using xtivreg: nonest option not allowed

[John Michael Perdue]

I am trying to cluster my standart errors (using xtivreg) on a variable that is not nested within a single cluster. After some research, I learned of the "nonest" option for accomplishing this. Nonest (seemed) to work fine with the "xtreg" command. However, in using the "xtivreg" I receive an error "nonest option not allowed". After further research I learned that the documentation on the "nonest" option is sparce at best.

I have learned that I can prefix my command with "Stat 9:" or something to that effect, but I have also learned that the predictions when using the "nonest" option are not always reliable.

Is there a protocol for data manipulation or otherwise when wanting to cluster on a variable that is not nested within a single cluster, i.e., is there anyway to work around the problem I'm having of wanting to cluster on a variable that is not nested within a single cluster?

[John Michael Perdue]

THis post is cross posted on cross validated.: https://www.stata.com/statalist/arch.../msg00126.html The answer is there but I'm having a problem still with the xtivreg:

I'm working with panel regression (fe = id year). I cluster on a regional variable, land, but me clusters are not nested. So I sorted out the ones that were not nesteed, i.e., where land observations vary in year (for id) using a a code posted by Nick Cox at Statalist [here]. I ran teh regression again, this time leaving out id observations where land varied over year, and it worked. However, when I tried the same using xtivreg, it did not work.
I'm wondering if this is due to differences in statistical requirements for the xtivreg.
Insights?