Special Education in Restructured Schools: Findings from Three Multi-Year Studies

Abstract: Pull-out services for students with learning disabilities (LD) have come under increasing attack, although empirical evidence of their ineffectiveness is scant. Calls for alternative, innovative models of special education services prompted three research studies that used a common data base to permit aggregation of findings on achievement of students with LD in these restructured programs. In this article, the three models are carefully described, as is the achievement of students with LD during model implementation. Taken together, the findings from the three research studies suggest that general education setting produce achievement outcomes for students with LD that are neither desirable nor acceptable.

Special education for students with learning disabilities (LD) has had a relatively short, but always controversial, history. Before the 1960s, few public schools concerned themselves with students who, despite normal intellectual abilities and opportunities to learn, demonstrated significant problems in school achievement and manifested other behavioral symptoms (hyperactivity, distractibility, perceptual problems) that have come to characterize the student with LD. When programs were started in public schools (e.g., the demonstration project in Syracuse, New York directed by Cruickshank and his colleagues (Cruickshank, Bentzen, Ratzeburg, & Tannhauser, 1961), they reflected the traditional service delivery model of the times -- the self-contained special education classroom. Until the passage of P.L. 94-142, the Education of All Handicapped Children Act (1975), few states recognized LD as a handicapping condition that required the provision of special education services. But EHA changed all that. With a mandate to serve and with federal guidelines for diagnosis, publicly funded special education programs for students with LD became commonplace; the numbers of students classified as LD and provided with special education services in public schools rose from 797,212 in 1976-77 to 2,214,326 in 1991-92.

With this increase in the LD population came a shift away from the self-contained class as the preferred model of service delivery; most students with LD received their special education services in part-time special education settings, usually resource rooms. Data in the Annual Reports to Congress illustrate this trend. In the 1984-85 school year, 15% of students with LD received the majority of their education in regular classes and received special education and related services for less than a quarter of the school day (in the Annual Report to Congress, this category of service is referred to as regular class) (U.S. Department of Education, 1988). In the 1988-89 school year, the percentage of students with LD served in this way was up to 19.6% (U.S. Department of Education, 1991); and in 1990-91, up again to 22.5% (US Department of Education, 1993).

The movement away from self-contained special education placements was strongly advocated by Dunn, (1968) in his critique of special education. Dunn grounded his call for the elimination of special classes and special schools for students with mild learning problems in the evidence provided by Coleman (1966) and others that academically disadvantaged African American children in racially segregated schools made less progress than those of comparable ability in integrated schools. Dunn called for more integrated service delivery models for students with disabilities, as well, and envisioned pull-out, remedial resource rooms, staffed by special education teachers, as the way to provide "a better education" (p. 5) for children with learning problems.

Eighteen years later, these very pullout services were, themselves, subject to criticism. "Although well intentioned, thus so-called 'pull out' approach to the educational difficulties of students with learning problems has failed in many instances to meet the educational needs of these students,'" wrote Will (1986, p. 413), then Assistant Secretary of Education and head of the Office of Special Education Program. Will and others (e.g., Gartner & Lipsky, 1987,1989; National Association of State Boards of Education, 1992) called for even more fully integrated educational experiences for children with learning problems to achieve "improved educational outcomes" (Will, p. 413), "academic growth" (Gartner & Lipsky, 1989, p. 21), and "better outcomes" (National Association of State Boards of Education, p. 4).

Even before Will's call for a change in special education service delivery, essays on the failure of special education had begun to proliferate and the theme was constant: Pullout special education is not working; fundamental changes in the delivery model for special education are needed to increase the academic accomplishments of students with LD. The Heritage Foundation report, in May 1984, proclaimed that "gains of the remediated educationally slower students have been modest at best" (p. 6). Biklin and Zollers (1986) asserted that "students do not benefit from this [pullout] special education" (p. 581), although they based their conclusions on a review of efficacy studies of special classes, not part-time, pullout resource room programs. Hagarty and Abramson (1987) concluded from their analysis of only two studies (Bloomer, Bates, Brown, & Norlander, 1982; Glass, 1983) that "this split scheduling approach for providing services ... is neither administratively nor instructionally supportable" (p. 316).

McKinney and Hocott, in a 1988 article calling for special education policy analysis and program change, explained that "part of the rationale for totally integrated [as compared to pullout] programs for mildly handicapped students is based on research that questions the efficacy of special education" (p. 15). But how solid was the research evidence indicating that more integrated programs could achieve better academic outcomes for students with mild disabilities? Dunn's literature review (Dunn, 1968) had focused on research conducted with students with mental retardation, not LD, and on special class placements, not resource rooms. Carlberg and Kavale's (1980) meta-analysis of efficacy studies corroborated Dunn's findings with respect to students with mental retardation; calculations of effect sizes showed that students with mental retardation in special class placements performed as well, academically, as those placed in regular grades. But the findings for students with LD were different. Carlberg and Kavale reported a modest academic advantage for students with LD in special classes over those placed in the regular class (effect size of .29); furthermore, their 'special classes' included both self-contained and resource programs. Madden and Slavin (1983) reviewed 7 studies on the efficacy of part-time resource placements compared to full-time special education classes and full-time placement in the mainstream, but only 2 of the 7 involved students with LD. One of these two studies (Sabatino, 1971) was dismissed by Madden and Slavin as methodologically flawed; the second (Wang, 1982), which compared resource placement with full-time integration, was characterized as reporting "nonsignificant trends" (Madden & Slavin, p. 529). Nevertheless, Madden and Slavin concluded that if increased academic achievement is the desired outcome, "the research favors placement in regular classes . . . supplemented by well designed resource programs (italics added)" (p. 630); such sentiment was hardly an endorsement for the elimination of pullout programs.

But despite repeated acknowledgment that research on the relative efficacy of full-time mainstream placement of students with LD was scarce, methodologically flawed and inconclusive (see Hallahan, Keller, McKinney, Lloyd, & Bryan, 1988; Singer, 1988; Singer & Butler, 1987; Tindal, 1985; Walberg & Wang, 1987; Wiederholt, 1989), advocates for a new service delivery model for special education, embodied in the rhetoric of a Regular Education Initiative (REI) (Will, 1986), won the day. These REI proponents argued that pull-out special education services were expensive and questioned the economic feasibility of operating several categorical programs (special education, bilingual education, Chapter 1) simultaneously (Spec. Ed. Costs, 1988). They argued further that diagnostic criteria for placement of students into the various pullout categorical programs were unclear (Ysseldyke & Algozzine, 1983). They emphasized the purported social stigma attached to being segregated from the mainstream environment (Dunn, 1968; Foster, Ysseldyke, & Reese, 1975). But most significantly, they noted the limited progress made by many students with mild disabilities assigned to pullout programs (Budoff & Gottlieb, 1976; Epps & Tindal, 1987; Idol-Maestas, 1983; Leinhardt, Bickel, & Pallay, 1982; Polloway, 1984) and the lack of opportunities for learning in self-contained special education classes (Leinhardt, Zigmond, & Cooley, 1981) and in resource rooms (Haynes & Jenkins, 1986). REI advocates claimed that special education programs and services provided exclusively in the general education classroom would achieve better educational outcomes for student with LD.

Supporters of the REI established a legal foundation in the Least Restrictive Environment (LRE) provision of EHA and subsequently the Individuals with Disabilities Education Act (IDEA). LRE requirements stipulated that special classes, separate schooling, or other removal of children with disabilities from the regular education environment will occur only when the nature or severity of a child's handicap is such that education in regular classes with the use of supplemental aids and services cannot be achieved satisfactorily (Turnbull, 1986, p. 134). But the REI was not a civil rights movement, demanding equal access to mainstream education for students with LD. It was a call for alignment of programs to achieve better academic outcomes than had been achieved with more traditional, fragmented pullout programs.

In 1985, the Division of Innovation and Development (DID) in the Office of Special Education Programs (OSEP), U. S. Department of Education, through a set of long-term research priorities (Kaufman, Kameenui, Birman, & Danielson, 1990), challenged the research community to identify instructional, administrative, and organizational strategies for delivering special education services within the regular classroom. In 1988, the agency issued a call for school building models for educating children with disabilities, consistent with the REI philosophy. The R&D efforts funded under both of these initiatives were to judge efficacy in terms of students' academic outcomes in academic achievement.

The purpose of this article is to describe three innovative models for providing special education services for students with LD which were designed, implemented, and evaluated as a result of these DID/OSEP research priorities. As Fuchs and Fuchs (1990) point out, "In the past there has been a handful of well documented examples of school-based R&D, [each contributing to more effective educational practice in general and special education ...] but a study here and another there will not provide robust solutions" (p. 106). In an effort to enhance the interpretability of findings from single studies, the three research programs described in this article added a unique dimension: They developed a common data base that would permit aggregation of findings across the three studies to increase the significance and power of the individual research efforts. The three school building models represented three quite different approaches to the provision of special education services, but within each model, academic achievement was valued as the critical outcome of schooling and the indicator of effective special education services. Within each model, the question posed was, "To what extent can the academic needs of students with LD be met under innovative service delivery models?"

In the context of these three research projects, six schools developed and implemented three models for (a) altering general education classroom conditions that previously had necessitated the referral of students to special education, (b) returning students with disabilities from special education settings to general education, and (c) accommodating students with disabilities more effectively within those mainstream classrooms. The University of Pittsburgh model was implemented in one suburban and three rural elementary schools, scattered across Pennsylvania; the University of Washington model was implemented in one small-town elementary school; the Vanderbilt University model was implemented in one urban middle school. In Table 1: Site Demographics, we provide descriptive information on the schools in which these reform efforts occurred. In the following section, we describe the similarities and differences among the three models. Because the development of each model occurred over time, we selected one particular school year as the focus for this article. The results reported here were derived from the 1990-91 school year, which was the second year of the three-year Pittsburgh and Vanderbilt initiatives and the third year of the four-year initiative in Washington.

In all three research projects, the school reform effort was initiated by university faculty interested in implementing and evaluating alternative, integrated models for serving students with disabilities. The University of Pittsburgh project was funded by the Pennsylvania Department of Education, Quality Education Initiatives program after the model had been established in a single site using DID/OSEP research funds under the agency's 1986 priority. The University of Washington and Vanderbilt University projects were funded under the 1988 priority. In all cases, the university faculties approached schools to discuss the possibility of cooperatively developing the specific nature of the model at those sites. As they agreed to participate, the schools became partners with the university staffs in designing and implementing the models. All three models relied on a planning process that involved, cooperatively, school faculty, the principal, and university staff. The structure, frequency, and duration of the planning process, however, differed across sites. Following are descriptions of the treatments incorporated into the three models; Table 2: Model Dimensions organizes the treatments by model along six key dimensions.

In the University of Pittsburgh model, a yearlong planning stage preceded the implementation of change. Planning activities were designed to provide teachers and administrators with a knowledge base from which to make planning and instructional decisions and through which to alleviate fears of integration; planning sessions also served to actively involve the entire staffs of each building in anticipating problems and generating solutions. Planning sessions involved both inservice workshops and the work of committees organized in each building. Workshop topics included an orientation to the Pittsburgh model, information on the characteristics and educational needs of students with LD, effective reading strategies, techniques for monitoring and accommodating hard-to-teach students, and steps in a problem-solving model. Workshops were held for 1.5 hours before or after school, during half-days or full days designated for inservice during the school year, or on summer workshop days. Planning committees developed procedures for making mainstreaming decisions, administering curriculum-based assessment measures, administering climate surveys, and implementing a problem-solving team. Other committees recommended guidelines for grading students with LD in the mainstream, reviewing curriculum coverage, and improving the climate of the school. Each committee was composed of volunteer regular education teachers, special education teachers and administrators, and met 3 or 4 times during the second semester (see Baker, Ingold, Roemele, & Zigmond, in press, for a more complete description of the planning process). The culminating planning activity occurred at the end of the planning year when specific decisions were made to identify the best available classroom to which to return each student with LD the following Fall.

During the summer preceding the implementation of change, the school that teamed with the University of Washington allocated several full-day workshops to planning and preparing the school staff for the reform effort. One workshop was devoted to revising the mission/vision of the school, reviewing instructional strategies, and planning scheduling changes. During a second workshop, University staff prepared teachers for the instructional treatments to be introduced. To assist teachers in their implementation of instructional strategies, weekly staff development sessions were held in lieu of staff meetings. In addition, weekly intervention team planning and problem-solving meetings occurred throughout the year.

The Vanderbilt University school building model relied on six meetings with the school's principal, sometimes alone sometimes accompanied by key staff he had invited to attend, and on two full-day workshops during the summer prior to the implementation of change. These workshops were devoted to the identification of the school's needs and potential solutions. An 8-member Advisory Committee then was formed; it met twice during the academic year and once during each subsequent summer, each time for a full day. At these meetings, the Advisory Committee formatively evaluated the school-change process and reform efforts and identified alternative directions as necessary. To introduce instructional treatments to classrooms, frequent 45-minute inservices occurred at the beginning of the school day, and University staff provided technical assistance to teachers in their classrooms on a routine basis.

The shared purpose of the three school-building models was to increase general education's capacity to accommodate student diversity and to increase the meaningful participation and the achievement outcomes of students with LD within the general education structure of the school. The explicit goal for the University of Pittsburgh and University of Washington school-building models was to eliminate pull-out special and remedial education services. In line with this goal, both schools terminated all forms of pull-out service when the implementation stage of the project began.

Over time, the University of Pittsburgh model continued to serve its students with LD full-time in the mainstream. Within the University of Washington model, however, after the first year of implementation, some limited special services were reintroduced. The school scheduled daily 20 minute lessons in decoding for selected primary grade students; support teachers and aides taught these lessons to small groups during non-instructional times usually within the general education classrooms. In addition, two types of special education pullout services were introduced. First, the special education teacher or aide provided an extended day program in reading or a 'lunch-bunch' program in math. Second, the special education aide acted as a reading partner for two students during the classroom reading period, either in the classroom or in the hallway depending on the classroom activities on a given day.

In contrast to the explicit goal of returning all students with disabilities to general education settings immediately, the Vanderbilt University model adopted the goal of gradually decreasing the amount of time students with LD spent in special education classes. Although the expectation was that the amount of pullout services would decrease over time, it was never anticipated that pullout services would be eliminated completely. To support the goal of a gradual downsizing of special education pullout services, a deliberate process of reintegrating students on a case-by-case basis was instituted, as restructuring efforts within general education occurred. This reintegration process relied on (a) ongoing assessment and intensive instruction in special education to insure that the reentering student's skills approximated those of non-disabled, low-achieving students who were successfully accommodated within the general education classroom to which the student with a disability would return, (b) transenvironmental programming to increase the similarity of the instruction, curriculum materials, motivational strategies, and expectations for classroom behavior between special and general education settings, and (c) frequent, structured meetings between the special education and general education teachers. This case-by-case approach to reintegration typically took 2 to 3 months.

In the University of Pittsburgh and the University of Washington models, traditional special education services were almost, if not entirely, discontinued. This allowed special and remedial education staff to provide support to general education teachers. The two models differed, however, in terms of how they redeployed these newly available staff.

In the Pennsylvania schools, the special education teachers in each building, freed from pullout teaching responsibilities, assumed the role of co-teacher and co-planner of instruction delivered in general education. Depending on the school enrollment and number of special education staff available, the special education teachers spent 30 to 45 minutes, 4 to 5 times per week, co-teaching with each general educator who had a student with LD assigned to his/her classroom (and often with teachers who did not have a student with LD assigned). Co-teaching did not necessarily mean pull-in remedial instruction for the students with LD assigned to the classroom, although it sometimes took that form. More often, co-teaching provided opportunities for small group instruction that was more explicit, more closely monitored, and more strategic than was possible with only one teacher in the classroom. To prepare for this co-teaching, the special education teachers spent at least 30 minutes per week planning with each general educator in whose classroom they worked. In addition to co-teaching, and co-planning, special education teachers took leadership roles in the weekly problem-solving meetings attended by school faculties and in the school-wide implementation and interpretation of curriculum-based assessments.

The University of Washington model redeployed 1 special education teacher and 1 special education aide as well as 1.5 compensatory education teachers and 1 compensatory education aide in three ways. First, special education (and remedial) personnel assisted individual students or small groups in general education during reading, language arts, and mathematics lessons. Second, as mentioned, these teachers and aides taught daily 20-minute small group phonics lessons using Levels I and II of the Reading Mastery program (Engelmann & Bruner, 1988) to approximately 30 primary-grade students who had been identified by their classroom teachers as experiencing extreme difficulty in decoding and word recognition. These lessons were supplemental to the regular reading program, and were delivered within the regular classroom or, when grouping students across different classrooms, in other locations. Finally, with the help of the aides, the special education teacher(s) managed and supervised a cross-age tutoring program in reading. Working in the school's multipurpose room, fourth-grade students taught letter sounds or sight words and provided oral reading practice to primary-grade students as a supplement to their classroom reading instruction. In all three versions (pull-in, supplementary decoding, and cross-age peer tutoring), services were given to any student, with or without an identified disability, who was experiencing difficulties in the general education classroom.

The Vanderbilt University model encouraged several special education reforms. First, special education instruction was intensified, with an increase in the amount of individually-tailored instruction, in an attempt to increase the number of students who had skills that would permit reintegration. Second, as students were reintegrated and the direct service caseloads of special educators began to decrease, these faculty began to serve as consultants to general educators to help strengthen pre-referral intervention and reintegration. Pre-referral activities within general education were-aimed at increasing mainstream teachers' capacity to accommodate student diversity and at reducing the frequency with which referrals were made for testing and possible special education placement. At the same time, special and general educators worked together to terminate pullout services for many children already identified with special needs. As general educators shouldered some of the responsibility for implementing classroom modifications, it was expected that their capacity to cope successfully with other problematic students would increase. Finally, because not all students with disabilities became appropriate cases for reintegration, the Vanderbilt model developed and implemented "reverse mainstreaming" classrooms in which special and general education teachers were paired. The caseloads of these paired classrooms were mixed to reduce the general educator's, but increase the special educator's class size and to distribute students with disabilities and other low-achieving students equitably. The activities in these classrooms were restructured to rely more on peer-mediation, more active learning strategies, as well as more structured, validated methods for presenting new material.

Each of the three school-building models focused on restructuring mainstream instruction to increase teachers' capacity to accommodate learning activities that met a greater range of student needs. Some methods, including building-based problem-solving teems, peer tutoring, ongoing curriculum-related assessment, and organizational/study skills instruction, were common across two or all three projects; others were unique to individual models. Most strategies were based on validated procedures. Some however, had been researched only in general education settings and with non-disabled students, and little was known about their potential benefits for students with LD. Other strategies had been researched only in special education settings and little was known about how feasible they might be for use in general education.

Within the University of Pittsburgh model, once special education services were eliminated, mainstream teachers were encouraged to revise their classroom practices. Teachers modified their grouping patterns for instruction so students with special needs could receive special attention and so use of the special education co-teacher could be maximized. Teachers implemented more effective approaches to teaching literacy skills using graphic organizers and cognitive strategies (Calfee & Wadleigh, 1992). The schools' focus on content coverage was reduced, so that teachers could make professional decisions about what to teach more intensively, what to skip, and how to order the objectives of the curriculum. Teachers also rethought criteria for awarding grades to reward effort and individual differences in achievement and to reduce potentially punitive effects of grades for students with special needs.

In addition, the University of Pittsburgh model incorporated ongoing monitoring of students' reading progress using curriculum-based measurement (Deno, 1985). Assisted by special educators, the mainstream reading teachers assessed every student in their classes each week. Special educators reviewed the assessment profiles, identified students whose progress was inadequate, and met each week with mainstream teachers to plan instructional alternatives for problematic students. Curriculum-based assessment information, disciplinary actions, and demographic data were managed by computer to assist the school staff in identifying students at risk for failure and to facilitate development and documentation of action plans. Finally, the University of Pittsburgh model relied on school-based problem solving meetings in which teachers shared descriptions of problematic students and cooperatively generated solutions.

Modifications introduced into mainstream classrooms within the University of Washington model included Cooperative Integrated Reading and Composition (CIRC) (Stevens, Madden. Slavin, & Famish, 1987) and class-wide peer tutoring (Greenwood, Delquadro, & Hall, 1989); Active Mathematics (Good, Grouws, & Ebmeier, 1983); Skills for Success (Archer & Gleason, 1989); and classroom assistance by special and remedial teachers and aides.

CIRC relies on cooperative learning in combination with restructured basal reading and language arts texts; workbooks and worksheets are eliminated. CIRC employs teacher-led word reading, vocabulary, story introduction, and discussion activities; partner- and group-assisted silent and oral reading, story comprehension, word meaning, writing, and spelling activities; and direct instruction in reading comprehension and writing. Teachers used CIRC in Grades 2-6, conducting instruction as a whole-class activity without ability groupings. In addition, at Grades 5 and 6, teachers supplemented CIRC's partner reading, which was conducted on basal texts 2 days each week, with class-wide peer tutoring on the other 3 days using library materials.

Active Mathematics shifted the emphasis of math instruction in general education from teaching the spiral curriculum to teaching grade-level concepts to 80% mastery, instead; the entire staff spent several months carefully delineating the concepts and strategies to be taught at each grade level. Active math relied on the teacher to provide daily systematic review, mental math activities that developed quantitative thinking, and practice in problem solving. Teachers led direct instruction in concept development, guided practice using partners and teams, and assigned daily homework to which they gave immediate feedback.

At all grades, students were taught the School Behaviors and Organizational Skills strand of the Skills for Success program. Students in Grades 3-6 also mastered six learning strategies (e.g., memorizing/studying information, answering questions). Students in Grades 4-6 also learned textbook reference skills during library instruction.

Finally, classroom teachers received support during classroom reading and language arts from one or more special or remedial education teacher(s) or aide(s). Typically, in-class services from the support staff comprised assisting small groups or individual students in CIRC reading or writing assignments, pre-teaching aspects of CIRC lessons to selected students, and providing additional small group instruction in math for intermediate students. At weekly meetings, classroom teachers and support staff cooperatively planned individual adjustments for students.

Within the Vanderbilt University model, across the entire school, teachers taught an organizational/study skills curriculum during the reading instructional period for the first 2 weeks of school year. This curriculum was a version of Archer and Gleason's (1989) Skills for Success program, which was modified by the university and school staff cooperatively to reflect a standard set of expectations for task-related behavior for all teachers in the building. In every class, for example, teachers required the same heading structure for papers and the use of the same assignment calendar. At frequent intervals, the principal or his designee would conduct spot checks and provide rewards for student compliance with these standard expectations.

Within reading, math, social studies, and science classes, teachers relied heavily on a class-wide peer-tutoring structure that extended the work of Greenwood et al. (1989) to incorporate partner reading, as well as paragraph summary and story retell in reading, scripted coaching and intensive practice activities in math, and key topic discussion and vocabulary work in science and social studies. In reading, class-wide peer tutoring was conducted three times weekly; in math, social studies, and science, twice weekly. In addition, in reading and math, curriculum-based measurement was employed in conjunction with class-wide peer tutoring; a computer managed system generated suggestions based on this assessment information about how to differentiate the content of class-wide peer tutoring to address individual student learning needs. In addition, as already mentioned, in social studies and science, reverse mainstreaming occurred within six classrooms.

Finally, as with the other two school-building models, problem-solving teams met regularly, separately at each grade level, to plan additional adaptations for students with or without identified disabilities who were experiencing difficulty in the mainstream.

All three projects administered reading achievement measures in the Autumn and Spring of the 1990-91 school year. Because each project had selected different achievement tests, we agreed to use one common instrument to permit aggregation of scores across projects. This reading measure, taken from the Basic Academic Skills Survey (BASS) (Espin, Deno. Maruyama, & Cohen, 1989), consists of three passages. From each passage, every seventh word is deleted and replaced with three possible choices, only one of which fits the meaning of the passage. Students have 1 minute to complete the blanks in each passage; the BASS reading score is the average across the three passages of correct responses minus half the incorrect responses. Although brief and group administered, this test yields scores that demonstrate stability and correlate well with longer, norm-referenced achievement tests (Jenkins & Jewell, 1992).

Of interest across the three model programs was the reading performance of students with LD. The sample comprised those for whom there were Fall and Spring BASS data for the target year. This included 145 students: 95 in the four schools in the Pittsburgh project (grades 2-6), 13 in the Washington site (grades 2-6), and 37 in the Vanderbilt site (grades 5-6).

Our first analysis focused on the magnitude of the reading gains registered by students with LD, specifically whether their gains exceeded the standard error of measurement of the reading test. We judged gains that surpassed the standard error of measurement as indicative of material improvements in reading performance (i.e., indicative that students had made real growth). Over the three projects, roughly half of the students with LD (54%) achieved gains in excess of one standard error of measurement, ranging from 38% of the students in Washington, to 55~o in Pittsburgh, to 57% in Vanderbilt. Of course this also means that in one academic year, approximately half of our target group failed to register a gain in reading achievement that was larger than the error associated with the test.

Typically, the achievement level of children with LD is far below that of their peers. One goal, or hope, of special education is to prevent further widening of the achievement gap between them and their peers. A desired outcome is that students with LD, once they are diagnosed and assigned to special education services, will begin to 'hold their own.' Typical IEP goals for such students (e.g., "Johnny will make a year's growth in reading," "Suzy will make a year's growth in spelling") reflect an implicit acceptance of the idea that such students, with the help of a special education, can match the typical (or average) achievement rates, if not the levels, of children without disabilities.

Consequently, our second analysis focused on the extent to which education in the restructured schools prevented further widening of the achievement gap between students with LD and their peers. We first calculated mean achievement gains for each grade level within each site. Then we compared the achievement gain of each student with LD with the average gain of the relevant grade-level peer group. To maintain or narrow the gap, students with LD would have had to register gains that matched or exceeded the average gain of their grade level peers. Results (See Figure 1) showed that the percentage of students with LD who made average or better gains varied by site (i.e., 33% for Pittsburgh; 23% for Washington; and 54% for Vanderbilt) with an average of 37% across sites. Thus, despite the school-University efforts, a majority of students with LD (63%) did not register average size achievement gains. Even more disconcerting, we found that for 40% of our LD sample, gains were less than half the size of the grade level averages. In other words, 40% of students with LD being educated full-time or primarily in general education settings not only were failing to make average gains, but also were slipping behind at what many would consider a disturbing rate.

Figure 1: Percent of students with LD whose gains match or exceed average gains of grade level peers.

In the previous analysis, we examined the relative progress of students with LD in relation to average grade-level progress, using the magnitude of achievement gains as the index for comparisons. Another way to conceptualize achievement compares students' test standing relative to the peer group at the beginning as opposed to the end of the school year, to determine whether students actually start to catch up to their average-achieving peers. Using reading test performance, we computed autumn and spring standard scores (i.e., z scores) for each grade level at each site, separately, and then determined the %age of target students with LD whose achievement status (i.e., their relative standing in the grade-level peer group) had improved during the school year. Results revealed differences across the three projects, with 53% (Pittsburgh), 38% (Washington), and 62% (Vanderbilt) of students with LD gaining ground on their peers. Overall, approximately half of the target group (51%) had moved up in relative standing; the other half lost ground (i.e., their standard scores, derived from the reading measure, had fallen further from the mean of its respective population).

Concern among politicians, media pundits, educators, researchers, funding and regulatory agency officials, and the general public continues to grow that education in the United States is not meeting the needs of a large number of children and youth. Evidence sparking this concern seems to be all around us; reported in scientific literature and in the popular press: Large numbers of students fail to achieve basic academic goals whether the measure is achievement test scores, or SAT scores, or scores on high school competency tests (Montague, 1987); alarming numbers of students leave high school before graduation (Hammock, 1986); and more and more students are labeled LD and pulled out of the general education to receive part-time or full-time instruction through special education (Gerber, 1984).

At the same time, there is growing criticism of the large-scale general education pullout program aimed at students educationally at risk, namely Chapter 1. This Compensatory Education system was designed as part of President Lyndon Johnson's War on Poverty to help students who are both poor and not achieving well in school. It is a massive program that provides direct support to school districts that are heavily impacted by high numbers of poor and low-achieving students, and by most accounts, it has not been a success (Pitsch, 1993). Critics of Chapter 1 pullout programs question the lack of congruence of instruction delivered in pullout settings with that provided in the developmental reading or math program (Arlington & McGill-Franzen, 1988; Allington, Steutzel, Shake, & Lamarche, 1986; Johnston, Allington, & Afflerbach, 1985), the loss of instructional time in moving from classroom to pullout setting and back (Arlington et al., 1986), and the negative effects of labeling students for eligibility (Glass & Smith, 1977; Leinhardt et al., 1982). Reformers of Chapter 1 have called for an end to pullout compensatory or remedial education services and conversion of Chapter 1 service delivery to an in-class model.

In this climate of criticism of educational services, in general, and of Chapter l pullout education programs, in particular, part-time special education programs for students with LD have also come under attack -although as documented earlier, the conclusion of relative ineffectiveness does not hold up under careful scrutiny of the empirical literature. Proponents of - new, more fully integrated service delivery model for special education argue that pullout special education services, whether part time a full-time, have not provided suitable programs for students with LD.

Moreover, critics of special education have condemned past evaluation strategies which have focused on compliance rather than quality, on head counts rather than achievement gains, on "education inputs... (rather than) on the outcomes of instruction" (National Association of State Boards of Education, 1992, p. 17). "An appropriate education is determined by student outcomes," contend Gartner and Lipsky (1989, p. 30). "The quality of education provided to students now in special education must be scrutinized with greater care and the focus of the scrutiny must be on outcomes for students" (p. 29).

We took this challenge to heart, and, as Will (1986) suggested, focused our evaluation of the three model programs on "accumulating data on the efficacy of these new instructional approaches on the 'outcomes' side..." (p. 414). The three 'experiments' in school restructuring described in this article met the challenge of the DID/OSEP 1988 funding priority to establish and evaluate school-building models for educating students with disabilities in general education settings. All three models included "strategies for (a) assisting teachers in analyzing and solving instructional and behavioral problems, (b) managing classrooms to maximize academic learning time for students with and without disabilities, (c) providing appropriate instructional and learning opportunities for students with disabilities at different academic levels and with heterogeneous instructional and curricular needs, (d) consistently monitoring the progress of students and adjusting instruction based on the results of the monitoring, and (e) appropriately delivering special education and related services designed to meet the individual needs of students with disabilities within the general education setting" (Kaufman et al., 1990. p. 113). All three required large investments of time and resources for preparation, planning, training, technical assistance, and support. All three were able to win the cooperation of general education school personnel in a genuine restructuring effort. And importantly, all three defined results at least partly in terms of academic growth.

Taken together, the findings from these three studies suggest that general education settings produce achievement outcomes for students with LD that are neither desirable nor acceptable. The three analyses of BASS reading data indicate that only approximately half of the students with LD educated in general education settings made reliable gains in reading achievement, a minority of these children made average size achievement gains? around 40% made gains that were less than half the magnitude of the grade level average, and only approximately half improved their standing in the achievement distribution. By extension, for approximately half of the students with LD in the six schools, achievement outcomes after a year of fully integrated educational programs and services were unsatisfactory.

It is important to remember that these three projects invested tremendous amounts of resources, financial and professional, into the enhancement of services for students with LD in the mainstream setting. Despite this investment, the achievement outcomes were disappointing. Furthermore, these models did not answer the question about how best to provide services for students with serious learning problems, for to do so one would have to compare enhanced general education as was accomplished by these three projects, to pull-out services that have had equal resources invested for the purpose of intensifying programs. Such a "fair" test has yet to be conducted.

"The successful academic and social integration of exceptional students goes beyond the mere acquisition of basic academic skills " (Maheady & Algozzine, 1991, p. 68), but it must certainly begin there. Students with LD first appear in special education because they do not initially benefit from general education learning environments. The data aggregated in these three studies indicate that for a significant proportion of students with LD, enhanced educational opportunities provided in the general education setting also do not produce desired achievement outcomes. In addition, these data indicate that there are no simple answers to the question of where appropriate and effective programs and services for students with serious learning problems, such as those with LD, should be delivered. Clearly, the research evidence to date, including the data aggregated in the three research programs reported here, provides no basis for eliminating the continuum of services for students with LD and no basis for the conclusion that satisfactory outcomes can be achieved in the general education setting.

Allington, R. L., and McGill-Franzen, A. (1988). Coherence or chaos? Qualitative dimensions of the literacy instruction provided low achievement children. State University of New York at Albany, (ERIC Document reproductions service No. ED 292 060).

Allington, R. L., Steutzel, H., Shake, M., Lamarche, S., (1986), What is remedial reading? A descriptive study. Reading Research and Instruction, 26, 15-30.

Archer, A., and Gleason, M. (1989). Skills for school success. North Billerica, MA: Curriculum Associates.

Baker, J. N., Ingold, A., Roemele, A., and Zigwood, N. (in press). Preparations for mainstreaming: Insemce and planning activities for school personnel. Teacher Education and Special Education.

Biklin, D., and Zollers, N. (1986). "The focus of advocacy in the LD field". Journal of Learning Disabilities, 19, 679-586

Bloomer, R., Bates, H., Brown, S., and Norlander, K. (1982). Mainstreaming in Vermont: A study of the identification process. Livonia, NY: Brador Publications.

Budoff, M. and Gottlieb, J. (1976). Special-class EMR children mainstreamed: study of an aptitude (learning potential) x treatment interaction. American Journal of Mental Deficiency, 81(1), 1-11.

Calfee, R.C., and Wadleigh, C. (1992). How Project READ builds inquiring schools. Educational Leadership, 50, 28-32.

Carlberg, C. and Kavale, K. (1980). The efficacy of special versus regular class placement for exceptional children: meta-analysis. The Journal of Special Education, 14, 295309.

Coleman, J. S., (1966). Equality of educational opportunity. Washington, DC: USPGO.

Cruickshank, W., Bentzen, F., Ratzeburgh, T., and Tannhauser,3I. (1961). Teaching methods for brain injured and hyperactive children. Syracuse, NY: Syracuse University Press.

Deno, S. L. (1985). Curriculum-based measurement: The emerging alternative. Exceptional Children, 52, 219-232

Dunn, L. M. (1968). Special education for the mildly retarded -- Is much of it justifiable? Exceptional Children, 35, 0-22.

Engelmann, S., and Bruner, E. (1988). Reading mastery I. Chicago: Science Research Associates.

Epps, S. and Tindal, G. (1987). The effectiveness of differential programming in serving students with mild handicaps: Placement options and instructional programming. In M. C. Wang, M. C. Reynolds, and H. J. Walberg (Eds.) Handbook of Special Education: Research and Practice, Volume I: Learner Characteristics and Adaptive Education. Oxford: Pergamon Press, pp. 213-248.

Espin, C., Deno, S., Maruyama, G. and Cohen, C. (1989). The basic academic skills samples (BASS): An instrument for the screening and identification of children at-risk for failure in regular education classrooms. Paper presented at the annual American Educational Research Association Conference, San Francisco, CA.

Foster, G.-G., Ysseldyke, J. E. and Reese, J. H. (1975). I wouldn't have seen it if I hadn't believed it. Exceptional Children, 41, 469-473.

Fuchs, D., and Fuchs, L. (1990). Making educational research more important. Exceptional Children, 57, 102-108.

Gartner, A., and Lipsky, D. K (1987). Beyond special education: Toward a quality system for all students. Harvard Educational Review, 57, 367395.

Gartner, A., and Lipsky, D. K. (1989). The yoke of special education: How to break it. Washington, DC: National Center on Education and the Economy.

Gerber, M. M. (1984). The Department of Education's sixth annual report to Congress on P.L. 94-142: Is Congress getting the full story? Exceptional Children, 51, 209-224.

Glass, G. (1983). Effectiveness of special education. Policy Studies Review, 2(1), 65-78.

Glass, G. B., and Smith, M. S. (1977) Pullout in compensatory education. Unpublished manuscript, The University of Colorado.

Good, T. C., Grouws, D. A., and Ebmeier, H. (1983). Active mathematics teaching. New York: Longman Publishing Co..

Greenwood, C. R., Delquadro, J., and Hall, R. V. (1989). Longitudinal effects of classwide peer tutoring. Journal of Educational Psychology, 81, 371-383.

Hagarty, G. J., and Abramson, M. (1987). Impediments to implementing national policy change for mildly handicapped students. Exceptional Children 53, 315-323

Hallahan, D. Keller, C., McKinney, J. Lloyd, J. and Bryan, T. (1988). Examining the research base of the regular education initiative: Efficacy studies and the Adaptive Learning Environments Model. Journal of Learning Disabilities 21(1), 29-35.

Hammock, F. M. (1986). Large school systems' dropout reports: An analysis of definitions, procedures, and findings. Teachers College Record, 87, 324-341.

Haynes, M. C. and Jenkins, J. R. (1986). Reading instruction in special education resource rooms. American Educational Research Journal, 23(23), 161-190.

Heritage Foundation (1984, May 11). The crisis: Washington shares the blame. The Heritage Foundation Backgrounder, Washington, DC: The Heritage Foundation.

Idol-Maestas, L. (1983). Special Educator's Consultation Handbook. Rockville, MA Aspen Publications.

Jenkins, J. R., and Jewell, M. (1992) An examination of the concurrent validit of the Basic Academic Skills Samples (BASS).Diagnostique, 17, 273-288.

Johnston, P., Allington, R., and Afflerbach, P. (1985)., The congruence of classroom and remedial reading instruction, The Elementary School Journal, 85, 465-477.

Kaufman, M., Kameenui, E., Birman, B., and Danielson, L. (1990), Special Education and the process of change: Victim or master of education reform? Exceptional Children, 57 (2), 109-116.

Leinhardt, G., Bickel, W. and Pallay, "A. (1982). Unlabeled but still entitled: Toward more effective remediation. Teachers College Record, 84(2) 391-422.

Leinhardt, G., Zigmond, N., and Cooley, W. W. (1981). Reading instruction and its effects. American Educational Research Journal, 18, 343-361.

Madden, N. A., and Slavin, R. E. (1983). Mainstreaming students with mild handicaps: Academic and social outcomes. Review of Educational Research, 53, 519-569.

Maheady, L., and Algozzine, B. (1991). The Regular Education Initiative -- Can we proceed in an orderly and scientific fashion? Teacher Education and Special Education, 14, 66-73.

McKinney, J. D. and Hocott, A. M. (1988). The need for policy analysis in evaluating the Regular Education Initiative. Journal of Learning Disabilities, 21, 12-18.

Montague, W. (1987). Bennett's 'Wallchart' shows student gains have reached plateau. Education Week, VI(21). 1.

National Association of State Boards of Education, (1992). Winners all: A call for inclusive schools. Alexandria VA: Author.

Pitsch, M., (1993, December 1). Test changes for Chapter 1 are predicted. Education Week, 13(13), 1, 14.

Heritage FoundatioPolloway, E. A.. (1984). The integration of mildly retarded students in the school: A historical review. Remedial and Special Education, 5(4), 1828:

Sabatino, D. (1971). An evaluation of resource rooms for children with learning disabilities. Journal of Learning Disabilities, 4, 84-93

Singer, J. D., and Butler, J. A. (1987). The Education for All Handicapped Children Act: Schools as agents of social reform. Harvard Educationc Review, 57, 125-152. Singer, J. D. (1988). Should special education merge with regular education? Educational Policy, 2, 409-424. Special education costs about twice the price of regular education. (1988). Education of the Handicapped, 14 (25), 1-4.

Stevens, R., Madden, N., Slavin, R., and Famish, A. (1987). Cooperative integrated reading and composition: Two field experiments. Reading Research Quarterly, 22, 433-454.

Tindal, G., (1985). Investigating the effectiveness of special education: An analysis of methodology. Journal of Learning Disabilities, 18, 101-112.

Turnbull, H. R. (1986), Free Appropriate Public Education: The law and children with disabilities, Denver, CO: Love Publishing Co.

U.S. Department of Education. (1988). Tenth annual report to Congress on the implementation of The Education of the Handicapped Act. Washington, DC: Author.

U.S. Department of Education. (1991). Thirteenth annual report to Congress on the implementation of the Individuals with Disabilities Education Act. Washington, DC: Author.

U.S. Department of Education. (1992). Fourteenth annual report to Congress on the implementation of the Individuals with Disabilities Education Act. Washington, DC: Author.

Wang, M. (1987). Toward achieving educational excellence for all students: Program design and student outcomes, Remedial and Special Education, 8 (3),25-34.

Wang, M. (1982). Effective mainstreaming is possible -- Provided that ... Pittsburgh PA: University of Pittsburgh, Learning Research and Development Center.

Wiederholt, J. (1989). Restructuring special education services: The past, the present, the future. Learning Disabilities Quarterly, 12, 181-191.

Will, M. C. (1986). Educating children with learning problems: A shared responsibility. Exceptional Children 52(5) 411-415.

Ysseldyke, J. E. and Algozzine, B. (1983). LD or not LD That's not the question. Journal of Learning Disabilities, 16, 29-31.

At the time this research was undertaken, Naomi Zigmond was affiliated with the University of Pittsburgh, Joseph Jenkins with the University of Washington, Lynn Fuchs with Vanderbilt University, Stan Deno with the University of Minnestota, Doug Fuchs with Vanderbilt University, Janice N. Baker with Vanderbilt University, Linda Jenkins with the Mt. Vernon School District in Mt. Vernon, WA, Martha Couthino with Central Florida University.

The report was furnished to us by Joseph Jenkins. You can reach him via e-mail at jjenkins@u.washington.edu.

PA Department of Education
Quality Education Initiative to the University of Pittsburgh
#H023F8005 from OSEP
Department of Education to Vanderbilt University
#H023F800013 from OESP, Department of Education to Mt. Vernon School District, WA.

This information is provided by:
Office of State Superintendent of Public Instruction
Special Education
P O Box 47200
Olympia, WA 98504-7200
(360) 725-6088
Fax (360)586-1631
E-mail: dgill@ospi.wednet.edu

	Sites
	Pittsburgh				Washington	Vanderbilt
Variable	Site One	Site Two	Site Three	Site Four
Setting	Rural	Suburban	Rural	Rural	Rural	Urban
Per Pupil Expenditure	$3,341	$3,147	$2,647	$3,722	$3,651	$3,663
School Size	360	920	360	466	392	492
Level	Elementary (K-6)	Elementary (K-6)	Elementary (K-6)	Elementary (K-6)	Elementary (K-6)	Middle (5-6)
Number Sp Ed Pupils	13	44	35	22	18	48
% Ethnic/Racial Minority	3%	3%	0%	0.4%	27%	46%
% Reduced/Free Lunch	7%	9%	32%	40%	47%	52%

	Sites
Dimensions	Pittsburgh	Washington	Vanderbilt
Collaborative Planning Mechanism	Year Preceding Full Staff	Summer Preceding Full Staff	Summer Preceding Representative Committee
Ongoing Technical Assistance	In-Class After school inservices Summer workshops	In-Class After school inservices School day meetings	In-Class Before school inservices
Reintegration Process	All students reintegrated (beginning Year 2) Eliminate pull-out	All students reintegrated (beginning Year 1) Eliminate pull-out	Case-by-case reintegration gradually deceased Maintain pull-out
Modifications in Plan	None	Limited special services reintroduced Year 2	None
Redeployment Sp Ed Staff	Co-teaching Co-planning	Pull-in services Limited pull-out Year 2 Manage cross-age tutoring	Reintesification Sp Ed Prereferral/reintegration consultants Reverse mainstreaming
Mainstream Information	New grouping patterns More effective literacy teaching Reduced content coverage Revised grading CBM for entire schools Problem-solving teams Sp Ed co-teaching	CIRC CWPT Active Mathematics Skills for Success Pull-in services Problem-solving teams	Study Skills CWPT CBM for selected students Reverse mainstreaming Problem-solving teams