I am running the following linear regression equation as follows:
y_{id} = \alpha + \beta_1 * x_d + \beta_2 * z_i + \beta_3 * z^2_i + \beta_4 * z_i*x_d + \beta_5*z^2_i*x_d
where x_d is a dummy variable whereas y, z and z^2 are continuous. I get very large coefficient and stardard error estimates for beta_1. When I am not interacting the variables, the size of the coefficients make more sense. I would think that high coefficient size signals overfitting of the model, but the adjusted R square is low. What could be causing such large coefficient sizes?
Thank you in advance to anyone who has an idea.
y_{id} = \alpha + \beta_1 * x_d + \beta_2 * z_i + \beta_3 * z^2_i + \beta_4 * z_i*x_d + \beta_5*z^2_i*x_d
where x_d is a dummy variable whereas y, z and z^2 are continuous. I get very large coefficient and stardard error estimates for beta_1. When I am not interacting the variables, the size of the coefficients make more sense. I would think that high coefficient size signals overfitting of the model, but the adjusted R square is low. What could be causing such large coefficient sizes?
Thank you in advance to anyone who has an idea.
