Message/Author 

Jielu Lin posted on Wednesday, September 24, 2008  6:51 am



Dear Drs. Muthen, I am doing a twolevel growth model for a continuous outcome using long data format (like example 9.16). I want to obtain the residuals but the "output: residual" command is not available for "type = random". Is there anyway to get the residuals? Thanks! This is my code: variables: names = y id time; cluster = id; within = time; analysis: type = twolevel random; model: %within% s  y on time; %between% y s 


The variance of y varies as a function of time. Therefore, there is not one covariance matrix. This is why the RESIDUAL option is not available. 

Jielu Lin posted on Wednesday, September 24, 2008  9:30 pm



Thanks for answering my question! I did a fixrandom effects model in STATA using the "xmixed" procedure and could obtain a global residual using "predict r, resid".I want to replicate this in mplus. Just want to make sureDo you mean that mplus won't produce a global residual like the structural equation model growth curves? Thanks! 


You say your STATA analysis used "a fixrandom effects model". I don't know if that means fixed slopes only or also random slopes. If there are no random slopes for covariates, Mplus does give a global residual for the outcome at each time point. 

Jielu Lin posted on Friday, September 26, 2008  6:11 am



Got it. Thank you! 

Jielu Lin posted on Monday, September 29, 2008  11:07 am



New question: Now I am doing random intercepts only and as you indicated, this gives me a global residual. Is there anyway to get the actual numbers of the residual? I want to do a histogram to see how these residuals are distributed. Thanks! 


We don't give individual residuals. 

Back to top 