I need help interpreting an SPSS output. There is a significant interaction. Now what?

Kinda tough to comment here...:confused:

Can you give us the background of the study and what you're trying to accomplish, and the nature of the interaction?

We don't have ready access to specific software, so you'll need to elaborate for us....

JohnM

my 2 DV are initiating violence and receiving violence (continuous measures) the IV are gender and prior sexual victimization y/n....there is a significant interaction for both initiating and receiving violence but the plots only indicate crossed lines for the initiating violence so I am somewhat confused...

You don't need crossed lines for there to be a significant interaction - all you need is for them to be "non-parallel" (enough to cause a pair of end-points to be far enough away from each other to be significantly different).

What does the interaction plot look like for receiving violence?

Are you doing a MANOVA or two separate ANOVAs? If it makes sense, from the underlying behavioral theory, to combine "initiating violence" and "receiving violence" into 1 composite measure, then use MANOVA.

However, if it doesn't make sense, then do separate ANOVAs - one for "initiating" and one for "receiving."

Ohhh! Thanks so much. The end points are significantly far apart.

Are both pairs of end points significantly far apart, or just one pair?

The key to determining the presence of an interaction is the "non-parallel" nature of the lines.

Also, it can be a bit tricky with MANOVA - basically this procedure tries to find interactions or significant effects for "some linear combination" of the multiple dependent variables - MAKE SURE YOU UNDERSTAND THAT MANOVA IS COMBINING THE DVs!

Although this is a rather old posting, I had two thoughts when I read it:

- If amanning runs two separate ANOVA's, one for each DV, he wouldn't get an interaction, right?

- I'm not sure I understand the concept of the MANOVA using "1 composite measure" that combines the two DV's. Are you suggesting that the two DV's should somehow be combined into 1 score (as an average, a classification, etc.), and entered as 1 field in the dataset before running the test? Or are you just saying that, when SPSS calculates a MANOVA, it used both DV's (e.g., "combines" them) for calculation purposes, but the dataset still has 2 separate fields, one for each DV?

Just trying to take the rust off my stats concepts :)

Roberto