Advocates and critics of universal government Pre-K seem to strongly disagree about what the research shows. Upjohn Institute economist and government Pre-K advocate Tim Bartik, for instance, claims to have a very different view of that research from Russ Whitehurst, an early education expert at the Brookings Institution who is critical of the case for universal government Pre-K.
At least some of that disagreement is illusory, because Bartik and Whitehurst are asking different questions. Bartik seeks to prove that at least a few high-quality early education programs have shown lasting success. Whitehurst wants to know about the long term effects of large scale Pre-K programs, particularly government programs.
Bartik is right that there are two early education programs in particular, High Scope/Perry and Abecedarian, that showed substantial long term benefits. But these were tiny programs operated by the people who had designed them and serving only a few dozen or a few score children. Since it is difficult to massively replicate any service without compromising its quality, the results of these programs cannot be confidently generalized to large scale government Pre-K programs.
In other words, Bartik is providing evidence that is largely irrelevant to the merits of universal government Pre-K, the policy he seems to be championing. Whitehurst and others focus on the results of large scale federal and state programs, because these are relevant to the present policy debate.
So far, there have been four randomized controlled trials of large-scale government Pre-K programs. The first two examined the same group of Head Start students, one observing them at the end of first grade and the other observing them at the end of the third grade. Both studies show initial effects enjoyed during the Head Start program to essentially vanish by the early elementary grades. The next examined Early Head Start and found much the same thing. The fourth looked at Tennesee's Pre-K program and found it to have a statistically significant negative effect (and the other, statistically insignificant point estimates were mostly negative as well).
When Bartik deals with the evidence on Head Start he juxtaposes the negative results of the gold standard randomized controlled experiments with several non-experimental studies of the same program. But Bartik selectively neglects to discuss the many other non-experimental studies of large scale government Pre-K programs that haven't shown lasting benefits. Indeed when the federal government reviewed a generation of non-experimental research in the 1980s, its own meta-analysis concluded that the consensus showed Head Start's effects fading out during the K-12 years.
To sum up, there is at best no favorable consensus among non-experimental studies of large scale government Pre-K programs, but there is a consensus among the more reliable experimental studies: program effects fade out by the elementary school years, sometimes by the end of kindergarten. That is the evidence that matters when discussing proposals for expanding government Pre-K.