INGOs and restrictions

Code

library(tidyverse)
library(targets)
library(scales)
library(glue)
library(gt)
library(ggmosaic)
library(ggtext)
library(patchwork)
library(tidybayes)
library(here)
library(ggpattern)
library(countrycode)

tar_config_set(store = here('_targets'),
               script = here('_targets.R'))

tar_load(c(survey_orgs, survey_countries, survey_all, dcjw, gwf))
tar_load(c(
  df_restrictions_regime, models_restrictions_regime,
  df_restrictions_issue, models_restrictions_issue,
  df_restrictions_reg_regime, models_restrictions_reg_regime
))

# Plotting functions
invisible(list2env(tar_read(graphic_functions), .GlobalEnv))
invisible(list2env(tar_read(table_functions), .GlobalEnv))
set_annotation_fonts()

Routine regulation vs. burdensome regulation

Code

dcjw_registration <- dcjw %>%
  mutate(question = recode(question,
    q_2a = "registration",
    q_2b = "burdensome",
    q_2c = "appeal",
    q_2d = "diff.foreign"
  )) %>%
  filter(question %in% c("registration", "burdensome"))

dcjw_registration_panel <- dcjw_registration %>%
  tidyr::expand(Country, question,
    year = min(.$year, na.rm = TRUE):2013
  ) %>%
  left_join(
    dcjw_registration,
    by = c("Country", "question", "year")
  ) %>%
  # Bring most recent legislation forward
  group_by(Country, question) %>%
  mutate(value = zoo::na.locf(value, na.rm = FALSE)) %>%
  ungroup() %>%
  mutate(value = ifelse(is.na(value), 0, value)) %>%
  mutate(cowcode = countrycode(Country, "country.name", "cown",
    custom_match = c(`Serbia` = 340L)
  ))

regime_types <- gwf %>%
  select(cowcode, year, gwf.ever.autocracy) %>%
  group_by(cowcode) %>%
  summarise(ever.autocracy = any(gwf.ever.autocracy, na.rm = TRUE))

dcjw_registration_summary <- dcjw_registration_panel %>%
  left_join(regime_types, by = "cowcode") %>%
  # There are a few countries that didn't have regime type from full.data
  mutate(ever.autocracy = case_when(
    .$Country %in% c(
      "Bahrain", "Equatorial Guinea", "Kuwait", "Oman",
      "Singapore", "United Arab Emirates"
    ) ~ TRUE,
    .$Country == "Serbia" ~ FALSE,
    TRUE ~ .$ever.autocracy
  )) %>%
  group_by(year, question, ever.autocracy) %>%
  summarise(prop.with.reg = sum(value) / n()) %>%
  ungroup() %>%
  mutate(
    question =
      factor(question,
        levels = c("registration", "burdensome"),
        labels = c(
          "Registration required",
          "Registration burdensome"
        )
      )
  ) %>%
  mutate(
    ever.autocracy =
      factor(ever.autocracy,
        levels = c(TRUE, FALSE),
        labels = c("Autocracies", "Democracies")
      )
  ) %>%
  mutate(ever.autocracy = factor(ever.autocracy,
    levels = rev(levels(ever.autocracy)),
    ordered = TRUE
  )) %>%
  filter(year > 1990)

ggplot(
  dcjw_registration_summary,
  aes(x = year, y = prop.with.reg, colour = ever.autocracy)
) +
  geom_line() +
  labs(x = NULL, y = "Proportion of countries\nwith regulations") +
  scale_y_continuous(labels = label_percent()) +
  coord_cartesian(ylim = c(0, 0.65)) +
  scale_colour_manual(values = c(clrs$Prism[4], clrs$Prism[6]), 
    name = NULL, na.translate = FALSE) +
  theme_ingo() +
  facet_wrap(vars(question))

Overall restriction

Regime type

Code

ggplot(data = df_restrictions_regime) +
  geom_mosaic(aes(x = product(target.regime.type, Q4.17_collapsed), 
                  fill = target.regime.type),
              divider = mosaic("v"), 
              offset = 0, color = "white", linewidth = 1) +
  scale_fill_manual(values = c(clrs$Prism[4], clrs$Prism[6]),
                    guide = "none") +
  labs(x = "Target country regime type",
       y = "Level of restriction in target country") +
  theme_ingo() +
  theme(panel.grid.major.x = element_blank(),
        panel.grid.major.y = element_blank(),
        plot.title = element_markdown(lineheight = 1.2))

Code

# Template: 60% of INGOs that reported feeling very restricted work in autocracies
p1 <- models_restrictions_regime$draws %>% 
  ggplot(aes(x = .epred, y = Q4.17_collapsed, fill = Q4.17_collapsed)) +
  stat_halfeye() +
  scale_x_continuous(labels = label_percent()) +
  scale_fill_manual(values = clrs$Peach[c(1, 3, 5, 7)],
                    guide = "none") +
  labs(x = "Posterior proportion", y = "Level of restriction in target country") +
  facet_wrap(vars("Proportion of INGOs that reported level\nof restriction that work in autocracies")) +
  theme_ingo()

plot_data <- models_restrictions_regime$diffs %>%
  mutate(diffs = factor(Q4.17_collapsed,
    levels = c(
      "Very restricted - Moderately restricted",
      "Moderately restricted - Slightly restricted",
      "Slightly restricted - Not restricted"
    )
  ))

diffs_summary <- plot_data %>% 
  group_by(diffs) %>% 
  median_qi(.epred, .width = c(0.5, 0.8, 0.95))

p2 <- models_restrictions_regime$diffs %>% 
  mutate(diffs = factor(Q4.17_collapsed,
    levels = c(
      "Very restricted - Moderately restricted",
      "Moderately restricted - Slightly restricted",
      "Slightly restricted - Not restricted"
    )
  )) %>% 
  ggplot(aes(x = .epred, fill = diffs, pattern_fill = diffs)) +
  geom_density_pattern(
    pattern = "stripe",
    pattern_density = 0.5,
    pattern_spacing = 0.2,
    pattern_size = 0,
    trim = TRUE,
    linewidth = 0
  ) +
  geom_pointinterval(data = diffs_summary, aes(x = .epred, xmin = .lower, xmax = .upper)) +
  geom_vline(xintercept = 0, linewidth = 0.5, color = clrs$Prism[8]) +
  facet_wrap(vars(diffs), ncol = 1) +
  scale_fill_manual(values = c(clrs$Peach[7], clrs$Peach[5], clrs$Peach[3]),
                    guide = "none") +
  scale_pattern_fill_manual(values = c(clrs$Peach[5], clrs$Peach[3], clrs$Peach[1]),
                            guide = "none") +
  scale_x_continuous(labels = label_number(scale = 100, style_negative = "minus",
                                           suffix = " pp."),
                     breaks = seq(-0.6, 0.2, by = 0.2)) +
  labs(x = "Percentage point differences", y = NULL) +
  theme_ingo() +
  theme(panel.grid.major.y = element_blank(),
        axis.text.y = element_blank())

p1 | p2

Code

models_restrictions_regime$draws %>% 
  median_qi(.epred) %>% 
  mutate(Q4.17_collapsed = as.character(Q4.17_collapsed),
         type = "Median proportions for INGOs working in autocracies") %>%
  rename(y = .epred, ymin = .lower, ymax = .upper) %>% 
  bind_rows(models_restrictions_regime$diffs_summary) %>% 
  mutate(across(c(y, ymin, ymax, p_greater_0), 
                list(nice = ~label_number(accuracy = 0.01)(.)))) %>% 
  mutate(nice_value = glue("{y_nice}<br>({ymin_nice} – {ymax_nice})")) %>% 
  mutate(Q4.17_collapsed = str_replace(Q4.17_collapsed, " - ", " − ")) %>% 
  select(type, Q4.17_collapsed, nice_value, p_greater_0_nice) %>% 
  group_by(type) %>% 
  gt() %>% 
  cols_label(
    Q4.17_collapsed = "",
    nice_value = "Posterior median",
    p_greater_0_nice = "p > 0"
  ) %>% 
  fmt_markdown(nice_value) %>% 
  sub_missing(p_greater_0_nice) %>% 
  tab_footnote(footnote = "95% credible intervals shown in parentheses") %>% 
  opts_theme()

	Posterior median	p > 0
Median proportions for INGOs working in autocracies
Very restricted	0.60 (0.47 – 0.72)	—
Moderately restricted	0.49 (0.39 – 0.59)	—
Not restricted	0.26 (0.20 – 0.31)	—
Slightly restricted	0.42 (0.34 – 0.50)	—
Differences
Moderately restricted − Slightly restricted	0.07 (-0.06 – 0.20)	0.86
Slightly restricted − Not restricted	0.16 (0.07 – 0.26)	1.00
Very restricted − Moderately restricted	0.11 (-0.05 – 0.27)	0.91
95% credible intervals shown in parentheses

Issue contentiousness

Code

ggplot(data = df_restrictions_issue) +
  geom_mosaic(aes(x = product(potential.contentiousness, Q4.17_collapsed), 
                  fill = potential.contentiousness),
              divider = mosaic("v"), 
              offset = 0, color = "white", linewidth = 1) +
  scale_fill_manual(values = c(clrs$Prism[2], clrs$Prism[8]),
                    guide = "none") +
  labs(x = "Contentiousness of INGO work",
       y = "Level of restriction in target country") +
  theme_ingo() +
  theme(panel.grid.major.x = element_blank(),
        panel.grid.major.y = element_blank(),
        plot.title = element_markdown(lineheight = 1.2))

Code

p1 <- models_restrictions_issue$draws %>% 
  ggplot(aes(x = .epred, y = Q4.17_collapsed, fill = Q4.17_collapsed)) +
  stat_halfeye() +
  scale_x_continuous(labels = label_percent()) +
  scale_fill_manual(values = clrs$Peach[c(1, 3, 5, 7)],
                    guide = "none") +
  labs(x = "Posterior proportion", y = "Level of restriction in target country") +
  facet_wrap(vars("INGOs with contentious issues")) +
  theme_ingo()

plot_data <- models_restrictions_issue$diffs %>% 
  mutate(diffs = factor(Q4.17_collapsed,
    levels = c(
      "Very restricted - Moderately restricted",
      "Moderately restricted - Slightly restricted",
      "Slightly restricted - Not restricted"
    )
  ))

diffs_summary <- plot_data %>% 
  group_by(diffs) %>% 
  median_qi(.epred, .width = c(0.5, 0.8, 0.95))

p2 <- plot_data %>% 
  ggplot(aes(x = .epred, fill = diffs, pattern_fill = diffs)) +
  geom_density_pattern(
    # aes(y = after_stat(density * 1e5)),
    pattern = "stripe",
    pattern_density = 0.5,
    pattern_spacing = 0.2,
    pattern_size = 0,
    trim = TRUE,
    linewidth = 0
  ) +
  geom_pointinterval(data = diffs_summary, aes(x = .epred, xmin = .lower, xmax = .upper)) +
  geom_vline(xintercept = 0, linewidth = 0.5, color = clrs$Prism[8]) +
  facet_wrap(vars(diffs), ncol = 1) +
  scale_fill_manual(values = c(clrs$Peach[7], clrs$Peach[5], clrs$Peach[3]),
                    guide = "none") +
  scale_pattern_fill_manual(values = c(clrs$Peach[5], clrs$Peach[3], clrs$Peach[1]),
                            guide = "none") +
  scale_x_continuous(labels = label_number(scale = 100, style_negative = "minus",
                                           suffix = "\n pp."),
                     breaks = seq(-0.4, 0.4, by = 0.2)) +
  labs(x = "Percentage point differences", y = NULL) +
  theme_ingo() +
  theme(panel.grid.major.y = element_blank(),
        axis.text.y = element_blank())

p1 | p2

Code

models_restrictions_issue$draws %>% 
  median_qi(.epred) %>% 
  mutate(Q4.17_collapsed = as.character(Q4.17_collapsed),
         type = "Median proportions for NGOs working on contentious issues") %>%
  rename(y = .epred, ymin = .lower, ymax = .upper) %>% 
  bind_rows(models_restrictions_issue$diffs_summary) %>% 
  mutate(across(c(y, ymin, ymax, p_greater_0), 
                list(nice = ~label_number(accuracy = 0.01)(.)))) %>% 
  mutate(nice_value = glue("{y_nice}<br>({ymin_nice} – {ymax_nice})")) %>% 
  mutate(Q4.17_collapsed = str_replace(Q4.17_collapsed, " - ", " − ")) %>% 
  select(type, Q4.17_collapsed, nice_value, p_greater_0_nice) %>% 
  group_by(type) %>% 
  gt() %>% 
  cols_label(
    Q4.17_collapsed = "",
    nice_value = "Posterior median",
    p_greater_0_nice = "p > 0"
  ) %>% 
  fmt_markdown(nice_value) %>% 
  sub_missing(p_greater_0_nice) %>% 
  tab_footnote(footnote = "95% credible intervals shown in parentheses") %>% 
  opts_theme()

	Posterior median	p > 0
Median proportions for NGOs working on contentious issues
Very restricted	0.47 (0.34 – 0.60)	—
Moderately restricted	0.29 (0.20 – 0.38)	—
Slightly restricted	0.33 (0.26 – 0.41)	—
Not restricted	0.28 (0.23 – 0.34)	—
Differences
Moderately restricted − Slightly restricted	-0.05 (-0.17 – 0.08)	0.23
Slightly restricted − Not restricted	0.05 (-0.05 – 0.15)	0.85
Very restricted − Moderately restricted	0.19 (0.02 – 0.34)	0.99
95% credible intervals shown in parentheses

Registration and restrictions

Code

ggplot(data = df_restrictions_reg_regime) +
  geom_mosaic(
    aes(
      x = product(regime_registered, Q4.17_collapsed),
      fill = regime_registered
    ),
    divider = mosaic("v"),
    offset = 0, color = "white", linewidth = 1, na.rm = TRUE
  ) +
  scale_x_productlist(
    expand = c(0, 0.02),
    labels = c("Registered", "", "Not registered", "")) +
  scale_y_productlist(expand = c(0, 0.02)) +
  scale_fill_manual(
    values = c(
      clrs$Prism[1], clrs$Prism[11],
      clrs$Prism[7], clrs$Prism[6]
    ),
    guide = "none"
  ) +
  facet_wrap(vars(target.regime.type), ncol = 2) +
  theme_ingo() +
  theme(
    panel.grid.major.x = element_blank(),
    panel.grid.major.y = element_blank(),
    axis.title.x = element_blank(),
    axis.title.y = element_blank()
  )

Code

models_restrictions_reg_regime %>% 
  unnest(diffs_summary) %>% 
  mutate(nice_value = glue::glue(
    "{diff}<br>({p})",
    diff = label_number(style_negative = "minus", accuracy = 0.01)(y),
    p = label_number(accuracy = 0.01)(p_greater_0))
  ) %>% 
  ungroup() %>% 
  select(Q4.17_collapsed, target.regime.type, nice_value) %>% 
  pivot_wider(names_from = "target.regime.type", values_from = "nice_value") %>% 
  gt() %>%
  fmt_markdown(columns = 2:3) %>%
  cols_align(align = "left", columns = Q4.17_collapsed) %>%
  cols_label(Q4.17_collapsed = "") %>% 
  tab_footnote(footnote = "Values show the differences in proportion of registered and non-registered INGOs providing each response (not registered proportion − registered proportion)") %>% 
  opts_theme()

Median differences in proportion of registered and non-registered INGOs selecting frequency of type of restriction. Probability that the proportion is greater than 0 given in parentheses.
	Democracy	Autocracy
Not restricted	0.19 (1.00)	0.09 (0.86)
Slightly restricted	−0.15 (0.02)	−0.08 (0.18)
Moderately restricted	−0.35 (0.00)	−0.39 (0.00)
Very restricted	0.04 (0.61)	0.21 (0.97)
Values show the differences in proportion of registered and non-registered INGOs providing each response (not registered proportion − registered proportion)

Gatekeeping and program capture restrictions

Reactions to restrictions

Size

Other instrumental concerns

--- title: "INGOs and restrictions" format: html: code-fold: true --- ```{r setup, include=FALSE} knitr::opts_chunk$set(fig.align = "center", fig.retina = 3, fig.width = 6, fig.height = (6 * 0.618), out.width = "80%", collapse = TRUE, dev = "ragg_png") options(digits = 3, width = 120, dplyr.summarise.inform = FALSE, knitr.kable.NA = "") ``` ```{r load-libraries-data, warning=FALSE, message=FALSE} library(tidyverse) library(targets) library(scales) library(glue) library(gt) library(ggmosaic) library(ggtext) library(patchwork) library(tidybayes) library(here) library(ggpattern) library(countrycode) tar_config_set(store = here('_targets'), script = here('_targets.R')) tar_load(c(survey_orgs, survey_countries, survey_all, dcjw, gwf)) tar_load(c( df_restrictions_regime, models_restrictions_regime, df_restrictions_issue, models_restrictions_issue, df_restrictions_reg_regime, models_restrictions_reg_regime )) # Plotting functions invisible(list2env(tar_read(graphic_functions), .GlobalEnv)) invisible(list2env(tar_read(table_functions), .GlobalEnv)) set_annotation_fonts() ``` # Routine regulation vs. burdensome regulation ```{r plot-dcjw-registration, fig.width=6, fig.height=2, warning=FALSE} dcjw_registration <- dcjw %>% mutate(question = recode(question, q_2a = "registration", q_2b = "burdensome", q_2c = "appeal", q_2d = "diff.foreign" )) %>% filter(question %in% c("registration", "burdensome")) dcjw_registration_panel <- dcjw_registration %>% tidyr::expand(Country, question, year = min(.$year, na.rm = TRUE):2013 ) %>% left_join( dcjw_registration, by = c("Country", "question", "year") ) %>% # Bring most recent legislation forward group_by(Country, question) %>% mutate(value = zoo::na.locf(value, na.rm = FALSE)) %>% ungroup() %>% mutate(value = ifelse(is.na(value), 0, value)) %>% mutate(cowcode = countrycode(Country, "country.name", "cown", custom_match = c(`Serbia` = 340L) )) regime_types <- gwf %>% select(cowcode, year, gwf.ever.autocracy) %>% group_by(cowcode) %>% summarise(ever.autocracy = any(gwf.ever.autocracy, na.rm = TRUE)) dcjw_registration_summary <- dcjw_registration_panel %>% left_join(regime_types, by = "cowcode") %>% # There are a few countries that didn't have regime type from full.data mutate(ever.autocracy = case_when( .$Country %in% c( "Bahrain", "Equatorial Guinea", "Kuwait", "Oman", "Singapore", "United Arab Emirates" ) ~ TRUE, .$Country == "Serbia" ~ FALSE, TRUE ~ .$ever.autocracy )) %>% group_by(year, question, ever.autocracy) %>% summarise(prop.with.reg = sum(value) / n()) %>% ungroup() %>% mutate( question = factor(question, levels = c("registration", "burdensome"), labels = c( "Registration required", "Registration burdensome" ) ) ) %>% mutate( ever.autocracy = factor(ever.autocracy, levels = c(TRUE, FALSE), labels = c("Autocracies", "Democracies") ) ) %>% mutate(ever.autocracy = factor(ever.autocracy, levels = rev(levels(ever.autocracy)), ordered = TRUE )) %>% filter(year > 1990) ggplot( dcjw_registration_summary, aes(x = year, y = prop.with.reg, colour = ever.autocracy) ) + geom_line() + labs(x = NULL, y = "Proportion of countries\nwith regulations") + scale_y_continuous(labels = label_percent()) + coord_cartesian(ylim = c(0, 0.65)) + scale_colour_manual(values = c(clrs$Prism[4], clrs$Prism[6]), name = NULL, na.translate = FALSE) + theme_ingo() + facet_wrap(vars(question)) ``` # Overall restriction ## Regime type ```{r plot-restrictions-regime, warning=FALSE} ggplot(data = df_restrictions_regime) + geom_mosaic(aes(x = product(target.regime.type, Q4.17_collapsed), fill = target.regime.type), divider = mosaic("v"), offset = 0, color = "white", linewidth = 1) + scale_fill_manual(values = c(clrs$Prism[4], clrs$Prism[6]), guide = "none") + labs(x = "Target country regime type", y = "Level of restriction in target country") + theme_ingo() + theme(panel.grid.major.x = element_blank(), panel.grid.major.y = element_blank(), plot.title = element_markdown(lineheight = 1.2)) ``` ```{r plot-restrictions-regime-bayes, fig.width=7.5, fig.height=4} # Template: 60% of INGOs that reported feeling very restricted work in autocracies p1 <- models_restrictions_regime$draws %>% ggplot(aes(x = .epred, y = Q4.17_collapsed, fill = Q4.17_collapsed)) + stat_halfeye() + scale_x_continuous(labels = label_percent()) + scale_fill_manual(values = clrs$Peach[c(1, 3, 5, 7)], guide = "none") + labs(x = "Posterior proportion", y = "Level of restriction in target country") + facet_wrap(vars("Proportion of INGOs that reported level\nof restriction that work in autocracies")) + theme_ingo() plot_data <- models_restrictions_regime$diffs %>% mutate(diffs = factor(Q4.17_collapsed, levels = c( "Very restricted - Moderately restricted", "Moderately restricted - Slightly restricted", "Slightly restricted - Not restricted" ) )) diffs_summary <- plot_data %>% group_by(diffs) %>% median_qi(.epred, .width = c(0.5, 0.8, 0.95)) p2 <- models_restrictions_regime$diffs %>% mutate(diffs = factor(Q4.17_collapsed, levels = c( "Very restricted - Moderately restricted", "Moderately restricted - Slightly restricted", "Slightly restricted - Not restricted" ) )) %>% ggplot(aes(x = .epred, fill = diffs, pattern_fill = diffs)) + geom_density_pattern( pattern = "stripe", pattern_density = 0.5, pattern_spacing = 0.2, pattern_size = 0, trim = TRUE, linewidth = 0 ) + geom_pointinterval(data = diffs_summary, aes(x = .epred, xmin = .lower, xmax = .upper)) + geom_vline(xintercept = 0, linewidth = 0.5, color = clrs$Prism[8]) + facet_wrap(vars(diffs), ncol = 1) + scale_fill_manual(values = c(clrs$Peach[7], clrs$Peach[5], clrs$Peach[3]), guide = "none") + scale_pattern_fill_manual(values = c(clrs$Peach[5], clrs$Peach[3], clrs$Peach[1]), guide = "none") + scale_x_continuous(labels = label_number(scale = 100, style_negative = "minus", suffix = " pp."), breaks = seq(-0.6, 0.2, by = 0.2)) + labs(x = "Percentage point differences", y = NULL) + theme_ingo() + theme(panel.grid.major.y = element_blank(), axis.text.y = element_blank()) p1 | p2 ``` ```{r tab-restrictions-regime-bayes} models_restrictions_regime$draws %>% median_qi(.epred) %>% mutate(Q4.17_collapsed = as.character(Q4.17_collapsed), type = "Median proportions for INGOs working in autocracies") %>% rename(y = .epred, ymin = .lower, ymax = .upper) %>% bind_rows(models_restrictions_regime$diffs_summary) %>% mutate(across(c(y, ymin, ymax, p_greater_0), list(nice = ~label_number(accuracy = 0.01)(.)))) %>% mutate(nice_value = glue("{y_nice}<br>({ymin_nice} – {ymax_nice})")) %>% mutate(Q4.17_collapsed = str_replace(Q4.17_collapsed, " - ", " − ")) %>% select(type, Q4.17_collapsed, nice_value, p_greater_0_nice) %>% group_by(type) %>% gt() %>% cols_label( Q4.17_collapsed = "", nice_value = "Posterior median", p_greater_0_nice = "p > 0" ) %>% fmt_markdown(nice_value) %>% sub_missing(p_greater_0_nice) %>% tab_footnote(footnote = "95% credible intervals shown in parentheses") %>% opts_theme() ``` ## Issue contentiousness ```{r plot-restrictions-issue, warning=FALSE} ggplot(data = df_restrictions_issue) + geom_mosaic(aes(x = product(potential.contentiousness, Q4.17_collapsed), fill = potential.contentiousness), divider = mosaic("v"), offset = 0, color = "white", linewidth = 1) + scale_fill_manual(values = c(clrs$Prism[2], clrs$Prism[8]), guide = "none") + labs(x = "Contentiousness of INGO work", y = "Level of restriction in target country") + theme_ingo() + theme(panel.grid.major.x = element_blank(), panel.grid.major.y = element_blank(), plot.title = element_markdown(lineheight = 1.2)) ``` ```{r plot-restrictions-issue-bayes, fig.width=7.5, fig.height=4} p1 <- models_restrictions_issue$draws %>% ggplot(aes(x = .epred, y = Q4.17_collapsed, fill = Q4.17_collapsed)) + stat_halfeye() + scale_x_continuous(labels = label_percent()) + scale_fill_manual(values = clrs$Peach[c(1, 3, 5, 7)], guide = "none") + labs(x = "Posterior proportion", y = "Level of restriction in target country") + facet_wrap(vars("INGOs with contentious issues")) + theme_ingo() plot_data <- models_restrictions_issue$diffs %>% mutate(diffs = factor(Q4.17_collapsed, levels = c( "Very restricted - Moderately restricted", "Moderately restricted - Slightly restricted", "Slightly restricted - Not restricted" ) )) diffs_summary <- plot_data %>% group_by(diffs) %>% median_qi(.epred, .width = c(0.5, 0.8, 0.95)) p2 <- plot_data %>% ggplot(aes(x = .epred, fill = diffs, pattern_fill = diffs)) + geom_density_pattern( # aes(y = after_stat(density * 1e5)), pattern = "stripe", pattern_density = 0.5, pattern_spacing = 0.2, pattern_size = 0, trim = TRUE, linewidth = 0 ) + geom_pointinterval(data = diffs_summary, aes(x = .epred, xmin = .lower, xmax = .upper)) + geom_vline(xintercept = 0, linewidth = 0.5, color = clrs$Prism[8]) + facet_wrap(vars(diffs), ncol = 1) + scale_fill_manual(values = c(clrs$Peach[7], clrs$Peach[5], clrs$Peach[3]), guide = "none") + scale_pattern_fill_manual(values = c(clrs$Peach[5], clrs$Peach[3], clrs$Peach[1]), guide = "none") + scale_x_continuous(labels = label_number(scale = 100, style_negative = "minus", suffix = "\n pp."), breaks = seq(-0.4, 0.4, by = 0.2)) + labs(x = "Percentage point differences", y = NULL) + theme_ingo() + theme(panel.grid.major.y = element_blank(), axis.text.y = element_blank()) p1 | p2 ``` ```{r tab-restrictions-issue-bayes} models_restrictions_issue$draws %>% median_qi(.epred) %>% mutate(Q4.17_collapsed = as.character(Q4.17_collapsed), type = "Median proportions for NGOs working on contentious issues") %>% rename(y = .epred, ymin = .lower, ymax = .upper) %>% bind_rows(models_restrictions_issue$diffs_summary) %>% mutate(across(c(y, ymin, ymax, p_greater_0), list(nice = ~label_number(accuracy = 0.01)(.)))) %>% mutate(nice_value = glue("{y_nice}<br>({ymin_nice} – {ymax_nice})")) %>% mutate(Q4.17_collapsed = str_replace(Q4.17_collapsed, " - ", " − ")) %>% select(type, Q4.17_collapsed, nice_value, p_greater_0_nice) %>% group_by(type) %>% gt() %>% cols_label( Q4.17_collapsed = "", nice_value = "Posterior median", p_greater_0_nice = "p > 0" ) %>% fmt_markdown(nice_value) %>% sub_missing(p_greater_0_nice) %>% tab_footnote(footnote = "95% credible intervals shown in parentheses") %>% opts_theme() ``` # Registration and restrictions ```{r plot-registration-restrictions, fig.width=6.5, fig.height=2} ggplot(data = df_restrictions_reg_regime) + geom_mosaic( aes( x = product(regime_registered, Q4.17_collapsed), fill = regime_registered ), divider = mosaic("v"), offset = 0, color = "white", linewidth = 1, na.rm = TRUE ) + scale_x_productlist( expand = c(0, 0.02), labels = c("Registered", "", "Not registered", "")) + scale_y_productlist(expand = c(0, 0.02)) + scale_fill_manual( values = c( clrs$Prism[1], clrs$Prism[11], clrs$Prism[7], clrs$Prism[6] ), guide = "none" ) + facet_wrap(vars(target.regime.type), ncol = 2) + theme_ingo() + theme( panel.grid.major.x = element_blank(), panel.grid.major.y = element_blank(), axis.title.x = element_blank(), axis.title.y = element_blank() ) ``` ```{r tab-registration-restrictions-diffs} #| tbl-cap: Median differences in proportion of registered and non-registered INGOs selecting frequency of type of restriction. Probability that the proportion is greater than 0 given in parentheses. models_restrictions_reg_regime %>% unnest(diffs_summary) %>% mutate(nice_value = glue::glue( "{diff}<br>({p})", diff = label_number(style_negative = "minus", accuracy = 0.01)(y), p = label_number(accuracy = 0.01)(p_greater_0)) ) %>% ungroup() %>% select(Q4.17_collapsed, target.regime.type, nice_value) %>% pivot_wider(names_from = "target.regime.type", values_from = "nice_value") %>% gt() %>% fmt_markdown(columns = 2:3) %>% cols_align(align = "left", columns = Q4.17_collapsed) %>% cols_label(Q4.17_collapsed = "") %>% tab_footnote(footnote = "Values show the differences in proportion of registered and non-registered INGOs providing each response (not registered proportion − registered proportion)") %>% opts_theme() ``` # Gatekeeping and program capture restrictions # Reactions to restrictions ## Size ## Other instrumental concerns