Brookings Papers on Education Policy 2005.1 (2005) 175-207
[Access article in PDF]
Education Reform and Content:
The Long View
E. D. Hirsch Jr
The most notable long-range study of educational programs ever attempted, conducted in the United States at a cost of half a billion 1970 dollars, was Project Follow Through. The study evaluated the effectiveness (through third grade) of several early childhood programs designed to educate disadvantaged students. The results show that programs like Direct Instruction, which emphasize explicit, systematic teaching of basic skills such as sounding out words and doing simple arithmetic, were much more effective than programs that put less time and effort into directly teaching those skills.1
It is a result that could have been predicted, since it has been a consistent finding of psychological research, well known in the 1970s when the Follow Through study was being conducted, that in learning, what comes out is highly dependent on what goes in, especially in the area of procedural skills. If basic skills are not emphasized, then mastery of those skills is unlikely to result. However, because the results did not fit the Romantic, naturalistic, antidrill prejudices of American experts in early childhood, the results were reanalyzed into oblivion and ultimately ignored, to the consternation of the proponents of Direct Instruction.2
Although the studies in Project Follow Through did not exhibit the methodological experimental rigor that is often demanded today, the results were highly credible on two grounds. First, and most important, they were consistent with what is known from the laboratory about human learning. Had the results been inconsistent with mainline psychology, they would have been more interesting but also more dubious. To call into question the what-comes-out-must-go-in principle regarding skills, a truly rigorous data gathering and [End Page 175] analysis—much more rigorous than Project Follow Through—would have been needed. Second, the effect sizes were quite large.
Today, the evaluation of educational programs is a significant industry. In contrast to the Follow Through studies, however, effect sizes in current studies tend to be smaller, and policy implications are correspondingly less certain.3 Moreover, the duration of current evaluative studies tends to be shorter than the three to four years of Project Follow Through. These two characteristics of current evaluation research, brevity and uncertainty, are connected. It takes time to achieve more significant and thus more certain differential effects in education, especially when evaluation reaches out into the middle school and high school years, during which academic achievement has to transcend the limited and explicit skills in reading and math that were probed in Project Follow Through. The need for longer-term and theory-driven evaluations of programs that are themselves long term and theory driven is a theme of this paper.
Educational Theories and Lack of Achievement in Later Grades
Besides the influence of Romantic naturalism that has dominated early childhood education, there may be another reason why so little attention has been paid to the results of Project Follow Through. The large academic differentials that could be detected in third grade when contrasting explicit and implicit instruction in basic skills did not result in large academic differentials in later grades. Subsequent evaluations have continued to show, as did Follow Through, the short-term effectiveness of explicit early programs in teaching the sounding out and writing down of words—programs such as Direct Instruction, Success for All, and Open Court Reading. After initial successes in early grades, however, these programs have not led to significant rises in reading achievement in later grades.4 The so-called fourth-grade slump in the performance of disadvantaged children still persists, ameliorated, no doubt, by these explicit early reading programs but with levels still essentially stagnant and, by general agreement, unsatisfactory.5 These good early programs have not produced the sort of long-range benefit in reading scores that was hoped for from the short-term benefits of explicit early instruction. There is, of course, every reason to persist in explicit early instruction in basic skills, because they...