Hi everyone,

I would like to have your opinion of the following. I am performing a regression on the change in news risk (independent) on weekly return (dependent variable).I would like to know how to visualize a positive underlying tendency for groups in the data (simpson's paradox) when this positive tendency does not outweigh the negative general tendency.

Dummies were included for ext1-ext11, ext1 respresenting the highest group of absolute changes in news risk and ext11 the lowest (dummies taking on 1 when group is equal to i with i taken from 1 to 11). When multiplying the dummies by the change in news risk variable, the coefficient is positive and more positive for higher absolute changes in news risk than for lower absolute changes as can be seen in the table. However, because the overall variable for change in news risk has a higher absolute coefficient than any of these positive coefficients, the overall effect and slope of these groups remains negative (as opposed to some graphs I found online). This is the reason why it is harder for me to visualize the positive tendency for groups behind the graph, simply because this effect does not outweigh the negative general tendency. For example the change in news risk has a coefficient of -30 and the first group (ext1) has a positive coefficient of 27. The positive coefficient does not outweigh the absolute value of the negative coefficient and hence the positive tendency is not visable in my graphs (when making a graph for a certain group).

Any ideas on how I can visualize this positive tendency ? And how I can come up with a formula that captures these positive tendencies of groups?
Array

** Note, i also included the dummies itself in the regression in the figure (without the multiplication but as my question is only about these variables, I only demonstrated these here so it is easier to compare)