Selection Bias and Covariate Imbalances in Randomized Clinical Trials