Geographic unit Ethiopia 2007 census

I want to use the 2007 Ethiopia census for analysis. I wonder, however, at which admin level the census provides representative information. The sample overview information states that the geographical unit is the Zone, which is admin level 2. However, when I look at various census documents and at a description of the census at the World Bank microdata website (Ethiopia - Population and Housing Census 2007 - IPUMS Subset), the refer to the wereda, which is a admin level 3. I suspect that it is the latter, which means there is a mistake on the IPUMS website. Can you please clarify?


Wereda (Admin Level 3) is available in IPUMS for the Ethiopia 2007 sample and provides representative information. It likely was not listed on the sample characteristics page because IPUMS International does not spatially harmonize Level 3 geography. If you use Wereda for your analysis, keep in mind that it is harmonized by name and does not take into consideration changes in boundaries over time.

Dear Grace,

Thanks for the explanation. I am currently processing the Ethiopia Woreda data but fail to locate a corresponding map. Looking on the internet, there are various sources where Ethiopia shapefiles are shared (e.g. GADM, geoboundaries, etc) but none of them provide maps corresponding with the Woreda classification as used in the 2007 Population census. Would IPUMS be able to share the 2007 Population census map for admin 3 units? Thanks.

We do not have a shapefile of Ethiopia Woreda data. You can join IPUMS data with geographic data you find in other sources if they include the administrative code; this will likely provide a more complete join than trying to join based on name. As you mentioned, there are multiple sources of Ethiopia shapefiles available online. This data source on Ethiopian subnational Administrative Divisions contains Woreda codes (admin3Pcode) that can be joined with the WERDET variable from IPUMS. Note, you will need to clean up the data to ensure the fields match before the join will work (for example, remove “ET” from the codes in the shapefile).

Dear Grace, Thanks for the suggestion. In the meanwhile I have explored your suggested map as well as other maps and arrrived at a solution. I briefly describe it here as it might be of use for others with the same interests. A comparison of the admin3Pcode and WERDET indicated there are possibilities to join a number of records. Unfortunately, the map has far more polygons/Weredas than are listed in the Ethiopia population census and therefore cannot easily be used. A better map is provided by the US Census Bureau, who present Wereda level population projections using various rounds of the Ethiopia population census. The accompanying map has nearly the same Wereda structure as in the census and can be found here: Subnational Population by Sex, Age, and Geographic Area. The only problem is that the map presents names (as opposed to codes) for the Weredas, which makes it difficult to join the two sources. After exploring the ‘fuzzyjoin’ package in R as well as a bit of manual mapping, I managed to link all but 13 Weredas of the population census.

1 Like