What's happening in self-contained special education classrooms?
Ability grouping in education (Evaluation)
Special education (Research)
Algozzine, Bob
Morsink, Catherine V.
Algozzine, Kate M.
Pub Date:
Name: Exceptional Children Publisher: Council for Exceptional Children Audience: Academic; Professional Format: Magazine/Journal Subject: Education; Family and marriage Copyright: COPYRIGHT 1988 Council for Exceptional Children ISSN: 0014-4029
Date: Nov, 1988 Source Volume: v55 Source Issue: n3

Accession Number:
Full Text:
What's Happening in Self-Contained Special Education Classrooms?

In recent years, more than 4 million handicapped students received all or part of their education in classrooms directed by a specially trained teacher. Conventional wisdom holds that categories used to classify individuals as elgibile for special services represent mutually exclusive groups of people and serve as the basis for some type of differentiated treatment (cf. Hallahan & Kauffman, 1986; Kirk & Gallagher, 1986). Hallahan and Kauffman (1986, p. 5) defined special education as "specially designed instruction that meets the unique needs of an exceptional child." Educators have reasoned that special materials as well as special teaching techniques, equipment, or facilities are required for special education to be effective.

Clearly, there is a need for specially designed instruction for some exceptional students. For example, it is difficult to imagine not providing specialized classroom interventions for individuals who are blind or deaf; in fact, special education programs for these people are the oldest, and maybe the best developed. Identifying the unique needs of learning disabled, emotionally handicapped, or mentally retarded students--or their educational treatments--is not so simply accomplished. There are many reasons why a student does not achieve commensurate with ability, or why abilities differ, or why a student fails to demonstrate expected ability, achievement, or adaptive behavior. Most of these reasons do not lead directly to specific interventions (Ysseldyke & Algozzine, 1984).

This failure to identify categorically specific characteristics or treatments has led some to argue for a decrease in categorization. For example, Hallahan and Kauffman (1977) provided "logical justification for considering children traditionally falling into three categories of mildly handicapped--learning disabled, emotionally disturbed, and educable mentally retarded--within a behavioral rather than a categorical framework" (p. 139). Their argument was supported by discussion of the confusing, imprecise definitions that are used to develop eligibility criteria in most states, the common historical perspective evident in studying the development of each of the mild handicaps, and the overlap of behavioral characteristics among these three groups of exceptional students. Similarly convincing rhetoric has been offered by others (cf. Gardner, 1977; Hewett & Forness, 1974; Lilly, 1979; Neisworth & Greer, 1975); in fact, as Edgar and Hayden (1984-1985) pointed out, "the literature is replete with statements that these three groups represent essentially the same population . . ." (p. 533).

Despite its logical support, the noncategorical perspective does not enjoy universal acceptance. For example, Epstein and Cullinan (1983) stated: "It is most unfortunate that, despite the intense and often persuasive advocacy of some of its proponents, there has been very little scientific study of the assumptions on which cross-categorical special education is based" (p. 306). They used academic performance of learning disabled (LD) and behaviorally disordered (BD) students to derive the following conclusion: "The results do not support the idea that intervention practices should necessarily be the same for students with LD, BD, and educable mental retardation" (p. 305). Again, the literature is replete with opinions and data supporting the separation of exceptional students into categorical groups to supply appropriate special education (cf. Becker, 1978; Cullinan, Epstein, & Dembinski, 1979; Lieberman, 1980; Phipps, 1982).

The purpose of this research was to describe instruction provided in categorical special education classrooms. Such information would be a valuable starting point for determining the appropriatness of categorical grouping of students.



Teachers of students classified as emotionally handicapped (EH), learning disabled (LD), or educable mentally retarded (EMR) were observed during different types of classroom instruction. Students classified as emotionally handicapped exhibited severe behavior problems that disrupted their own (or others') school progress; students classified as educable mentally retarded had significantly subaverage intellectual ability and adaptive behavior problems, and students classified as learning disabled showed a "severe" discrepancy between their ability and achievement. Generalizations to teachers in special classes containing similar students are warranted.

Classroom observations were completed in 40 self-contained special education rooms; exceptional students were assigned to these rooms for at least 25 hours each week. Students in 16 of the classrooms were classified as EMR, students in 13 of the classrooms were classified as LD, and students in the remaining 11 rooms were classified as EH. This distribution roughly approximates the overall demography for students with high-incidence handicaps in the state in which the data were collected (i.e., 37% 17%, and 10% are classified as LD, EMR, and EH, respectively).

Participating teachers were selected by special education administrators familiar with their experience and previous performance. All the participants had been teaching in self-contained classrooms for at least 3 years, and most had more than 5 years of special education teaching experience. Ninety percent of the participating teachers were female; all were categorically certified to teach in the areas in which they were currently teaching, and none were teaching in classes with temporary credentials.

Most (72%) of the observations took place during language arts, reading, or math instruction. The average number of students in the room during the observation was 9; over half of the data were collected in classes containing from 6 to 12 students, and about one fourth were collected in rooms containing less than 6 or more than 12 students.

Observation System

All observational data were collected by trained observes using the Classroom Observation Keyed for Effectiveness Research (COKER) observation system. The COKER consists of 389 items organized into two sections. Specific groups of teacher behaviors (e.g., presenting, questioning, responding) and students behaviors (e.g., compliance, asking questions, answering questions, making voluntary comments, being off-task) are the basis for coding interactions that occur during the initial phase of the observations. Additionally, occurrences of 33 student behaviors that promote or disrupt learning (e.g., following classroom routines, teasing, criticizing others) and 86 teacher behavior related to methodlogy, grouping, classroom affect, and control are also coded during the observation intervals. The COKER is considered a low-inference observation instrument, and using it involves observing and recording teacher/student behaviors as they occur during 5-minute observation periods.

Validity, reliability, and norming information presented in the COKER user's manual (Coker & Coker, 1982, pp. 13-19: 31-37) indicate that the system has adequate technical characteristics for use in this type of research. For example, construct validity was supported when expected dimensions of classroom instruction were confirmed by a factor analysis of data from more than 150 student teachers. When using an observation system such as the COKER, one is concerned with reliability of scoring keys rather than reliability of the entire instrument. Item cluster reliability coefficients obtained during development of the COKER ranged from .38 to .83; coefficients computed for 22 scoring keys ranged from 0.00 to .80. As Coker and Coker (1982) indicated. "Coefficients of zero are possible because of unreliable measurement and/or sheer absence of the behaviors' that comprise a key. "Generally, those scoring keys with the lower realiabilities had fewer numbers of items" (p. 37). Median reliabilities for 22 scoring keys ranged from .38 to .47 in four independent studies; in each case, most (36%-77%) of the coefficients were greater than .30. In another study, estimates of consistency between observations gathered for the same teacher in different classes on 12 scoring keys ranged from .47 to .85; multi-class and single-class realiabilities for these teachers were generally moderate to high (e.g., 58% were greater than .35). Additionally, in "studies of observer agreement, in which the results of pairs of observers are compared, the extent of agreement tends to be in the .80s with occasional agreement approaching .90" (Coker & Coker, 1982, p. 31).

Observer Training

All observations were completed by university faculty members or graduate students trained in the use of the COKER observation system by its developers or by staff trained by them. The training sessions were structured to provide practice using the observation forms with with videotaped presentations of classroom interactions. All observers completed a minimum of six training observations of at least 1 hour in length and were considered proficient in the use of the COKER when introbserver agreements between themselves and trained observers reached 90%.

Observation and Scoring Procedures

Each observer conducted approximately twelve 5-minute observations in each classroom over a 2-day period of time. Teachers were informed that the observations were for purposes of evaluating another measurement system and that they were not being evaluated. Observers also explained that the research was intended to help state department personnel develop observation competencies for use in exceptional student education. Observers are instructed to abstain from involvement in any classroom activities, to meet the teacher early, to determine the classroom setting, activities, and other demographic information before entering the room and to keep a "low profile" during the observations.

Scores on the COKER represent occurrence-nonoccurence of items that are being observed; combinations of these items are referred to as "competency keys" (Coker & Coker, 1982). The use of these keys is analogous to combining 10 items measuring computational skills with fractions and 15 items measuring computational skills with decimals into scores reflecting "competency" with fractions and "competency" with decimals. Scoring the COKER is a two-step process that involves determining the number of times an item is observed relative to the number of times a teacher was observed and then summing these values across keys to obtain competency scores. In this way, individual item values summed across competencies and represented as percentage reflect the proportion of items (0 to 100) that are observed within a competency. The greater the competency score, the more the behaviors indicative of effective teaching were observed (Coker & Coker, 1982; Dickson & Wiersma, 1984). Put another way, scores on the COKER reflect the extent to which expected behaviors actually occurred. Because keys contain different numbers of items, comparisons are not made across competencies when using proportionally scores; these scores are considered appropriate for comparing individuals within a group or when comparing groups to each other on separate competencies.

Data Analysis

More than 50 sets of keys are available for use with the COKER. The dimensions of competence developed for use with special education teachers (cf. Dickson & Wiersma, 1984) were evaluated as dependent variables in this research; descriptors for competencies within each of these domains are presented in Figure 1. The independent variable of interest was the type of self-contained classroom teacher being observed.

Classroom observations were completed in 40 special classes. Scores on effective teaching competencies were compiled, analyzed, and compared through a series of one-way analysise of variance; a statistical level of 0.50 was used in evaluating main effects in these analyses.


Proportionality scores for 22 teaching competencies were available for analysis. Internal consistency estimates (Cronbach's alpha) were obtained as a measure of the reliability of these data. The reliability estimates for the nine keys that comprise the instructional strategies, techniques, or methods that are scored from COKER data ranged from 0.00 to 0.68. Keys with unacceptable (i.e., less than 0.10) reliability (i.e., those representing the extent to which teachers were using a variety of resources or establishing varied transitions and instructional sequences) were eliminated from subsequent analyses. Three keys (i.e., motivates students to ask questions, uses verbal and nonverbal skills, and uses questions that lead students to synthesize) were eliminated within the learner communication domain for similar reasons; the reliabilities for the remaining five keys in this domain ranged from 0.17 to 0.58. Reliability estimates for the five keys indicative of learner reinforcement and involvement by these teachers ranged from 0.14 to 0.76. As in previous analyses (cf. Dickson & Wiersma, 1984), lower reliabilities were obtained for keys comprised of smaller numbers of items. Comparisons of the 17 keys with acceptable internal consistency were completed to evaluate the similarities and differences in instruction provided by teachers of self-contained special classes for LD, EH, or EMR students.

Means and standard deviations for instructional competencies exhibited in self-contained special classes are presented in Table 1. Univariate analyses of variance indicated that differences in each domain, except one were nonsignificant. The extent to which the observed teachers modified instruction to meet learner needs was different (F (2, 37) = 4.23, p < 0.05). Follow-up analysis indicated that teachers of EMR students modified instruction less (M = 11.06) than did teachers of EH students (M = 24.23) or teachers of LD students (M = 23.50). Of the instructional competencies that were similarly observed in different types of classrooms, behaviors related to providing experiences that facilitate transfer of learning were least likely to be observed; and those related to structuring student time to facilitate learning were more likely to be observed. Low occurrence of behaviors (less than 20% rates) related to working with individuals and groups, using convergent/divergent inquiry, and developing problem-solving skills was observed. Moderate (30%-40%) use of a variety of instructional strategies was evident.

Means and standard deviations for communication competencies exhibited in self-contained special classes are presented in Table 2. Univariate analyses of variance indicated that differences in each domain were nonsignificant. Of those teaching behaviors that were similarly observed in different types of classrooms, behaviors related to accepting varied view-points were least likely to be observed, while those related to demonstrating proper listening skills and giving clear directions were the most common actions. Low occurrence (less than 10% rates) of providing learner feedback was observed, and moderate (more than 40%) rates of providing group communication were evident.

Means and standard deviations for involvement competencies exhibited in self-contained special classes are also presented in Table 2. Univariate analyses of variance indicated differences in each domain were nonsignificant. Behaviors using positive reinforcement were the least likely to be observed; and those maintaining active involvement were the most common actions. Moderate (20%-60%) rates of developing student self-feedback, assisting in error correction, and using effective classroom management were evident in different types of self-contained classrooms.

Classroom observations of 40 teachers of self-contained special education classrooms were completed. More than five hundred 5-minute "snapshots" of teacher-pupil interaction were compiled and analyzed. In general, it appears that instruction for EMR, LD, or EH studens is more similar than different. It appears that teachers of EMR students modify the instruction to meet individual learner needs, about half as frequently as do teachers of LD or EH students.


While no comparisons of the effectiveness of categorical versus noncategorical instruction or special versus regular education were completed in this research, questioning the overall effectiveness of special education cannot logically be completed without first documenting the nature of instruction that takes place in all types of special classes. To document the characteristics of classroom instruction in self-contained special classes, teachers of LD, EH, or EMR students were observed. This description of what was going on in this one type of special class was considered a logical first step in the continuing search for an answer to the following question: What's special about special education?

Similar degrees of variability were evident in more than 90% of the instructional activities observed in the classrooms of teachers working with three different types of students. The magnitude of important competencies of effective teachers, such as structuring student time, providing learner feedback, and giving clear directions, were also similar. It appears that this group of special education teachers was performing adequately relative to effective teaching domains, but they were not performing differently relative to the type of student in their self-contained special classrooms.

Almost all states support the use of category-specific sligibility criteria and report figures for placement of students in classes for the emotionally disturbed, mildly mentally retarded, and learning disabled (Algozzine & Korinek, 1985). Despite arguments supporting a decrease in categorization (cf. Hallahan & Kauffman, 1977), authors of special education introductory textbooks continue to present categorical perspectives because they believe such books should reflect "the general viewpoint of the special education profession, and the prevailing approach to educating exceptional children remains categorical" (Hallahan & Kauffman, 1986, p. xiii). It seems that a type of professional schizophrenia permeates theory, research, and practice in contemporary special education; and some argue that the conceptual pathology results partly from the absence of data.

Students are placed in categorical programs despite limited evidence that the groupings lead to differential treatment. Students of special education learn the history, definitions, and supposed characteristics of different types of students despite limited evidence that this information is useful in planning programs for most exceptional students. Teachers are certified to teach categorically different students despite limited evidence that most of the groupings are a source of differential treatment methods. Data from these self-contained special classrooms support the scientific basis for taking the noncategorical side of special education personality more seriously. Observations of teachers in self-contained classrooms containing LD, EH, or EMR students did not support conclusions about differentiated instruction on the basis of category. Though this does not mean that categorical instruction is better or worse than noncategorical instruction or that special education is better or worse than regular education for these students, it does seem a reasonable beginning for continued research to determine the appropriateness of placing and teaching students in disability groups.


Algozzine, B., & Korinek, L. (1985). Where is special education for students with high prevalence handicaps going? Exceptional Children, 51, 388-394.

Becker, L. D. (1978). Learning characteristics of educationally handicapped and retarded children. Exceptional Children, 44, 502-511.

Coker, J.G., & Coker, H. (1982). Classroom observations keyed for effectiveness research: User's manual. Carrollton, GA: Author.

Cullinan, D., Epstein, M. H., & Dembinski, R. J. (1979). Behavior problems of educationally handicapped and normal pupils. Journal of Abnormal Child Psychology, 7, 495-502.

Dickson, G. E., & Wiersma, W. (1984). Empirical measurement of teacher performance. Toledo, OH: The Center for Educational Research and Services.

Edgar, E., & Hayden, A. H. (1984-1985). Who are the children special education should serve and how many children are there? Journal of Special Education, 18, 523-539.

Epstein, M. H., & Cullinan, D. (1983). Academic performance of behaviorally disordered and learning-disabled pupils. Journal of Special Education, 17, 303-307.

Gardner, W. I. (1977). Learning and behavior characteristics of exceptional children and youth. Boston: Allyn and Bacon.

Hallahan, D. P., & Kauffman, J. M. (1977). Labels, categories, behaviors: ED, LD, and EMR reconsidered. The Journal of Special Education, 11, 139-149.

Hallahan, D. P., & Kauffman, J. M. (1986). Exceptional children: Inroduction to special education (3rd ed.). Englewood Cliffs, NJ: Prentice-Hall.

Hewett, F. M., & Forness, S. R. (1974). Education of exceptional learners. Boston: Allyn and Bacon.

Kirk, S. A., & Gallagher, J. J. (1986). Educating exceptional children (5th ed.). Boston: Houghton Mifflin.

Liberman, L. M. (1980). The implications of noncategorical special education. Journal of Learning Disabilities, 13, 65-68.

Lilly, M. S. (1979). Children with exceptional needs. New York: Holt.

Neisworth, J. T., & Greer, J. G. (1975). Functional similarities of learning disability and mild retardation. Exceptional Children, 42, 17-21.

Phipps, P. M. (1982). The merging of categories: Appropriate education or administrative convenience. Journal of Learning Disabilities, 15, 153-154.

Ysseldyke, J. E., & Algozzine, B. (1984). Introduction to special education. Boston: Houghton Mifflin.

BOB ALGOZZINE is Professor of Curriculum and Instruction, University of North Carolina at Charlotte.

CATHERINE V. MORSINK is Professor of Special Education, University of Florida, Gainsville. KATE M. ALGOZZINE is a Teacher of Gifted Students.
Gale Copyright:
Copyright 1988 Gale, Cengage Learning. All rights reserved.