Selection of Sites

The site selection process was necessarily purposeful. Sites were selected because they were purported to be exemplars of the chosen strategies. Although the routes through which sites were identified varied, the common theme was nomination by persons regarded as experts within the program type, often the programs developer or a well-known disseminator. In most cases, after a program specialist had nominated a school, both the state and local Chapter 1 directors were asked to verify the exemplary status of the schools program. In the rare instances where state Chapter 1 coordinators expressed doubt as to the exemplary status of a nominated schools program, the sites were dropped and other sites identified. In 23 of 25 instances, selected schools received multiple nominations.

In retrospect, it is clear that not all nominated sites were instructionally exemplary within their program types.³ As with previous large-scale studies, this initially undesired natural variation was both a strength and limitation of the sample, depending on the question being addressed. It proved to be a strength regarding questions of implementation and the effects of various levels of implementation. However, the final sample posed limitations for researchers addressing questions of the effects of full-scale program implementation.

The Hopkins/Abt proposals called for three-year studies of cohorts of students in two schools of each type. In one case the longitudinal sample was expanded to three schools out of a concern, never realized, that a district desegregation settlement might eliminate the program during year one of the study. This resulted in a 25-school core sample. (See Appendix C for an overview of the 25 longitudinal sites.)

The grade of the cohort identified for each program is shown in Figure 2. Each schools cohort was followed over a three-year period beginning in the fall of 1990. For example, in the case of first graders, Special Strategies followed the majority of the students through the spring of their third- grade experience. A few students were retained in a grade for one year, and a smaller number "skipped" a grade and moved ahead. So long as students remained at their Special Strategies school, they were considered to be in their cohort, regardless of their academic progress over the three years. Due to limits on funding, students who transferred into or out of the 25 sites were not followed. In the Special Strategies sites and the larger Prospects samples, this typically resulted in the loss of a third to a half of initial samples.

In addition to the longitudinal sample, the design called for visiting multiple replication sites for each of the programs. The replicates provided valuable data on diverse contextual variables that may facilitate or impede implementation for specific strategy types.

No replication sites were arranged for either Schoolwide Projects or the Coalition of Essential Schools, because each program was represented by two or three sites in both the urban and the suburban/rural studies.

Computer-assisted instruction (CAI) provided a unique replication circumstance. Neither the Department of Education nor the Special Strategies research team wished to be perceived as endorsing a particular technological product line. Therefore, CAI replicates were deliberately drawn from non-CCC sites. Replicates included Jostens, HOTS, and locally developed combinations of writing, spreadsheet, database, and reading/writing/math skills software. This resulted in a total of four CAI replicates. Each was a different exploration on the theme of CAI.

An effort was made to maximize the number of sites representing each program. The study design included two to four replication sites for programs having fewer than four longitudinal sites. The result of the sampling scheme was four to six sites (longitudinal plus replication) for each of the major program types.

Figure 3 illustrates that the sites and replicates were drawn from across the country. Site locations ranged from near the Mexican and Canadian borders, from the West to the East Coast, and from the "rust belt" to the "sun belt." The sample included schools in which the majority of students were of Native American, Hispanic, African, Asian, or European origin.

  Figure 3: Special Strategies Case Study Site by State

A history of research related to compensatory education generally, and to promising programs for educating disadvantaged students particularly, was produced during the first year of Special Strategies (Slavin, Stringfield, & Winfield, 1990). The program-specific components of that review are updated in the Special Strategies Final Report (Stringfield et al., 1997). Those reviews indicated the following:

• Although Chapter 1/Title I has been a funding program, not a specific intervention, it has tended to have positive impact on students academic achievement. The size of Chapter 1/Title Is positive impact appears to have grown over time. ⁴

• Innovative programs, under certain replicable conditions, tend to have more positive impact than traditional Chapter 1/Title I programs.

• Considering the level of national interest in adopting educational innovations, there is a dearth of third-party research on most of the more widely discussed and implemented innovations.

  ³ Both the 1960s' "Planned Variation Studies" (Goodlad & Klein, 1970) and the more recent studies of the Center on Organization and
Restructuring of Schools (Newmann & Wehlage, 1995) experienced the same problem of reputational nominations producing widely varying
levels of actual implementation.  Weber (1971) found that upon actual obeservation, only four of seventeen repeatedly-nominated-as-exemplary
urban schools met two simple criteria.
  ⁴ For a meta-analysis of studies examining the effects of Title I generally, see Borman and D'Agostino (1995).