education assignment among the study sites and across time. Nevertheless, there are sufficient data to conclude that early intervention services for children living in poverty that are provided during the first five years of life can reduce subsequent rates of grade retention and use of special education services in middle childhood. The important research question is to determine why some programs are more successful than others. Comparable longitudinal studies have not been conducted on children with diagnosed developmental disabilities.
Assessments of school achievement provide another set of criteria by which the impact of early intervention services may be measured. Once again, the literature demonstrates positive program effects but the patterns of impact are variable and not detected universally. Graduates of the Abecedarian Project scored significantly higher than controls in reading and knowledge on the Woodcock-Johnson Test of Achievement at age 12 and in mathematics and reading at age 15 (Campbell and Ramey, 1994, 1995). Perry Preschool participants achieved significantly higher scores in reading, arithmetic, and language on the California Achievement Test (Schweinhart et al., 1993). Follow-up studies of children served by ParentChild Development Centers indicate positive trends in reading, vocabulary, and language on the Iowa Tests of Basic Skills, but the differences did not reach statistical significance (Johnson and Walker, 1991). At age 8, there were no overall differences on the Woodcock-Johnson Test between the intervention and follow-up groups from the Infant Health and Development Program, but the heavier of the low-birthweight intervention group had significantly higher mathematics scores than a matching subset of the control group (McCarton et al., 1997).
Taken together, the follow-up literature provides abundant evidence of intervention-control group differences in academic achievement during middle childhood, but no consistent or distinctive pattern of advantage associated with a particular type of preschool curriculum or program format. Moreover, the nature of the outcomes (i.e., grade retention, special education placements, and academic achievement scores) do not lend themselves to analyses that address questions regarding growth, as described in Chapter 4. Perhaps of greater concern is the possibility that the absence of reproducible patterns of outcomes across studies is a reflection of the extent to which published reports focus primarily on those variables for which statistically significant differences are found, with little attention given to the much larger number of measured outcomes that demonstrate no program-control differences. This criticism was raised by Locurto (1991) in an analysis of data from the Perry Preschool Project and the Milwaukee Project, which noted their mutually inconsistent and counterintuitive findings regarding the relation between IQ scores and academic achievement.