Research note
Replication materials
library(tidyverse) # for manipulating data
library(srvyr) # for dealing with survey weights
library(sjlabelled) # for using labels
library(flextable) # for formatting output tables
library(haven) # for reading SPSS files
library(broom) # for dealing with annoying output chunks
library(scales) # for percentage labelling
library(countrycode) # for working with country names
library(ggthemes) # for plot formatting
library(ggpubr) # for plot arrangements
library(cowplot) # for plot arrangements
library(gridExtra) # for plot arrangements
library(egg) # for plot arrangements
library(glue) # because paste() can work in annoying waysSUPLEMENTARY MATERIALS
As accompanying material to “…”, this document presents details of data transformations, statistical procedures and visualisation techniques. It also includes additional material, which has not been included in the main body of the paper for reasons of brevity, but support specific statements made in the paper. Code chunks are hidden by default but can be accessed by clicking the relevant buttons. All calculations rely on the publicly available integrated spss data file.
EMPIRICAL BASE
Data_harmonised_EP_trust <- haven::read_sav("Data_harmonised_EP_trust.sav")
country_vector <- c("BE", "DE", "EE", "ES", "FI", "FR", "GB", "HU", "IE", "NL", "PL", "PT", "SE", "SI")
trust_survey_data <- Data_harmonised_EP_trust %>%
select(T_SURVEY_ID, T_SURVEY_NAME, T_SURVEY_COUNTRY, T_SURVEY_YEAR, T_SURVEY_EDITION, Respondent_ID_harmonised, EP_trust_harmonised, NP_trust_harmonised, EP_trust_harmonised_item_nonresponse, NP_trust_harmonised_item_nonresponse, Weight_total_harmonised) %>%
filter(T_SURVEY_NAME %in% c("EB", "ESS")) %>%
filter(T_SURVEY_COUNTRY %in% country_vector) %>%
mutate(weights = sjlabelled::remove_all_labels(Weight_total_harmonised))Figure 0
trust_survey_data %>%
filter(T_SURVEY_COUNTRY %in% country_vector) %>%
group_by(T_SURVEY_COUNTRY, T_SURVEY_YEAR, T_SURVEY_NAME) %>%
count() %>%
mutate(T_SURVEY_COUNTRY =
countrycode::countrycode(sjlabelled::remove_all_labels(T_SURVEY_COUNTRY), origin = "iso2c", destination = "country.name")) %>%
ggplot(aes(x = T_SURVEY_COUNTRY, y = T_SURVEY_YEAR, label = n, color = T_SURVEY_NAME)) +
geom_label(show.legend = F) +
scale_color_manual(values = c("#56B4E9", "#D55E00"), labels = c("Eurobarometer", "European Social Survey")) +
coord_flip() + theme_bw() +
xlab("") + ylab("") + ggtitle("Survey projects - surveyed countries - survey waves - survey respondents") +
theme(text = element_text(colour = "black"),
legend.position = "bottom",
legend.title = element_text(size = 10, face = "bold"),
strip.background = element_rect(fill = "gray98"),
strip.text = element_text(size = 10, face = "bold", margin = margin(1,0,1,0, "pt")),
panel.spacing.x = unit(2, "pt"),
panel.spacing.y = unit(2, "pt"),
axis.text = element_text(color = "black", size = 9, face = "bold"),
plot.title = element_text(size = 10, face = "bold")) + facet_wrap(~T_SURVEY_NAME, nrow = 2)
FIGURE 1
#EP_NP_TRUST_PLOT
#EP_trust
#data_transformations
prop_EP_value_plot1 <- trust_survey_data %>%
filter(!is.na(EP_trust_harmonised)) %>%
filter(EP_trust_harmonised != 88) %>%
group_by(T_SURVEY_ID, T_SURVEY_NAME, T_SURVEY_COUNTRY, T_SURVEY_YEAR) %>%
count(EP_trust_harmonised, wt = weights) %>%
group_by(T_SURVEY_ID) %>%
mutate(prop_EP = n/sum(n))
prop_EP_DK_plot1 <- trust_survey_data %>%
filter(!is.na(EP_trust_harmonised)) %>%
group_by(T_SURVEY_ID, T_SURVEY_NAME, T_SURVEY_COUNTRY, T_SURVEY_YEAR) %>%
count(EP_trust_harmonised, wt = weights) %>%
group_by(T_SURVEY_ID) %>%
mutate(prop_EP = n/sum(n)) %>%
filter(EP_trust_harmonised == 88)
prop_EP_plot1 <- full_join(prop_EP_value_plot1, prop_EP_DK_plot1)
survey_means_ESS_EP <- trust_survey_data %>% srvyr::as_survey(weights = weights) %>%
mutate(T_SURVEY_COUNTRY = countrycode::countrycode(sjlabelled::remove_all_labels(T_SURVEY_COUNTRY),
origin = "iso2c",
destination = "country.name")) %>%
filter(EP_trust_harmonised != 88) %>%
filter(T_SURVEY_NAME == "ESS") %>%
group_by(T_SURVEY_ID, T_SURVEY_COUNTRY, T_SURVEY_YEAR) %>%
summarise(mean_EP_trust = srvyr::survey_mean(EP_trust_harmonised, na.rm = T))
survey_means_EB_EP <- trust_survey_data %>% srvyr::as_survey(weights = weights) %>%
mutate(T_SURVEY_COUNTRY = countrycode::countrycode(sjlabelled::remove_all_labels(T_SURVEY_COUNTRY),
origin = "iso2c",
destination = "country.name")) %>%
filter(EP_trust_harmonised != 88) %>%
filter(T_SURVEY_NAME == "EB") %>%
group_by(T_SURVEY_ID, T_SURVEY_COUNTRY, T_SURVEY_YEAR) %>%
summarise(mean_EP_trust = srvyr::survey_mean(EP_trust_harmonised, na.rm = T))
EB_DK_data_EP <- prop_EP_plot1 %>%
filter(T_SURVEY_NAME == "EB") %>%
mutate(T_SURVEY_COUNTRY = countrycode::countrycode(sjlabelled::remove_all_labels(T_SURVEY_COUNTRY),
origin = "iso2c",
destination = "country.name")) %>%
filter(!is.na(EP_trust_harmonised)) %>%
filter(EP_trust_harmonised == 88)
#EP_trust plot
trust_ep <- prop_EP_plot1 %>% ungroup() %>%
mutate(T_SURVEY_COUNTRY = countrycode::countrycode(sjlabelled::remove_all_labels(T_SURVEY_COUNTRY),
origin = "iso2c",
destination = "country.name")) %>%
filter(T_SURVEY_NAME == "ESS") %>%
filter(!is.na(EP_trust_harmonised)) %>%
mutate(EP_trust_recoded = ifelse(EP_trust_harmonised == 88, -0.3, EP_trust_harmonised)) %>%
ggplot(aes(x = as.numeric(T_SURVEY_YEAR))) +
geom_point(aes(y = EP_trust_recoded, size = prop_EP,
fill = ifelse(EP_trust_recoded == -0.3, "1", "2"),
alpha = prop_EP, shape = ifelse(EP_trust_recoded == -0.3, "1", "2")),
color = "gray20") +
geom_point(data = EB_DK_data_EP, aes(y = -0.15, size = prop_EP, alpha = prop_EP, fill = "1", shape = "1")) +
geom_smooth(data = survey_means_ESS_EP, aes(y = mean_EP_trust, linetype = "1"),
se = F, color = "gray20", alpha = 0.8) +
geom_smooth(data = survey_means_EB_EP, aes(y = mean_EP_trust, linetype = "2"),
se = F, color = "gray20", alpha = 0.8) +
geom_hline(yintercept = -0.05, linetype = "solid", size = 0.5) +
xlab("") +
ylab("") +
ggtitle("Trust in the European Parliament") +
scale_shape_manual(values = c(21, 22), labels = c("Don't know", "Scale-point indicaiton")) +
scale_linetype_discrete(labels = c("ESS 11-point scale average", "EB 2-point scale average")) +
scale_fill_manual(values = c("#56B4E9", "#D55E00"), labels = c("Don't know", "Scale-point indicaiton")) +
scale_size_continuous(breaks = seq(0.05, 0.3, by = 0.05),
labels = scales::label_percent(accuracy = 5L),
limits = c(0, 0.34)) +
scale_alpha_continuous(breaks = seq(0.05, 0.3, by = 0.05),
labels = scales::label_percent(accuracy = 5L),
limits = c(0, 0.34)) +
scale_y_continuous(labels = c("ESS", "EB", seq(0.0, 1.0, by = 0.1)),
breaks = c(-0.3, -0.15, seq(0, 1.0, by = 0.1))) +
scale_x_continuous(labels = c("'04", "'06", "'08", "'10", "'12", "'14", "'16", "'18"),
breaks = seq(2004, 2018, by = 2)) +
coord_cartesian(ylim = c(-0.31, 1.01)) +
guides(shape = FALSE,
fill = FALSE,
size = guide_legend("Percentage of\nDon't Knows",
override.aes = list(shape = 21, fill = "#56B4E9"),
label.position = "right",
reverse = T),
linetype = guide_legend("Trend-lines", reverse = T, label.position = "bottom"),
alpha = guide_legend("Percentage of\nDon't Knows",
reverse = T)) +
facet_wrap(~ T_SURVEY_COUNTRY, nrow = 2) +
theme_bw() +
theme(text = element_text(colour = "black"),
legend.position = "none",
plot.margin = margin(-4, 3, 0, 0, unit = "pt"),
legend.title = element_text(size = 10, face = "bold"),
strip.background = element_rect(fill = "gray98"),
strip.text = element_text(size = 10, face = "bold", margin = margin(1,0,1,0, "pt")),
panel.spacing.x = unit(2, "pt"),
panel.spacing.y = unit(2, "pt"),
axis.text = element_text(color = "black", size = 9, face = "bold"),
plot.title = element_text(size = 10, face = "bold"))
#NP_trust
#data_transformations
prop_NP_value <- trust_survey_data %>%
filter(!is.na(NP_trust_harmonised)) %>%
filter(NP_trust_harmonised != 88) %>%
group_by(T_SURVEY_ID, T_SURVEY_NAME, T_SURVEY_COUNTRY, T_SURVEY_YEAR) %>%
count(NP_trust_harmonised, wt = weights) %>%
group_by(T_SURVEY_ID) %>%
mutate(prop_NP = n/sum(n))
prop_NP_DK <- trust_survey_data %>%
filter(!is.na(NP_trust_harmonised)) %>%
group_by(T_SURVEY_ID, T_SURVEY_NAME, T_SURVEY_COUNTRY, T_SURVEY_YEAR) %>%
count(NP_trust_harmonised, wt = weights) %>%
group_by(T_SURVEY_ID) %>%
mutate(prop_NP = n/sum(n)) %>%
filter(NP_trust_harmonised == 88)
prop_NP_plot1 <- full_join(prop_NP_value, prop_NP_DK)
survey_means_ESS_NP <- trust_survey_data %>% srvyr::as_survey(weights = weights) %>%
mutate(T_SURVEY_COUNTRY = countrycode::countrycode(sjlabelled::remove_all_labels(T_SURVEY_COUNTRY),
origin = "iso2c",
destination = "country.name")) %>%
filter(NP_trust_harmonised != 88) %>%
filter(T_SURVEY_NAME == "ESS") %>%
group_by(T_SURVEY_ID, T_SURVEY_COUNTRY, T_SURVEY_YEAR) %>%
summarise(mean_NP_trust = srvyr::survey_mean(NP_trust_harmonised, na.rm = T))
survey_means_EB_NP <- trust_survey_data %>% srvyr::as_survey(weights = weights) %>%
mutate(T_SURVEY_COUNTRY = countrycode::countrycode(sjlabelled::remove_all_labels(T_SURVEY_COUNTRY),
origin = "iso2c",
destination = "country.name")) %>%
filter(NP_trust_harmonised != 88) %>%
filter(T_SURVEY_NAME == "EB") %>%
group_by(T_SURVEY_ID, T_SURVEY_COUNTRY, T_SURVEY_YEAR) %>%
summarise(mean_NP_trust = srvyr::survey_mean(NP_trust_harmonised, na.rm = T))
EB_DK_data_NP <- prop_NP_plot1 %>%
filter(T_SURVEY_NAME == "EB") %>%
mutate(T_SURVEY_COUNTRY = countrycode::countrycode(sjlabelled::remove_all_labels(T_SURVEY_COUNTRY),
origin = "iso2c",
destination = "country.name")) %>%
filter(!is.na(NP_trust_harmonised)) %>%
filter(NP_trust_harmonised == 88)
#NP_trust plot
trust_np <- prop_NP_plot1 %>% ungroup() %>%
mutate(T_SURVEY_COUNTRY = countrycode::countrycode(sjlabelled::remove_all_labels(T_SURVEY_COUNTRY),
origin = "iso2c",
destination = "country.name")) %>%
filter(T_SURVEY_NAME == "ESS") %>%
filter(!is.na(NP_trust_harmonised)) %>%
mutate(NP_trust_recoded = ifelse(NP_trust_harmonised == 88, -0.3, NP_trust_harmonised)) %>%
ggplot(aes(x = as.numeric(T_SURVEY_YEAR))) +
geom_point(aes(y = NP_trust_recoded, size = prop_NP,
fill = ifelse(NP_trust_recoded == -0.3, "1", "2"),
alpha = prop_NP, shape = ifelse(NP_trust_recoded == -0.3, "1", "2")),
color = "gray20") +
geom_point(data = EB_DK_data_NP, aes(y = -0.15, size = prop_NP, alpha = prop_NP, fill = "1", shape = "1")) +
geom_smooth(data = survey_means_ESS_NP, aes(y = mean_NP_trust, linetype = "1"),
se = F, color = "gray20", alpha = 0.8) +
geom_smooth(data = survey_means_EB_NP, aes(y = mean_NP_trust, linetype = "2"),
se = F, color = "gray20", alpha = 0.8) +
geom_hline(yintercept = -0.05, linetype = "solid", size = 0.5) +
xlab("") +
ylab("") +
ggtitle("Trust in the National Parliament") +
scale_shape_manual(values = c(21, 22), labels = c("Don't know", "Scale-point indicaiton")) +
scale_linetype_discrete(labels = c("ESS 11-point scale average", "EB 2-point scale average")) +
scale_fill_manual(values = c("#56B4E9", "#D55E00"), labels = c("Don't know", "Scale-point indicaiton")) +
scale_size_continuous(breaks = seq(0.05, 0.3, by = 0.05),
labels = scales::label_percent(accuracy = 5L),
limits = c(0, 0.35)) +
scale_alpha_continuous(breaks = seq(0.05, 0.3, by = 0.05),
labels = scales::label_percent(accuracy = 5L),
limits = c(0, 0.35)) +
scale_y_continuous(labels = c("ESS", "EB", seq(0.0, 1.0, by = 0.1)),
breaks = c(-0.3, -0.15, seq(0, 1.0, by = 0.1))) +
scale_x_continuous(labels = c("'04", "'06", "'08", "'10", "'12", "'14", "'16", "'18"),
breaks = seq(2004, 2018, by = 2)) +
coord_cartesian(ylim = c(-0.31, 1.01)) +
guides(shape = FALSE,
fill = FALSE,
size = guide_legend("Percentage of ESS\nvalid responses",
override.aes = list(shape = 22, fill = "#D55E00"),
label.position = "right",
reverse = T),
linetype = FALSE,
alpha = guide_legend("Percentage of ESS\nvalid responses",
reverse = T)) +
facet_wrap(~ T_SURVEY_COUNTRY, nrow = 2) +
theme_bw() +
theme(text = element_text(colour = "black"),
plot.margin = margin(3, 3, 0, 0, unit = "pt"),
legend.position = "none",
strip.background = element_rect(fill = "gray98"),
legend.title = element_text(size = 10, face = "bold"),
strip.text = element_text(size = 10, face = "bold", margin = margin(1,0,1,0, "pt")),
axis.text = element_text(color = "black", size = 9, face = "bold"),
panel.spacing.x = unit(2, "pt"),
panel.spacing.y = unit(2, "pt"),
plot.title = element_text(size = 10, face = "bold"))
#Merging EP_plot with NP_plot
legend_1 <- cowplot::get_legend(trust_np +
theme(legend.position = "bottom",
legend.box.margin = margin(-5, 1, 1, 50, unit = "pt"),
legend.margin = margin(-5, 1, 1, 50, unit = "pt"),
legend.justification = "left"))
legend_2 <- cowplot::get_legend(trust_ep +
theme(legend.position = "bottom",
legend.box.margin = margin(-5, 2, 1, 60, unit = "pt"),
legend.margin = margin(-5, 2, 1, 60, unit = "pt"),
legend.justification = "left"))
plot_dot_1 <- ggpubr::ggarrange(trust_np, trust_ep, nrow = 2)
gridExtra::grid.arrange(plot_dot_1, legend_1, legend_2,
layout_matrix = rbind(c(1, 1),c(2, 3)),
heights = c(5.5, 0.25),
widths = c(10, 35, 1))
FIGURE 2
#PLOT_COR
#data transformaitons
survey_means_1 <- trust_survey_data %>% srvyr::as_survey(weights = weights) %>%
filter(EP_trust_harmonised != 88) %>%
filter(T_SURVEY_NAME == "EB") %>%
group_by(T_SURVEY_ID, T_SURVEY_COUNTRY, T_SURVEY_YEAR) %>%
summarise(EPtrust_EB = srvyr::survey_mean(EP_trust_harmonised, na.rm = T)) %>%
select(-EPtrust_EB_se) %>%
mutate(join_id = glue::glue(T_SURVEY_COUNTRY, T_SURVEY_YEAR))
survey_means_EP_ESS_1 <- trust_survey_data %>% srvyr::as_survey(weights = weights) %>%
filter(EP_trust_harmonised != 88) %>%
filter(T_SURVEY_NAME == "ESS") %>%
group_by(T_SURVEY_ID, T_SURVEY_COUNTRY, T_SURVEY_YEAR) %>%
summarise(EPtrust_ESS = srvyr::survey_mean(EP_trust_harmonised, na.rm = T)) %>%
mutate(join_id = glue::glue(T_SURVEY_COUNTRY, T_SURVEY_YEAR)) %>%
ungroup() %>%
select(join_id, EPtrust_ESS)
survey_means_ESS_NP_1 <- trust_survey_data %>% srvyr::as_survey(weights = weights) %>%
filter(NP_trust_harmonised != 88) %>%
filter(T_SURVEY_NAME == "ESS") %>%
group_by(T_SURVEY_ID, T_SURVEY_COUNTRY, T_SURVEY_YEAR) %>%
summarise(NPtrust_ESS = srvyr::survey_mean(NP_trust_harmonised, na.rm = T)) %>%
mutate(join_id = glue::glue(T_SURVEY_COUNTRY, T_SURVEY_YEAR)) %>%
ungroup() %>%
select(join_id, NPtrust_ESS)
survey_means_EB_NP_1 <- trust_survey_data %>% srvyr::as_survey(weights = weights) %>%
filter(NP_trust_harmonised != 88) %>%
filter(T_SURVEY_NAME == "EB") %>%
group_by(T_SURVEY_ID, T_SURVEY_COUNTRY, T_SURVEY_YEAR) %>%
summarise(NPtrust_EB = srvyr::survey_mean(NP_trust_harmonised, na.rm = T)) %>%
mutate(join_id = glue::glue(T_SURVEY_COUNTRY, T_SURVEY_YEAR)) %>%
ungroup() %>%
select(join_id, NPtrust_EB)
#plot
plot_x1 <- inner_join(survey_means_1, survey_means_EP_ESS_1, by = c("join_id")) %>%
inner_join(survey_means_ESS_NP_1, by = c("join_id")) %>%
inner_join(survey_means_EB_NP_1, by = c("join_id")) %>%
select(-join_id) %>%
pivot_longer(4:7, names_to = c("variable", "survey"), names_sep = "_", values_to = "trust") %>%
mutate(year = as.numeric(T_SURVEY_YEAR), country = sjlabelled::remove_all_labels(T_SURVEY_COUNTRY)) %>%
ungroup() %>%
select(country, year, survey, variable, trust) %>%
pivot_wider(names_from = survey, values_from = trust) %>%
ggplot(aes(x = EB, y = ESS, group = variable, fill = variable)) +
geom_point(aes(shape = variable, alpha = ifelse(variable == "EPtrust", "1", "2"))) +
geom_line(aes(color = variable, linetype = variable),
stat = "smooth",
alpha = 0.8,
linetype = "solid",
method = "lm",
se = FALSE) +
ggpubr::stat_cor(aes(color = variable, label = ..r.label..),
method = c("pearson"),
cor.coef.name = "r",
label.y = c(0.25, 0.6),
label.x = c(0.6, 0.15),
r.accuracy = 0.001,
output.type = "text") +
scale_alpha_manual(values = c(0.9, 0.5), breaks = c("1", "2")) +
scale_shape_manual(values = c(22, 21),
labels = c("Trust in the European Parliament", "Trust in the National parliament")) +
scale_fill_manual(values = c("#56B4E9", "#D55E00"),
labels = c("Trust in the European Parliament", "Trust in the National parliament")) +
scale_color_manual(values = c("#56B4E9", "#D55E00"),
labels = c("Trust in the European Parliament", "Trust in the National parliament")) +
scale_x_continuous(breaks = c(0, 0.25, 0.5, 0.75, 1),
labels = c(0, 0.25, 0.5, 0.75, 1)) +
scale_y_continuous(breaks = c(0, 0.25, 0.5, 0.75, 1),
labels = c(0, 0.25, 0.5, 0.75, 1)) +
coord_cartesian(xlim = c(0, 1), ylim = c(0, 1)) +
guides(shape = guide_legend("Type of trust measured",
label.position = "right",
reverse = T),
fill = guide_legend("Type of trust measured",
label.position = "right",
reverse = T),
label = FALSE,
color = FALSE,
linetype = FALSE,
alpha = FALSE) +
facet_wrap(.~year, nrow = 2) +
xlab("Eurobarometer: country-level mean-aggregates (rescaled: [0,1])") +
ylab("ESS: country-level mean-aggregates (rescaled: [0,1])") +
theme_bw() +
theme(text = element_text(colour = "black"),
legend.position = "bottom",
legend.direction = "horizontal",
legend.text = element_text(size = 9),
legend.title = element_text(size = 9, face = "bold"),
axis.title.y = element_text(size = 9, face = "bold"),
axis.title.x = element_text(size = 9, face = "bold"),
legend.margin = margin(-1, -1, -1, -1, unit = "pt"),
axis.text = element_text(color = "black", size = 9, face = "bold"),
panel.spacing.x = unit(2, "pt"),
panel.spacing.y = unit(2, "pt"),
strip.background = element_rect(fill = "gray98"),
strip.text = element_text(size = 9, face = "bold"))
#facet editing
egg::tag_facet(plot_x1,
x = Inf, y = Inf,
open = "", close = "",
hjust = 1.6,
vjust = 2.7,
tag_pool = seq(2004, 2018, by = 2))
#DATA TRANSFORMATIONS
prop_EP_DK_h1a <- trust_survey_data %>%
filter(!is.na(EP_trust_harmonised)) %>%
group_by(T_SURVEY_ID, T_SURVEY_NAME, T_SURVEY_COUNTRY, T_SURVEY_YEAR) %>%
count(EP_trust_harmonised, wt = weights) %>%
group_by(T_SURVEY_ID) %>%
mutate(prop_EP = n/sum(n)) %>%
filter(EP_trust_harmonised == 88) %>%
mutate(T_SURVEY_COUNTRY = countrycode::countrycode(sjlabelled::remove_all_labels(T_SURVEY_COUNTRY),
origin = "iso2c",
destination = "country.name")) %>%
select(T_SURVEY_ID, T_SURVEY_NAME, T_SURVEY_COUNTRY, T_SURVEY_YEAR, prop_EP)
prop_NP_DK_h1a <- trust_survey_data %>%
filter(!is.na(NP_trust_harmonised)) %>%
group_by(T_SURVEY_ID, T_SURVEY_NAME, T_SURVEY_COUNTRY, T_SURVEY_YEAR) %>%
count(NP_trust_harmonised, wt = weights) %>%
group_by(T_SURVEY_ID) %>%
mutate(prop_NP = n/sum(n)) %>%
filter(NP_trust_harmonised == 88) %>%
mutate(T_SURVEY_COUNTRY = countrycode::countrycode(sjlabelled::remove_all_labels(T_SURVEY_COUNTRY),
origin = "iso2c",
destination = "country.name")) %>%
select(T_SURVEY_ID, prop_NP)
pqr <- inner_join(prop_EP_DK_h1a, prop_NP_DK_h1a, by = c("T_SURVEY_ID"))
pqr1 <- pqr[ , 2:6] %>%
select(T_SURVEY_COUNTRY, T_SURVEY_YEAR, T_SURVEY_NAME, prop_EP, prop_NP) %>%
pivot_longer(cols = 4:5, names_to = "key", values_to = "value") %>%
unite(3:4, col = "name", sep = "*") %>%
pivot_wider(names_from = name, values_from = value)
#t_tests
ttest1 <- t.test(pqr1$`EB*prop_EP`, pqr1$`ESS*prop_EP`, paired = T, alternative = "greater", conf.level = 0.95) %>% broom::tidy()
ttest2 <- t.test(pqr1$`EB*prop_NP`, pqr1$`ESS*prop_NP`, paired = T, alternative = "greater", conf.level = 0.95) %>% broom::tidy()
ttest3 <- t.test(pqr1$`EB*prop_EP`, pqr1$`EB*prop_NP`, paired = T, alternative = "greater", conf.level = 0.95) %>% broom::tidy()
ttest4 <- t.test(pqr1$`ESS*prop_EP`, pqr1$`ESS*prop_NP`, paired = T, alternative = "greater", conf.level = 0.95) %>% broom::tidy()
#flexTable
column <- c("EB_EP_trust - ESS_EP_trust", "EB_NP_trust - ESS_NP_trust", "EB_EP_trust - EB_NP_trust", "ESS_EP_trust - ESS_NP_trust")
rbind(ttest1, ttest2, ttest3, ttest4) %>% cbind(column) %>% select(column, estimate, statistic, p.value, method, alternative) %>%
rename("Mean differences" = column) %>% as_tibble() %>%
qflextable() %>%
theme_booktabs() %>%
align_nottext_col(align = "center") %>%
colformat_double(j = 2:3, digits = 3)Mean differences | estimate | statistic | p.value | method | alternative |
EB_EP_trust - ESS_EP_trust | 0.056 | 10.123 | 9.3e-18 | Paired t-test | greater |
EB_NP_trust - ESS_NP_trust | 0.042 | 13.663 | 7.9e-26 | Paired t-test | greater |
EB_EP_trust - EB_NP_trust | 0.073 | 17.476 | 6.0e-34 | Paired t-test | greater |
ESS_EP_trust - ESS_NP_trust | 0.059 | 15.207 | 3.3e-29 | Paired t-test | greater |
H2aH2b
#H2ah2b
rty <- inner_join(survey_means_1, survey_means_EP_ESS_1, by = c("join_id")) %>%
inner_join(survey_means_ESS_NP_1, by = c("join_id")) %>%
inner_join(survey_means_EB_NP_1, by = c("join_id")) %>%
select(-join_id) %>%
mutate(EP_abs_diff = abs(EPtrust_EB - EPtrust_ESS),
NP_abs_diff = abs(NPtrust_EB - NPtrust_ESS))
#ttests
ttestx <- t.test(rty$EP_abs_diff, mu = 0, alternative = "greater", conf.level = 0.95) %>% broom::tidy()
ttesty <- t.test(rty$NP_abs_diff, mu = 0, alternative = "greater", conf.level = 0.95) %>% broom::tidy()
ttestz <- t.test(rty$EP_abs_diff, rty$NP_abs_diff, paired = T, alternative = "greater", conf.level = 0.95) %>% broom::tidy()
#flexTable
column1 <- c("EP_abs_difference - 0", "NP_abs_difference - 0", "EP_abs_difference - NP_abs_difference")
rbind(ttestx, ttesty, ttestz) %>% cbind(column1) %>% select(column1, estimate, statistic, p.value, method, alternative) %>%
rename("Mean differences" = column1) %>% as_tibble() %>%
qflextable() %>%
theme_booktabs() %>%
align_nottext_col(align = "center") %>%
colformat_double(j = 2:3, digits = 3)Mean differences | estimate | statistic | p.value | method | alternative |
EP_abs_difference - 0 | 0.167 | 20.387 | 1.3e-39 | One Sample t-test | greater |
NP_abs_difference - 0 | 0.082 | 14.598 | 6.9e-28 | One Sample t-test | greater |
EP_abs_difference - NP_abs_difference | 0.085 | 7.446 | 1.1e-11 | Paired t-test | greater |
H2c
inner_join(survey_means_1, survey_means_EP_ESS_1, by = c("join_id")) %>%
inner_join(survey_means_ESS_NP_1, by = c("join_id")) %>%
inner_join(survey_means_EB_NP_1, by = c("join_id")) %>%
select(-join_id) %>%
rename("Survey year" = T_SURVEY_YEAR) %>%
group_by(`Survey year`) %>%
summarise(cor_pearson_ep = cor(EPtrust_EB, EPtrust_ESS, method = "pearson"),
cor_spearman_ep = cor(EPtrust_EB, EPtrust_ESS, method = "spearman"),
cor_pearson_np = cor(NPtrust_EB, NPtrust_ESS, method = "pearson"),
cor_spearman_np = cor(NPtrust_EB, NPtrust_ESS, method = "spearman"), .groups = "drop") %>%
as_tibble() %>%
qflextable() %>%
theme_booktabs() %>%
align_nottext_col(align = "center") %>%
colformat_double(digits = 3)Survey year | cor_pearson_ep | cor_spearman_ep | cor_pearson_np | cor_spearman_np |
2004 | 0.728 | 0.609 | 0.928 | 0.780 |
2006 | 0.836 | 0.710 | 0.905 | 0.815 |
2008 | 0.642 | 0.200 | 0.920 | 0.846 |
2010 | 0.721 | 0.666 | 0.876 | 0.767 |
2012 | 0.619 | 0.644 | 0.920 | 0.899 |
2014 | 0.570 | 0.600 | 0.928 | 0.974 |
2016 | 0.760 | 0.811 | 0.941 | 0.938 |
2018 | 0.858 | 0.851 | 0.953 | 0.960 |