Dear all on Statalist,

I am posting to ask for your help on interpretation of marginsplot graph, especially on the issue of overlapping CI.

I am trying to explain how x1's influence to y changes as x2 gets larger. (x1, x2 are both continuous variables)

So here I got a marginsplot graph.

Array

Here are some of my interpretation and questions.

1. the regression output with interaction term of x1 and x2 shows insignificance for the interaction term. Should I just ignore it? Or should I pick an interesting value of x2 to show the effect of x1 under certain conditions?

2. the margin effects shows sometimes there is significance. So is it correct to say, the effect is significant when x2 equals to certain values but we cannot tell the result when x2 exceeds this range?
Array


3. I was told by one audience during my presentation that the CI should not be overlapped. I have been extremely confused on it.

4. it may not show on this graph. But when the line on the marginsplot graph keeps going and cross the x-axis becoming negative, how should I interpret the negative part on the marginsplot graph.


Thank you very much!