I have a dataset of households.

The level of the data is as follows: households < district < state < country

The dataset gives me household weights for the state and country level but not at the district level.

The treatment I am exploiting is at the district level and hence I want to do the analysis at the district level.

Can anyone suggest (I) a way to construct district weights; (ii) potential weights I can use (say log population of the district at the baseline); (iii) a different strategy where I don't have to aggregate at the district level.


Thanks a lot!

Dhruv