I have a clustered model which I run regress on unbalanced panel data.
I am not sure which assumptions to test for... the cluster "allows" for 4) and 5) right? and I have written how I would test in Stata, but I'm hoping to hear if these are the right approaches.
1) Linearity - do a scatter plot
2) Collinearity: VIF
3) Zero conditional mean: can't really be tested?
4) White's test for heteroskedasticity
5) Durbin Watson test for autocorrelation/serial correlation
Code:
regress log_spending rand_plan_group2 rand_plan_group3 rand_plan_group4 rand_plan_group5 rand_plan_group6 `controls' demeaned_fam_start_month_site* demeaned_cal_year* if median_income_fam ==0, cluster(ifamily)
I am new to Stata and not the best at econometrics, so I apologize if these questions are quite basic.
0 Response to unbalanced pandel data model - assumptions
Post a Comment