Estimation of Causal Effects with a Binary Treatment Variable: A Unified M-Estimation Framework