RS = 4.885 + β (M of A +1SD) + β (M of P +1SD) + β (A*P)

First, my output data is coming from SPSS. In the last part of the equation A*P is that the main effect mean of A times the main effect mean of P. OR is that the mean for the interaction term? I am pulling these means from the descriptives.

Also, should i be using standardized or unstandardized Betas?

P.S. I copied and pasted the formula as I am using where A and P are A/B. This is the 'high-high' one (have the other 3 variations, i.e. high-low, low-high, and low-low.

I don't understand your question. Are you trying to compute the predicted value and trying to figure out what to plug in for A*P?

And A & P are A/B - does that mean binary?

Karen

Sorry. Let me give it another shot. I thought in the regression formula the A*B is the mean for the interaction term.

What i was trying to understand is what value must be put in this place.

Is this the interaction mean as the output in SPSS generates or is this the mean of A's main effect multiplied by B's main effect. I suspect it is the mean of the interaction and not the multiplication of A main effect x B main effect.

Does that clarify my question? Also, when inputting the betas should these be the standardized or unstandardized or does it matter?

I have been going through Aiken, West, Cohen, and Cohen (2008) trying to better understand this but can't get this last part nailed down. :(

But what are you trying to do? You didn't answer that. Predict the mean value of Y at the mean values of A and P? If so, plug in the mean values of A and P.

If not, be explicit (to yourself, at least) about your aim.:)

Karen

I know what I am trying to do guess i just can't explain what i need. We are predicting an interaction for A-attachment and P-narcissism with the DV. So, my question is in evaluating the interaction part of the formula is the mean of P multiplied by the mean of A or is it the value in the SPSS output that indicates a mean for the interaction term A X P?

You say "plug in the mean values" does this mean after plugging in the values you multiply them? The confusion is coming with the multiple use of the asterisk, by, and X.

If this doesn't clarify then i guess i will try elsewhere. I don't know how else to explain it.

Hi Daniel,

So it seems you are trying to compute predicted values of Y, for some values of X (X the vector).

This is one of those situations where there is no wrong way, but one way will be easier to interpret.

I would suggest plugging in the means values for A and P (and the beta coefficient estimate), then multiplying them. Then multiply that product by their estimated beta coeffiicient.

Then the predicted value of Y will be the predicted value at the mean of A and P.

This predicted value with be the expected value of Y when both A and P are at their mean. Most interaction terms are skewed (at least if both component variables are on a positive scale), so the mean value of A*P is likely to be much higher.

And unless A*P actually means something in its own right (which it usually doesn't), you won't get a meaningful answer.

Karen (who hopes she made sense before coffee :))