Home » Data » Weighting data » Clustered Standard Errors
Re: Clustered Standard Errors [message #22269 is a reply to message #22242] |
Mon, 22 February 2021 09:50 ![Go to previous message Go to previous message](/theme/default/images/up.png) ![Go to next message Go to previous message](/theme/default/images/down.png) |
analyst_till
Messages: 10 Registered: January 2021
|
Member |
|
|
Thanks again for the answer.
Indeed, when I try to run districts instead of clusters in the "svyset" framework, STATA is giving me missings for the F-statistics.
To sum up my issue: I have districts as groups and clusters as PSUs in my framework. I also understood from the literature that one should first of all cluster SE at the PSU-level, corresponding to the survey design.
But is it not an issue to cluster standard errors at another (lower) level than the groups that I use in my diff-in-diff?
May I finally ask, how you would proceed in this case? I read your abovementioned paper about the SMGL in Uganda and if I am not wrong you also compare changes in districts over time. So, I guess you clustered SEs at the PSU-level here and this is in general a valid approach to proceed (although groups are districts) ?
Thanks a lot for your time!
Till
[Updated on: Mon, 22 February 2021 14:29] Report message to a moderator
|
|
|
Goto Forum:
Current Time: Sat Aug 31 01:27:18 Coordinated Universal Time 2024
|