In a multiple regression, what means homoskedasticity of the error term?
1
All error terms ui have the same variance across all observations i = 1, 2, 3, ... ,n
2
The regressors have each the same variance across all random observations i = 1, 2, 3, ..., n
3
Two error terms ui and uj with j = 1, 2, 3, ..., n
4
The regressors and the error term are literally dependent