Graphing simple slopes of an interaction

[frankelgar]

I am attempting to plot a line graph of a regression slope representing the association between an IV (affluence) and predicted outcome (symptoms). I have had no luck so far. Attempts to produce even the simplest line graph results in gibberish. Doesn't seem to matter if I use one of the tutorial data sets or my own.
The screen shot linked below shows a discontinuous line graph, with uninterpretable slopes between each point of my affluence scale. There should be just one line yes?
What I really want is simple slopes of the 2-way interaction of 2 affluence variables, with 2 lines running across the graph. Attempts to do this result in filling the graph window with a solid blue, with lines going every which way. Perhaps I am missing something obvious.

Thank you!
Frank
https://www.dropbox.com/s/79ye6b11n1s20 ... nplots.jpg

[ChrisCharlton]

In your prediction you have chosen to include all the parameters, including the residuals (you can see which parameters are selected as they become grey when unselected and black when selected). This essentially means that your prediction is the same as your response. You can confirm this by comparing the data in your prediction column (c21) with the response column (psychosomatic). For what you want to do I believe that you need to choose a subset of the parameters to create predictions and then plot these. This is covered in section 4.2 of the MLwiN manual (http://www.bris.ac.uk/cmm/software/mlwi ... ection.4.2).