How to overcome conflicts over population figures?

Share

Jan 03 2020 Print

When organizations have conflicting data, arguing on who has the only perfect total number is not useful.  Keep a professional approach: gain the trust of other organizations colleagues, be honest and detailed, transparent on methodologies, explain challenges and limitations of data, recognize necessary improvements.  Approach this exercise with the realistic and professional understanding that most likely nobody has the perfect figures, and therefore your data are not necessarily going to always be the best data.

Two practical approaches developed and used by DTM and UNHCR to constructively address population datasets that do not match are suggested below. They guide the cooperation, identify reasons for different numbers and opportunities for better understanding of population data.

One consensus-based dataset - Comparing data and reach consensus at the lowest admin level possible

From an idea of Shelley Gornall, UNHCR

  1. Approach this exercise with the realistic and professional understanding that most likely nobody has the perfect figures, and therefore your data are not necessarily going to always be the best data.
  2. Gather around a table those actors with first-hand information on population figures. Participants will only be data and IM experts of those organizations owning the population figures, not all clusters or all agencies. Ask these organizations to bring their data at the lowest admin level possible, together with complete metadata (e.g., date of data, source and methodology).
  3. Compare population figures for each of these locations. Proceed jointly to look at figures and accompanying metadata in relation to each specific locations, at the lowest possible admin level (do not discuss the total figure).
  4. Agree jointly on the most reliable data at each of the lowest admin level (do not discuss the total figure).
  5. Retain all the metadata for the figures that are selected in the end.  If the figure is a combination of multiple figures include the explanation of how the figure was calculated in the metadata.
  6. Document sources, methodology and data, among other metadata, for each of the locations. Always include these detailed metadata when disseminating the figures.
  7. Identify gaps in knowledge for some locations and document them. If possible, agree on what organization could fill that gap for the next update (think of various methodologies, including remote sensing if access is a challenge).
  8. Add the agreed figures to obtain the total figure. If somebody disagrees with the total, ask them to point at which specific location figures they disagree with, and discuss changes only to the ‘small’ location figures, never to the total. The total is always the sum of the ‘small’ locations.
  9. Ensure official endorsement of the agreed figures (the detailed compilation of locations, not only the total) through HCT or other appropriate mechanism, explaining the process.
  10. Agree on periodical updates to the locations figures, to be carried out in the same way (the first time the process will be long, however, updating is usually much easier, based on experience).
  11. For example, if an organization has an old total figure for one location, and another has a more up-to-date figure of an influx in that location, it is possible to adjust the old figure adding the more recent influx. This choice must be detailed in the metadata.

Comparing data at the lowest admin level possible  

(From an idea of Nuno Nunes, IOM).

The population data working group or the coordinating entity (e.g., OCHA or authorities) can compile and disseminate to humanitarian responders a table with population figures reported by each agency, at the minimum administrative level possible.  


This will enable data users to consider the data and draw their own conclusions relative to their programmes. They will likely also identify some locations where number are similar and others where their programmes will be able to assess further.  For example, the table could have a similar structure to the one below: 

Example

Addressing differences in population data-Practical tips, in: https://dtm.iom.int/dtm-partners-toolkit/population-data