Testing Group Invariance in PLSEM - Stats subcommand not running

[May Smith]

Hello,

I am using -plssem- to run an sem model with group invariance in a sample. The sample has missing data, and while there is a -missing- option in the plssem command, for some reason, the -group- subcommand does not run the -stats- part of the code. Here is an abbreviated example of the code:

plssem (LV 1 > x1 x2 x3) (LV2 > x4 x5 x6) (LV3 > x7 x8 x9), ///
structural (LV1 LV2 LV3, LV2 LV3) ///
group (groupname) ///
missing (knn) k (5)///
stats correlate(lv)

One solution to not getting the descriptive stats I need is to use -by- before the plssem command. I get the descriptive stats I need but I am not sure how to test for group invariance between the two different models. This code is:

sort group
by group: plssem (LV 1 > x1 x2 x3) (LV2 > x4 x5 x6) (LV3 > X7 x8 x9), ///
structural (LV1 LV2 LV3) ///
missing (knn) k (5)///
stats correlate(lv)

I am not clear on: 1) why they first code does not provide descriptive stats in the code 2) if possible and not too labor intensive, how can I test for measurement and structural invariance witht the second code? I know -plssem- is a user written command, but I thought folks here might be able to help, especially with the second part of the question.

Any suggestions would be appreciated.

[Nick Cox]

plssem is from SSC. I have never used it. (The name is as in your code, not in your title.)

There isn't here a reproducible example. Your syntax here is unlikely to run as written as there are spaces between some option names and the arguments.

 structural (LV1 LV2 LV3, LV2 LV3) group (groupname) missing (knn) k (5)

However, if i understand the question correctly, you are not getting syntax errors, but just less output than expected.

This is a cross-posting of https://www.reddit.com/r/stata/comme...n_stata_stats/ to which there are no useful answers as I write, Please note that it is a rule there on Reddit and a request here on Statalist to tell people about cross-posting. This is so that (1) people who might answer, even with feeble advice as you're getting from me, can see what has already been said (2) people interested in reading an answer can again see what has been said elsewhere.

I would reach out to the program authors for advice.