Planned contrasts for linear mixed model

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

Planned contrasts for linear mixed model

nanditav
Dear SPSS experts,

I'm running a linear mixed model investigating time (5 levels) by group (2 levels) interactions. Time is modelled as a repeated measures categorical factor. I am trying to create a contrast that investigates whether the change between time 1 and 5 is significantly different between group 1 and 2.

I have tried to use the following syntax

/test time*group -1/2 1/2 0 0 0 0 0 0 1/2 -1/2

but it does not work, stating that the L matrix is not estimable. I have no idea if I've used the correct syntax, or if another error is occurring. Any help would be really really appreciated.

Thank you!

Nandi

 
Reply | Threaded
Open this post in threaded view
|

Re: Planned contrasts for linear mixed model

Ryan
Your design matrix is 2(groups)x5(time points):

g1t1 g1t2 g1t3 g1t4 g1t5
g2t1 g2t2 g2t3 g2t4 g2t5

You want to construct the following contrast:

[mean(g1t1) - mean(g1t5)]
-
[mean(g2t1) - mean(g2t5)]

Therefore, the coefficients for this contrast would be:

(1 0 0 0 -1) - (1 0 0 0 -1)

If we want to eliminate the parentheses, we could apply the distributive property, resulting in:

1 0 0 0 -1 -1 0 0 0 1

Therefore, the TEST statement should be:

/test 'grp diff in change from t1 to t5' group*time 1 0 0 0 -1 -1 0 0 0 1

Best,

Ryan

p.s. I'm responding by phone so apologies for any typos/spacing issues.

Sent from my iPhone

> On Aug 22, 2014, at 12:30 AM, nanditav <[hidden email]> wrote:
>
> Dear SPSS experts,
>
> I'm running a linear mixed model investigating time (5 levels) by group (2
> levels) interactions. Time is modelled as a repeated measures categorical
> factor. I am trying to create a contrast that investigates whether the
> change between time 1 and 5 is significantly different between group 1 and
> 2.
>
> I have tried to use the following syntax
>
> /test time*group -1/2 1/2 0 0 0 0 0 0 1/2 -1/2
>
> but it does not work, stating that the L matrix is not estimable. I have no
> idea if I've used the correct syntax, or if another error is occurring. Any
> help would be really really appreciated.
>
> Thank you!
>
> Nandi
>
>
>
>
>
> --
> View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Planned-contrasts-for-linear-mixed-model-tp5727058.html
> Sent from the SPSSX Discussion mailing list archive at Nabble.com.
>
> =====================
> 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
Reply | Threaded
Open this post in threaded view
|

Re: Planned contrasts for linear mixed model

nanditav
Great, thanks for your help Ryan!

Nandi