Estimate random slope with multiple v...
Message/Author
 Taylor B Scott posted on Monday, August 25, 2014 - 11:35 am
When conducting a multilevel regression, I computed a random slope that was essentially a regression between time 2 (the DV of interest) on time 1 and several control variables. Example below:
S | T2 on T1 cov1 cov2

I realized later that slopes are typically estimated between just two variables (e.g., S | T2 on T1) with no covariates. Is it incorrect to include multiple variables in the estimation of a slope?
 Linda K. Muthen posted on Monday, August 25, 2014 - 12:07 pm
The specification above creates three random slopes that are held equal. You could specify

S1 | T2 on T1;
S2 | T2 on cov1;
S3 | T2 on cov2;

or

S1 | T2 on T1;
T2 on cov1;
T2 on cov2;
 Taylor B Scott posted on Tuesday, August 26, 2014 - 8:46 am
Thank you for your quick response.

A follow-up question:
If I run the analysis that creates slopes using one command:
S | T2 on T1 cov1 cov2

My output includes only one mean for S.
Is that the mean of all slopes?
i.e., [(T2 on T2) + (T2 on cov1) + (T2 on cov2)]/3
 Linda K. Muthen posted on Tuesday, August 26, 2014 - 11:32 am
Yes, they are held equal so they constitute one random slope.
 Taylor B Scott posted on Tuesday, August 26, 2014 - 12:37 pm
Thank you!
 Taylor B Scott posted on Tuesday, August 26, 2014 - 1:03 pm
Thank you!
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: