This meta-analysis summarizes teaching effectiveness studies of the past decade and investigates the role of theory and research design in disentangling results. Compared to past analyses based on the process-product model, a framework based on cognitive models of teaching and learning proved useful in analyzing studies and accounting for variations in effect sizes. Although the effects of teaching on student learning were diverse and complex, they were fairly systematic. The authors found the largest effects for domain-specific components of teaching--teaching most proximal to executive processes of learning. By taking into account research design, the authors further disentangled meta-analytic findings. For example, domain-specific teaching components were mainly studied with quasi-experimental or experimental designs. Finally, correlational survey studies dominated teaching effectiveness studies in the past decade but proved to be more distal from the teaching-learning process. (Contains 11 tables, 1 figure, and 6 notes.)