BookmarkSubscribeRSS Feed
Choose Language Hide Translation Bar

Community Member


Mar 23, 2012

Mixed model random slope and covariance structure

I'm trying to migrate to jmp from R so I'm using the same platform as my colleagues.

I'm having trouble figuring out how to specify a mixed model with a random slope. Is this even possible in jmp?

The example I'm using has 4 variables; the indepentent variable is estimates number of jobs (JobsK), dependent variables are one continuous predictor (Time), and two categorical predictors (Rural, and County). My best guess for how to do this in jmp is,

Fit Model(

     Y( :JobsK ),




          :Rural * :Time,

          :County & Random,

          :County * Time & Random


     Personality("Standard Least Squares"),

     Emphasis( Effect Leverage ),

     Method( REML),


In R, this would be

lme(JobsK ~ Time + Rural + Time:Rural, data = AlabJobs,

           random = ~ Time | County, method = "REML")

I'm getting different results from the two programs, and I'm not sure where I'm going wrong.

Also, I can't figure out the default covariance structure jmp uses with multiple random variables. I assume unstructured. Does anyone know if this is documented, and where?

Thanks for any help.