how to decompose forecast results of binary regression

[capuchin]

I am running some binary regressions and would like to plot the results by factor. This is to say, I want to create a chart showing that x% of the projected change is due to factor A and y% of the chance is due to factor B, as it is quite interesting in my case to be able to say what exactly is driving the changes in modelled chance.

I am running into the problem that the Representations tab of the estimated equation has the @CNORM function, which does not appear to follow the distributive property (i.e., X*(A+B) = X*A + X*B). Taking @CNORM of each regressor individually using the model's coefficients gives a much different shape than the results of the model's equation.

However, removing @CNORM from the equation gives a fairly similar shape as the results of the binary model, though these results sometimes dip much lower when the binary model's output is close to 0%. Is there a way to decompose a binary model's output or reasonably transform the @CNORM-free results? It seems like I could potentially impose a transformation causing the maximum to equal 1 and impose a rule that values below a certain level equal 0, but I would ideally like to avoid =if rules.

[EViews Glenn]

Hi. To be honest, I'm not certain that I understand the proposed exercise, in particular, what you mean by the "projected change" and what distributive property you want your results to follow.