One of the big questions we get asked a lot during the setup process is how many layers of administrative divisions "should" a client use? For some countries, this is not a big issue, since publicly available data may be limited. But other countries (China, for example), have layers upon layers that are available. So how do you know how far to go? Well, the short answer is: It Depends. However, we have a few general rules of thumb:
Only Use the Layers You Need
This seems wildly obvious, but true: if you currently only collect data for individual locations/facilities and aggregate that up to the country level, and you have NO indicators that report at a geographic disaggregation of, say, province or district--why would you add anything? If you feel like you should add the layers just in case, etc., ask yourself these questions:
- Will I ever want to see location-level indicators aggregated to a district or province level?
- If the answer is yes, it's worth considering adding those layers.
- If the answer is no, you probably don't need this layer--ask yourself the next question.
- Do I anticipate having indicators that would report per district or per province?
- If the answer is yes, you must add those layers for data to ever be reported at them!
- If the answer is no, it's likely not worth adding this layer.
- And finally: do I truly need the layer I'm considering adding? For example, you might have a municipality and a municipality border layer. If you only ever have indicator performance reported per municipality, and you never have reports that you want to run that would be displayed by municipality border, there is absolutely no need for having that municipality border layer.
My Country's Shapefiles Allow a Municipality Layer, but I also have cities/municipalites as Locations. What are the advantages/disadvantages of using the municipality administrative division layer rather than just locations?
The lowest level of Administrative Divisions is, in many countries, the most complex because it contains the most individual shapes and the most complicated shapes--often they might be municipality or village boundaries. Some clients want these as shapes; some want them as locations. We often get asked about the advantages/disadvantages of each approach. There is no hard and fast rule here, though the same rules as above apply; if you never disaggregate indicators at this layer and have no interest in aggregating data to this layer, you have no need to add it.
If, however, you do disaggregate indicators at this layer, then the big question becomes: should I use locations or should I use administrative divisions?
Think about the other types of locations you might have in the system. If you ever have individual locations--such as schools, hospitals, or health clinics--that are IN a specific municipality and that you'd want to aggregate INTO municipalities within reports--you need the administrative division layer.
However, if the municipality or village is the only type of location you ever report at, and you just need these to aggregate up to province, district, etc. layers--then you don't need the administrative division layer.
Let me use the city of Buffalo, New York, as an example:
- If I use the administrative division layer, I define Buffalo as a shape on the map, and I could report on 6 different schools (locations) within that shape. I could run reports showing indicator performance data for all of Buffalo that would aggregate together those 6 schools, while also having reports that show indicators per school. Data would further aggregate up into the state of New York and the country of the United States.
- If I define Buffalo as a location, it's just a point on the map. I would report all indicators here to Buffalo. If I don't have individual schools here, this completely meets my needs, and I can still aggregate Buffalo data up into the state of New York and the country of the United States.
In both scenarios, though, the data in the location or that administrative division will automatically aggregate up to broader administrative divisions/the country overall. So both options will allow you to aggregate further up flawlessly--it is only a question of how granular you need to be with your data.