Message/Author 

Arena posted on Thursday, November 08, 2012  12:32 pm



I am comparing two ways of measuring classroomsindividual student perception of the classroom vs. a classroom (cluster) average of student perception. I am using latent factors to measure individual student perception of the classroom. For the classroom average, I want to create a cluster average of the classmates' latent factor scores. I have tried using this syntax but isn't worked for me in the "Define" and "Model" sections: newvar = CLUSTER_MEAN (latentvar); Where does this command belong in the input file? Thank you in advance. 

Arena posted on Thursday, November 08, 2012  1:05 pm



Also, I would like to create an interaction term between the studentlevel latent factors. I tried using the XWITH command with two continuous latent factors, however that requires TYPE = RANDOM. I'm currently using TYPE=COMPLEX because I would like to take student nesting within classrooms into account and to adjust the standard errors. Is there a way of creating interaction terms between continuous latent variables without sacrificing the CLUSTER command or TYPE=COMPLEX? Thank you again, Arena 


The CLUSTER_MEAN option is for observed variables only. It is part of the DEFINE command. You can use TYPE=COMPLEX RANDOM. 

Arena posted on Sunday, November 11, 2012  6:17 pm



Thank you for the prompt responses. Is there another way to create cluster averages from latent variables? 


What you should do is create a factor on between using the between part of the factor indicators. This latent variable decomposition is described in Example 9.2. It is used in a CFA in Example 9.6. 

Back to top 