re-sampling

[Jakob Petersenn]

I have a data set with property adverts by small area and year. There can be between 0 and over 100 adverts per small area and year. Some areas undergo an intervention and simple area aggregation would lead to a loss of precision in terms of exposure. There is in a way both too little and too much data at the same. Could I re-sample the data set to create a balanced panel data set with similar properties? Any suggestions?

[Felix Bittmann]

I am not really sure what the problem is. Maybe a weighting approach could help if you want to make your sample similar to a population. Check out ebalfit from ssc.

[Jakob Petersenn]

Thanks Felix - Entropy balancing by year and a higher level of geography does seem viable to me for a repeated cross-sectional design