Mapping poverty

8 Mapping poverty

Our final step is to map the results. We can do this by joining the estimated poverty rates to the admin3 shapefile. Let’s first load the shapefile and then join the results:

admin3 <- vect("data/mw3.shp")
admin3 <- admin3 |>
  left_join(resultsall, by = "TA_CODE")
admin3

 class       : SpatVector 
 geometry    : polygons 
 dimensions  : 76, 6  (geometries, attributes)
 extent      : 493675.9, 691460, 8591761, 8964834  (xmin, xmax, ymin, ymax)
 coord. ref. : Arc 1950 / UTM zone 36S 
 names       : TA_CODE DIST_CODE poor_ebp poor_ebp_var poor_dir poor_dir_var
 type        :   <chr>     <chr>    <num>        <num>    <num>        <num>
 values      :   10120       101   0.2175      0.07543   0.3177       0.1326
                 10110       101   0.4475       0.3654       NA           NA
                 10102       101   0.3263      0.05838   0.3463      0.08992

Finally, we can map poverty rates and standard errors. Specifically, let’s map the poverty rate and the coefficient of variation, which is defined as the standard error divided by the mean. Here is the code:

g1 <- ggplot() +
  geom_spatvector(data = admin3, aes(fill = poor_ebp), color = NA) +
  scale_fill_distiller("", palette = "Spectral") + 
  labs(subtitle = "A. Poverty rate") +
  theme_bw()
g2 <- ggplot() +
  geom_spatvector(data = admin3, aes(fill = poor_ebp_var/poor_ebp), color = NA) +
  scale_fill_distiller("", palette = "Spectral") + 
  labs(subtitle = "B. Poverty CV") +
  theme_bw()

And the output is in Figure 1. One thing to note is that the CV can be somewhat misleading in areas with very low poverty rates. This is particularly true in the capital of Northern Malawi, where poverty rates are below 0.1, leading to some of the highest CVs in the region.

Figure 1: Mapping poverty in Northern Malawi