Does any living statistician come close to the level of Donald Rubin in terms of research impact? Missing data analysis, causal inference, EM algorithm, any probably more. He just walks around creating new subfields.
ps. Gelman was Rubin's doctoral student.