21 December 2017 11 6K Report

I encountered a problem in plotting the predicted probability of multiple logistic regression over a single variables. For example, my model is Prob = - 0.727 + -0.002*X1+ -0.022*X2+ -0.002*X3+ 0.001*X4+0.144*X5. I want to plot Prob over X4 only. As I can see it has positive coefficient, I expect to get a plot with increasing probability for increasing values of X4. But it is just reverse. It could be the result of small effect of X4 to offset larger (-) values of intercept and other variables but I am not sure. Could you suggest me that a variable with positive coefficient can result a inverse probability for a range of values of the variable while plotting them separately ?

I am not sure If I asked the question properly but hope that you would understand the essence.

Any suggestion would be highly appreciated

Lila

More Lila Puri's questions See All
Similar questions and discussions