  kable(caption = "Ethnicity and institutional Trust", digits = 3)%>%
  kable_styling(bootstrap_options = c("striped", "hover"))

Ethnicity and institutional Trust

ethnicity	doctor	Ministers/Religious Leaders	Community Leaders	Local Health Clinic Worker	Community-Based Organizations	Department of Health	African scientists	International Scientists	International NGOs	WHO
Black or African	3.928	2.779	2.523	3.652	3.063	3.809	3.379	3.523	3.251	4.071
White or Caucasian	3.700	2.250	1.900	3.050	2.850	3.150	2.450	2.850	2.600	3.400
Asian or Indian	3.667	3.667	2.667	3.889	3.556	3.667	3.444	3.222	3.222	4.222
Other	3.600	2.600	2.160	3.640	2.400	3.200	2.880	2.880	2.760	3.600
Prefer not to say	3.833	3.500	2.750	4.250	3.333	4.417	3.417	3.917	3.750	4.583

2. Religion

df_retarget$demog_religion[!df_retarget$demog_religion %in% c("Christianity","Islam", "African traditional", "Hinduism", "Judaism", "No religion", "Other", "Prefer not to say")] = NA
df_retarget$demog_religion = factor(df_retarget$demog_religion, levels=c("Christianity","Islam", "African traditional", "Hinduism", "Judaism", "No religion", "Other", "Prefer not to say"))
religion_df = df_retarget %>% 
  filter(!is.na(demog_religion)) %>% 
  select(doctor, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo,
         demog_religion) %>%
  group_by(demog_religion) %>%
  drop_na() %>%
  dplyr::summarise(
    doctor = mean(doctor),
    `Ministers/Religious Leaders` = mean(`Ministers/Religious Leaders`),
    `Community Leaders` = mean(`Community Leaders`),
    `Local Health Clinic Worker` = mean(`Local Health Clinic Worker`),
    `Community-Based Organizations` = mean(`Community-Based Organizations`),
    `Department of Health` = mean(`Department of Health`),
    `African scientists` = mean(`African scientists`),
    `International Scientists` = mean(`International Scientists`),
    `International NGOs` = mean(`International NGOs`),
    WHO = mean(WHO)
  )

religion_df %>%
  kable(caption = "Religion and institutional Trust", digits = 3)%>%
  kable_styling(bootstrap_options = c("striped", "hover"))

Religion and institutional Trust

demog_religion	doctor	Ministers/Religious Leaders	Community Leaders	Local Health Clinic Worker	Community-Based Organizations	Department of Health	African scientists	International Scientists	International NGOs	WHO
Christianity	3.920	2.816	2.502	3.664	3.066	3.815	3.331	3.522	3.238	4.092
Islam	4.007	2.941	2.787	3.676	3.147	3.963	3.596	3.735	3.434	4.162
African traditional	3.640	2.120	2.280	3.320	2.720	3.360	3.520	2.960	2.720	3.400
Hinduism	3.000	4.000	2.750	3.250	3.250	3.500	2.750	2.500	2.500	3.750
Judaism	5.000	4.000	5.000	5.000	5.000	4.000	4.000	5.000	4.000	5.000
No religion	3.920	1.920	1.960	3.560	2.840	3.280	3.600	3.400	2.960	3.600
Other	3.000	1.750	1.500	3.250	1.500	3.250	2.750	2.750	2.500	3.750
Prefer not to say	3.533	1.800	2.000	3.200	2.400	3.200	2.667	2.733	2.867	3.400

3. Income

South Africa

df_SA = df_retarget %>% filter(country_answer == "South Africa")
df_SA$demog_income[!df_SA$demog_income %in% c("< R5,000","R5,000 - R9,999", "R10,000 - R29,999", "R30,000 - R49,999", "R50,000 - R99,999", "> R100,000")] = NA
df_SA$demog_income = factor(df_SA$demog_income, levels=c("< R5,000","R5,000 - R9,999", "R10,000 - R29,999", "R30,000 - R49,999", "R50,000 - R99,999", "> R100,000"))
SA = df_SA %>% 
  filter(!is.na(demog_income)) %>% 
  select(doctor, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo,
         demog_income) %>%
  group_by(demog_income) %>%
  drop_na() %>%
  dplyr::summarise(
    doctor = mean(doctor),
    `Ministers/Religious Leaders` = mean(`Ministers/Religious Leaders`),
    `Community Leaders` = mean(`Community Leaders`),
    `Local Health Clinic Worker` = mean(`Local Health Clinic Worker`),
    `Community-Based Organizations` = mean(`Community-Based Organizations`),
    `Department of Health` = mean(`Department of Health`),
    `African scientists` = mean(`African scientists`),
    `International Scientists` = mean(`International Scientists`),
    `International NGOs` = mean(`International NGOs`),
    WHO = mean(WHO)
  )

SA %>%
  kable(caption = "Income and institutional Trust", digits = 3)%>%
  kable_styling(bootstrap_options = c("striped", "hover"))

Income and institutional Trust

demog_income	doctor	Ministers/Religious Leaders	Community Leaders	Local Health Clinic Worker	Community-Based Organizations	Department of Health	African scientists	International Scientists	International NGOs	WHO
< R5,000	3.813	2.488	2.130	3.805	2.821	3.618	3.098	2.976	2.675	3.797
R5,000 - R9,999	4.030	2.576	2.030	3.606	2.818	3.848	3.697	3.364	3.091	4.061
R10,000 - R29,999	4.045	2.591	2.182	4.000	3.136	4.227	3.909	3.818	3.136	4.182
R30,000 - R49,999	3.714	3.571	2.857	3.857	3.000	3.429	3.571	3.571	3.429	3.571
R50,000 - R99,999	4.333	2.583	2.750	3.583	2.917	3.500	3.333	3.000	3.083	3.417
> R100,000	3.800	2.200	2.200	3.700	2.700	3.300	3.100	3.300	2.900	3.700

Kenya

df_Kenya = df_retarget %>% filter(country_answer == "Kenya")
df_Kenya$demog_income[!df_Kenya$demog_income %in% c("< 50k KSh","50k KSh - 100k KSh", "100k KSh - 300k KSh", "300k KSh - 500k Ksh", "500k Ksh-1,000k KSh", "> 1,000k KSh")] = NA
df_Kenya$demog_income = factor(df_Kenya$demog_income, levels=c("< 50k KSh","50k KSh - 100k KSh", "100k KSh - 300k KSh", "300k KSh - 500k Ksh", "500k Ksh-1,000k KSh", "> 1,000k KSh"))
Kenya = df_Kenya %>% 
  filter(!is.na(demog_income)) %>% 
  select(doctor, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo,
         demog_income) %>%
  group_by(demog_income) %>%
  drop_na() %>%
  dplyr::summarise(
    doctor = mean(doctor),
    `Ministers/Religious Leaders` = mean(`Ministers/Religious Leaders`),
    `Community Leaders` = mean(`Community Leaders`),
    `Local Health Clinic Worker` = mean(`Local Health Clinic Worker`),
    `Community-Based Organizations` = mean(`Community-Based Organizations`),
    `Department of Health` = mean(`Department of Health`),
    `African scientists` = mean(`African scientists`),
    `International Scientists` = mean(`International Scientists`),
    `International NGOs` = mean(`International NGOs`),
    WHO = mean(WHO)
  )

Kenya %>%
  kable(caption = "Income and institutional Trust", digits = 3)%>%
  kable_styling(bootstrap_options = c("striped", "hover"))

Income and institutional Trust

demog_income	doctor	Ministers/Religious Leaders	Community Leaders	Local Health Clinic Worker	Community-Based Organizations	Department of Health	African scientists	International Scientists	International NGOs	WHO
< 50k KSh	3.851	2.817	2.691	3.566	3.229	3.686	3.069	3.343	3.297	3.931
50k KSh - 100k KSh	3.735	2.985	2.868	3.779	3.456	3.721	3.250	3.353	3.250	3.882
100k KSh - 300k KSh	4.214	2.893	2.821	3.750	3.429	3.786	3.179	3.786	3.464	4.214
300k KSh - 500k Ksh	3.333	2.000	2.000	2.500	2.167	3.667	3.333	3.500	3.000	3.833
500k Ksh-1,000k KSh	3.857	2.429	3.143	3.714	3.143	3.714	3.429	4.000	3.857	4.000
> 1,000k KSh	3.900	2.500	2.600	4.000	3.100	3.900	3.000	3.800	3.800	4.400

Nigeria

df_Nigeria = df_retarget %>% filter(country_answer == "Nigeria")
df_Nigeria$demog_income[!df_Nigeria$demog_income %in% c("< 50,000₦","50,000 - 99,999₦", "100,000₦  - 299,999₦", "300,000₦  - 499,999₦", "500,000₦  - 999,999₦", "> 1,000,000₦")] = NA
df_Nigeria$demog_income = factor(df_Nigeria$demog_income, levels=c("< 50,000₦","50,000 - 99,999₦", "100,000₦  - 299,999₦", "300,000₦  - 499,999₦", "500,000₦  - 999,999₦", "> 1,000,000₦"))
Nigeria = df_Nigeria %>% 
  filter(!is.na(demog_income)) %>% 
  select(doctor, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo,
         demog_income) %>%
  group_by(demog_income) %>%
  drop_na() %>%
  dplyr::summarise(
    doctor = mean(doctor),
    `Ministers/Religious Leaders` = mean(`Ministers/Religious Leaders`),
    `Community Leaders` = mean(`Community Leaders`),
    `Local Health Clinic Worker` = mean(`Local Health Clinic Worker`),
    `Community-Based Organizations` = mean(`Community-Based Organizations`),
    `Department of Health` = mean(`Department of Health`),
    `African scientists` = mean(`African scientists`),
    `International Scientists` = mean(`International Scientists`),
    `International NGOs` = mean(`International NGOs`),
    WHO = mean(WHO)
  )

Nigeria %>%
  kable(caption = "Income and institutional Trust", digits = 3)%>%
  kable_styling(bootstrap_options = c("striped", "hover"))

Income and institutional Trust

demog_income	doctor	Ministers/Religious Leaders	Community Leaders	Local Health Clinic Worker	Community-Based Organizations	Department of Health	African scientists	International Scientists	International NGOs	WHO
< 50,000₦	3.963	3.157	2.787	3.556	3.120	4.028	3.685	3.926	3.556	4.278
50,000 - 99,999₦	3.521	2.854	2.208	3.333	2.708	3.438	3.167	3.458	3.333	4.271
100,000₦ - 299,999₦	4.364	3.061	2.424	3.606	2.939	3.939	3.545	3.788	3.394	4.303
300,000₦ - 499,999₦	4.400	3.467	2.600	3.533	3.200	4.067	3.667	4.067	3.867	4.600
500,000₦ - 999,999₦	4.077	2.769	2.000	3.385	2.769	4.077	3.308	4.000	3.846	4.615
> 1,000,000₦	4.833	3.500	3.000	4.000	4.000	4.333	4.167	4.667	4.167	4.667

Ghana

df_Ghana = df_retarget %>% filter(country_answer == "Ghana")
df_Ghana$demog_income[!df_Ghana$demog_income %in% c("< 5,000GH₵","5,000GH₵ - 9,999GH₵", "10,000GH₵ - 29,999GH", "30,000GH₵ - 49,999GH", "50,000GH₵-99,999GH₵", "> 100,000GH₵")] = NA
df_Ghana$demog_income = factor(df_Ghana$demog_income, levels=c("< 5,000GH₵","5,000GH₵ - 9,999GH₵", "10,000GH₵ - 29,999GH", "30,000GH₵ - 49,999GH", "50,000GH₵-99,999GH₵", "> 100,000GH₵"))
Ghana = df_Ghana %>% 
  filter(!is.na(demog_income)) %>%
  select(doctor, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo,
         demog_income) %>%
  group_by(demog_income) %>%
  drop_na() %>%
  dplyr::summarise(
    doctor = mean(doctor),
    `Ministers/Religious Leaders` = mean(`Ministers/Religious Leaders`),
    `Community Leaders` = mean(`Community Leaders`),
    `Local Health Clinic Worker` = mean(`Local Health Clinic Worker`),
    `Community-Based Organizations` = mean(`Community-Based Organizations`),
    `Department of Health` = mean(`Department of Health`),
    `African scientists` = mean(`African scientists`),
    `International Scientists` = mean(`International Scientists`),
    `International NGOs` = mean(`International NGOs`),
    WHO = mean(WHO)
  )

Ghana %>%
  kable(caption = "Income and institutional Trust", digits = 3)%>%
  kable_styling(bootstrap_options = c("striped", "hover"))

Income and institutional Trust

demog_income	doctor	Ministers/Religious Leaders	Community Leaders	Local Health Clinic Worker	Community-Based Organizations	Department of Health	African scientists	International Scientists	International NGOs	WHO
< 5,000GH₵	3.904	2.637	2.555	3.610	3.007	3.849	3.534	3.699	3.219	4.048
5,000GH₵ - 9,999GH₵	3.926	2.037	2.148	4.000	3.000	3.741	3.296	3.407	3.185	4.259
10,000GH₵ - 29,999GH	4.333	3.167	2.917	4.167	3.667	4.167	4.333	4.000	3.667	4.333
30,000GH₵ - 49,999GH	4.333	3.000	3.000	4.000	3.000	3.333	3.000	3.000	3.000	4.000

6. Other Variables

Takeaway:

• Surprisingly, people who have heard of ‘herd immunity’ have a lower institutional trust level

Heard “herd immunity” OR Not

Question: Have you heard of “herd immunity? - yes(1); no(0)

Correlation Matrix

df_retarget$herd_immunity[df_retarget$herd_immunity == "yes"] = 1
df_retarget$herd_immunity[df_retarget$herd_immunity == "no"] = 0
df_retarget$herd_immunity[!df_retarget$herd_immunity %in% c("0","1")] = NA
df_retarget$herd_immunity = strtoi(df_retarget$herd_immunity, base = 0L)

df_heard_corr = df_retarget %>%
  select(doctor, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo,
         herd_immunity)

df_heard_corr = data.frame(cor(df_heard_corr, use = "pairwise.complete.obs"))
df_heard_corr = df_heard_corr %>% 
  select(herd_immunity)
#df_future_corr = df_future_corr[!(row.names(df_future_corr) %in% c("vax_next_year")),]

ggcorrplot(df_heard_corr, type = "lower", lab = TRUE, lab_size = 5/.pt, tl.cex = 7, colors= c(cb_colors[1], "white", cb_colors[2])) + 
  labs(y = "", x = "", title = "Correlation Matrix: \nInstitutional Trust Variables vs 'herd immunity'", 
       caption = paste("Number of Observations:", scales::comma(nrow(df_heard_corr))))+
  theme(axis.text.x = element_text(angle = 45, hjust=1))

Descriptive statistics

df_retarget %>%
  select(vax_status_retarget, herd_immunity) %>% 
  split(.$vax_status_retarget) %>% 
  map(describe) %>% 
  do.call(rbind, .) %>%
  data.frame() %>%
  select(n, mean , se) %>%
  rownames_to_column() %>%
  separate(rowname, c("vax_status_retarget", "variable"), sep = "\\.") %>%
  filter(variable != "vax_status_retarget*") %>%
  pivot_longer(cols = c("n", "mean", "se")) %>%
  mutate(column = paste(vax_status_retarget, name, sep = "_")) %>%
  select(column, variable, value) %>%
  pivot_wider(names_from = column) %>%
  arrange(desc(`0_mean`)) %>%
  kable(digits = 3,caption = "Summary Statistics for Trust Variables by Vaccination Status", 
        col.names = c("", "N", "Mean", "SE", "N", "Mean", "SE")) %>%
  kable_styling(bootstrap_options = c("striped", "hover"))%>%
  add_header_above(c(" " = 1, "Unvaccinated" = 3, "Vaccinated" = 3))

Summary Statistics for Trust Variables by Vaccination Status

	Unvaccinated			Vaccinated
	N	Mean	SE	N	Mean	SE
herd_immunity	279	0.272	0.027	760	0.353	0.017

7. Aggregate trust variables - Updated on 4/11

According to the Trust Factor Analysis script here, we aggregate the following variables into one and perform analysis in this section:

• health_department
• WHO
• doctor
• clinic_worker

df_retarget$institutional_trust_raw = (df_retarget$WHO+df_retarget$doctor+df_retarget$clinic_worker+df_retarget$health_department)/4
df_retarget$institutional_trust_raw = as.factor(df_retarget$institutional_trust_raw)
institutional_trust_df = df_retarget %>% filter(!is.na(institutional_trust_raw))
institutional_trust_plot <- institutional_trust_df %>% count(institutional_trust_raw) %>% mutate(percentage = n / nrow(WHO_df))
ggplot(institutional_trust_plot, aes(x = fct_inorder(institutional_trust_raw), y = n, fill = institutional_trust_raw)) +
  geom_bar(stat = "identity") +
  geom_text(aes(label = paste0(" (", round(percentage * 100,2), "%)")), vjust = -0.25, size=3) +
  ggtitle(paste0("Institutional Trust - Aggregated (n = ", sum(institutional_trust_plot$n), ")")) +
  theme(plot.title = element_text(face = "bold"), legend.position = "none") +
  labs(x = "trust score", y = "count")

Assign levels

We classify the institutional_trust into 3 levels: - High: score \(\geq\) 4.0 - Median: 2.5 \(\geq\) score \(\geq\) 3.75 = Low: score \(\leq\) 2.25

df_retarget$institutional_trust_raw = as.numeric(as.character(df_retarget$institutional_trust_raw))
df_retarget$institutional_trust = with(df_retarget,
                                       ifelse(institutional_trust_raw > 3.75, "High",
                                              ifelse(institutional_trust_raw > 2.49, "Medium",
                                                     "Low")))

institutional_trust_df = df_retarget %>% filter(!is.na(institutional_trust))
institutional_trust_plot <- institutional_trust_df %>% count(institutional_trust) %>% mutate(percentage = n / nrow(WHO_df))
ggplot(institutional_trust_plot, aes(x = fct_inorder(institutional_trust), y = n, fill = institutional_trust)) +
  geom_bar(stat = "identity") +
  geom_text(aes(label = paste0(" (", round(percentage * 100,2), "%)")), vjust = -0.25, size=3) +
  ggtitle(paste0("Institutional Trust - Aggregated (n = ", sum(institutional_trust_plot$n), ")")) +
  theme(plot.title = element_text(face = "bold"), legend.position = "none") +
  labs(x = "trust score", y = "count")

Heterogeneity analysis

Other trust variables

Correlation Matrix

df_heard_corr = df_retarget %>%
  select(institutional_trust_raw, family, friends, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Community-Based Organizations` = community_based_org, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo, `Local News/Media` = local_news_media, `International News/Media` = international_news_media)

df_heard_corr = data.frame(cor(df_heard_corr, use = "pairwise.complete.obs"))
df_heard_corr = df_heard_corr %>% 
  select(institutional_trust_raw)
#df_future_corr = df_future_corr[!(row.names(df_future_corr) %in% c("vax_next_year")),]

ggcorrplot(df_heard_corr, type = "lower", lab = TRUE, lab_size = 5/.pt, tl.cex = 7, colors= c(cb_colors[1], "white", cb_colors[2])) + 
  labs(y = "", x = "", title = "Correlation Matrix: \nAggregated Institutional Trust vs Other Trust variables", 
       caption = paste("Number of Observations:", scales::comma(nrow(df_heard_corr))))+
  theme(axis.text.x = element_text(angle = 45, hjust=1))

Descriptive statistics

df_retarget %>%
  select(institutional_trust, family, friends, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Community-Based Organizations` = community_based_org, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo, `Local News/Media` = local_news_media, `International News/Media` = international_news_media) %>% 
  split(.$institutional_trust) %>% 
  map(describe) %>% 
  do.call(rbind, .) %>%
  data.frame() %>%
  select(n, mean , se) %>%
  rownames_to_column() %>%
  separate(rowname, c("institutional_trust", "variable"), sep = "\\.") %>%
  filter(variable != "institutional_trust*") %>%
  pivot_longer(cols = c("n", "mean", "se")) %>%
  mutate(column = paste(institutional_trust, name, sep = "_")) %>%
  select(column, variable, value) %>%
  pivot_wider(names_from = column) %>%
  arrange(desc(`High_mean`)) %>%
  kable(digits = 3,caption = "Summary Statistics for Trust Variables by Aggregated Institutional Trust", 
        col.names = c("", "N", "Mean", "SE", "N", "Mean", "SE", "N", "Mean", "SE")) %>%
  kable_styling(bootstrap_options = c("striped", "hover")) %>%
  add_header_above(c(" " = 1, "High" = 3, "Low" = 3, "Medium" = 3))

Summary Statistics for Trust Variables by Aggregated Institutional Trust

	High			Low			Medium
	N	Mean	SE	N	Mean	SE	N	Mean	SE
International Scientists	828	4.037	0.042	268	1.694	0.059	412	2.927	0.063
African scientists	830	3.808	0.044	272	1.765	0.061	414	2.841	0.065
International NGOs	826	3.723	0.042	262	1.634	0.055	411	2.766	0.059
family	830	3.573	0.047	273	2.095	0.081	417	3.151	0.072
Community-Based Organizations	830	3.567	0.042	272	1.688	0.058	417	2.504	0.056
International News/Media	826	3.489	0.044	258	1.655	0.056	408	2.502	0.059
Local News/Media	826	3.249	0.046	260	1.592	0.059	409	2.369	0.058
Ministers/Religious Leaders	830	3.164	0.049	273	1.758	0.069	417	2.374	0.062
Community Leaders	830	2.902	0.047	273	1.659	0.061	417	2.067	0.055
friends	830	2.708	0.046	273	1.861	0.071	417	2.235	0.055

Sorting Questions

Correlation Matrix

df_heard_corr = df_retarget %>%
  select(institutional_trust_raw, covid_is_a_problem, covid_is_problem_my_country,vax_is_safe,vax_prevent_sick_death, need_vax_for_protection,vax_safer_covid,developer_want_to_help,health_worker_want_to_help,gov_want_to_help,important_to_protect_myself,important_to_protect_other,vax_moral,worry_short_term_side_effect,worry_long_term_side_effect,covid_is_real,probably_exposed,afraid_needle,deeply_religious,important_moral,fit_in_group_importance,responsible_importance)

df_heard_corr = data.frame(cor(df_heard_corr, use = "pairwise.complete.obs"))
df_heard_corr = df_heard_corr %>% 
  select(institutional_trust_raw)
#df_future_corr = df_future_corr[!(row.names(df_future_corr) %in% c("vax_next_year")),]

ggcorrplot(df_heard_corr, type = "lower", lab = TRUE, lab_size = 5/.pt, tl.cex = 7, colors= c(cb_colors[1], "white", cb_colors[2])) + 
  labs(y = "", x = "", title = "Correlation Matrix: \nAggregated Institutional Trust vs Sorting Questions", 
       caption = paste("Number of Observations:", scales::comma(nrow(df_heard_corr))))+
  theme(axis.text.x = element_text(angle = 45, hjust=1))

Variable Descriptions:

• covid_is_a_problem: People are still dying from COVID
• covid_is_problem_my_country: COVID is a problem in [my country]
• vax_is_safe: I think the COVID vaccines are safe
• vax_prevent_sick_death: COVID vaccines help prevent serious sickness and death
• need_vax_for_protection: You need a vaccine for protection from sickness (not because you are sick)
• vax_safer_covid: Getting the vaccine is much safer than getting COVID
• developer_want_to_help: I think the people who developed the vaccine wanted to help people
• health_worker_want_to_help: I think my local healthcare workers want me to be healthy and well
• gov_want_to_help: I think my government’s department of health workers want me to be healthy and well
• important_to_protect_myself: It is important to me that I protect myself from the effects of COVID
• important_to_protect_other: It is important to me that I protect others from the effects of COVID
• vax_moral: Getting vaccinated is a moral issue
• worry_short_term_side_effect: I worry about short-term side effects of the COVID vaccine
• worry_long_term_side_effect: I worry about long-term side effects of the COVID vaccine
• covid_is_real: COVID has killed millions of people worldwide
• probably_exposed: I will probably be exposed to someone with COVID over the next year
• afraid_needle: I am very afraid of needles
• deeply_religious: I am a deeply religious person
• important_moral: It is important to me that I be a moral person
• fit_in_group_importance: It is important to me to feel like I “fit in” in with my group
• responsible_importance: It is important to me to feel like I am a responsible member of my community

Descriptive statistics

df_retarget %>%
  select(institutional_trust, covid_is_a_problem, covid_is_problem_my_country,vax_is_safe,vax_prevent_sick_death, need_vax_for_protection,vax_safer_covid,developer_want_to_help,health_worker_want_to_help,gov_want_to_help,important_to_protect_myself,important_to_protect_other,vax_moral,worry_short_term_side_effect,worry_long_term_side_effect,covid_is_real,probably_exposed,afraid_needle,deeply_religious,important_moral,fit_in_group_importance,responsible_importance) %>% 
  split(.$institutional_trust) %>% 
  map(describe) %>% 
  do.call(rbind, .) %>%
  data.frame() %>%
  select(n, mean , se) %>%
  rownames_to_column() %>%
  separate(rowname, c("institutional_trust", "variable"), sep = "\\.") %>%
  filter(variable != "institutional_trust*") %>%
  pivot_longer(cols = c("n", "mean", "se")) %>%
  mutate(column = paste(institutional_trust, name, sep = "_")) %>%
  select(column, variable, value) %>%
  pivot_wider(names_from = column) %>%
  arrange(desc(`High_mean`)) %>%
  kable(digits = 3,caption = "Summary Statistics for Trust Variables by Aggregated Institutional Trust", 
        col.names = c("", "N", "Mean", "SE", "N", "Mean", "SE", "N", "Mean", "SE")) %>%
  kable_styling(bootstrap_options = c("striped", "hover")) %>%
  add_header_above(c(" " = 1, "High" = 3, "Low" = 3, "Medium" = 3))

Summary Statistics for Trust Variables by Aggregated Institutional Trust

	High			Low			Medium
	N	Mean	SE	N	Mean	SE	N	Mean	SE
important_to_protect_myself	659	4.595	0.027	144	3.771	0.115	312	4.295	0.052
important_to_protect_other	657	4.542	0.028	144	3.792	0.109	310	4.290	0.050
vax_safer_covid	661	4.365	0.036	148	3.095	0.116	315	3.854	0.064
covid_is_real	655	4.344	0.037	142	3.423	0.119	306	3.987	0.064
responsible_importance	647	4.338	0.035	140	3.679	0.110	303	4.013	0.061
health_worker_want_to_help	660	4.312	0.033	146	3.281	0.110	314	3.844	0.054
gov_want_to_help	661	4.300	0.031	146	3.048	0.108	314	3.653	0.060
important_moral	650	4.229	0.036	140	3.679	0.112	303	4.030	0.058
need_vax_for_protection	662	4.192	0.041	149	3.208	0.112	316	3.696	0.072
developer_want_to_help	661	4.150	0.038	146	2.959	0.110	314	3.669	0.058
vax_is_safe	663	3.840	0.042	151	2.576	0.116	316	3.282	0.061
worry_long_term_side_effect	657	3.828	0.045	141	3.504	0.115	307	3.831	0.067
vax_moral	658	3.653	0.043	143	2.825	0.112	310	3.161	0.067
deeply_religious	652	3.647	0.048	140	3.264	0.122	304	3.467	0.073
vax_prevent_sick_death	662	3.613	0.045	151	2.689	0.112	315	3.098	0.063
fit_in_group_importance	650	3.558	0.052	140	3.043	0.114	303	3.333	0.075
worry_short_term_side_effect	657	3.531	0.045	142	3.352	0.114	308	3.640	0.064
covid_is_a_problem	665	3.325	0.046	152	2.638	0.109	319	3.100	0.063
covid_is_problem_my_country	664	3.280	0.052	151	2.424	0.113	318	2.909	0.074
afraid_needle	653	2.758	0.057	141	2.730	0.123	305	2.856	0.084
probably_exposed	654	2.622	0.048	142	2.599	0.110	306	2.490	0.069

Variable Descriptions:

• covid_is_a_problem: People are still dying from COVID
• covid_is_problem_my_country: COVID is a problem in [my country]
• vax_is_safe: I think the COVID vaccines are safe
• vax_prevent_sick_death: COVID vaccines help prevent serious sickness and death
• need_vax_for_protection: You need a vaccine for protection from sickness (not because you are sick)
• vax_safer_covid: Getting the vaccine is much safer than getting COVID
• developer_want_to_help: I think the people who developed the vaccine wanted to help people
• health_worker_want_to_help: I think my local healthcare workers want me to be healthy and well
• gov_want_to_help: I think my government’s department of health workers want me to be healthy and well
• important_to_protect_myself: It is important to me that I protect myself from the effects of COVID
• important_to_protect_other: It is important to me that I protect others from the effects of COVID
• vax_moral: Getting vaccinated is a moral issue
• worry_short_term_side_effect: I worry about short-term side effects of the COVID vaccine
• worry_long_term_side_effect: I worry about long-term side effects of the COVID vaccine
• covid_is_real: COVID has killed millions of people worldwide
• probably_exposed: I will probably be exposed to someone with COVID over the next year
• afraid_needle: I am very afraid of needles
• deeply_religious: I am a deeply religious person
• important_moral: It is important to me that I be a moral person
• fit_in_group_importance: It is important to me to feel like I “fit in” in with my group
• responsible_importance: It is important to me to feel like I am a responsible member of my community

Demographics

Continuous Demographic Variables

Starred variables have a corresponding variable, with the suffix _num (ie education_num) that provides a numeric ordering of the variables. This order is listed in parenthesis.

Variable	Description
gender*	male (0) or female (1)
age	integer 18 to 99
education*	< high school (1), high school (2), some college (3), 2-year degree (4), 4-year degree (5), graduate degree (6), other, prefer not to say
location*	rural(1), suburban (2), urban(3), other, prefer not to say
religiosity*	not very religious (1), somewhat religious (2), very religious (3), other, prefer not to say
white	1 if the participant is a white or caucasian, 0 if not
ethnicity	asian or indian, black or african, coloured, white or caucasian, other, prefer not to say

df_retarget_num = numeric_cols(df_retarget)
df_heard_corr = df_retarget_num %>%
  select(institutional_trust_raw, age, education = education_num, location=location_num, religiosity = religiosity_num, gender = gender_num, white)

df_heard_corr = data.frame(cor(df_heard_corr, use = "pairwise.complete.obs"))
df_heard_corr = df_heard_corr %>% 
  select(institutional_trust_raw)
#df_future_corr = df_future_corr[!(row.names(df_future_corr) %in% c("vax_next_year")),]

ggcorrplot(df_heard_corr, type = "lower", lab = TRUE, lab_size = 5/.pt, tl.cex = 7, colors= c(cb_colors[1], "white", cb_colors[2])) + 
  labs(y = "", x = "", title = "Correlation Matrix: \nAggregated Institutional Trust vs Demographics", 
       caption = paste("Number of Observations:", scales::comma(nrow(df_heard_corr))))+
  theme(axis.text.x = element_text(angle = 45, hjust=1))

Other Demographic Variables

1. Ethnicity

df_retarget %>%
  drop_na(institutional_trust, ethnicity) %>%
  count(institutional_trust, ethnicity) %>%
  arrange(institutional_trust) %>%
  ggplot(aes(institutional_trust, n, fill = ethnicity)) +
  geom_col(position = "dodge", alpha = 0.8, color="black") +
  theme_bw() +
  theme(legend.position = "bottom") +
  labs(
    x = "institutional trust level",
    y = "Count",
    fill = "ethnicity",
    title = "Distribution of ethnicity by each institutional trust level"
  )

2. Religion

df_retarget %>%
  drop_na(institutional_trust, demog_religion) %>%
  count(institutional_trust, demog_religion) %>%
  arrange(institutional_trust) %>%
  ggplot(aes(institutional_trust, n, fill = demog_religion)) +
  geom_col(position = "dodge", alpha = 0.8, color="black") +
  theme_bw() +
  theme(legend.position = "bottom") +
  labs(
    x = "institutional trust level",
    y = "Count",
    fill = "religion",
    title = "Distribution of religion by each institutional trust level"
  )

3. Income

South Africa

df_SA = df_retarget %>% filter(country_answer == "South Africa")
df_SA$demog_income[!df_SA$demog_income %in% c("< R5,000","R5,000 - R9,999", "R10,000 - R29,999", "R30,000 - R49,999", "R50,000 - R99,999", "> R100,000")] = NA
df_SA$demog_income = factor(df_SA$demog_income, levels=c("< R5,000","R5,000 - R9,999", "R10,000 - R29,999", "R30,000 - R49,999", "R50,000 - R99,999", "> R100,000"))

df_SA %>%
  drop_na(institutional_trust, demog_income) %>%
  count(institutional_trust, demog_income) %>%
  arrange(institutional_trust) %>%
  ggplot(aes(institutional_trust, n, fill = demog_income)) +
  geom_col(position = "dodge", alpha = 0.8, color="black") +
  theme_bw() +
  theme(legend.position = "bottom") +
  labs(
    x = "institutional trust level",
    y = "Count",
    fill = "income",
    title = "Distribution of income by each institutional trust level"
  )

Kenya

df_Kenya = df_retarget %>% filter(country_answer == "Kenya")
df_Kenya$demog_income[!df_Kenya$demog_income %in% c("< 50k KSh","50k KSh - 100k KSh", "100k KSh - 300k KSh", "300k KSh - 500k Ksh", "500k Ksh-1,000k KSh", "> 1,000k KSh")] = NA
df_Kenya$demog_income = factor(df_Kenya$demog_income, levels=c("< 50k KSh","50k KSh - 100k KSh", "100k KSh - 300k KSh", "300k KSh - 500k Ksh", "500k Ksh-1,000k KSh", "> 1,000k KSh"))

df_Kenya %>%
  drop_na(institutional_trust, demog_income) %>%
  count(institutional_trust, demog_income) %>%
  arrange(institutional_trust) %>%
  ggplot(aes(institutional_trust, n, fill = demog_income)) +
  geom_col(position = "dodge", alpha = 0.8, color="black") +
  theme_bw() +
  theme(legend.position = "bottom") +
  labs(
    x = "institutional trust level",
    y = "Count",
    fill = "income",
    title = "Distribution of income by each institutional trust level"
  )

Nigeria

df_Nigeria = df_retarget %>% filter(country_answer == "Nigeria")
df_Nigeria$demog_income[!df_Nigeria$demog_income %in% c("< 50,000₦","50,000 - 99,999₦", "100,000₦  - 299,999₦", "300,000₦  - 499,999₦", "500,000₦  - 999,999₦", "> 1,000,000₦")] = NA
df_Nigeria$demog_income = factor(df_Nigeria$demog_income, levels=c("< 50,000₦","50,000 - 99,999₦", "100,000₦  - 299,999₦", "300,000₦  - 499,999₦", "500,000₦  - 999,999₦", "> 1,000,000₦"))

df_Nigeria %>%
  drop_na(institutional_trust, demog_income) %>%
  count(institutional_trust, demog_income) %>%
  arrange(institutional_trust) %>%
  ggplot(aes(institutional_trust, n, fill = demog_income)) +
  geom_col(position = "dodge", alpha = 0.8, color="black") +
  theme_bw() +
  theme(legend.position = "bottom") +
  labs(
    x = "institutional trust level",
    y = "Count",
    fill = "income",
    title = "Distribution of income by each institutional trust level"
  )

Ghana

df_Ghana = df_retarget %>% filter(country_answer == "Ghana")
df_Ghana$demog_income[!df_Ghana$demog_income %in% c("< 5,000GH₵","5,000GH₵ - 9,999GH₵", "10,000GH₵ - 29,999GH", "30,000GH₵ - 49,999GH", "50,000GH₵-99,999GH₵", "> 100,000GH₵")] = NA
df_Ghana$demog_income = factor(df_Ghana$demog_income, levels=c("< 5,000GH₵","5,000GH₵ - 9,999GH₵", "10,000GH₵ - 29,999GH", "30,000GH₵ - 49,999GH", "50,000GH₵-99,999GH₵", "> 100,000GH₵"))

df_Ghana %>%
  drop_na(institutional_trust, demog_income) %>%
  count(institutional_trust, demog_income) %>%
  arrange(institutional_trust) %>%
  ggplot(aes(institutional_trust, n, fill = demog_income)) +
  geom_col(position = "dodge", alpha = 0.8, color="black") +
  theme_bw() +
  theme(legend.position = "bottom") +
  labs(
    x = "institutional trust level",
    y = "Count",
    fill = "income",
    title = "Distribution of income by each institutional trust level"
  )

Vaccination Status/Attitudes

Vaccination Status

df_retarget %>%
  drop_na(institutional_trust, vax_status_retarget) %>%
  count(institutional_trust, vax_status_retarget) %>%
  arrange(institutional_trust) %>%
  ggplot(aes(institutional_trust, n, fill = vax_status_retarget)) +
  geom_col(position = "dodge", alpha = 0.8, color="black") +
  theme_bw() +
  theme(legend.position = "bottom") +
  labs(
    x = "institutional trust level",
    y = "Count",
    fill = "vaccine status",
    title = "Distribution of vaccine status by each institutional trust level"
  )

Vaccination Dose

df_retarget %>%
  drop_na(institutional_trust, vax_status_retarget_4) %>%
  count(institutional_trust, vax_status_retarget_4) %>%
  arrange(institutional_trust) %>%
  ggplot(aes(institutional_trust, n, fill = vax_status_retarget_4)) +
  geom_col(position = "dodge", alpha = 0.8, color="black") +
  theme_bw() +
  theme(legend.position = "bottom") +
  labs(
    x = "institutional trust level",
    y = "Count",
    fill = "vaccine dose",
    title = "Distribution of vaccine dose by each institutional trust level"
  )

Intention to get vaccine in future

df_retarget$vax_next_year = as.factor(df_retarget$vax_next_year)
df_retarget %>%
  drop_na(institutional_trust, vax_next_year) %>%
  count(institutional_trust, vax_next_year) %>%
  arrange(institutional_trust) %>%
  ggplot(aes(institutional_trust, n, fill = vax_next_year)) +
  geom_col(position = "dodge", alpha = 0.8, color="black") +
  theme_bw() +
  theme(legend.position = "bottom") +
  labs(
    x = "institutional trust level",
    y = "Count",
    fill = "intention to get vaccine",
    title = "Distribution of intention to get vaccine in future by each institutional trust level"
  )

8. Regularized regressions - Updated on 4/11

library(glmnet)

Predict intentions to vaccinate

Regularized regressions predicting intentions to vaccinate from all the different institutional trust variables and sorting questions

without aggregation

df_Regularized = df_retarget %>%
  select(vax_next_year,
         doctor, family, friends, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Local Health Clinic Worker` = clinic_worker, `Community-Based Organizations` = community_based_org, `Department of Health` = health_department, WHO, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo, `Local News/Media` = local_news_media, `International News/Media` = international_news_media,
         covid_is_a_problem, covid_is_problem_my_country,vax_is_safe,vax_prevent_sick_death, need_vax_for_protection,vax_safer_covid,developer_want_to_help,health_worker_want_to_help,gov_want_to_help,important_to_protect_myself,important_to_protect_other,vax_moral,worry_short_term_side_effect,worry_long_term_side_effect,covid_is_real,probably_exposed,afraid_needle,deeply_religious,important_moral,fit_in_group_importance,responsible_importance) %>%
  na.omit()

df_Regularized$vax_next_year = strtoi(df_Regularized$vax_next_year)
df_Regularized$vax_next_year[!df_Regularized$vax_next_year %in% c("1","2","3","4","5")] = NA

x = df_Regularized %>%
  select(doctor, `Local Health Clinic Worker`, `Department of Health`, WHO, family, friends, `Ministers/Religious Leaders`, `Community Leaders`, `Community-Based Organizations`, `African scientists`, `International Scientists`, `International NGOs`, `Local News/Media`, `International News/Media`,
         covid_is_a_problem, covid_is_problem_my_country,vax_is_safe,vax_prevent_sick_death, need_vax_for_protection,vax_safer_covid,developer_want_to_help,health_worker_want_to_help,gov_want_to_help,important_to_protect_myself,important_to_protect_other,vax_moral,worry_short_term_side_effect,worry_long_term_side_effect,covid_is_real,probably_exposed,afraid_needle,deeply_religious,important_moral,fit_in_group_importance,responsible_importance) %>% 
  mutate_if(is.character, as.numeric)

#x = as.numeric(x)
x = data.matrix(x)

y = df_Regularized$vax_next_year
  
cvfit = cv.glmnet(x, y)

print(cvfit)

## 
## Call:  cv.glmnet(x = x, y = y) 
## 
## Measure: Mean-Squared Error 
## 
##      Lambda Index Measure      SE Nonzero
## min 0.05802    27   1.606 0.04254      15
## 1se 0.14710    17   1.644 0.04045      10

coef(cvfit, s = "lambda.min")

## 36 x 1 sparse Matrix of class "dgCMatrix"
##                                         s1
## (Intercept)                   -0.048807649
## doctor                         .          
## Local Health Clinic Worker     .          
## Department of Health           0.066824688
## WHO                            .          
## family                         .          
## friends                        .          
## Ministers/Religious Leaders    .          
## Community Leaders              .          
## Community-Based Organizations  0.056618375
## African scientists             .          
## International Scientists       .          
## International NGOs             .          
## Local News/Media               .          
## International News/Media       0.002843443
## covid_is_a_problem             .          
## covid_is_problem_my_country    0.024929571
## vax_is_safe                    0.214425733
## vax_prevent_sick_death         0.080354398
## need_vax_for_protection        0.043472237
## vax_safer_covid                0.078846333
## developer_want_to_help         0.096599587
## health_worker_want_to_help     .          
## gov_want_to_help               0.041978636
## important_to_protect_myself    .          
## important_to_protect_other     .          
## vax_moral                      0.075961889
## worry_short_term_side_effect   .          
## worry_long_term_side_effect    .          
## covid_is_real                  .          
## probably_exposed               0.041808640
## afraid_needle                  .          
## deeply_religious               0.011900046
## important_moral                .          
## fit_in_group_importance        0.005327352
## responsible_importance         0.042700674

with aggregation

df_Regularized = df_retarget %>%
  select(vax_next_year,
         institutional_trust_raw, family, friends, `Ministers/Religious Leaders` = minister_religious_leader, `Community Leaders` = community_leader, `Community-Based Organizations` = community_based_org, `African scientists` = african_scientist, `International Scientists` = international_scientist, `International NGOs` = international_ngo, `Local News/Media` = local_news_media, `International News/Media` = international_news_media,
         covid_is_a_problem, covid_is_problem_my_country,vax_is_safe,vax_prevent_sick_death, need_vax_for_protection,vax_safer_covid,developer_want_to_help,health_worker_want_to_help,gov_want_to_help,important_to_protect_myself,important_to_protect_other,vax_moral,worry_short_term_side_effect,worry_long_term_side_effect,covid_is_real,probably_exposed,afraid_needle,deeply_religious,important_moral,fit_in_group_importance,responsible_importance) %>%
  na.omit()

df_Regularized$vax_next_year = strtoi(df_Regularized$vax_next_year)
df_Regularized$vax_next_year[!df_Regularized$vax_next_year %in% c("1","2","3","4","5")] = NA

x = df_Regularized %>%
  select(institutional_trust_raw, family, friends, `Ministers/Religious Leaders`, `Community Leaders`, `Community-Based Organizations`, `African scientists`, `International Scientists`, `International NGOs`, `Local News/Media`, `International News/Media`,
         covid_is_a_problem, covid_is_problem_my_country,vax_is_safe,vax_prevent_sick_death, need_vax_for_protection,vax_safer_covid,developer_want_to_help,health_worker_want_to_help,gov_want_to_help,important_to_protect_myself,important_to_protect_other,vax_moral,worry_short_term_side_effect,worry_long_term_side_effect,covid_is_real,probably_exposed,afraid_needle,deeply_religious,important_moral,fit_in_group_importance,responsible_importance) %>% 
  mutate_if(is.character, as.numeric)

#x = as.numeric(x)
x = data.matrix(x)

y = df_Regularized$vax_next_year
  
cvfit = cv.glmnet(x, y)

print(cvfit)

## 
## Call:  cv.glmnet(x = x, y = y) 
## 
## Measure: Mean-Squared Error 
## 
##      Lambda Index Measure      SE Nonzero
## min 0.04817    29   1.607 0.06993      15
## 1se 0.17718    15   1.671 0.06407       9

coef(cvfit, s = "lambda.min")

## 33 x 1 sparse Matrix of class "dgCMatrix"
##                                         s1
## (Intercept)                   -0.188344863
## institutional_trust_raw        0.054739125
## family                         .          
## friends                        .          
## Ministers/Religious Leaders    .          
## Community Leaders              .          
## Community-Based Organizations  0.062431561
## African scientists             .          
## International Scientists       .          
## International NGOs             .          
## Local News/Media               .          
## International News/Media       0.011182500
## covid_is_a_problem             .          
## covid_is_problem_my_country    0.029398872
## vax_is_safe                    0.216371920
## vax_prevent_sick_death         0.077431535
## need_vax_for_protection        0.045570901
## vax_safer_covid                0.084175135
## developer_want_to_help         0.099093909
## health_worker_want_to_help     .          
## gov_want_to_help               0.046951354
## important_to_protect_myself    .          
## important_to_protect_other     .          
## vax_moral                      0.076755843
## worry_short_term_side_effect   .          
## worry_long_term_side_effect    .          
## covid_is_real                  .          
## probably_exposed               0.047743904
## afraid_needle                  .          
## deeply_religious               0.016622301
## important_moral                .          
## fit_in_group_importance        0.009360354
## responsible_importance         0.047646207

Predict trust in WHO

Regularized regressions predicting trust in WHO from all the different sorting questions and continuous demographic variables

df_retarget_num = numeric_cols(df_retarget)
df_Regularized = df_retarget_num %>%
  select(WHO,
         covid_is_a_problem, covid_is_problem_my_country,vax_is_safe,vax_prevent_sick_death, need_vax_for_protection,vax_safer_covid,developer_want_to_help,health_worker_want_to_help,gov_want_to_help,important_to_protect_myself,important_to_protect_other,vax_moral,worry_short_term_side_effect,worry_long_term_side_effect,covid_is_real,probably_exposed,afraid_needle,deeply_religious,important_moral,fit_in_group_importance,responsible_importance,
         age, education = education_num, location=location_num, religiosity = religiosity_num, gender = gender_num, white) %>%
  na.omit()

x = df_Regularized %>%
  select(covid_is_a_problem, covid_is_problem_my_country,vax_is_safe,vax_prevent_sick_death, need_vax_for_protection,vax_safer_covid,developer_want_to_help,health_worker_want_to_help,gov_want_to_help,important_to_protect_myself,important_to_protect_other,vax_moral,worry_short_term_side_effect,worry_long_term_side_effect,covid_is_real,probably_exposed,afraid_needle,deeply_religious,important_moral,fit_in_group_importance,responsible_importance,
         age, education, location, religiosity, gender, white) %>% 
  mutate_if(is.character, as.numeric)

#x = as.numeric(x)
x = data.matrix(x)

y = as.numeric(df_Regularized$WHO)
  
cvfit = cv.glmnet(x, y)

print(cvfit)

## 
## Call:  cv.glmnet(x = x, y = y) 
## 
## Measure: Mean-Squared Error 
## 
##      Lambda Index Measure      SE Nonzero
## min 0.03251    31   1.412 0.08034      15
## 1se 0.19041    12   1.488 0.07547       5

coef(cvfit, s = "lambda.min")

## 28 x 1 sparse Matrix of class "dgCMatrix"
##                                        s1
## (Intercept)                   1.529528958
## covid_is_a_problem            0.015983094
## covid_is_problem_my_country   0.040545196
## vax_is_safe                   0.090509791
## vax_prevent_sick_death        .          
## need_vax_for_protection       0.010718201
## vax_safer_covid               0.141460171
## developer_want_to_help        0.100097074
## health_worker_want_to_help    .          
## gov_want_to_help              0.196787543
## important_to_protect_myself   0.002515641
## important_to_protect_other    .          
## vax_moral                     .          
## worry_short_term_side_effect  .          
## worry_long_term_side_effect   .          
## covid_is_real                 0.105457155
## probably_exposed             -0.087973193
## afraid_needle                 .          
## deeply_religious              0.010433955
## important_moral               .          
## fit_in_group_importance       .          
## responsible_importance        0.053436090
## age                          -0.009197975
## education                     .          
## location                      .          
## religiosity                   .          
## gender                        0.119538942
## white                        -0.111750891