Printer Friendly

Review of comparative studies in the interaction of students with moderate and severe handicaps.

ABSTRACT: Systematic instructional strategies that investigators have used with students who display moderate to severe handicaps were identified and defined. The investigations which directly compared two or more strategies were analyzed, and summary statements/ recommendations about the relative effectiveness and efficiency of the strategies were made. Analyzing the effectiveness of the procedures involved determining which strategies successfully taught skills to subjects. The efficiency of the strategies was analyzed on measures which indicated the skills were taught in a productive and timely manner (i.e., trials to criterion, sessions to criterion, errors to criterion, direct instruction time). Recommendations for future research and practice include (a) conduct more studies comparing the effectiveness and efficiency of instructional strategies, (b) conduct investigations ofthe specific variables of single strategies to identify the most efficient use of each procedure, (c) expand the efficiency measures to assess whether students learn information not directly targeted for instruction, and (d) conduct research to determine which strategy is best to use with given types of students and skills.

The technology of teaching new behaviors to moderately and severely handicapped learners has progressed to the point that a wide range of behaviors have been taught successfully. A number of instructional strategies (i.e., defined as systematic, replicable approaches to providing instruction addressing both antecedent and consequent events) have emerged. These strategies were the focus of a literature review conducted by the Comparison of Instructional Strategies (CIS) research project Wolery, Ault, Doyle, & Gast, 1986). The literature review progressed through four phases: (a) identifying relevant articles, (b) coding articles, (c) checking the reliability of the coding, and (d) analyzing the coded articles to derive recommendations for research and practice. To identify articles, each issue of 10 journals from 1975 until early 1987 were reviewed; these included American Journal of Mental Deficiency; Analysis and Intervention in Developmental Disabilities; Applied Research in Mental Retardation; Behavior Research of Severe Developmental Disabilities; Education and Training of the Mentally Retarded; Education and Treatment of Children; Journal of Applied Behavior Analysis; Journal of Autism and Developmental Disorders; Journal of the Association for Persons with Severe Handicaps: and Mental Retardation. Identified articles were then coded in terms of age and diagnosis of subjects, target tasks, trial sequences, procedural variables, and effects. Interobserver agreement was assessed on 10% of the coded articles and ranged from 92% to 100% for each of the coded variables. Finally, recommendations for future research and practice were developed.

From the review, 13 instructional strategies were identified as being repeatedly and effectively reported in the literature. The strategies were categorized based on their common procedural descriptions rather than the labels applied by various authors. A listing and description of each strategy are shown in Figure 1; for more complete descriptions and flow charts see Wolery et al. i 986).

The purpose of the article is to review and analyze the investigations where two or more of the strategies were compared directly. The rationale for this review is that the results from direct comparisons should lead to guidelines for selecting one instructional strategy over another. Similar reviews also exist in the literature; for example, Billingsley and Romer (1983) reviewed 6 studies which compared the prompt-fading strategies of decreasing assistance, graduated guidance, time delay, and increasing assistance. Schoen i 986) reviewed I I studies which involved these same procedures and included the instructional strategies of stimulus fading, stimulus shaping, and integrated strategies. The current article expands these two reviews by including studies which systematically compared two or more of the instructional strategies described in Figure 1. TYPES OF COMPARISONS in our review, 31 studies were identified, and the comparisons made and related citations are shown in Table 1. Error correction was compared most frequently to other instructional strategies followed by the stimulus modification procedures (stimulus shaping and fading) and the system of least prompts. The antecedent prompt and fade procedure, constant time delay, and naturalistic teaching procedures received the least attention in the comparative literature. Of these 31 studies, 26 compared two strategies and 5 compared three strategies. In addition to the instructional strategy comparison, 9 of the studies also included the comparison of an instructional variable. For example, Dorry (I 976) compared three types of stimulus fading with an antecedent prompt and test strategy, and Haught, Walls, and Crist (1984) compared error correction and antecedent prompting when both procedures were used to teach a long and short task. One study (Mosk & Bucher, 1984) compared a single instructional strategy (system of least prompts) with an integration of two strategies system of least prompts and stimulus shaping). A complete tabled description of each study is available from the authors, and identifies the number of subjects; subjects' gender, ages, and diagnoses; the instructional setting; behaviors taught; strategies used; and the effectiveness and efficiency of that use. RESULTS Error Correction Versus Antecedent Prompting Strategies Error correction was compared to an antecedent prompt and test strategy in four studies. In each of these studies, an antecedent prompting strategy, in which a model was given before a trial, was compared with an error correction strategy, in which a model was given only after an error. Instructional variables were also investigated in these studies; for example, Haught et al. (1984) compared a short task with a long task, where both were conducted with each strategy. Effectiveness. Two of these studies (Haught et al., 1984; Zane, Walls, & Thvedt, 1981) taught subjects with one strategy for a specific number of minutes, then discontinued training at that time if criterion was not met; therefore, statements about effectiveness are not possible. in the other two studies Ellis, Walls, & Zane, 1980; Walls, Zane, & Thvedt, 1980) all strategies were effective in establishing criterion level performance with the exception of one subject in the Walls et al. study. Efficiency. In terms of efficiency, all four studies reported instructional time and error measures. Ellis et al. (1980) and Walls et al. (1980) reported no significant differences between strategies in terms of instructional time. The antecedent model strategy with total task presentation required less time than any of the other strategies (Zane et al., 198 1 ), and Haught et al. (1984) stated that the short task took less time than the long task regardless of the instructional strategy used. In terms of errors during training, the antecedent model strategy produced fewer errors than the error correction strategy in all four studies. Stimulus Modification Versus Other Strategies Eleven studies compared some kind of stimulus manipulation strategy with one or more strategies. A variety of stimulus modifications were compared with a variety of other strategies such as most-to-least prompting, trial-and-error, progressive time delay, error correction, and antecedent prompt that was not faded. Effectiveness. Stimulus shaping was more effective than superimposition and stimulus fading Walsh & Lamberts, 1979) and was more effective than a trial-and-error procedure Zawlocki & Walls, 1983). Stimulus fading was more effective than most-to-least prompting (Schreibman, 1975; Wolfe & Cuvo, 1978), trial-and-error Aeschleman & Higgins, 1982; Richmond & Bell, 1986; Strand & Morris, 1986), error correction McGee & McCoy, 1981), progressive time delay (Aeschleman & Higgins, 1982), and an antecedent prompting procedure (Dorry, 1976). Mixed results were found in the McGee and McCoy i 98 1) study comparing stimulus fading and progressive time delay; subjects' history with either procedure appeared to increase the effectiveness of that procedure. Finally, a combined strategy of fading cues on the relevant and irrelevant dimension was more effective than either strategy in isolation (Irvin & Bellamy, 1977). Efficiency. Nine of the 11 stimulus modification procedures reported some efficiency data. Stimulus shaping or stimulus fading produced a lower number or lower percentage of errors than most-to-least (Schreibman, 1975), trial-and-error (Aeschleman & Higgins, 1982; Richmond & Bell, 1986; Strand & Morris, 1986; Zawlocki & Walls, 1983), time delay or error correction McGee & McCoy, 1981), and most-to-least or trial-and-error (Richmond & Bell, 1983; Richmond & Bell, 1986). However, no significant differences existed in the number of errors between most-to-least and stimulus-fading strategies in one study (Strand & Morris, 1986). The combined stimulus-fading strategy used in the Irvin and Bellamy (1977) study produced fewer errors than the stimulus-fading strategies used in isolation, and the fading strategy on the irrelevant dimension had fewer errors than fading on the relevant dimension.

In terms of trials to criterion, stimulus fading had fewer trials to criterion than most-to-least (Wolfe & Cuvo, 1978) and trial-and-error (Strand & Morris, 1986). However, Strand and Morris (1986) reported no significant differences in the number of trials between most-to-least and stimulus fading. Aeschleman and Higgins (1982) reported no significant differences in trials to criterion between stimulus fading, progressive time delay, and trial and error. Error correction had fewer trials followed by progressive delay, and finally, stimulus fading had the most trials in the McGee and McCoy (1981) investigation. Irvin and Bellamy (1977) reported that the combined stimulus manipulation strategy had the fewest trials to criterion followed by stimulus fading on the irrelevant dimension and then fading on the relevant dimension. Time Delay Versus Other Strategies In addition to the McGee and McCoy (1981) and Aeschleman and Higgins (1982) studies, progressive time delay was compared with the system of least prompts and with an unlimited delay (i.e., error correction) strategy. Constant time delay has been compared to the system of least prompts and progressive time delay. Effectiveness. McGee and McCoy (1981) found that subjects' history determined whether progressive time delay or stimulus fading was more effective, and found that progressive time delay was more effective than error correction. In the other studies, all compared strategies were effective. Efficiency. Progressive time delay was more efficient than the system of least prompts in terms of sessions, trials, errors, and direct instructional time to criterion (Bennett, Gast, Wolery, & Schuster, 1986; Godby, Gast, & Wolery, 1987). Constant time delay was more efficient than system of least prompts in terms of sessions, trials, percent of errors, and direct instruction time (Ault, Wolery, Gast, Doyle, & Eizenstat, in press; Doyle, Wolery, Gast, Ault, & Wiley, in press; Gast, Ault, Wolery, Doyle, & Belanger, 1988). Progressive time delay and constant delay were essentially equal in terms of these efficiency measures (Ault, Gast, & Wolery, 1988). Progressive time delay produced fewer errors than error correction (Aeschleman & Higgins, 1982; Walls, Dowler, Haught, & Zawlocki, 1984). System of Least Prompts Versus Other Strategies The system of least prompts strategy has been compared not only to time delay and stimulus shaping, but also to a most-to-least strategy, an error correction strategy, and an integrated strategy of stimulus shaping plus system of least prompts. Effectivness. With the system of least prompts and most-to-least prompting, Glendenning, Adams, and Stemberg (1983) and Walls, Crist, Sienicki, and Grant i 98 1) trained students for a specific number of trials or minutes rather than to criterion; therefore, no effectiveness statements can be made. However, Csapo (1981) found both strategies were effective in building fluency, and Day (1987) reported that both procedures produced an increase in correct responding; more improvement resulted when subjects were taught using the most-to-least procedure. Mosk and Bucher (1984) reported the integrated strategy of stimulus shaping plus the system of least prompts was more effective than the system of least prompts alone. Rynders, Behlen, and Horrobin (1979) found that neither the least-prompts system nor the error-correction system were effective with all students, but the system of least-prompts strategy resulted in more problems being solved than the error-correction strategy. Efficiency. The Rynders et al. (1979) study did not report efficiency measures, and the efficiency measures with the most-to-least and system of least-prompting strategies are mixed. The system of least prompts produced faster learning than most-to-least prompts (Csapo, 198 1), but it also produced more errors than the most-to-least prompt strategy (Csapo, 1981; Day, 1987) and less self-initiated responding (Glendenning, Adams, & Stemberg, 1983). However, Walls et al. (1984) found no differences between the two procedures in terms of errors or minutes of instructional time. Mosk and Bucher (1984) stated that use of the system of least prompts plus stimulus shaping resulted in fewer errors, trials, and prompts than the system of least prompts alone. Naturalistic Versus Other Strategies Three studies in the literature were found which compared incidental teaching with either a question-label" strategy or an error-correction strategy. Effectiveness. Cavallaro and Bambara (1982) compared incidental teaching with a question-label strategy which involved the teacher asking questions and describing objects in the environment, but with the child not required to speak before gaining access to materials. The incidental teaching strategy was more effective than the question-label condition. When compared with error correction, incidental teaching was found to have no difference in effectiveness, but produced greater generalization than error correction (McGee, Krantz, & McClannahan, 1985). However, in the Neef, Walters, and Egel (1984) investigation, incidental teaching was effective whereas error correction was not. Efficiency. The only efficiency measure found in these studies is reported by McGee et al. (1985), who stated that no significant difference in terms of training time existed between incidental teaching and error correction. Experimental Design A variety of designs have been used in these comparative investigations. A group design involving statistical analyses of data was the most frequently used design (e.g., Ellis et al., 1980; Strand & Morris, 1986). Other designs used in order of frequency of use have been the parallel treatments design (e.g., Bennett et al., 1986; Godby et al., 1987), multiple baseline design (McGee et al., 1985; Neef et al., 1984), alternating treatments design Cavallaro & Bambara, 1981; Day, 1987), repeated measures design (Walsh & Lamberts, 1979; Zane et al., 1981), and an ABA withdrawal design (Csapo, 198 1).

When comparing two or more instructional strategies with learners who exhibit moderate to profound handicaps, investigators are faced with several problems. Frequently, sufficient numbers of similar subjects are not available to use traditional, between-group designs. To control for the heterogeneity of subjects investigators have employed single-subject research designs. However, when single-subject designs are used, investigators must control for multi-treatment interference and sequence effects (Gast & Wolery, 1988; Sindelar, Rosenberg, & Wilson, 1985). Further, when the investigator is interested in the efficiency of each strategy, the design must allow a single instructional strategy to be applied to the behavior(s) from baseline levels to criterion levels. These problems and the manner in which different single subject designs control for these issues are discussed by Wolery and Gast (in press). DISCUSSION Findings and Recommendations for Practice In this review, 31 studies included direct comparisons between instructional strategies in teaching students with moderate to severe handicaps. Of the strategies investigated, all were effective in teaching students new behaviors in at least some of the studies. However, there were differences in terms of efficiency. These findings are summarized in Figure 2 and discussed in the following paragraphs.

In short, the stimulus modification procedures (stimulus shaping, stimulus fading) appear to be more efficient in terms of errors and direct instruction time than trial-and-error, error correction, antecedent prompt and test, most-to-least prompting, and progressive time delay. However, subjects' history with the time delay procedure can reverse this conclusion. Despite this apparent superiority of the stimulus modification procedures, recommendations for their use should be tempered by (a) the extensive material preparation time required, (b) the difficulty in modifying the stimuli for some behaviors and tasks, and (c) their general complexity.

The response prompting strategies (most-to-least prompting and progressive time delay, system of least prompts, antecedent prompt and test) that have been compared to trial-and-error and error-correction procedures are more effective and were more efficient in terms of errors to criterion. This finding suggests that providing prompts prior to students' responses may decrease the probability of errors and increase the probability of error-free learning. This is particularly desirable with students who tend to display maladaptive behaviors when presented with difficult tasks (i.e., those where errors are frequent) (cf. Carr & Durand, 1985). Given the superiority of stimulus-modification procedures and response-prompting strategies over trial-and-error and error-correction procedures, there is little justification for using trial-and-error or error-correction procedures. However, if for given students these procedures appear to be as efficient as more complex procedures, then their simplicity would suggest that their use is justified.

The system of least prompts appears to be less efficient in some cases than most-to-least prompting in terms of errors to criterion, and produces more errors, trials, and minutes of direct instruction time than progressive time delay and constant time delay. Thus, despite frequent use of the system of least prompts in the literature (Doyle, Wolery, Ault, & Gast, 1988) little justification exists for its continued use. This statement is particularly true with imitative students and discrete responses. Recent data also suggest that constant time delay may be as efficient as the system of least prompts for chained tasks (McDonnell, 1987).

Based on one study with students who had moderate handicaps, it appears that progressive time delay and constant time delay are both effective and produce similar efficiency data. However, for isolated students a definite advantage may exist for o he strategy over the other. These findings are consistent with two studies with students having mild handicaps (Precious, 1985; Thomas, 1986).

Finally, combining stimulus shaping and stimulus fading or combining one of these stimulusmodification procedures with a response-prompting strategy appears to be more effective and efficient than any single strategy. This finding suggests that providing a massive amount of support for learning is more likely to increase the probability of learning than providing smaller amounts of support. Recommendations for Future Research Additional statements or recommendations from research for practice cannot be made because of the insufficient number of comparative and/or replication investigations. However, four statements can be made concerning additional research.

First, additional comparative investigations are warranted. Despite finding 31 investigations that compared two or more of the 13 instructional strategies, some strategies have not been compared to others. Of the possible 78 comparisons between the 13 identified strategies, only 19 have been studied, and 8 of these 19 have been addressed in only one study. The most studied strategies were error correction, most-to-least prompting, system of least prompts, progressive time delay, and stimulus fading. Less frequently studied strategies include antecedent prompt and test, constant time delay, stimulus shaping, and incidental teaching. Strategies that were not included in any of the comparative studies included antecedent prompt and fade, mandmodel procedure, and naturalistic time delay. Some of the summary statements were made on the basis of a growing data base, but others were made on a relatively limited set of studies. For example, additional studies testing the effectiveness of stimulus-modification procedures to response-prompting prOcedures probably are not needed, nor are additional studies comparing stimulus modification or response prompting strategies to error correction and trial-and-error.

The data are fairly consistent that the stimulus-modification procedures will be more efficient than response-prompting strategies, and response-prompting-strategies will be more efficient than error-correction and trial-and-error procedures. However, within the response prompting strategies, more replications are needed. These replications should occur across populations (e.g., different ages, functioning levels, and disabilities), across the types of tasks being taught (e.g., chained or discrete responses), and across relevant curricular domains (e.g., communication, social, self-care, domestic, community living, leisure, and cognitive skills). Further, given the frequent recommendations urging practitioners to use naturalistic teaching strategies (cf. Halle, Alpert, & Anderson, 1984; Kaiser, Alpert, & Warren, 1987), more comparative research is needed with these procedures and the response prompting strategies. Specifically, incidental teaching, naturalistic time delay, and the mand-model procedure should be compared to constant and progressive time delay, antecedent prompt and test, and most-to-least prompting.

Second, while some of the strategies have been investigated extensively in isolation (e.g., system of least prompts, Doyle et al., 1988; time delay, Handen & Zane, 1987), the parameters of the procedures have not been investigated fully. For example, is the system of least prompts more efficient if reinforcement is provided only for unprompted correct responses than if it is provided for unprompted and prompted correct responses? Does learning occur more rapidly with the progressive time delay procedure when the delay interval is increased by trial rather than by blocks of trials, and can teachers in the field reliably implement progressive delay when the interval is increased by trial? What criterion levels are most appropriate when the most-to-least prompting strategy is used? What adaptations must be made, if any, in these procedures to implement them in group instructional arrangements and/or in distributed trial formats? Answers to these and similar questions would identify the "most efficient" manner for implementing each strategy. Perhaps these studies should first be conducted and the most efficient forms of two or more strategies should be compared to each other.

Third, the efficiency measures used in comparative research should be expanded. In the investigations described in this review, efficiency measures included sessions, prompts, trials, errors, and minutes of direct instruction time to criterion. While these are useful measures, other efficiency measures may be more important. Of particular importance is whether students can learn either a rule or incidental information that will be useful in later instruction. For example, if a rule were inserted into the prompt levels of the system of least prompts or most-to-least prompting strategies, would students incidentally learn the rule and apply it to new materials or problems where it is needed? Or would students learn to classify stimuli based on the information presented in prompt levels or in the descriptive praise statements following correct responses? If teachers can get "two new behaviors for the price of one," then true efficiency would be demonstrated. Related to this recommendation is the notion of studying whether differential generalization occurs when one procedure is compared to another.

Fourth, additional research is needed to identify which strategy is most efficient with different types of students and skills. For example, one study (Ault et al., 1988) indicates that progressive and constant time delay are roughly equivalent in terms of efficiency; however, for individual subjects, one of the two strategies was clearly more efficient. If further research continues to indicate that this is the case, then additional research should be conducted to develop an "assessment procedure" to identify which strategy will be most efficient for individual students. Similarly, additional research should identify which subject variables appear to predict superior efficiency with a given instructional strategy. SUMMARY When selecting instructional strategies, instructors should consider several variables. These include the potential harmfulness of each procedure, their intrusiveness and restrictiveness, the response patterns of individual students, the phase of learning, the law of parsimony, social validity of the procedure, and the existing data base (Wolery et al., 1986). This review presents the current data base and provides recommendations for practice based on that research. Further, directions for future research are discussed. The applications of what is currently known and the resolution of these research issues will lead to an instructional technology where the selection of strategies is based on empirical evidence rather than teacher preference or other subjective determinants.
COPYRIGHT 1989 Council for Exceptional Children
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1989 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Ault, Melinda Jones; Wolery, Mark; Doyle, Patricia Munson; Gast, David L.
Publication:Exceptional Children
Date:Jan 1, 1989
Words:3784
Previous Article:Is AIDS a biasing factor in teacher judgment?
Next Article:Disciplining handicapped students: legal issues in light of Honig v. Doe.
Topics:


Related Articles
Effects of social integration on preschool children with handicaps.
Goal structure effects on social interaction: nondisabled and disabled elementary students.
Cooperative learning: does it improve the academic achievement of students with handicaps?
A response to Esposito and Koorland: a bias in search of supporting data.
When cooperative learning improves the achievement of students with mild disabilities: a response to Tateyama-Sniezek.
Effects of preschool integration for children with disabilities.
The relationship between instructional variables and problem behavior: a review.
The sociometric status of students with disabilities in a full-inclusion school.
Paraprofessional Support of Students With Disabilities: Literature From the Past Decade.
Increasing peer interactions for students with severe disabilities via paraprofessional training.

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