MIXED longitudinal contrast interactions

[Charla]

Help!

I'm constantly hoping that SPSS will evolve to allow for MIXED model contrasts before another PI presents with data that have primary aims involving longitudinal analyses with contrasts, but SPSS does not seem to be getting more flexible in this area.  

Here's my scenario:  We have a data set with 4 time points and 4 treatment groups and we are interested in the linear "GROUP by TIME" effect for each of the following GROUP contrasts:

GROUP Contrast(1)  -.5 -.5  0  1
GROUP Contrast(2)  -.5 .5 -.5 .5
GROUP Contrast(3)  .5 -.5 -.5 .5

I've tried many models- and this is the closest I can get:

MIXED OUTCOME BY GROUP WITH TIME
/FIXED GROUP TIME GROUP*TIME
/test= GROUP  -1 -1 0 2
/test=GROUP -1 1 -1 1
/test= GROUP 1 -1 -1 1
 /random intercept | subject (ID) /print solution testcov cps r g
  /SAVE PRED.

However, the contrast estimates seem to be based on the GROUP effect and not the linear GROUP by TIME effect, which is what is of interest.

Can you either advise me or direct me to a good resource on this type of model?

Thanks!

Charla Nich

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Bruce Weaver]

Have you seen this page on the UCLA website?

http://stats.idre.ucla.edu/spss/faq/how-can-i-test-contrasts-and-interaction-contrasts-in-a-mixed-model/

Charla wrote

Help!

I'm constantly hoping that SPSS will evolve to allow for MIXED model contrasts before another PI presents with data that have primary aims involving longitudinal analyses with contrasts, but SPSS does not seem to be getting more flexible in this area.  

Here's my scenario:  We have a data set with 4 time points and 4 treatment groups and we are interested in the linear "GROUP by TIME" effect for each of the following GROUP contrasts:

GROUP Contrast(1)  -.5 -.5  0  1
GROUP Contrast(2)  -.5 .5 -.5 .5
GROUP Contrast(3)  .5 -.5 -.5 .5

I've tried many models- and this is the closest I can get:

MIXED OUTCOME BY GROUP WITH TIME
/FIXED GROUP TIME GROUP*TIME
/test= GROUP  -1 -1 0 2
/test=GROUP -1 1 -1 1
/test= GROUP 1 -1 -1 1
 /random intercept | subject (ID) /print solution testcov cps r g
  /SAVE PRED.

However, the contrast estimates seem to be based on the GROUP effect and not the linear GROUP by TIME effect, which is what is of interest.

Can you either advise me or direct me to a good resource on this type of model?

Thanks!

Charla Nich

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Ryan]

Charla,

Here is some sample code.

 /TEST = 'Linear Trend Contrast c1 vs c2' Condition*Time -3 -1  1   3 

                              3  1 -1 -3

                              0  0  0  0

                              0  0  0  0.

 /TEST = 'Linear Trend Contrast c1 vs c3' Condition*Time -3 -1  1   3 

                              0  0  0  0

                              3  1 -1 -3

                              0  0  0  0.

HTH.

Ryan 

Sent from my iPhone

On Feb 28, 2017, at 10:13 AM, Charla <[hidden email]> wrote:

Help!

I'm constantly hoping that SPSS will evolve to allow for MIXED model contrasts before another PI presents with data that have primary aims involving longitudinal analyses with contrasts, but SPSS does not seem to be getting more flexible in this area.  

Here's my scenario:  We have a data set with 4 time points and 4 treatment groups and we are interested in the linear "GROUP by TIME" effect for each of the following GROUP contrasts:

GROUP Contrast(1)  -.5 -.5  0  1
GROUP Contrast(2)  -.5 .5 -.5 .5
GROUP Contrast(3)  .5 -.5 -.5 .5

I've tried many models- and this is the closest I can get:

MIXED OUTCOME BY GROUP WITH TIME
/FIXED GROUP TIME GROUP*TIME
/test= GROUP  -1 -1 0 2
/test=GROUP -1 1 -1 1
/test= GROUP 1 -1 -1 1
/random intercept | subject (ID) /print solution testcov cps r g
 /SAVE PRED.

However, the contrast estimates seem to be based on the GROUP effect and not the linear GROUP by TIME effect, which is what is of interest.

Can you either advise me or direct me to a good resource on this type of model?

Thanks!

Charla Nich

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD