For part c, given that we've only established E(ui) =0 and not E(ui|x) =0 in part B how can we know that the transformed model satisfies mlr. 1-ml4 and thus easily say the new estimator is unbiased without further technical proofs for part c.