error message using epctile when medians are zero

[Jen Mark]

I am using Stas Kolenikov's epctile command in Stata 17 with survey data in order to report weighted medians for continuous variables. I have first run the following command: svyset secluster2021 [pweight=coreimmfamweight2021], singleunit(scaled) strata(sestratum2021)

My issue is that when using the epctile command to report medians and IQRs for my continuous variables, I am getting the following error for only certain variables: estimates post: matrix has missing values

An example of where it works fine is the following: epctile healthstatus2021, p (25 50 75 ) svy over(signdebt_last2waves)
-healthstatus2021 is a Likert scale where values are 1-5.

An example of where it doesn't work and the error I get is when I run the following: epctile imptotalhealthexp2021, p (25 50 75 ) svy over(signdebt_last2waves)
The error is: estimates post: matrix has missing values

I have confirmed that there are no missing values, and I am fairly sure this is happening in situations where I have a lot of 0s (i.e., 25th & 50th % are both 0). I have tried previous suggestions to change slightly (i.e., 49.5 or 50.1), but that doesn't help in this situation.

Are there any suggestion for how to get this command to run?

Thanks.
Jen

[Andrew Musau]

In cases where a statistic cannot be calculated, you likely end up with missing values, and the command fails. I'm speculating here, but for a question like yours, it's important to provide a reproducible example using the dataex command. See

help dataex

Additionally, examining the output with:

set trace on

will likely help identify why and where the command fails.