Obtaining district-level estimates using geographically masked location from Demographic and Health Survey data

All analysis files for Emily Wilson et al.’s paper Obtaining district-level estimates using geographically masked location from Demographic and Health Survey data have been made available for replication or additional analysis. There are two .R files to re-run the methods, three .R files to reproduce the figures, 13 .do files to recreate the tables, for each survey year – 2000, and 2004 – and there is one .R file to reformat the tables, and one .R file which combines data from 2000 and 2004 to produce Figure 5. We cannot directly provide the geographic location of DHS sampling units, the DHS household data, or the administrative boundaries from FAO, as these require access permission. However, we encourage all who are interested in obtaining these data to register for accounts with dhsprogram.com and fao.org, to obtain the raw data files which are needed. We used these 2000 DHS files: MWPR41FL.DTA MWKR41FL.DTA MWHW41FL.DTA MWHR41FL.DTA MWGE43FL(.shx, .shp.xml, .shp, .sbx, .sbn, .prj, .dbf) MWBR41FL.DTA We used these 2004 DHS files: MWPR4DFL.DTA MWKR4DFL.DTA MWHW4CFL.DTA MWHR4DFL.DTA MWGE4BFL(.shx, .shp.xml, .shp, .sbx, .sbn, .prj, .dbf) MWBR4DFL.DTA We also used second-order administrative unit (ADM2) files for Malawi for two years – 2000 and 2004. We recommend preserving the naming convention of the DHS as well as the file structure in this downloadable zip. We also recommend accessing the .R analyses by first opening the DistrictAssignment.Rproj file. In order to run the analysis files, svmatf needs to be installed from ssc, on Stata, and the R packages listed at the top of each .R file need to be installed on R.