-
Notifications
You must be signed in to change notification settings - Fork 2
/
prepare_analysis_data.R
240 lines (201 loc) · 12.1 KB
/
prepare_analysis_data.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
# Setup -------------------------------------------------------------------
library(magrittr)
library(tidyverse)
library(lubridate)
library(sf)
library(sp)
library(rgeos)
library(econometr)
source("analysis_util.R")
wgs.84 <- "+proj=longlat +datum=WGS84 +no_defs +ellps=WGS84 +towgs84=0,0,0"
kenya.proj4 <- "+proj=utm +zone=36 +south +ellps=clrk80 +units=m +no_defs"
# Load the data ----------------------------------------------------------
load(file.path("data", "takeup_village_pot_dist.RData"))
## Ed note: this produces error "unknown input format" as of 2023/04/13 so
## we use the RData file and some env munging instead
# This data was prepared in takeup_field_notebook.Rmd
# census.data <- read_rds(file.path("data", "takeup_census.rds")) %>%
# rename(census.consent = consent) # Rename this to reduce chance of error
census_data_env = new.env()
with_env = function(f, e = parent.frame()) {
stopifnot(is.function(f))
environment(f) = e
f
}
load_census_function = function(){
load(file.path("data", "takeup_census.RData"))
return(census.data)
}
census.data = with_env(load_census_function, census_data_env)() %>%
rename(census.consent = consent) # Rename this to reduce chance of error
baseline.data <- read_rds(file.path("data", "takeup_baseline_data.rds"))
takeup.data <- read_rds(file.path("data", "takeup.rds"))
all.endline.data <- read_rds(file.path("data", "all_endline.rds"))
reconsent.data <- read_rds(file.path("data", "reconsent.rds"))
cluster.strat.data <- read_rds(file.path("data", "takeup_processed_cluster_strat.rds")) # Study clusters metadata
sms.content.data <- read_rds(file.path("data", "takeup_sms_treatment.rds"))
pot.info <- read_rds(file.path("data", "pot_info.rds")) # Point of treatment information
wtp.data <- read_rds(file.path("data", "takeup_wtp.rds")) # Willingness-to-pay experiment data
# Geographic data for clusters in the study
pot_geo_info <- pot.info %>%
filter(!is.na(wave)) %>%
transmute(cluster.id,
pot.lon = lon.post.rct.update,
pot.lat = lat.post.rct.update)
# HH distance to PoT ------------------------------------------------------
# census.data %>%
# group_by(cluster.id) %>%
# group_modify(function(households, ...) {
# hh_sf <- st_as_sf(households, coords = c("lon", "lat"), crs = wgs.84) %>%
# st_transform(kenya.proj4)
#
# # pot_sf <- st_as_sf(pot_geo_info, coords = c("pot.lon", "pot.lat"), crs = wgs.84)
#
# return(households)
# }, pot_geo_info = pot_geo_info) %>%
# ungroup()
#
#
#
# nest_join(pot_geo_info, ., by = "cluster.id", name = "households") %>%
# # st_as_sf(coords = c("pot.lon", "pot.lat"), crs = wgs.84) %>%
# mutate(households = map(households, st_as_sf, coords = c("lon", "lat"), crs = wgs.84) %>%
# map2(geometry, ~ mutate(.x,
# dist.to.pot = st_distance(.x %>% st_transform(kenya.proj4),
# st_sfc(.y, crs = wgs.84) %>% st_transform(kenya.proj4)) %>%
# as.vector())))
# Calculate the distance between households and their assigned point-of-treatment
census.data %<>%
left_join(pot_geo_info, by = "cluster.id") %>%
group_by(cluster.id) %>%
group_modify(~ {
if (is.na(first(.y$cluster.id))) {
individ.dist.to.pot <- NA
} else {
cluster.pot.location <- convert.to.sp(slice(.x, 1), ~ pot.lon + pot.lat, wgs.84) %>%
spTransform(kenya.proj4)
individ.locations <- convert.to.sp(.x, ~ lon + lat, wgs.84) %>%
spTransform(kenya.proj4)
individ.dist.to.pot <- c(gDistance(individ.locations, cluster.pot.location, byid = TRUE))
}
mutate(.x, dist.to.pot = individ.dist.to.pot)
}) %>%
ungroup()
# Village center distance to PoT ------------------------------------------
village.centers %<>%
select(cluster.id, lon, lat) %>%
left_join(pot_geo_info, by = "cluster.id") %>% {
cluster.pot.locations <- convert.to.sp(., ~ pot.lon + pot.lat, wgs.84) %>%
spTransform(kenya.proj4)
cluster.center.locations <- convert.to.sp(., ~ lon + lat, wgs.84) %>%
spTransform(kenya.proj4)
mutate(., dist.to.pot = diag(gDistance(cluster.center.locations, cluster.pot.locations, byid = TRUE)))
} %>%
left_join(select(cluster.strat.data, cluster.id, assigned.treatment, dist.pot.group), by = "cluster.id")
# Core data prep --------------------------------------------------------------
baseline.data %<>%
prepare.baseline.data(cluster.strat.data)
endline.data <- prepare.endline.data(all.endline.data, census.data, cluster.strat.data)
consent.dewormed.reports <- prepare.consent.dewormed.data(all.endline.data, reconsent.data)
analysis.data <- prepare.analysis.data(census.data, takeup.data, endline.data, baseline.data, consent.dewormed.reports, cluster.strat.data, max.name.match.cost = 1)
# Outliers ----------------------------------------------------------------
cell.takeup.data <- prepare.cluster.takeup.data(analysis.data, monitored.only = FALSE,
add_group_by = c("phone_owner", "mon_status"))
outlier.cells <- cell.takeup.data %>%
filter(outlier)
# WTP prep ----------------------------------------------------------------
wtp.data %<>%
mutate(first_choice = factor(first_choice, levels = 1:2, labels = c("bracelet", "calendar")),
second_choice = if_else(first_choice == "bracelet", cal_plus_ksh, bra_plus_ksh) %>%
factor(levels = 1:2, labels = c("switch", "keep"))) %>%
select(-county) %>% # I want to use the more reliable one in cluster.strat.data
left_join(select(cluster.strat.data, cluster.id, assigned.treatment, dist.pot.group, county), by = "cluster.id")
# Social info report in SMS -----------------------------------------------
social.info.data <- sms.content.data %>%
filter(!is.na(social.info)) %>%
left_join(select(census.data, cluster.id, village, assigned.treatment, sms.treatment, KEY.individ), by = "KEY.individ") %>%
filter(!is.na(cluster.id)) %>%
left_join(select(cluster.strat.data, cluster.id, dist.pot.group), by = "cluster.id") %>%
group_by(deworming.day, assigned.treatment, dist.pot.group) %>%
do({
interq.info <- quantile(.$social.info/10, c(0.25, 0.75))
summarize(.,
cumul = mean(social.info)/10,
num.obs = n(),
qant.25 = interq.info[1],
qant.75 = interq.info[2])
}) %>%
group_by(deworming.day, assigned.treatment) %>%
mutate(all.dist.cumul = weighted.mean(cumul, num.obs)) %>%
ungroup %>%
rename(dewormed.day = deworming.day)
# Knowledge tables prep ---------------------------------------------------
# endline.know.table.data <- read_rds(file.path("data", "know_tables.rds"))
endline.know.table.data <- bind_rows(table.A = read_csv(file.path("data", "raw-data", "Endline Survey-survey-sec_D-tableA.csv")) %>%
transmute(num.recognized = recogniseA,
dewormed = dewormedA,
second.order = order_2ndA,
second.order.reason = order_2nd_reason,
relationship = relationshipA,
relationship.other = relationshipA_other,
times.seen = times_seenA,
visited = visitedA,
know.other.index = instanceA,
PARENT_KEY, KEY),
table.B = read_csv(file.path("data", "raw-data", "Endline Survey-survey-sec_D-tableB.csv")) %>%
transmute(num.recognized = recogniseB,
dewormed = dewormedB,
dewormed.know.only = dewormedBB,
know.other.index = instanceB + 0:9,
know.other.index.2 = know.other.index + 1,
PARENT_KEY, KEY),
.id = "know.table.type") %>%
mutate(recognized = num.recognized > 0)
survey.know.list <- file.path("instruments", "SurveyCTO Forms", "Endline Survey", "Deployed Form Version",
c("knowledge_list.csv", "kak_knowledge_list.csv")) %>%
map_df(read_csv, col_types = cols(knowledge_person_key = col_character(), survey.type = col_character()), .id = "wave") %>%
mutate(wave = as.integer(wave)) %>%
filter(wave == 1 | !is.na(survey.type))
endline.know.table.data %<>%
inner_join(select(endline.data, KEY.individ, wave, cluster.id, assigned.treatment, sms.treatment, dist.pot.group, KEY),
by = c("PARENT_KEY" = "KEY")) %>%
inner_join(select(survey.know.list, person_key, know.other.index, KEY.individ.other),
by = c("KEY.individ" = "person_key", "know.other.index")) %>%
left_join(select(survey.know.list, person_key, know.other.index, KEY.individ.other), # Get the second person (table B)
by = c("KEY.individ" = "person_key", "know.other.index.2" = "know.other.index"),
suffix = c(".1", ".2")) %>%
# mutate_at(vars(recognized, dewormed, dewormed.know.only),
# list(~ factor(., levels = c(0:1, 98), labels = c("no", "yes", "don't know")))) %>%
mutate(across(c(recognized, dewormed, dewormed.know.only), factor, levels = c(0:1, 98), labels = c("no", "yes", "don't know"))) %>%
# mutate_at(vars(second.order),
# list(~ factor(., levels = c(1:2, 97:98), labels = c("yes", "no", "prefer not say", "don't know")))) %>%
mutate(across(second.order, factor, levels = c(1:2, 97:98), labels = c("yes", "no", "prefer not say", "don't know"))) %>%
# mutate_at(vars(relationship),
# list(~ factor(.,
# levels = c(1:5, 99),
# labels = c("hh member", "extended family", "friend", "neighbor", "church", "other")))) %>%
mutate(across(relationship, factor, levels = c(1:5, 99), labels = c("hh member", "extended family", "friend", "neighbor", "church", "other"))) %>%
mutate(relationship = if_else(fct_match(relationship, "other") & fct_match(str_to_lower(relationship.other), c("village member", "village mate", "same village", "village elder", "village mates", "villager")),
"village member", as.character(relationship)) %>% factor,
dewormed = if_else(fct_match(know.table.type, "table.B") & is.na(dewormed), dewormed.know.only, dewormed)) %>%
left_join(filter(analysis.data, monitored) %>% transmute(KEY.individ, respondent.dewormed.any = dewormed.any),
by = "KEY.individ") %>%
left_join(filter(analysis.data, monitored) %>% transmute(KEY.individ, actual.other.dewormed.any.1 = dewormed.any),
by = c("KEY.individ.other.1" = "KEY.individ")) %>%
left_join(filter(analysis.data, monitored) %>% transmute(KEY.individ, actual.other.dewormed.any.2 = dewormed.any),
by = c("KEY.individ.other.2" = "KEY.individ")) %>%
mutate(actual.other.dewormed.any.either = actual.other.dewormed.any.1 | (fct_match(know.table.type, "table.B") & actual.other.dewormed.any.2))
# left_join(select(cluster.strat.data, cluster.id, assigned.treatment, dist.pot.group), "cluster.id") # Get it from endline data
# Save data ---------------------------------------------------------------
save(all.endline.data, endline.data, consent.dewormed.reports, baseline.data,
analysis.data,
cell.takeup.data, outlier.cells,
census.data, reconsent.data, takeup.data, sms.content.data, social.info.data, village.centers, wtp.data, endline.know.table.data,
file = file.path("data", "analysis.RData"))
endline.data %>%
set_names(str_replace_all(names(.), "\\.|-", "_")) %>%
mutate_if(is.list, list(~ map_chr(., ~ str_c(.x, collapse = " ")))) %>%
haven::write_dta(file.path("data", "endline_data.dta"))
analysis.data %>%
set_names(str_replace_all(names(.), "\\.|-", "_")) %>%
haven::write_dta(file.path("data", "analysis_data.dta"))