More Varying Covariate

[Justin Hulbert]

Hi all. I've been plugging away with Levine's guide to SPSS for ANOVA and various online
documents, but got stuck in various places. I'm hoping you could help out.


I took a repeated measures design with a varying covariate and turned it into a one-line-per-level
format, with participants as random subjects (see below syntax). While I can specify the correct
error terms in the /test lines and get the custom hypothesis test results, the /emmeans and /
lmatrix commands don't produce any results (it says they can't be computed).


Is there anything I'm doing wrong? More to the point, I would like to run simple effect contrasts
using the error terms specified in the /test lines -- not the incorrectly specified error terms
utilized in the standard between subjects effects table. Additionally I'd like to get the adjusted
means (adjusted for the covariate, 'cv').


For example, I'd to run simple contrasts on the 4 level RP factor at each level of the test factor
(e.g., rp0 vs rp1 within test1, rp0 vs rp6 within test1, etc.).  I'd also like to report the means of
RP0, RP1, RP6, and RP12 for Test1 and Test2, adjusted for the covariate.


Could I use a /contrast=simple(rp) command but then what would I specify in the /design line? /
design = rp within test(1), rp within test(2)? Do I have to specify all the other interactions? I would
like to get the 3-way interaction with ip_sp. My worry is that these contrasts won't use the correct
error terms.


Let me know if you need any additional information to understand/answer my question.


Any guidance would be greatly appreciated,


Justin Hulbert
[hidden email]




UNIANOVA
   dv  BY ip_sp conditio rp test particip  WITH cv
   /METHOD = SSTYPE(3)
   /random = particip
   /INTERCEPT = INCLUDE
   /CRITERIA = ALPHA(.05)
   /EMMEANS=TABLES(rp)
   /EMMEANS=TABLES(test)
   /EMMEANS=TABLES(rp*test)
   /EMMEANS=TABLES(ip_sp)
   /EMMEANS=TABLES(rp*test*ip_sp)
   /EMMEANS TABLES (test*ip_sp) COMPARE(ip_sp)
   /LMATRIX 'rp0 vs rp1 within test1_sp' rp 1 -1 0 0
     rp*test 1 0 -1 0 0 0 0 0
   /LMATRIX 'rp0 vs rp6 within test1_sp' rp 1 0 -1 0
     rp*test 1 0 0 0 -1 0 0 0
   /LMATRIX 'rp0 vs rp12 within test1_sp' rp 1 0 0 -1
     rp*test 1 0 0 0 0 0 -1 0
   /LMATRIX 'rp0 vs rp1 within test2_ip' rp 1 -1 0 0
     rp*test 0 1 0 -1 0 0 0 0
   /LMATRIX 'rp0 vs rp6 within test2_ip' rp 1 0 -1 0
     rp*test 0 1 0 0 0 -1 0 0
   /LMATRIX 'rp0 vs rp12 within test2_ip' rp 1 0 0 -1
     rp*test 0 1 0 0 0 0 0 -1
/test conditio vs particip(conditio*ip_sp)
/test ip_sp vs particip(conditio*ip_sp)
/test conditio*ip_sp vs particip(conditio*ip_sp)
/test rp vs rp BY particip(conditio*ip_sp)
/test conditio*rp vs rp BY particip(conditio*ip_sp)
/test ip_sp*rp vs rp BY particip(conditio*ip_sp)
/test conditio*rp*ip_sp vs rp BY particip(conditio*ip_sp)
/test test vs test BY particip(conditio*ip_sp)
/test conditio*test vs test BY particip(conditio*ip_sp)
/test ip_sp*test vs test BY particip(conditio*ip_sp)
/test ip_sp*conditio*test vs test BY particip(conditio*ip_sp)
/test test*rp vs test BY rp BY particip(conditio*ip_sp)
/test conditio*test*rp vs test BY rp BY particip(conditio*ip_sp)
/test ip_sp*test*rp vs test BY rp BY particip(conditio*ip_sp)
/test conditio*ip_sp*test*rp vs test BY rp BY particip(conditio*ip_sp)
/design = cv
conditio ip_sp conditio*ip_sp
rp conditio*rp ip_sp*rp conditio*ip_sp*rp
test conditio*test ip_sp*test conditio*ip_sp*test
test*rp conditio*test*rp ip_sp*test*rp conditio*ip_sp*test*rp
particip(conditio*ip_sp)
rp BY particip(conditio*ip_sp)
test BY particip(conditio*ip_sp)
test BY rp BY particip(conditio*ip_sp) .