Census-to-DHS Geography Crosswalk: Second Administrative Level

The second-level crosswalk variable, GEOLEV2, enables users to link DHS sample clusters to low-level census geography in the IPUMS International database. GEOLEV2 provides consistent second-level administrative units for each country across sample years to facilitate comparisons over time.

Users can summarize high-density census data from IPUMS International to the units in GEOLEV2 and attach those measures to IPUMS DHS as contextual information. For example, users might summarize educational attainment in each second-level geographic area using IPUMS International, merge that result onto their IPUMS DHS extract using GEOLEV2 as the linking key, and assess the effect of local educational attainment net of the respondent's own characteristics.

Computation

Census-to-DHS geography crosswalks are based on a common methodology using Esri's ArcGIS software suite. Geography codes are computed within a 2 to 5 kilometer buffer around a GPS cluster to account for cluster displacement (a confidentiality measure employed by DHS). A 2-kilometer buffer is used around the urban clusters and a 5-kilometer buffer is used around the rural clusters. 10-kilometer buffers are deliberately not used in this analysis because they constitute only 1% of rural clusters.

Next, GPS cluster areas from DHS are overlaid on harmonized second-level administrative units (GEOLEV2). If the entire buffer is contained within the administrative unit, we assign a placement probability index of 100%. If the buffer crosses boundaries of two or more second-level administrative units, we list the code for the unit with the highest placement probability index along with its index value. Countries with small second-level administrative units (e.g. Egypt, Nigeria) have lower placement probability index percentages on average compared to countries with larger second-level administrative units (e.g. Bangladesh, Mali).

For further details on GPS clusters, see the documentation on cluster displacement. Users interested in performing spatial analysis with GPS cluster datasets may obtain DHS cluster shapefiles from the DHS Program website.

Using census-to-DHS crosswalks for research

Researchers intending to utilize the second-level geography crosswalk for contextual analysis should include the GEOLEV2 variable in their customized IPUMS DHS extract and in the corresponding IPUMS International extract that they intend to summarize. There are sufficient cases at the second administrative level in IPUMS International to support many summary statistics, but users should be cognizant of confidence intervals. DHS data, however, should NOT be summarized using the units identified in GEOLEV2. The DHS sample designs do not support analyses at that geographic level.

Downloading second administrative level crosswalks

Researchers may include GEOLEV2 as part of their customized data file (extract) and use the codes from GEOLEV2 as crosswalk codes to link second level administrative units from IPUMS International.

Users could also download the entire .csv files for all countries that list DHSID, corresponding GEOLEV2 codes that link to IPUMS International and the placement probability index (GEOLEV2PROB) of the cluster within the administrative boundary.

Download CSV file from HERE