Printer Friendly

Who leaves and who stays in special education: a 2-year follow-up study.

Who Leaves and Who Stays in Special Education: A 2-Year Follow-up Study

Under the federal reporting requirements of the Education for All Handicapped Children Act, Public Law (P.L.) 94-142, local and state education agencies have provided a large, and annually growing, corpus of information regarding total numbers of children in special education and their distribution according to disability (e.g., Office of Special Education and Rehabilitative Services, OSERS, 1983, 1984, 1985). These reports have documented the increase in the size of the special education population, which grew by 16.2% from 1976-77 to 4.3 million in 1983-84. In addition, they have provided valuable data on the changing characteristics of those served, specifically the increase in the proportion of the special education population labeled learning disabled and the corresponding decrease in the proportion labeled speech impaired and mildly mentally retarded (Algozzine & Korinek, 1985; Edgar & Hayden, 1984-85; Zill, 1985).

The changing size and composition of the special education population can be attributed to many factors. For example, within the special education population itself, reclassification of some children formerly labeled mentally retarded to other categories such as the learning disabled and emotionally disturbed accounts for part of the compositional shift (Gerber, 1984; Polloway & Smith, 1983; Reynolds, 1984).

But care must be exercised in assessing the real degree of change associated with variations in the size and composition of the special education population because these conclusions have been based on data collected on repeated cross sections. Knowing that the size of the special education population has increased does not tell us whether fewer children have been terminated or more children have been identified. Similarly, an increase in the numbers of children labeled learning disabled may stem from the identification of new children or from reclassification of children from other disability classifications. Use of recurrent cross-sectional estimates precludes the possibility of distinguishing between these alternate explantions. And although the pooled evidence from cross-sectional studies has begun to suggest patterns, no studies reported to date have tracked special education students over time. Such information would be of considerable value to school administrators and policy makers, who recognize that the decision to place a child in special education may represent a commitment to a multiyear cost and curriculum stream.

Toi fill this need, we conducted a 2-year follow-up study in three large school systems to address the following questions:

1. What proportion of special education students remain in their school district? Who is most likely to leave?

2. What proportion are terminated from the program? Who is most likely to be terminated?

3. What proportion remain in special education but under a different primary handicap designation? Who is most likely to be reclassified?


Sample Selection

This study was completed by reviewing the special education records of 1,184 children during the 1982-83 and 1984-85 school years in three sites of the Collaborative Study of Children with Special Needs: Charlotte Mecklenburg Schools, North Carolina; Milwaukee Public Schools, Wisconsin; and Rochester City School District, New York. Initial samples were drawn from district lists of all elementary school students enrolled in special education in the fall of 1982. A stratified random selection technique was used to ensure adequate numbers of children with the more severe but less common problems (Cochran, 1980). In each site, the special education population in kindergarten through sixth grade was divided into three strata based upon the school's designation of the student's primary handicap: (a) those with speech impairments or learning disabilities; (b) those with emotional and behavioral problems or mental impairments; and (c) those with physical, sensory, or health impairments. Stratification was used for sampling purposes only, and permits generalizations both for subgroups of children and for the special education population as a whole.

An initial sample of 1,829 children was selected and divided approximately equally across the three sites and three strata. Nine percent of these children were ineligble to participate because they had moved out of the district, were no longer in special education, were siblings of others in the sample, or had died. Consent was granted for 74% of those eligible, from which a random sample of 1,184 was selected for data collection (429 in Charlotte, 397 in Milwaukee, and 360 in Rochester). A comparison of the ineligible students with the remainder of the initial sample revealed that the ineligible children were more likely to be speech impaired or learning disabled rather than one of the other strata. The ineligible students did not differ significantly, however, with regard to age, grade, gender, race, or ethnicity. Because of the slightly decreased representation of speech and learning disabled children, which is directly attributable to their higher rates of termination from special education, the results generalized to thosee enrolled in special education for a minimum of 6 months. After 2 years, data were collected on all 1,184 in the initial sample.

All three school systems had large special education programs and a socioeconomically diverse student body (Table 1). The percentage of the elementary school population enrolled in special education ranged from 7.6% in Charlotte to 13.4% in Rochester. Each district had a somewhat higher proportion of Black than White students in special education, which reflected similar proportions in the general education population in Milwaukee and Rochester, although not in Charlotte. Approximately half to two-thirds of the special education students came from poor and near-poor families. The proportions of special education students whose mothers had not graduated from high school ranged from 35% in Charlotte to 53% in Rochester.


Records Reviews. In the Spring of 1983 and the fall of 1984, record reviews were carried out by trained personnel, using data in the child's permanent files, primarily the Individual Educational Plan (IEP). The initial review averaged 30 minutes per child and included information on classroom placement, grade level, related services recommended, and parent attendance at the most recent IEP meeting. The follow-up review averaged 15 minutes per child and included information on whether the child was still receiving special education services, and if so, under which primary handicap classification. Depending on the child's age and grade in fall 1982, a child was coded as being in an appropriate grade if he was no more than 7 years older than the value of his grade (e.g., a child no older than 10 years of age enrolled in the third grade).

Parent Interviews. In the spring of 1983, parents were interviewed for 40 minutes over the telephone in either English or Spanish by personnel from the University of Illinois Survey Research Laboratory. The questionnaire included items about the child's disabilities and functional level adopted from national health surveys (National Center for Health statistics, 1982; Walker, Gortmaker, & Weitzman, 1981), parent satisfaction with the child's educational program, and descriptors of the family's socio-economic status.

The parent's designation of the child's primary handicap was derived from the parent's response to the open question: "What is the name of the major problem or condition that (index child) has?" Parents were then asked whether the child had each of 20 medical or health conditions, and if he or she did, whether it "limits his ability to do things all other children his age can do." The number of major handicaps was defined as the toal number of conditions that limited the child's ability.

Parent satisfaction with the child's "overall educational program" was assessed using a 4-point scale ranging from "very dissatisfied" to "very satisfied."

Poverty status was based on the family's size and income relative to the 1982 poverty line (U.S. Bureau of the Census, 1982). For statistical analyses, the continuous form of this measure was used. For presentation, this measure is divided into four groups: poor (below the poverty line); near poor (between 100% and 150% of poverty); low income (between 150% and 200% of poverty); and not poor (above 200% of the poverty line).

Statistical Analyses

Estimates presented in this article are based on a weighting procedure that compenseates statistically for the oversampling of low-prevalence disability groups. Within each site, weights were computed to generalize the results to the special education population of that site; these weights then were calibrated to total the actual number of respondents in that site. All percentages reported incorporate these weights, so that estimates generalized to the average special education population of the three sites.

Multiple linear regression was not appropriate because of the dichotomous nature of the independent variables for the analyses that identified factors associated with mobility, termination, and reclassification (Fox, 1984; Press & Wilson, 1978). Therefore, multiple logistic regression analysis was used to adjust for confounding factors such as student age and family background. The analysis strategy was to estimate the additional effects of school-related characteristics, such as appropriate grade for age and type of services received, after controlling for maximal variation explained by child and family background characteristics. Statistical software limitations precluded the possibility of weighting these analyses. Because they were conducted within primary handicap groups, however, the weights should have had little influence on the findings (Kish & Frankel, 1974).

The total sample size of 1,184 is large enough to provide ample statistical power (greater than .80) to detect small effects (Cohen, 1977). Within subgroups, power does diminish, but the estimates of standard errors remain relatively small.


Table 2 presents the special education status of the sample in fall 1984, overall and by initial primary handicap. Over the 2-year period, the special education population of these three sites showed a moderate degree of change. Almost 8% of students were no longer in their school district. Of those remaining, 17.2% were terminated from special education, 12.3% were still receiving special education services, but under a different primary classification, and 70.5% had no change in their special education status.

Who Leaves the School System?

The small percentages leaving the school systems varied significantly by initial primary handicap and ranged from 5.7% of those with mental impairments and 6.8% of those with speech impairments to 14.7% of those with physical/multiple handicaps and 18.3% of those with hearing impairments. Bivariate analyses revealed that children with more major handicaps had a higher probability of leaving than those with fewer, as did children who were White (versus Black or Hispanic) or who were from more affluent families. No school-related characteristics, such as study site, initial grade for age, parent attendance at the IEP meeting, and parent satisfaction with the child's overall educational program were associated with the probability of leaving the school district.

In a multiple logistic regression model identifying associated factors, leaving the school system was primarily related to the child's family background and not the school program. Poverty status, race, and number of major handicaps remained statistically significant predictors, whereas differences by initial primary classification disappeared once these background variables were controlled, suggesting that the variations observed in Table 2 are artifacts of correlations between initial primary and socioeconomic status. Thus, leaving the school system was primarily related to the child's family background and not the school program.

Who is Terminated from Special Education?

Unlike leaving the school system, termination from special education was strongly associated with the child's initial primary handicap classification (Table 2). Those initially classified as speech impaired were most likely to be terminated (33.1%), followed by those initially classified as learning disabled (14.9%), emotionally/behaviorally disturbed (9.1%) or vision impaired (8.6%). Children initially classified as hearing impaired, physical/multiply handicapped, or mentally retarded were rarely, if ever, terminated.

An analysis of the factors associated with termination could be performed only for the two groups with adequate numbers of students terminating--the speech impaired and the learning disabled. Children in both groups were significantly more likely to be terminated during the 2-year period if, in the fall of 1982, they were in the upper elementary grades (4-6) or were receiving only related services, not special instructional services. In addition, children initially classified as learning disabled were more likely to be terminated if they were in Charlotte (versus the other two districts) or were placed in an appropriate grade for their age.

These bivariate relationships tended to persist even when subjected to simultaneous control in logistic regression models (Table 3). Children in both groups were more likely to be terminated during the 2-year period if they were only receiving related services or were in grades 4-6 in fall 1982. Among those initially classified as speech impaired, two additional variables--parent report of no learning or emotional problems and being non-Black--also were significantly related to termination after controlling for the other factors. Only one other variable--being in the appropriate grade for age--was significantly related to the probability of termination for those originally classified as learning disabled. All site differences observed in the bivariate analyses disappeared after controlling for the factors.

Table 4 presents the probability of termination for children initially classified as speech impaired or learning disabled by different combinations of the measures identified in table 3, after adjusting for all other factors.

Among speech impaired children, those receiving special education in fall 1982 were unlikely to be terminated (probability less than .20) regardless of their grade or parent report of learning or emotional problems. Those receiving only speech therapy, on the other hand, were far more likely to be terminated, especially if they were in grades 4-6 initially and/or had no accompanying learning or emotional problem.

A similar pattern was found among those initially classified as learning disabled. Those receiving special instruction in fall 1982 were unlikely to be terminated during the 2-year period, especially if they were in the early elementary grades or were behind in grade for age (probability less than .30).

Who is Reclassified within Special Education?

Approximately 12% of those remaining in their school district were still in special education two years later, but under a new primary handicap classification (Table 2). Those most likely to be reclassified were children originally labeled physically/multiply handicapped (25.2%), speech impaired (20.5%), or emotionally/behaviorally disturbed (14.8%). Reclassification was uncommon among those initially classified as learning disabled, mentally retarded, and hearing or visually impaired (less than 8.0%).

Primary handicap designations in the fall of 1984 versus the fall of 1982 among those children still in the school district are displayed in Table 5. Percentages on the diagonal (in bold type) represent those children for whom there was no change; off-diagonal percentages reflect patterns of reclassification. Regardless of initial classification, the two most common categories to move into were learning disabilities and mental retardation. Those initially classified as physically/multiply handicapped were most often reclassified as mentally retarded (15.8%), and speech impaired children were usually reclassified as learning disabled (13.2%).

In general, the less congruence between the parent's independent report of the child's primary handicap and the school's report of the child's primary handicap in the fall of 1982 the more likely the child was to have been reclassified during the 2-year-period (Table 6). When the parent and school agreed, only 6.7% of students were reclassified; when the parent agreed the child had the problem but did not think it was his primary handicap, 20% were reclassified; and when the parent totally disagreed with the scool, 22.7% were reclassified. These differences were most pronounced among those initial classifications most likely to experience change: physical/multiple handicaps and speech impairments.

Because of limited sample sizes, further analyses could be conducted only for those originally classified as speech impaired. Reclassification was most likely if a speech impaired child was receiving special insetruction in fall 1982 (in addition to or instead of speech theraphy), if his or her parents were dissatisfied with the school's overall educational program, if the parents reported a learning problem, or if the child was not in the appropriate grade for his or her age. These differntials tended to persist, even when examined simultaneously in a logistic regression model. Thus, those speech impaired children who were reclassified often could be identified on the basis of characteristics assessed before their reclassification--specifically, their instructional program, their grade assignment, parent dissatisfaction with the overall educational program, and parent disagreement as to what was wrong with the child.


Although each of the three samples is a representative probability sample of that district, their combined results may not generalize to all school districts around the country. At best, findings may be reflective of national urban experience, but not small-city or rural experience. Furthermore, the precise meaning of observed changes over time must be considered cautiously because we have not directly assessed what happened to each child during the 2-year period. Plausible explanations of changes include the following:

1. Achievement of the special education goals specified in the child's IEP.

2. Maturation and growth of the child such that the original condition of the child is better (or worse).

3. "Falling off the cliff," because of limited placement options for elementary school graduates.

4. Changes in school district budgets or policies, such as a decrease in funding for Chapter I or a change in program eligibility criteria.

5. Assessment changes, including the use of different criteria by different evaluators and improved evaluation techniques overall.

Although these alternative explanations for changes over time cannot be differentiated in our analyses, the practical implications of changes, regardless of the reasons, are discussed.

Children Leaving School Districts

Very few special education students (7.7%) left their school districts during the 2-year period. Although no national school mobility data are available for direct comparison, this percentage is less than the overall mobility figures reported by the U.S. Bureau of the Census (1983) for 5-17-year-olds in the United States in 1980-81.

One explanation why children receiving special education services might be somewhat less mobile than the general school age population is that parents make a conscious decision not to move once their handicapped child is placed in an acceptable special education program. Other analyses of these data have documented the high degree of parent satisfaction with their child's special education program; such satisfaction may provide a compelling reason for not moving (Singer & Butler, 1987).

An alternative explanation rests with the socioeconomic status of the special education population in these urban sites. Since these children tend to come from poor homes, and we have shown that poorer families were less likely to move, the low mobility rates may not reflect conscious parental decisions but rather an inability to be able to afford to move.

Children Terminated or Reclassified

Those most likely to change--through either termination or reclassification--were those originally classified as speech impaired. Only 46% of the initial group remained in special education 2 years later under the same rubric. Whereas 33% were terminated and the remaining 21% were reclassified. Many of those terminated probably have improved because of therapy or natural maturation (Edgar & Hayden, 1984-85; Zill, 1985). But the higher termination rates for those in grades 4-6 suggest that some children may be terminated for a structural reason--the lack of services after elementary school.

Reclassification, too, may be the result of more appropriate or accurate disgnoses, specifically later developing learning problems. However, the higher reclassification rates for those children receiving special instruction at the time of the initial data collection suggest that some of these children had additional problems to begin with. School professionals and parents may prefer to use the label "speech impaired" as a less stigmatizing preliminary special education designation until it is clearer what the child's problems really are. Whatever the reason, the category of "speech impairment" is the most fluid special education designation.

Children initially classified as learning disabled and emotionally or bahaviorally disturbed were the next most likely group to experience a change during the 2-year period, with termination more common among the former and reclassification more common among the latter. Without comparisons from other longitudinal studies, it is difficult to judge if these rates are high or low. It is not surprising, however, that termination was most likely among those learning disabled children initially in grades 4-6 who were not receiving special instructional services. Some of this movement may parallel similar phasing out of children from compensatory education programs such as Chapter I and bilingual education. But a larger component is likely due to difficulties in distinguishing between those special education students labeled learning disabled and emotionally and behaviorally disturbed, and those regular education students traditionally called low achievers (Algozzine & Ysseldyke, 1981, 1983; Shepard, Smith, & Vojir, 1983; Ysseldyke, Algozzine, & Richey, 1982). thus a child may begin in one classification and be terminated if his or her problems diminish, or be reclassified if they escalate.

Children Usually Not Terminated or Reclassified

Few of the remaining children, those initially classified as mentally retarded, visually impaired, hearing impaired, or physically or multiply handicapped, were terminated from special education during the 2-year period. The lack of termination for these children reflects the permanency of their problems; for example, a mentally retarded child does not "get better" through treatment or maturation, although the child may certainly be better able to function with his or her impairment. Other researchers, including Algozzine and Korinek (1985), Edgar and Hayden (1984-85) and Zill (1985) have used the OSERS annual compliance reports and other cross-sectional data to argue that this subpopulation within special education is stable; these longitudinal findings provide strong corroboration for such arguments.

Reclassification Patterns

During the 2-year study period, 25% of those initially classified as physically or multiply handicapped were reclassifed, usually as mentally retarded. On the one hand, reclassification may reflect a more realistic appraisal of the children's abilities, especially as they mature and the cognitive load of school becomes more stressful and difficult. Alternatively, this shift may simply reflect the decisions of school administrators in a financially constrained environment where it costs less to classify a child mentally retarded, often because occupational and physical therapy are not provided to this group (Raphael, Singer, & Walker, 1985).

Over and above this shift between physical or multiple impairments and mental retardation, a more general pattern of reclassification was found: children were more likely to be reclassified if their parent's independent report of the child's major problem did not correlate with the school's assessment. Several plausible explanations of this finding need to be explored in further research. Congruence between parents and schools suggests that both parties share an understanding of the child's handicaps and needs. More important, though, is what the lack of congruence means, and whether it stems from the parent's inability to communicate at an IEP meeting, the school's nominal preference for a category that may not perfectly characterize the child's problem, or the school's inability to communicate clearly to the parent what precisely is wrong with the child.

In summary, the present research documents a continuum of stability with the special education population: movement was especially pronounced among those initially classified as speech impaired; moderate among those initially classified as learning disabled, emotionally or behaviorally disturbed and physically or multiply handicapped; and minimal among the hearing impaired, visually impaired, and mentally retarded. It has also identified a series of programatic factors associated with movement, including grade and services received, and suggested why those factors may be predictive. Other longitudinal studies should be undertaken to further explore the stability of the special education population.


Algozzine, B., & Korinek, L. (1985). Where is special education for students with high prevalence handicaps going? Exceptional Children, 51(5), 388-396.

Algozzine, B., & Ysseldyke, J. E. (1981). Special education services for normal children: Better safe than sorry? Exceptional Children, 48(3), 238-243.

Algozzine, B., & Ysseldyke, J. E. (1983). Learning disabilities as a subset of school failure: The oversophistication of a concept. Exceptional Children, 50(3), 242-246.

Cochran, W. G. (1980). Sampling techniques (3rd ed.). New York: Wiley.

Cohen, J. (1977). Statistical analyses for the behavioral sciences (2nd ed.). New York: Academic Press.

Edgar, E., & Hayden, A. H. (1984-85). Who are the children special education should serve and how many children are there? Journal of Special Education, 18(4), 523-539.

Fox, J. (1984). Linear statistical models and related methods. New York: Wiley.

Gerber, M. (1984). The Department of Education's Sixth Annual Report to Congress on PL 94-142: Is Congress getting the full story? Exceptional Children, 51, 209-224.

Kish, L., & Frankel, M. R. (1974). Inference from complex samples. Journal of The Royal Statistical Society, Series B, 36, 1-37.

National Center for Health Statistics. (1982). Current estimates from the national health interview survey: United States, 1981. Vital and Health Statistics (Series 10, No. 141). Hyattsville, MD: Department of Health and Human Services.

Office of Special Education and Rehabilitative Services (1983). Fifth annual report to Congress on the implementation of the Education for the Handicapped Act. Washington: U.S. Dept of Education.

Office of Special Education and Rehabilitative Services (1984). Sixth annual report to Congress on the implementation of the Education for the Handicapped Act. Washington: U.S. Dept of Education.

Office of Special Education and Rehabilitative Services (1985). Seventh annual report to Congress on the implementation of the Education for the Handicapped Act. Washington: U.S. Dept of Education.

Polloway, E. A., & Smith, J. D. (1983). Changes in mild mental retardation: Population programs and perspectives. Exceptional Children, 50(2), 149-159.

Press, S. J., & Wilson, S. (1978). Choosing between logistic regression and discriminant analysis. Journal of the American Statistical Association, 73, 699-705.

Raphael, E. S., Singer, J. D., & Walker, D. K. (1985). Per child expenditures in special education in three metropolitan school districts. Journal of Educational Finance, 11(1), 60-88.

Reynolds, M. C. (1984). Classification of students with handicaps. In E. W. Gordon (Ed.), Review of Research in Education (Vol. 11). Washington, DC: American Educational Research Association.

Shepard, L. A., Smith, M. L., & Vojir, C.P. (1983). Characteristics of pupils identified as learning disabled. American Educational Research Journal, 20(3), 309-331.

Singer, J. D. & Butler, J. A. (1987). The Education for All Handicapped Children Act: Schools as agents of social reform. Harvard Educational Review, 57(2), 125-152.

U.S. Bureau of the Census. (1982). Money, income and poverty status of families and persons in the United States, 1982. Washington, DC: U. S. Government Printing Office.

U. S. Bureau of the Census. (1983). Statistical abstract of the United States: 1984 (104th ed.). Washington, DC: U. S. Government Printing Office.

Walker, D. K., Gortmaker, S. L., & Weitzman, M. (1981). Chronic illness and psychosocial problems among children in Genesee County. Boston, MA: Community Child Health Studeis, Harvard School of Public Health.

Ysseldyke, J. E., Algozzine, B., & Richey, L. (1982). Judgment under certainty: How many children are handicapped? Exceptional Children, 48(6), 531-534.

Zill, N. (1985). The school-aged handicapped. Washington, DC: U. S. Government Printing Office.

DEBORAH K. WALKER is Associate Professor, Department of Behavioral Sciences, Harvard School of Public Health, Boston. JUDITH D. SINGER is Assistant Professor, Harvard Graduate School of Education, Cambridge. JUDITH S. PALFREY is Assistant Professor, Department of Pediatrics, Harvard Medical School, Boston. MICHELE ORZA is a Doctoral Candidate, Dapartment of Health Policy and Management, Harvard School of Public Health, Boston. MARTA WENGER IS A Fellow, Bush Institute for Child and Family Policy, Frank Porter Graham Child Development Center, University of North Carolina at Chapel Hill. JOHN A. BUTLER is Assistant Professor, Department of Social Medicine and Health Policy, Harvard Medical School, Boston.
COPYRIGHT 1988 Council for Exceptional Children
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1988 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:includes bibliography
Author:Walker, Deborah K.; Singer, Judith D.; Palfrey, Judith S.; Orza, Michele; Wenger, Marta; Butler, Joh
Publication:Exceptional Children
Date:Feb 1, 1988
Previous Article:Means and ends.
Next Article:The consulting teacher model: risks and opportunities.

Related Articles
The National Longitudinal Surveys of Labor Market Experience: an annotated bibliography of research.
Survey of the states: special education knowledge requirements for school administrators.
Categorical and programming changes in special education services.
Academic, social, and behavioral adjustment for students declassified from special education.
Strategies for inclusion: an annotated bibliography.
Working in Special Education: Factors that Enhance Special Educators' Intent to Stay.
Effect of different methods of reporting and reclassification on trends in test scores for students with disabilities.
Legislative update: Washington politicos focus on IDEA reauthorization.
Extent of parental involvement in improving the students' levels in special education programs in Kuwait.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters