-
Notifications
You must be signed in to change notification settings - Fork 0
/
ipl_lirs_whatsapp_paper.Rmd
305 lines (251 loc) · 9.01 KB
/
ipl_lirs_whatsapp_paper.Rmd
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
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
---
title: "Analysis: IPL LIRS WhatsApp Paper"
date: "Updated: `r format(Sys.Date(), '%B %d, %Y')`"
output:
html_document:
df_print: paged
toc: yes
toc_depth: 4
number_sections: no
---
```{r setup, include=FALSE}
knitr::opts_chunk$set(message = FALSE, warning = FALSE, echo = TRUE, fig.retina = 4)
knitr::opts_chunk$set(fig.pos = 'H')
```
\newpage
## Data loading
```{r}
# Libraries
pacman::p_load(tidyverse, readxl, janitor, lubridate, extrafont, estimatr, fastDummies, zoo, hrbrthemes, compareGroups)
extrafont::loadfonts()
```
### Data source 1 (LIRS baseline survey data)
```{r}
# clean baseline data
df_baseline <- read_csv("data/baseline_data_lirs.csv")
```
### Data source 2 (Twilio contact sheets)
```{r}
# binding twilio whatsapp and sms data
df_twilio <-
bind_rows(
read_xlsx("data/whatsapp_twilio.xlsx") %>% rename(number_from = whatsAppNumber_final),
read_xlsx("data/sms_twilio.xlsx") %>% rename(number_from = smsNumber_final),
.id = "medium"
) %>%
mutate(
medium = recode(medium, `1` = "WhatsApp", `2` = "SMS", .default = NA_character_),
) %>%
distinct(study_ID, .keep_all = TRUE) %>%
distinct(number_from, .keep_all = TRUE)
```
### Data source 3 (LIRS panel data)
```{r}
# import google sheet with panel data
df_panel <-
read_xlsx("data/panel_data_lirs.xlsx") %>%
mutate(
number_from = parse_number(number_from) %>% as.character(),
date = as_datetime(date),
wave = str_c(month(date), year(date), sep = "-")
) %>%
mutate(
wave = if_else(wave == "8-2020", "9-2020", wave),
wave = if_else(wave == "12-2020", "11-2020", wave),
wave = if_else(wave == "3-2021", "2-2021", wave),
wave = if_else(wave == "5-2021", "6-2021", wave)
) %>% # edited misnumbered waves here if needed
arrange(number_from, desc(date)) %>%
distinct(wave, number_from, .keep_all = TRUE) %>% # keep most recent response in case a phone number responds multiple times
drop_na(number_from)
# manually handling 3 post failures
df_panel_manual <-
read_xlsx("data/panel_data_lirs.xlsx") %>%
mutate(
number_from = parse_number(number_from) %>% as.character(),
date = as_datetime(date),
wave = str_c(month(date), year(date), sep = "-")
) %>%
mutate(wave = if_else(wave == "8-2020", "9-2020", wave)) %>%
arrange(number_from, desc(date)) %>%
distinct(wave, number_from, .keep_all = TRUE) %>%
drop_na(number_from) %>%
select(number_from, date, wave, q0_response, q1_response:q18_response) %>%
mutate_at(vars(starts_with("q")), as.character)
df_panel <-
bind_rows(df_panel, df_panel_manual) %>%
arrange(number_from, desc(date)) %>%
distinct(wave, number_from, .keep_all = TRUE)
# removing redundant datasets
rm(df_panel_manual)
```
### Merging datasets
```{r}
# merging it all together
df_main <-
df_twilio %>%
inner_join(df_baseline, by = "study_ID") %>%
left_join(df_panel, by = "number_from") %>%
mutate(
survey_started = (q1_response != "No Data"),
survey_completed = (q18_response != "No Data"),
id_matched = if_else(yearArrive == q18_response, 1L, 0L),
wave = as.yearmon(wave, "%m-%Y") %>% as.character() %>% factor(levels = c("Sep 2020", "Oct 2020", "Nov 2020", "Feb 2021", "Jun 2021"))
) %>%
mutate_at(vars(q1_response:q18_response), str_to_lower) %>%
mutate_at(vars(q1_response:q18_response), ~ str_replace_all(., "[.]", "")) %>%
mutate(
female = recode(gender, "Male" = "0", "Female" = "1", .default = NA_character_) %>% as.character %>% parse_integer(),
unemploymentBeneBaseline = if_else(unemploymentBeneBaseline == "Yes", "1", unemploymentBeneBaseline),
unemploymentBeneBaseline = if_else(unemploymentBeneBaseline == "No", "0", unemploymentBeneBaseline),
unemployment = unemploymentBeneBaseline %>% as.integer(),
any_income = 1 - unemployment,
food_insecurity = if_else(str_detect(foodSecurity1Baseline, "^Enough"), "0", foodSecurity1Baseline),
food_insecurity = if_else(str_detect(food_insecurity, "not enough"), "1", food_insecurity),
food_insecurity = if_else(food_insecurity %in% c("0", "1"), food_insecurity, NA_character_),
food_insecurity = food_insecurity %>% as.integer(),
)
```
## Table 1: WhatsApp (LIRS)
```{r}
df_main %>%
filter(medium == "WhatsApp", wave %in% c("Sep 2020", "Feb 2021", "Jun 2021")) %>%
select(wave, survey_started, survey_completed, id_matched) %>%
mutate(completed_and_matched = (survey_completed & id_matched)) %>%
group_by(wave) %>%
summarize(
`Survey started` = sum(survey_started, na.rm = TRUE),
`Survey completed` = sum(survey_completed, na.rm = TRUE),
`Completed & matched` = sum(completed_and_matched, na.rm = TRUE),
`Total` = df_twilio %>% filter(medium == "WhatsApp") %>% nrow()
) %>%
mutate(
`% Started` = round(100 * `Survey started`/`Total`, 1),
`% Completed` = round(100 * `Survey completed`/`Total`, 1),
`% Completed & matched` = round(100 * `Completed & matched` / `Total`, 1),
`% Completed if started` = round(100 * `Survey completed` / `Survey started`, 1),
)
```
## Table A.3: SMS (LIRS)
```{r}
df_main %>%
filter(medium == "SMS", wave %in% c("Sep 2020", "Feb 2021", "Jun 2021")) %>%
select(wave, survey_started, survey_completed, id_matched) %>%
mutate(completed_and_matched = (survey_completed & id_matched)) %>%
group_by(wave) %>%
summarize(
`Survey started` = sum(survey_started, na.rm = TRUE),
`Survey completed` = sum(survey_completed, na.rm = TRUE),
`Completed & matched` = sum(completed_and_matched, na.rm = TRUE),
`Total` = df_twilio %>% filter(medium == "SMS") %>% nrow()
) %>%
mutate(
`% Started` = round(100 * `Survey started`/`Total`, 1),
`% Completed` = round(100 * `Survey completed`/`Total`, 1),
`% Completed & matched` = round(100 * `Completed & matched` / `Total`, 1),
`% Completed if started` = round(100 * `Survey completed` / `Survey started`, 1),
)
```
## Table A.2 (Panel B for LIRS)
### Wave 1
```{r}
# Sep 2020
df_main %>%
filter(medium == "WhatsApp", wave == "Sep 2020") %>%
mutate(
responded = if_else(survey_completed & id_matched, 1L, 0L),
responded = if_else(is.na(responded), 0L, responded)
) %>%
compareGroups(responded ~ female + birthYear + householdSize + any_income + food_insecurity, data = .) %>%
createTable(show.all = F)
```
```{r}
# standard errors of difference in means
round(sqrt((0.48^2 / 543) + (0.43^2 / 258)), 2)
round(sqrt((10.9^2 / 543) + (9.93^2 / 258)), 2)
round(sqrt((2.44^2 / 543) + (2.20^2 / 258)), 2)
round(sqrt((0.42^2 / 543) + (0.43^2 / 258)), 2)
round(sqrt((0.41^2 / 543) + (0.38^2 / 258)), 2)
```
```{r}
# extra check for birth years
df_main %>%
filter(medium == "WhatsApp", wave == "Sep 2020") %>%
mutate(
responded = if_else(survey_completed & id_matched, 1L, 0L),
responded = if_else(is.na(responded), 0L, responded)
) %>%
group_by(responded) %>%
summarize(
mean_birth = round(mean(birthYear, na.rm = T), 2),
sd_birth = round(sd(birthYear, na.rm = T), 2),
)
```
### Wave 2
```{r}
# Feb 2021
df_main %>%
filter(medium == "WhatsApp", wave == "Feb 2021") %>%
mutate(
responded = if_else(survey_completed & id_matched, 1L, 0L),
responded = if_else(is.na(responded), 0L, responded)
) %>%
compareGroups(responded ~ female + birthYear + householdSize + any_income + food_insecurity, data = .) %>%
createTable(show.all = F)
```
```{r}
# standard errors of difference in means
round(sqrt((0.47^2 / 580) + (0.46^2 / 197)), 2)
round(sqrt((10.6^2 / 580) + (10.5^2 / 197)), 2)
round(sqrt((2.37^2 / 580) + (2.29^2 / 197)), 2)
round(sqrt((0.42^2 / 580) + (0.43^2 / 197)), 2)
round(sqrt((0.39^2 / 580) + (0.41^2 / 197)), 2)
```
```{r}
# extra check for birth years
df_main %>%
filter(medium == "WhatsApp", wave == "Feb 2021") %>%
mutate(
responded = if_else(survey_completed & id_matched, 1L, 0L),
responded = if_else(is.na(responded), 0L, responded)
) %>%
group_by(responded) %>%
summarize(
mean_birth = round(mean(birthYear, na.rm = T), 2),
sd_birth = round(sd(birthYear, na.rm = T), 2),
)
```
### Wave 3
```{r}
# Jun 2021
df_main %>%
filter(medium == "WhatsApp", wave == "Jun 2021") %>%
mutate(
responded = if_else(survey_completed & id_matched, 1L, 0L),
responded = if_else(is.na(responded), 0L, responded)
) %>%
compareGroups(responded ~ female + birthYear + householdSize + any_income + food_insecurity, data = .) %>%
createTable(show.all = F)
```
```{r}
# Jun 2021
round(sqrt((0.47^2 / 624) + (0.46^2 / 146)), 2)
round(sqrt((10.5^2 / 624) + (10.7^2 / 146)), 2)
round(sqrt((2.36^2 / 624) + (2.32^2 / 146)), 2)
round(sqrt((0.43^2 / 624) + (0.41^2 / 146)), 2)
round(sqrt((0.40^2 / 624) + (0.38^2 / 146)), 2)
```
```{r}
# extra check for birth years
df_main %>%
filter(medium == "WhatsApp", wave == "Jun 2021") %>%
mutate(
responded = if_else(survey_completed & id_matched, 1L, 0L),
responded = if_else(is.na(responded), 0L, responded)
) %>%
group_by(responded) %>%
summarize(
mean_birth = round(mean(birthYear, na.rm = T), 2),
sd_birth = round(sd(birthYear, na.rm = T), 2),
)
```