D

Randomization mainly protects internal validity by balancing known and unknown confounders and preventing selection bias, thereby making the comparison groups comparable; it does not, by itself, broaden how well the results generalize to other settings or populations (external validity) [Sessler & Imrey, 2015, PMID 26378705; Akobeng, 2008, PMID 18728521].