Fgt_ci

[Mukesh Punia]

Dear members,

I am using FGT_CI to decompose the depth and severity of BMI.

I run this code.

FGT_CI BMIndx  social_grop_1 social_grop_3, rankvar(iwi) method(threshold) cutoff(24.99) power(2) modpart1(logit) modpart2(glm, family(gamma) link(log)) boot_reps(20)

The issue I am not clear about is when I use option

 boot_reps(20)

there is no warning. But any deviation

boot_reps(19)

Or

boot_reps(21)

there is Warning
Balancing within strata:
Warning: cannot balance for stratum 1 (n_h = 200, m_h = 199)
.

Rescaling weights
.....

Warning: the first-order balance was not achieved

Any thoughts on that?

[Carlo Lazzaro]

Mukesh:
is it that both strata should have the same number of observations to be balanced?

[Mukesh Punia]

I haven’t defined any Svyset.

Then how one should proceed with 600000 obs out of which 125000 are above cut off.

could you please elaborate more on strata here?

I can share raw data. I am using DHS data

[Carlo Lazzaro]

Mukesh:
I'd contact the author of the community-contributed modul -FGT_CI- and explain, with data, what's going on.

[Mukesh Punia]

Dear Prof. Carlo Lazzaro, Were we able to contact the author of FGT_CI?

I can share a dataex-created example here, but I think that will not help to clarify the issues as it needs STRATA and PSU to be included. An example with 100 or 200 observations will not capture the whole issue.

I can share full data but that is possible in personal communication due to data sharing policy.

Thank you!

[Carlo Lazzaro]

Mukesh:
just type -search FGT_CI- and open its help file.
Its author is: Marcel Bilger, Duke-NUS Medical School, Singapore (email [email protected]).

[Mukesh Punia]

Dear Prof. Carlo Lazzaro,

I have checked that, and Marcel Bilger is currently associated with the Vienna University of Economics and Business, Austria. I already mailed him & waiting for a reply.
I think the issue is related to bootstrap weight. So, discussing here.

If you can find some time, I can share data with you.

Thank you!
Mukesh

[Mukesh Punia]

Mukesh:
just type -search FGT_CI- and open its help file.
Its author is: Marcel Bilger, Duke-NUS Medical School, Singapore (email [email protected]).

I tried to send a mail, but the given mail ID is not valid!

[Carlo Lazzaro]

Mukesh:
I confirm that Prof. Marcel Bilger moved to Wien in the meantime ([email protected]).
There's no other email address to contact him.
Be patient and let's hope for his reply to your query.

[Mukesh Punia]

Dear Professor Carlo Lazzaro I mailed to Prof Marcel Bilger but not received any response so far. I also contacted his coauthor but he replied econometric part was done by Marcel Bilger.

I have cleaned the data and prepared a do file for fgi_ci. If I share the same with you can you find some time to check it.

Thank you!

[Carlo Lazzaro]

Mukesh:
I am not familiar with the community-contributed module -FGT_CI- and I cannot help.
That said, is there any other way to tackle your analysis?

[Mukesh Punia]

Dear Andrew Musau, can you please find some time to look into it? If needed I can share cleaned data and and .do file.

Thank you!

[Andrew Musau]

Dear Andrew Musau, can you please find some time to look into it? If needed I can share cleaned data and and .do file.

I have no experience with this command and therefore I'm unable to provide any advice.