Added-Variable Plot for Panel Data


New Member
Dear all,

in Frees (2004): "Longitudinal and Panel Data" there is a short description of a procedure for Added-Variable or partial regression plots for Fixed Effects models:
1) Run Y on X leaving out x_j
2) Run x_j on "the rest" of X
3) Plot the respective residuals.

My question: When I apply a fixed or random effects specification will I have to use it also for the second step (to regress x_j on X)? Or is this a simple regression? To my opinion it does not make sense to include fixed effects in step 2 as only explanatory variables are involved.
Do you have any ideas and/or references? Do you know of an application where they use an addad variable plot for panel data?
Any help is highly appreciated,

Kind regards