A large and flourishing literature documents the effectiveness of early childhood interventions on a variety of outcomes. It has become common practice to conduct meta-analyses of these interventions. Meta-analysis implicitly assumes that the programs synthesized are identical in terms of content and scope, populations targeted, and measurements taken, or can be adjusted to be so through conditioning on observed characteristics. Yet these programs are often very different in content and scope, populations targeted, and measurements taken. For the major programs shaping current policy discussions, there are no studies that document and compare (a) the exact nature of the curriculum and its actual implementation, (b) the qualifications of the staff, (c) the composition of the targeted populations, (d) the outcome measures analyzed, and (e) the rates of return. We seek support to make these comparisons using data on the four major early childhood programs with long-term follow-up up that are evaluated by the method of random assignment. Evidence from these programs shape the policy debate. We also seek support to conduct dynamic mediation analyses that examine how each program affected parenting, parent-child interactions, child cognitive and non cognitive skills and health, and the response of schools. A group of originators and early implementers of these programs, academics, and early childhood practitioners will participate in all stages of this activity as analysts and as commentators. A variety of methodologies will be utilized. The proposal builds on and will strengthen collaborations already in place. It will also inform and be informed by an ongoing independent study of the Head Start program (based on the Head Start Impact Study (HSIS) data) being conducted by some of the researchers in our group.