Put the hierarchy in a single reference table and read from it with a cascading reference question. Every survey then draws on the same list, and renaming a district becomes one edit instead of twenty.
Video: Using Cascading Reference questions in mWater on YouTube
Administrative lists are the most copied and least maintained data most organisations hold. Each new survey gets its own pasted version, the copies drift apart, and a year later the same district appears under four spellings so nothing aggregates. A shared list fixes the aggregation problem at its source, which is far cheaper than cleaning it afterwards.