Interaction Versus Moderation

I really do believe a lot of controversial findings are because people probe interactions out side of the range of the data that they observe. And so you have some continuous predictor z, and you automatically probe at plus or minus one standard deviation. What you don't know if you actually have cases at one standard deviation above or one standard deviation below,. when your distribution of your z predictor may not be symmetric at all. I think it's even further made a mess of because there's measurement error in our moderateces. Think there's a lot of things that are operating, and might be onfiscating our ability to detect these kinds of really important relations.