finding the effect of X on the relationship between Y and Z

I had a huge explanation and it got deleted, so here's a briefer version:

I'm trying to find whether high trait empathy predicts a greater effect of self-other differences in status on state empathy and helping.

I have differences coded as -1/1 (diff/sim).

In the zero-order correlations, trait empathy predicts state empathy, and state empathy predicts helping, but lower victim status predicted greater helping overall. There was no correlation between status difference and state empathy or helping.

But based on previous research, it might be that higher empathy predicts more egalitarian helping OR more help to someone of similar status (higher trait empathy has predicted differential effects of the physical distance of a victim on helpfulness). Hence the original question.

I don't think it would make sense to do a partial correlation where trait empathy is controlled for to see if there are effects of status difference on helpfulness, since it doesn't show the difference between high and low trait empathy. I feel like this calls for a regression analysis of some sort-- but multiple or single?

Should I maybe set it up as a stepwise? I keep reading that you need to try and figure out the order to put them in but I'm not sure what that would be; I just need to see if there is a correlation between T. empathy and the effect of the victim on state empathy and helpfulness (and what direction that effect is in). I've also been told not to spit a scaled group because people in the middle are arbitrarily dichtomized, and I don't have enough in my n's to justify throwing some out for analysis.

So does anyone have any ideas? I would greatly appreciate it!