Visualizing the DHS Data

Let us load the data we will need:

dhs14_design <- readRDS("data/dhs14_design.RDS")
dhs.df14 <- readRDS("data/dhs.df14.RDS")

Visualizing Data

Three common ways to generate basic graphics in R are via - base R - lattice - ggplot2

We will skip base R graphics since ggplot2 will be the graphics package for this class. Let us see how we use it, starting with a simple bar-chart

Remember the typical options…

• If you have one qualitative/categorical variables: use a bar-chart
• If you have one quantitative/continuous variables: use a histogram/box-plot/area-chart
  
• If you have two quantitative/continuous variables: use a scatter-plot/hex-bin
  
• If you want to add a third or fourth variable, that is easily achieved

ggplot2 can work with unit-level data (i.e., one row per observation) AND it can also work with calculated values stored in a data-frame (like our survey-weighted estimates from BRFSS/DHS)

we will see both ways in action

Bar-charts

Educational Attainment (educ)

library(ggplot2)
ggplot(data = dhs.df14, aes(x = educat)) + geom_bar()

Educational attainment by Wealth quintiles

ggplot(data = dhs.df14, aes(x = educat, group = wealthq, fill = wealthq)) +
  geom_bar()

ggplot(data = dhs.df14, aes(x = educat, group = wealthq, fill = wealthq)) +
  geom_bar(position = "dodge")

ggplot(data = dhs.df14, aes(x = educat, group = wealthq, fill = wealthq)) +
  geom_bar(position = "dodge") +
  facet_wrap(~ educat)

ggplot(data = dhs.df14, aes(x = educat, group = wealthq, fill = wealthq, y = ..prop..)) +
  geom_bar(position = "dodge") +
  facet_wrap(~ wealthq, ncol = 2) +
  scale_y_continuous(labels = scales::percent) +
  theme(legend.position = "bottom")

Weighted estimates

These are all unweighted estimates; how would we plot the weighted estimates?

library(survey)
library(srvyr)
tab.1 <- svymean(~ educat, dhs14_design, na.rm = TRUE)
tab.1 <- as.data.frame(tab.1) * 100
tab.1 <- round(tab.1, digits = 1)
tab.1

                           mean  SE
educatno education         24.0 0.6
educatincomplete primary    6.1 0.2
educatcomplete primary      4.1 0.2
educatincomplete secondary 13.3 0.3
educatcomplete secondary   38.5 0.6
educathigher               13.9 0.5

tab.1p <- tab.1 %>% 
  mutate(educ = rownames(tab.1),
         educ = stringr::str_replace(educ, "educat", ""),
         educ = stringr::str_to_title(educ),
         educ = ordered(educ, 
                        levels = c("No Education", "Incomplete Primary",
                                   "Complete Primary", "Incomplete Secondary", 
                                   "Complete Secondary", "Higher")
                        )
         )

knitr::kable(tab.1p)

mean	SE	educ
24.0	0.6	No Education
6.1	0.2	Incomplete Primary
4.1	0.2	Complete Primary
13.3	0.3	Incomplete Secondary
38.5	0.6	Complete Secondary
13.9	0.5	Higher

library(ggplot2)
ggplot(data = tab.1p, aes(x = educ, y = mean)) +
  geom_bar(stat = "identity") 

Weighted Educational attainment by Wealth quintiles?

tab.ew <- svyby(~educat, ~wealthq, dhs14_design, svymean, na.rm = TRUE)
tab.ew <- tab.ew[, c(1:7)] 
tab.ew2 <-  tab.ew[, c(2:7)] * 100  
tab.ew2 <- round(tab.ew2, digits  = 1)

knitr::kable(tab.ew2)

	educatno education	educatincomplete primary	educatcomplete primary	educatincomplete secondary	educatcomplete secondary	educathigher
poorest	47.7	8.6	4.3	13.8	22.7	3.0
poorer	37.4	8.9	4.9	14.8	30.4	3.6
middle	18.9	6.1	3.7	14.6	46.5	10.2
richer	14.0	5.1	5.0	13.8	44.8	17.3
richest	5.4	2.3	2.8	9.4	45.3	34.8

library(tidyr)
tab.ew3 <- tab.ew2 %>% 
  mutate(Wealth = rownames(tab.ew2)) %>% 
  gather(Education, Percent, 1:6)  %>% 
  mutate(Education = substring(Education, 7),
         Education = stringr::str_to_title(Education)
         ) %>% 
  mutate(Education2 = ordered(Education, 
                        levels = c("No Education", "Incomplete Primary",
                                   "Complete Primary", "Incomplete Secondary", 
                                   "Complete Secondary", "Higher")
                        ),
         Wealth = stringr::str_to_title(Wealth)
         ) %>% 
  mutate(Wealth2 = ordered(Wealth, 
                           levels = c("Poorest", "Poorer",
                                      "Middle", "Richer",
                                      "Richest")
                           )
         )

knitr::kable(tab.ew3)

Wealth	Education	Percent	Education2	Wealth2
Poorest	No Education	47.7	No Education	Poorest
Poorer	No Education	37.4	No Education	Poorer
Middle	No Education	18.9	No Education	Middle
Richer	No Education	14.0	No Education	Richer
Richest	No Education	5.4	No Education	Richest
Poorest	Incomplete Primary	8.6	Incomplete Primary	Poorest
Poorer	Incomplete Primary	8.9	Incomplete Primary	Poorer
Middle	Incomplete Primary	6.1	Incomplete Primary	Middle
Richer	Incomplete Primary	5.1	Incomplete Primary	Richer
Richest	Incomplete Primary	2.3	Incomplete Primary	Richest
Poorest	Complete Primary	4.3	Complete Primary	Poorest
Poorer	Complete Primary	4.9	Complete Primary	Poorer
Middle	Complete Primary	3.7	Complete Primary	Middle
Richer	Complete Primary	5.0	Complete Primary	Richer
Richest	Complete Primary	2.8	Complete Primary	Richest
Poorest	Incomplete Secondary	13.8	Incomplete Secondary	Poorest
Poorer	Incomplete Secondary	14.8	Incomplete Secondary	Poorer
Middle	Incomplete Secondary	14.6	Incomplete Secondary	Middle
Richer	Incomplete Secondary	13.8	Incomplete Secondary	Richer
Richest	Incomplete Secondary	9.4	Incomplete Secondary	Richest
Poorest	Complete Secondary	22.7	Complete Secondary	Poorest
Poorer	Complete Secondary	30.4	Complete Secondary	Poorer
Middle	Complete Secondary	46.5	Complete Secondary	Middle
Richer	Complete Secondary	44.8	Complete Secondary	Richer
Richest	Complete Secondary	45.3	Complete Secondary	Richest
Poorest	Higher	3.0	Higher	Poorest
Poorer	Higher	3.6	Higher	Poorer
Middle	Higher	10.2	Higher	Middle
Richer	Higher	17.3	Higher	Richer
Richest	Higher	34.8	Higher	Richest

ggplot(data = tab.ew3, aes(x = Education2, y = Percent, group = Wealth2, fill = Education2)) +
  geom_bar(stat = "identity") +
  facet_wrap(~Wealth2, ncol = 5) + 
  theme(legend.position = "bottom", axis.text.x = element_text(angle = 90))

Histograms with hemoglobin concentration

hemoglobinData <- read.csv(url("http://whitlockschluter.zoology.ubc.ca/wp-content/data/chapter02/chap02e3cHumanHemoglobinElevation.csv"))

ggplot(data = hemoglobinData, aes(x = hemoglobin)) + geom_histogram(fill="cornflowerblue", color = "white") + 
  labs(x = "Hemoglobin in grams (gm) per deciliter (dL)",  
       y = "Frequency", 
       title = "Histogram of Hemoglobin Concentration in Adult Males")