Latent between level variable? PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 Xu, Man posted on Tuesday, June 18, 2013 - 12:33 am
I am trying to update some old model results from Mplus 5.1 for a manuscript. When I ran the same input file in the Mplus 7.1, I got the below warning:

"A latent between-level variable is created for an x-variable but not referred to on the between level. Please check that this is what is intended. If this is not intended, specify the variable as a within variable. Problem with: PVIQUAD"

The old syntax originally ran in Mplus 5.1 is pasted below. I was wondering if there is a way to reproduce the old results in Mplus 7.1? Thanks!

WEIGHT IS STUWT;
BWEIGHT IS W_FSCHWT;
WTSCALE = CLUSTER;

USEVARIABLES ARE
Zkst16 ZPVSCIE pviquad;
cluster is SCHOOLID;
missing are all (999);

ANALYSIS:
TYPE is TWOLEVEL;

MODEL:
%WITHIN%

Zkst16 ON ZPVSCIE (b_within);
Zkst16 ON pviquad;

%BETWEEN%

Zkst16 ON ZPVSCIE (b_betwn);
 Linda K. Muthen posted on Tuesday, June 18, 2013 - 7:42 am
You need to mention pviquad in the between level as a covariate or put it on the WITHIN list.
 Xu, Man posted on Tuesday, June 18, 2013 - 8:42 am
Thank you. I just tried putting pviquad at within level (within is pviquad;) - because it was not estimated at between level previously. The model ran, and had the same number of free parameters as in the above script from v5.1, but it still doesn't seem to give the same results as from v5.1.

Is this due to updates between versions?
 Bengt O. Muthen posted on Tuesday, June 18, 2013 - 11:31 am
I would have to see the old 5.1 output to see what was done then, but not mentioning pviquad on between runs the risk that it is not handled right. The new warning is therefore helpful.

I would think that you want to include pviquad as yet another predictor of Zkst16 on between given that context often matters.
 Adam posted on Tuesday, May 23, 2017 - 5:23 am
Dear Linda and Bengt,

I have a multilevel dataset. There is a level-1 dependent latent variable with 9 indicators which is collected from store employees and a level-2 independent variable with 15 indicators which is collected from store managers. I have no level-1 independent variable. All I want to do is to run a regression analysis where between-level latent variable predicts within-level dependent variable.

Since my independent and dependent variable are not in the same level, I receive the following error: “A latent variable declared on the within level cannot be used on the between level. Problem with variable: F1"

Could you please give me a hint how I can run this model? Below is the syntax I used;

VARIABLE: NAMES ARE store x1-x15 y1-y9
USEV ARE store x1-x15 y1-y9;
CLUSTER = store;
BETWEEN = x1-x15;
ANALYSIS: TYPE= TWOLEVEL RANDOM;
MODEL:
%WITHIN%
f1 BY y1-y9;
%BETWEEN%
f2 BY x1-x15;
f1 ON f2;

Kind regards
 Bengt O. Muthen posted on Tuesday, May 23, 2017 - 5:46 pm
Define a factor on between for y1-y9 with the same loadings as the within factor:

MODEL:
%WITHIN%
f1 BY y1-y9*(p1-p9);
%BETWEEN%
f2 BY x1-x15;
f1b by y1-y9* (p1-p9);
f1b ON f2;
 Adam posted on Sunday, June 04, 2017 - 10:07 am
Dear Bengt,

Thank you very much for your answer. With the modification you suggest, model runs without a problem.

I have a follow up question about the results. Factor loadings for f1 and f1b are different in the results. Is it normal? If it is, which one I should take into account?

Kind regards
 Bengt O. Muthen posted on Sunday, June 04, 2017 - 4:54 pm
They are the same in their unstandardized form.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: