Hi all,

I am running a simple multilevel model to determine whether a group of people had a statistically significant change in an outcome variable over time. Each person was assessed at 5 time points. The outcome variable is continuous.

My idea is that I would run a multilevel model with a random intercept for the person and a random slope for time, using the following code:

Code:
. mixed dep time || person: time

Performing EM optimization:

Performing gradient-based optimization:

Iteration 0:   log likelihood = -191.64297  
Iteration 1:   log likelihood = -191.49105  
Iteration 2:   log likelihood = -191.48975  
Iteration 3:   log likelihood = -191.48974  

Computing standard errors:

Mixed-effects ML regression                     Number of obs     =         75
Group variable: person                          Number of groups  =         15

                                                Obs per group:
                                                              min =          5
                                                              avg =        5.0
                                                              max =          5

                                                Wald chi2(1)      =       0.02
Log likelihood = -191.48974                     Prob > chi2       =     0.8878

------------------------------------------------------------------------------
         dep |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
        time |   .0095594   .0677819     0.14   0.888    -.1232906    .1424094
       _cons |   78.29368   .7173816   109.14   0.000     76.88763    79.69972
------------------------------------------------------------------------------

------------------------------------------------------------------------------
  Random-effects Parameters  |   Estimate   Std. Err.     [95% Conf. Interval]
-----------------------------+------------------------------------------------
person: Independent          |
                   var(time) |   4.50e-12   5.45e-11      2.21e-22    .0917611
                  var(_cons) |   5.107832   2.402147      2.032018    12.83943
-----------------------------+------------------------------------------------
               var(Residual) |   7.128148   1.301701      4.983509    10.19573
------------------------------------------------------------------------------
LR test vs. linear model: chi2(2) = 17.59                 Prob > chi2 = 0.0002

Note: LR test is conservative and provided only for reference.
In this case, given the fixed variable for time was not statistically significant (p=0.89), I can conclude there was no (statistically) significant change over time in the outcome variable.


Does this sound correct?


Any comments much appreciated.


Thanks!

MJ