Interaction with continuous variables

[NomiW]

Hi All,
I have run a hierarchical linear regression model which includes an
interaction term between 2 continuous variables. The variables are centered
and the main effects are included in an earlier step of the model.
The interaction is significant but I'm not sure how to interpret this. Any
suggestion of a way to graph (or break down) this interaction would be
appreciated.
Thanks,


Nomi Werbeloff, PhD

=====================
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]

I just noticed this post now.  In a nutshell, what you want to do is save fitted values of Y at several selected combinations of the two interacting variables, and then plot them.  (This will yield a graph that is quite similar to what you'd get for a two-way ANOVA.)

For more information, I recommend that you check your library to see if it has the 1991 book by Aiken & West on interactions in regression models.  I also have some notes I could send.  If you want them, send me a message at the e-mail address shown in my signature file below.  (The hotmail address I use for posting to this list is not my everyday e-mail, and I do not check it very often.)

Aiken, L. S., & West, S. G. (1991). Multiple Regression: Testing and interpreting interactions. Thousand Oaks: Sage.

Oh, one other thing:  You said "the main effects are included in an earlier step of the model".  It's not clear if that means you took them out after adding the product term.  If you did take them out, you should put them back in.  For any two-way interaction, you should include both first-order terms; for any 3-way interaction, you should include all 3 first order terms, and all possible 2-way interactions involving those first-order terms.  And so on.

HTH.

NomiW wrote

Hi All,
I have run a hierarchical linear regression model which includes an
interaction term between 2 continuous variables. The variables are centered
and the main effects are included in an earlier step of the model.
The interaction is significant but I'm not sure how to interpret this. Any
suggestion of a way to graph (or break down) this interaction would be
appreciated.
Thanks,


Nomi Werbeloff, PhD

=====================
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]

I was also concerned when I read that statement, along with "hierarchical" regression. As Bruce said, make sure all lower-order terms are in the final model from which you will obtain the fitted values.

 

@Bruce: I was wondering what was taking you so long to respond to this post. This is right up your alley, my friend, especially given how quiet it is these days on SPSS-L...

 

:-)

 

Ryan

On Wed, May 1, 2013 at 9:54 PM, Bruce Weaver <[hidden email]> wrote:

I just noticed this post now.  In a nutshell, what you want to do is save
fitted values of Y at several selected combinations of the two interacting
variables, and then plot them.  (This will yield a graph that is quite
similar to what you'd get for a two-way ANOVA.)

For more information, I recommend that you check your library to see if it
has the 1991 book by Aiken & West on interactions in regression models.  I
also have some notes I could send.  If you want them, send me a message at
the e-mail address shown in my signature file below.  (The hotmail address I
use for posting to this list is not my everyday e-mail, and I do not check
it very often.)

Aiken, L. S., & West, S. G. (1991). Multiple Regression: Testing and
interpreting interactions. Thousand Oaks: Sage.

Oh, one other thing:  You said "the main effects are included in an earlier
step of the model".  It's not clear if that means you took them out after
adding the product term.  If you /did/ take them out, you should put them
back in.  For any two-way interaction, you should include both first-order
terms; for any 3-way interaction, you should include all 3 first order
terms, and all possible 2-way interactions involving those first-order
terms.  And so on.

HTH.



NomiW wrote

> Hi All,
> I have run a hierarchical linear regression model which includes an
> interaction term between 2 continuous variables. The variables are
> centered
> and the main effects are included in an earlier step of the model.
> The interaction is significant but I'm not sure how to interpret this. Any
> suggestion of a way to graph (or break down) this interaction would be
> appreciated.
> Thanks,
>
>
> Nomi Werbeloff, PhD
>
> =====================
> To manage your subscription to SPSSX-L, send a message to

> LISTSERV@.UGA

>  (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
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Interaction-with-continuous-variables-tp5719846p5719873.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