Message/Author 

yezi posted on Monday, April 16, 2012  12:01 am



Hi, How can add additional parameter in ESEM for factor loadings? Thank you! Best wishes! 


ESEM is EFA. You can't add any factor loadings. 

yezi posted on Monday, April 16, 2012  5:51 pm



Hi, Thank you! In CFA, I can set F1 BY x1x6*(p1p6) in the following example. How to do in ESEM? Thank you! DATA: FILE IS p.dat; VARIABLE: NAMES ARE x1x6; MODEL: F1 BY x1x6*(p1p6); F1@1; MODEL CONSTRAINT: new(H1); H1=(p1+p2+p3+p4+p5+p6); OUTPUT: CINTERVAL; 


You cannot put constraints on ESEM factors. Your model has only one factor so CFA and EFA are the same. 

yezi posted on Wednesday, April 18, 2012  6:00 am



Hi, Thank you! Is there a method to compute the sum of factor loadings in mplus software as the same as H1 in the CFA example? Thank you! Best wishes! 


I don't understand your question. 

yezi posted on Wednesday, April 18, 2012  7:54 pm



Hi, We can compute the sum of factor loadings in mplus software in the CFA example by setting (F1 "BY x1x6*(p1p6); MODEL CONSTRAINT: new(H1); H1=(p1+p2+p3+p4+p5+p6);" How can we do in ESEM? Best wishes! 


This cannot be done with ESEM. 

yezi posted on Friday, April 20, 2012  6:10 pm



Thank you! 

yezi posted on Monday, June 04, 2012  3:02 am



Hi, We can compute the sum of factor loadings in mplus software in the CFA example by setting (F1 "BY x1x6*(p1p6); MODEL CONSTRAINT: new(H1); H1=(p1+p2+p3+p4+p5+p6);" How can we do in ESEM? I eagerly need this function. I think that other researcher also need this function. Can you write code to achieve this function. Thank you! Best wishes! 


You cannot do this with ESEM. You could do it with EFA in a CFA framework. You will find the input for this in the Topic 1 course handout on the website. 

yezi posted on Tuesday, June 05, 2012  8:23 am



Hi, I can not find the input for this in the Topic 1 course handout on the website. Can you explain clearly? Thank you! Best wishes! 


Go to Slide 133 in http://www.statmodel.com/download/Topic%201.pdf 


I have a dataset where I want to correct for nonindependent observations, in this case siblings. On another thread, you wrote, "if the SEs are fairly similar to those of regular MLR estimation, you could then not bother with the clustering." May I ask if there is a quantitative guideline for "fairly similar"? Also, even if SEs and results of the model were similar, what would the explanation and advantage be for using the nonadjusted model rather than the adjusted model? 


I would say to check if the pattern of significance changes. If so, I would correct for nonindependence of observations. If not, you can go either way. 

Back to top 