|
|
 |
| Syntax for LCA of complex survey data? |
 
|
| Message/Author |
|
|
| Jean Flores posted on Wednesday, October 03, 2018 - 11:43 am
|
|
|
| I'm wondering if a community member can share sample syntax that can be used for latent class analysis for complex survey data (adjusting for weight, cluster and strata). I essentially want to accomplish what was done in this example, https://stats.idre.ucla.edu/mplus/dae/latent-class-analysis/ , but need to also adjust for the complex survey design. My understanding is that I should use "complex", not "mixture" analysis type, but when I try this I am unable to use the "class" option to specify the number of classes or groups I would like to divide the survey sample population into. |
|
|
|
|
Here you go
variable:
names are u1-u6 s cl w;
categorical = u1-u6;
usevar are u1-u6;
strat=s;
cluster=cl;
weight=w;
classes = c(2);
data: file is StratSamp1.dat;
analysis:
type = complex mixture;
model:
%overall%
%c#1%
[u1$1-u6$1*-1];
%c#2%
[u1$1-u6$1*1];
Take a look at this relevant article
https://www.statmodel.com/download/Chi2Complex4.pdf |
|
|
| Jean Flores posted on Saturday, October 06, 2018 - 10:56 pm
|
|
|
| Thanks so much for your help Tihomir! I was able to use your syntax example to get my analysis back on track! |
|
|
| Back to top |
|
|
