Can several thin reeds ever make a strong policy boat?

Thomas Cook and Vivian Wong, Northwestern University
The universal pre-K movement seems to be winning its political campaign, thanks in part to social science and to the dominant, empirically-supported theory in education that when students are truly and enduringly engaged in learning, individual achievement increases and the national stock of knowledge and national productivity prospers.

To increase 'engaged time on task', we cannot easily extend the school day or the school year, though attempts are being made to do both of these in some countries. Nor can we easily induce students to do more homework, or persuade teachers to deepen their students’ engagement in classroom learning, or get struggling students to respond better to 'College for All' rhetoric.

Of the few remaining possibilities, one is for children to begin their school career earlier. Empirical findings seem to support this option. Neurological results indicating greater brain plasticity in younger children suggest that the pre-K years deserve an especially high priority.

In addition, evaluations of many different pre-K programs have shown short-term cognitive gains even in random assignment studies. Some studies have even indicated social and economic benefits in adulthood. Pre-K would seem to be a robustly effective intervention whose long-term financial benefits even outweigh its costs.

But – and it's a big but – the studies indicating positive pre-K effects are not strong when examined individually. The Perry Preschool Project (1978), for example, involves a very small and local sample exposed to an unusually expensive intervention evaluated according to control group criteria that could not be reproduced today. Moreover, most of the program’s financial benefits are due to a few incarcerations registered during the current 20-year pro-imprisonment policy in the US, which may or may not continue.

Reynolds’s Chicago study (2001) depends on an opaque matching procedure and on data analyses that have routinely failed to recreate effect sizes similar to those of an experiment on the same topic. This implies the possibility of a selection 'confound' that has not been fully controlled. That's to say some of the difference attributed to the program under evaluation may instead be due to differences between the kind of children in the program versus those in the non-program comparison group.

The Abecedarian Project also involves a very local intervention even more intensive and expensive than Perry Preschool. Cognitive gains in the early twenties were indicated, but there was no clear evidence of reduced incarceration or improvements in the other adult outcomes assessed.

The national Head Start evaluation has a strong sampling and random assignment design, and short-term effects are evident in some domains. But they are patchy even in treatment-on-treated analyses, and we have no idea how the effects will hold up across elementary school, let alone into adulthood. We do have one long-term study of Head Start using a strong design (Ludwig & Philips, 2007). But this study was of the program as it was 40 years ago, and no long-term effects were observed for test scores, graduation rates or college enrollments, only for mortality. Short-term positive results have also been claimed for Early Head Start, but only after heroic analytic effort.

Finally, regression-discontinuity results show clearly that five state programs have raised achievement. But the five states have better than average pre-K programs. Effects were stronger for alphabet learning than for more general pre-reading or mathematical skills; long-term effects cannot yet be ascertained.

These findings are all the more limited because of a temporal mismatch built into almost all available long-term benefit-cost calculations. We are most interested in the long-term results from current programs implemented in the immediate future; but it is self-evident that such results cannot be directly observed. Instead, an indirect case has to be cobbled together from long-term studies implemented in a past that does not match today’s conditions, let alone any realistically imaginable future.

All pre-K policy has to be based on extrapolative leaps of faith from data and on other factors – principally the data themselves, education and human development theory and the political realities operating at any given time.

Fortunately, prevailing theory and current findings are at least consistent, leading us to revise our preconceptions and to believe that short-term cognitive effects of national pre-K are very likely and effects into adulthood are plausible. But we are not yet sure that these thin reeds from past empirical research can be woven together into a truly sturdy pre-K boat capable of weathering most future storms and consistently generating important long-term effects for children.

One such storm is already on the horizon. Latino children are currently under-represented in pre-K and would doubtless remain so under a universal pre-K program. Since Latino children already do very poorly throughout their school careers, the implication is that universal but voluntary pre-K may cause them to fall even further behind.

This would not only be due to lower enrollments. It would also occur if elementary school teachers raised their standards to accommodate the more numerous and better trained pre-K graduates they encounter. These pre-K graduates come disproportionately from non-Latino groups, and so the Latinos would be further disadvantaged as a population.

To hope that these teachers will not raise elementary school standards because of universal pre-K is perverse, for this would reduce the benefits to other students and the nation at large. Latino access to pre-K presents a serious problem that may get worse. Enrollment campaigns targeted at Latino families may reduce the problem, but they are unlikely to achieve what a mandatory pre-K program would. But mandatory pre-K opens up a large can of cacophonously strident political worms which advocates of pre-K would doubtless prefer to avoid.

• for a fully referenced version of this article, see The Benefits and Costs of Head Start [1].

See also:
The pros and cons of early years programs – where to start! [2]

Getting the measure of the 'mileage' of preschool care [3]