Can SPSS plot the interaction and compute the effect size for interaction terms?

Hi all,

I am trying to plot interactions for moderated regression and compute the effect size for interaction terms by following the recommendation by Aiken & West. But I couldn't find the function that helps me do that in SPSS. Is SPSS actually able to do this? Or if not, what could be the way to plot interactions, I mean, drawing graphs to show different slopes?

Thanks in advance.

[Bruce Weaver]

Here's one approach you could use for generating plots.

1. Create a new data set (via DATA LIST, for example) that has desired combinations of the explanatory variables, but nothing for the Y-variable.  It should also include a filter variable you can use to pick out these records later.

2. Append that data set to the raw data (via ADD FILES).

3. Run your model, saving the fitted/predicted value of Y.  

4. Use the filter from step 1 to pick out the records with the desired combinations of explanatory variables, and plot your graph.

If you run your model via UNIANOVA rather than REGRESSION, you can most likely get the same fitted values via a series of EMMEANS sub-commands.  (Note the use of WITH to set the values of any continuous variables.)

HTH.

[e-mail] wrote

Hi all,

I am trying to plot interactions for moderated regression and compute the effect size for interaction terms by following the recommendation by Aiken & West. But I couldn't find the function that helps me do that in SPSS. Is SPSS actually able to do this? Or if not, what could be the way to plot interactions, I mean, drawing graphs to show different slopes?

Thanks in advance.