Benjamin-Samuel Schlüter, Bruno Masquelier, December 2020
Paper published in PLOS ONE scientific journal link
This file and all other files in the repository should enable you to reproduce our results, plots, tables, ... from our joint paper. Our paper consists in the estimation of sub-national mortality rates for children aged 5-14 for 20 countries in Sub-Saharan Africa, using the method developped by Mercer et al. (2015).
There should be a single parent directory on your file system, where you have to set your R working directory and which will be divided into sub-directories named:
/code
/functions
/data
/fbh
/gps cluster
/shapefiles
/HIV_Adjustments
/auxiliary
/tidy
/richardli_github
The repository available on Github where this README stands is /code. In order to obtain /richardli_github sub-directory, go here, download the repo to create the sub-directory as I did above. Within this downloaded sub-directory, cut/copy HIV_Adjustments into /data in order to have the same directory structure as mine. Make sure to exactly reproduce this structure on your laptop before running the R scripts.
The analysis includes 96 DHS surveys. Each country has a certain number of DHS surveys (see table below). For each country we needed to select a shapefile which, for some countries, is used to relocate cluster within coherent sub-national areas over time. To do so we also needed GPS cluster location of the DHS surveys that needed to be re-locate in coherent sub-national areas. Each DHS survey used in our analysis will be stored in /fbh for Full-Birth History data (96 folders). There will be one shapefile per country hence, /shapefiles should include 20 folders. The number of data sets containing cluster GPS locations depends on the country and are in /gps cluster (32 folders).
In each of these three sub-directories, unzipped downloaded data for a given DHS survey will be stored in a folder named "IDcountryYearDHS"(i.e Full-Birth history data, shapefile and cluster GPS location for Ethiopian 2016 DHS survey will be in ET2016DHS. Hence, in our example, the folder ET2016DHS will appear three times, one time in each of the three sub-directories). Unzipped files downloaded come from two sources (DHS FBH with GPS location and shapefiles).
Below is the table showing the list of files you need to download in order to reproduce our analysis:
| Country | DHS | Shapefile | Cluster GPS location |
|---|---|---|---|
| Benin | 1996,2001,2006,2012,2018 | 2001 | |
| Burkina Faso | 1993,1999,2003,2010 | 1999 | 2003,2010 |
| Cameroon | 1998,2004,2011,2018 | 1998 | |
| Ethiopia | 2000,2005,2011,2016 | 2016 | |
| Ghana | 1993,1998,2003,2008,2014 | 2014 | |
| Guinea | 1999,2005,2012,2018 | 1999 | 2005,2012,2018 |
| Kenya | 1993,1998,2003,2008,2014 | 2014 | 2003,2008,2014 |
| Lesotho | 2004,2009,2014 | 2014 | |
| Madagascar | 1992,1997,2004,2009 | 2004 | 2009 |
| Mali | 1996,2001,2006,2013,2018 | 1996 | 1996,2001,2006,2013, 2018 |
| Malawi | 1992,2000,2004,2010,2016 | 2016 | |
| Nigeria | 1990,2003,2008,2013,2018 | 2018 | 1990 |
| Niger | 1992,1998,2006,2012 | 1998 | |
| Namibia | 2000,2007,2013 | 2013 | |
| Rwanda | 2005,2008,2010,2015 | 2015 | |
| Senegal | 1993,1997,2005,2011,2013,2014,2015,2016,2017,2018,2019 | 2017 | All selected DHS except 2017,2018 |
| Tanzania | 1996,1999,2005,2010,2016 | 1996 | 1999,2010,2016 |
| Uganda | 1995,2001,2006,2011,2016 | 1995 | 2006,2011,2016 |
| Zambia | 1992,1996,2002,2007,2014,2018 | 2002 | 2014,2018 |
| Zimbabwe | 1994,1999,2006,2011,2015 | 2015 |
For the selectivity assessment shown in the Supporting Information, we also downloaded UN IGME 10q5 estimates here and stored the data set in /data/auxiliary.
Within /code folder, script main.R allows you to run all R scripts on the 20 countries at once. Note that the loop might take time to run on all countries (approx. 3 hours). At each iteration associated to a country, main.R will store results in /data/tidy/country_name.RDS. After 20 iterations, you will have a .RDS file for each country. Then use outputs all ctry.R (script within /code) in order to obtain plots and tables presented in the paper.
If you want to create the output for one country at a time, only put one country in the loop.
An alternative, if you do not want to run the loop, is to start with prepare data.R and add these two lines of code at the beginning of the script:
ctry <- "country_name"
ID <- "country_ID"
where country_ID should be the ID of the country of interest
(i.e ctry <- "Malawi" and ID <- "MW"). Then run prepare data.R,
followed by get estimates.R and finally store results.R. Note however that
outputs all ctry.R relies on the fact that you have already stored the output for each country.