0
$\begingroup$

I'm looking at data from a health intervention study done in one middle school (16 classes/"clusters" at that school). Half assigned to control, half to intervention.

Is this an appropriate way to include the classes/clusters in the mixed model syntax in SPSS? I am unsure how to include this in the model. I have read that it would be a random effect, but I have also seen "schools" used as a fixed effect in others' models.

DV1 = my main dependent variable
Time = 4 time points
RandomizedGrp = 0 (intervention), 1 (control)
Class = coded 1-16 individual classes
(x) used in some places to denote multiplication

MIXED DV1 BY Time RandomizedGrp Class

  /CRITERIA=CIN(95) MXITER(100) MXSTEP(10) SCORING(1) SINGULAR(0.000000000001) HCONVERGE(0, 
    ABSOLUTE) LCONVERGE(0, ABSOLUTE) PCONVERGE(0.000001, ABSOLUTE) 

  /FIXED=Time RandomizedGrp Class Time(x)Randomized Time(x)Randomized(x)Class  | SSTYPE(3) 

  /METHOD=ML 

  /PRINT=SOLUTION TESTCOV 

  /REPEATED=Time | SUBJECT(ID) COVTYPE(UN) 

  /EMMEANS=TABLES(Time) 

  /EMMEANS=TABLES(RandomizedGrp)

  /EMMEANS=TABLES(Time*RandomizedGrp)

   /EMMEANS=TABLES(Time*RandomizedGrp*Class).
$\endgroup$
0
$\begingroup$

Are classes a (presumably random) sample of classes from a larger population, or all the classes of interest? If the former, typically class would be treated as a random effect. If the latter, typically it would be treated as a fixed effect (as in your MIXED command).

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.