-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy path05_table_5.R
267 lines (191 loc) · 10.3 KB
/
05_table_5.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
#
# Delib. in Kirkuk
# Attitudes, Opinions etc. split by ethnicity
#
# 1. list-wise delete missing only on the current wave
# 2. list-wise delete missing on any of the 2 waves (Table 5)
#
# set directory
setwd(githubdir)
setwd("kirkuk/")
# Load libs
library(tidyr)
library(dplyr)
library(broom)
# Read in the data
source("scripts/01_recode.R")
# 1. subset by ethnicity; list-wise delete missing only on the current wave
# --------------------------------------------------------------------------
t4_vars <- c("local_trust", "central_trust", "fed", "ind_kirkuk", "ca_3", "cb_5")
t4_ethnic <- all_dat[, c(t4_vars, "cond", "ethnicity", "wave")] %>%
group_by(ethnicity, cond, wave) %>%
summarise_all(funs(mean(., na.rm = TRUE)))
# Get Condition/Wave concat
t4_ethnic$cond <- paste0(t4_ethnic$cond, t4_ethnic$wave)
t4_ethnic_t <- t4_ethnic %>%
gather(key = var_name, value = value, 3:8) %>%
spread_(key = "cond", value = "value") %>%
filter(!is.na(ethnicity) & ethnicity != "Other") %>%
filter(var_name != "wave") %>%
arrange(var_name)
t4_ethnic_t$diff_delib <- t4_ethnic_t$delib2 - t4_ethnic_t$delib1
t4_ethnic_t$diff_delib_info <- t4_ethnic_t$delib_info2 - t4_ethnic_t$delib_info1
wave_wise_res <- t4_ethnic_t[, c("ethnicity", "var_name", "control1", "delib1", "delib2", "delib_info1", "delib_info2", "diff_delib", "diff_delib_info")]
# Pooled t1
# -----------------------------
ethnic_t1_res <- all_dat[, c(t4_vars, "ethnicity", "wave")] %>%
group_by(ethnicity, wave) %>%
summarise_all(funs(mean(., na.rm = TRUE))) %>%
filter(wave == 1) %>%
filter(!is.na(ethnicity) & ethnicity != "Other") %>%
select(-wave)
ethnic_t1_res <- melt(ethnic_t1_res)
# Table 5: list-wise delete missing on any of the 2 waves
# ----------------------------------------------------------
# Convenient subsets
ewall_dat <- subset(wall_dat, ethnicity_t1 %in% c("Arab", "Kurd", "Turkman"))
arab_d_set <- ewall_dat[ewall_dat$cond_t1 == "delib" & ewall_dat$ethnicity_t1 == "Arab", ]
arab_di_set <- ewall_dat[ewall_dat$cond_t1 == "delib_info" & ewall_dat$ethnicity_t1 == "Arab", ]
kurd_d_set <- ewall_dat[ewall_dat$cond_t1 == "delib" & ewall_dat$ethnicity_t1 == "Kurd", ]
kurd_di_set <- ewall_dat[ewall_dat$cond_t1 == "delib_info" & ewall_dat$ethnicity_t1 == "Kurd", ]
turk_d_set <- ewall_dat[ewall_dat$cond_t1 == "delib" & ewall_dat$ethnicity_t1 == "Turkman", ]
turk_di_set <- ewall_dat[ewall_dat$cond_t1 == "delib_info" & ewall_dat$ethnicity_t1 == "Turkman", ]
# List of tab 4 vars
tab_4_t2 <- paste0(c("ethnocentrism", "ca_3", "fed", "ind_kirkuk", "local_trust", "central_trust", "cb_5"), "_t2")
tab_4_t1 <- paste0(c("ethnocentrism", "ca_3", "fed", "ind_kirkuk", "local_trust", "central_trust", "cb_5"), "_t1")
# 5a: Arab
# ----------------
# No NA on either wave/List-wise deletion estimates
arab_res_nona <- data.frame(var_name = NA, t1_delib_nona = NA, t2_delib_nona = NA, t1_di_nona = NA, t2_di_nona = NA)
for (i in 1:length(tab_4_t2)) {
nona_delib <- !is.na(arab_d_set[, tab_4_t1[i]]) & !is.na(arab_d_set[, tab_4_t2[i]])
nona_delib_info <- !is.na(arab_di_set[, tab_4_t1[i]]) & !is.na(arab_di_set[, tab_4_t2[i]])
t1_delib <- mean(arab_d_set[, tab_4_t1[i]][nona_delib], na.rm = T)
t2_delib <- mean(arab_d_set[, tab_4_t2[i]][nona_delib], na.rm = T)
t1_delib_i <- mean(arab_di_set[, tab_4_t1[i]][nona_delib_info], na.rm = T)
t2_delib_i <- mean(arab_di_set[, tab_4_t2[i]][nona_delib_info], na.rm = T)
arab_res_nona[i, ] <- c(tab_4_t2[i], t1_delib, t2_delib, t1_delib_i, t2_delib_i)
}
# Arab P-values
diff_delib <- arab_d_set[, tab_4_t2] - arab_d_set[, tab_4_t1]
diff_delib <- subset(diff_delib, select = tab_4_t2)
res_delib <- do.call(rbind, lapply(diff_delib, function(x) tidy(t.test(x, mu = 0))))
names(res_delib) <- paste0(names(res_delib), "_d")
res_delib$var_name <- gsub("_t2", "", rownames(res_delib))
diff_delib_info <- arab_di_set[, tab_4_t2] - arab_di_set[, tab_4_t1]
diff_delib_info <- subset(diff_delib_info, select = tab_4_t2)
res_delib_info <- do.call(rbind, lapply(diff_delib_info, function(x) tidy(t.test(x, mu = 0))))
names(res_delib_info) <- paste0(names(res_delib_info), "_di")
res_delib_info$var_name_di <- gsub("_t2", "", rownames(res_delib_info))
arab_res <- res_delib %>%
left_join(res_delib_info, by = c("var_name" = "var_name_di"))
arab_res <- arab_res[, c("var_name", "estimate_d", "p.value_d", "estimate_di", "p.value_di")]
arab_res_all <- arab_res %>%
mutate(var_name = paste0(var_name, "_t2")) %>%
left_join(arab_res_nona)
# 5b. Kurds
# -----------------
# No NA on either wave/List-wise deletion estimates
kurd_res_nona <- data.frame(var_name = NA, t1_delib_nona = NA, t2_delib_nona = NA, t1_di_nona = NA, t2_di_nona = NA)
for (i in 1:length(tab_4_t2)) {
nona_delib <- !is.na(kurd_d_set[, tab_4_t1[i]]) & !is.na(kurd_d_set[, tab_4_t2[i]])
nona_delib_info <- !is.na(kurd_di_set[, tab_4_t1[i]]) & !is.na(kurd_di_set[, tab_4_t2[i]])
t1_delib <- mean(kurd_d_set[, tab_4_t1[i]][nona_delib], na.rm = T)
t2_delib <- mean(kurd_d_set[, tab_4_t2[i]][nona_delib], na.rm = T)
t1_delib_i <- mean(kurd_di_set[, tab_4_t1[i]][nona_delib_info], na.rm = T)
t2_delib_i <- mean(kurd_di_set[, tab_4_t2[i]][nona_delib_info], na.rm = T)
kurd_res_nona[i, ] <- c(tab_4_t2[i], t1_delib, t2_delib, t1_delib_i, t2_delib_i)
}
# kurd P-values
diff_delib <- kurd_d_set[, tab_4_t2] - kurd_d_set[, tab_4_t1]
diff_delib <- subset(diff_delib, select = tab_4_t2)
res_delib <- do.call(rbind, lapply(diff_delib, function(x) tidy(t.test(x, mu = 0))))
names(res_delib) <- paste0(names(res_delib), "_d")
res_delib$var_name <- gsub("_t2", "", rownames(res_delib))
diff_delib_info <- kurd_di_set[, tab_4_t2] - kurd_di_set[, tab_4_t1]
diff_delib_info <- subset(diff_delib_info, select = tab_4_t2)
res_delib_info <- do.call(rbind, lapply(diff_delib_info, function(x) tidy(t.test(x, mu = 0))))
names(res_delib_info) <- paste0(names(res_delib_info), "_di")
res_delib_info$var_name_di <- gsub("_t2", "", rownames(res_delib_info))
kurd_res <- res_delib %>%
left_join(res_delib_info, by = c("var_name" = "var_name_di"))
kurd_res <- kurd_res[, c("var_name", "estimate_d", "p.value_d", "estimate_di", "p.value_di")]
kurd_res_all <- kurd_res %>%
mutate(var_name = paste0(var_name, "_t2")) %>%
left_join(kurd_res_nona)
# 5c. Turkoman
# -----------------
# No NA on either wave/List-wise deletion estimates
turk_res_nona <- data.frame(var_name = NA, t1_delib_nona = NA, t2_delib_nona = NA, t1_di_nona = NA, t2_di_nona = NA)
for (i in 1:length(tab_4_t2)) {
nona_delib <- !is.na(turk_d_set[, tab_4_t1[i]]) & !is.na(turk_d_set[, tab_4_t2[i]])
nona_delib_info <- !is.na(turk_di_set[, tab_4_t1[i]]) & !is.na(turk_di_set[, tab_4_t2[i]])
t1_delib <- mean(turk_d_set[, tab_4_t1[i]][nona_delib], na.rm = T)
t2_delib <- mean(turk_d_set[, tab_4_t2[i]][nona_delib], na.rm = T)
t1_delib_i <- mean(turk_di_set[, tab_4_t1[i]][nona_delib_info], na.rm = T)
t2_delib_i <- mean(turk_di_set[, tab_4_t2[i]][nona_delib_info], na.rm = T)
turk_res_nona[i, ] <- c(tab_4_t2[i], t1_delib, t2_delib, t1_delib_i, t2_delib_i)
}
# turk P-values
diff_delib <- turk_d_set[, tab_4_t2] - turk_d_set[, tab_4_t1]
diff_delib <- subset(diff_delib, select = tab_4_t2)
res_delib <- do.call(rbind, lapply(diff_delib, function(x) tidy(t.test(x, mu = 0))))
names(res_delib) <- paste0(names(res_delib), "_d")
res_delib$var_name <- gsub("_t2", "", rownames(res_delib))
diff_delib_info <- turk_di_set[, tab_4_t2] - turk_di_set[, tab_4_t1]
diff_delib_info <- subset(diff_delib_info, select = tab_4_t2)
res_delib_info <- do.call(rbind, lapply(diff_delib_info, function(x) tidy(t.test(x, mu = 0))))
names(res_delib_info) <- paste0(names(res_delib_info), "_di")
res_delib_info$var_name_di <- gsub("_t2", "", rownames(res_delib_info))
turk_res <- res_delib %>%
left_join(res_delib_info, by = c("var_name" = "var_name_di"))
turk_res <- turk_res[, c("var_name", "estimate_d", "p.value_d", "estimate_di", "p.value_di")]
turk_res_all <- turk_res %>%
mutate(var_name = paste0(var_name, "_t2")) %>%
left_join(turk_res_nona)
res_all <- rbind(cbind(arab_res_all, ethnicity = "Arab"),
cbind(kurd_res_all, ethnicity = "Kurd"),
cbind(turk_res_all, ethnicity = "Turkman"))
# Merge
res_all$var_name <- gsub("_t2", "", res_all$var_name)
res_all <- res_all %>%
left_join(ethnic_t1_res, by = c("var_name" = "variable", "ethnicity" = "ethnicity"))
res_all <- res_all %>%
arrange(var_name)
res_all <- res_all[, c("var_name", "ethnicity", "value",
"t1_delib_nona", "t2_delib_nona", "estimate_d", "p.value_d",
"t1_di_nona", "t2_di_nona", "estimate_di", "p.value_di" )]
write.csv(res_all, file = "tabs/04_table_5.csv", row.names = F)
# Inverse Propensity Score Weighting
# Compare Change in D to DI.
# ---------------------------------------
# Subset on D and DI
dat_d_di <- subset(all_dat, cond %in% c("delib", "delib_info"))
# Create dummy vectors for missing
dat_d_di$nona_age <- nona(dat_d_di$age)
dat_d_di$age_miss <- is.na(dat_d_di$age)
dat_d_di$nona_lived_in_kirkuk_1 <- nona(dat_d_di$lived_in_kirkuk == 1)
dat_d_di$nona_lived_in_kirkuk_2 <- nona(dat_d_di$lived_in_kirkuk == 2)
dat_d_di$lived_in_kirkuk_miss <- is.na(dat_d_di$lived_in_kirkuk)
dat_d_di$nona_sunni <- nona(dat_d_di$sunni == 1)
dat_d_di$relig_miss <- is.na(dat_d_di$sunni)
# Dependent Variable so that it is clear
dat_d_di$d_or_di <- I(dat_d_di$cond == "delib")
# Vector of covariates
covars <- c("nona_age", "age_miss",
"female", "kirkuk",
"nona_lived_in_kirkuk_1", "nona_lived_in_kirkuk_2", "lived_in_kirkuk_miss",
"nona_relig", "relig_miss")
d_di_reg <- glm(d_or_di ~ nona_age +
age_miss +
female +
kirkuk +
nona_lived_in_kirkuk_1 +
nona_lived_in_kirkuk_2 +
lived_in_kirkuk_miss +
nona_sunni +
relig_miss,
data = dat_d_di[dat_d_di$wave == 1, ],
family = "binomial")
prs_df <- data.frame(pr_score = predict(d_di_reg, type = "response"),
d_or_di = d_di_reg$model$d_or_di)