Hello, I am fitting an HLM model using a complex survey (DHS) data . My data does not have Level 2 weights, hence I am unable to use - svy- commands like svy: melogit (which requires weights for all levels to be specified). Instead, I am using the plan - melogit - command
Code:
melogit condomless_spouse [pw=hiv05] || idhspsu:
. This produces the following result
Code:

Fitting fixed-effects model:

Iteration 0:   log likelihood = -2.213e+11  
Iteration 1:   log likelihood = -2.213e+11  
Iteration 2:   log likelihood = -2.213e+11  

Refining starting values:

Grid node 0:   log likelihood =          .
Grid node 1:   log likelihood =          .
Grid node 2:   log likelihood =          .
Grid node 3:   log likelihood =          .
(note: Grid search failed to find values that will yield a log likelihood value.)

Fitting full model:

initial values not feasible

One approach suggested here (https://www.stata.com/statalist/arch.../msg00906.html) is to fit a basic logit model and get the starting values therefrom. I fitted the model the following model, which seems to converge:
Code:
logit condomless_spouse sex [pw=hiv05]
mat a=e(b)
mat a1=(a,0)
xtmelogit condomless_spouse sex || idhspsu:,laplace from(a1, copy)
I want to be able to incorporate the survey weights in the melogit model.
Code:
 melogit condomless_spouse sex [pw=hiv05] || idhspsu:, laplace from(a1, copy)
, generates this error
Code:
invalid from() option;
option copy not allowed
I am wondering if there is anyway to include survey weights in the melogit command, when avoiding initial values problem.

Thanks - cY