|
|
3-level complex survey design estimate |
|
Message/Author |
|
|
Dear Drs. Muthen, Currently, I have a set of data collected under a complex survey design. Meanwhile, the data has 3 time points, and I hope to apply the growth curve model to model it. I also want to model the time variance nested within each individual, siblings nested within households, and households within neighborhoods. For each individual, I want to control for age and gender. So, I wrote: cluster= mapas fmid ; WITHIN=huid; Model: %WITHIN% iw sw | intsxs1@0 intsxs2@1 intsxs3@2; iw sw on age girl; intsxs1 on sensk1; intsxs2 on sensk2; intsxs3 on sensk3; %BETWEEN famid% ib2 sb2 | intsxs1@0 intsxs2@1 intsxs3@2; %BETWEEN mapas% ib3 sb3 | intsxs1@0 intsxs2@1 intsxs3@2; intsxs1 - intsxs3@0; The WITHIN level has both time-varying and time-invariance covariates? But the error message gives me : Unrestricted x-variables for analysis with TYPE=THREELEVEL must be a variable on only one level. I'm not sure how to work it out, and appreciate your help. |
|
|
The variables age girl sensk1 sensk2 sensk3 have to be on the Within= list. |
|
Back to top |
|
|