Fitted multiple regression line with one continuous and many dummy variables in the regressors

[Naveed Ahmed Usask]

Hello everyone,

I am working with the following model -

_Yi = β₁ + β₂ Xi + β₃ Z_i + β₄ P_i + β₅ Q_i + β₆ S_i + ε_i

where Y is a binary dependent variable and among the independent variables - only X is continuous and the rest are dummies (Z, P, Q and S). I am working with Linear Probability Model, Logit and Probit model.

What I am trying to do is

a) plot the fitted values of X using its coefficients from LPM model. Here all of the dummies will be at their respective averages. Next in the same plane, I want to plot X again using its coefficient from LPM but here only one of the of the dummies will take the value of 1 (i.e. only Z) and rest will be at their averages (P, Q and S).

b) Next in a similar manner, I would like to plot the fitted values of X using its marginal effects from Logit model. Here again all of the dummies will be at their respective averages. Next in the same plane, I want to plot X using its marginal effects from Logit but similarly here, only one of the of the dummies will take the value of 1 (i.e. only Z) and rest will be at their respective averages (P, Q and S).

Can anybody please help me with the commands? I am using STATA 14.2.

Thank you very much.

[Phil Bromiley]

You need to look at the margins command and marginsplot.