dyadic analysis question using spss mixed models

[Todd Kashdan]

Hi all,

I am currently dealing with an issue while applying Kenny's Actor-Partner Independence Model with spss mixed models.  I am investigating an interaction effect between a binary IV (experimental condition) and a continuous IV(trait curiosity) on closeness felt during a social
interaction.  I used contrast coding to evaluate the interaction
effects and plot them (-1 for a small-talk experimental condition and+1 for an intimacy condition).  The syntax looks like this:

MIXED
 close BY condeffcoding WITH Zprecei Zp_precei
 /FIXED = condeffcoding Zprecei Zp_precei condeffcoding*Zprecei
condeffcoding*Zp_precei  | SSTYPE(3)
 /METHOD = REML
 /PRINT = SOLUTION TESTCOV
 /REPEATED = sex | SUBJECT(dyad#) COVTYPE(CSH) .

close is the DV, a continuous variable.
condeffcoding is the binary IV with effects coding
Zprecei is the centered personality trait IV (actor effect)
Zp_precei is the centered personality trait IV (partner effect)
and then there are the interactions

The question I have concerns the BY command in the first line of the syntax.  This allows for a binary IV with effects coding (-1 for a small-talk condition and +1 for an intimacy condition).  I am not
sure what spss is doing with the binary variables under the BY command and whether there is any cost to treating it under the WITH
command to examine the nature of the interaction.  I ask this
because an odd thing happened when I looked at the simple effects.
Following Aiken and West, I conducted two models with a dummy coded
variable for condition.  The first model had intimacy=0 and
small-talk=1 so I could examine the effects of personality in the
intimacy condition.  The second model had small-talk=0 and intimacy=1
so I could examine the effects of personality in the small-talk
condition.  Well, an odd thing happened and I double-checked
everything in the dataset.  All of the betas, in fact all model
parameters, were identical between the contrast effects coded model
(1=intimacy) and the dummy coded model when 1=intimacy.  This leads me
to believe that spss is focusing on 1 when the variable is under the
BY command.

any help would be incredibly appreciated so I can find the slope and intercept for each group in my model.

gratefully,
Todd

**************************
Todd B. Kashdan, Ph.D.
Associate Professor
Associate Editor- J of Personality
Associate Editor- J of Positive Psychology
Department of Psychology
George Mason University
Mail Stop 3F5
Fairfax, VA  22030
http://mason.gmu.edu/~tkashdan

Confidentiality Notice: If you have reason to believe that you are not the intended recipient of this communication, please contact the sender immediately, and delete it from your system.  Any unauthorized review, use, disclosure or distribution is prohibited.

********************************************

=====================
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