Assuming you used Mplus, could you post your complete Mplus MODEL statement? Did you specify a latent class (latent profile) model at both levels (i.e., WITHIN and BETWEEN)? If you did, then you should get class counts (class proportions) for both levels. See the following Mplus example:

https://www.statmodel.com/usersguide/chap10/ex10.7.html

Dear Christian Geiser,

thank you very much for your prompt answer!

It helped a lot.

I guess we then misspecified the model.

Here is the complete MODEL statement for MPlus:

MODEL:

%BETWEEN%

%OVERALL%

ERday_ac1_mean-ERday_sab1_mean;

%bc#1%

[ERday_ac1_mean-ERday_sab1_mean] (1-12);

%bc#2%

[ERday_ac1_mean-ERday_sab1_mean] (13-24);

%bc#3%

[ERday_ac1_mean-ERday_sab1_mean] (25-36);

%bc#4%

[ERday_ac1_mean-ERday_sab1_mean] (37-48);

%bc#5%

[ERday_ac1_mean-ERday_sab1_mean] (49-60);

%WITHIN%

%OVERALL%

ERday_ac1_mean-ERday_sab1_mean WITH ERday_ac1_mean-ERday_sab1_mean;