Module 11 Student Led Discussion Barlow, D.H., & Hayes, S.C. (1979). Alternating treatments design: One strategy for comparing the effects of two treatments in a single subject. Journal of Applied Behavior Analysis, 12, 199-210. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1311363/pdf/jaba00105-0045.pdfLinks to an external site. Read article and answer 1 question from each group. A 1a. How can multiple treatment interference cause problems with external validity? 2a. How can carryover effects be minimized during ATD intervention? 3a. What are advantages to using ATD compared to withdrawal designs? B 1b. Do you think the design conducted in this study was appropriate for comparing these treatments? Why or why not? 2b. What are your thoughts regarding the quick alternation of treatments, counterbalancing the times and forming discriminative stimulus? 3b. How would you change a component of this study and how would it minimize the carryover effects? C 1c. In your current role and setting, can you create an example of how you would use Alternating Treatment Design? 2c. How would you increase internal validity in an Alternating Treatment Design? 3c. What are some of the disadvantages of using an Alternating Treatment Design? D. 1d. What are the advantages and disadvantages of Alternating Treatment Design? 2d. What are your thoughts on counterbalancing? 3d. What is one suggestion to help prevent a carryover effect? NUMBER 2 (SUMMER 1979) 1979, 1% 199-210 ALTERNATING TREATMENTS DESIGN: ONE STRATEGY FOR COMPARING THE EFFECTS OF TWO TREATMENTS IN A SINGLE SUBJECT DAVID H. BARLOW AND STEVEN C. HAYES JOURNAL OF APPLIED BEHAVIOR ANALYSIS STATE UNIVERSITY OF NEW YORK AT ALBANY AND UNIVERSITY OF NORTH CAROLINA AT GREENSBORO A little used and often confused design, capable of comparing two treatments within a single subject, has been termed, variously, a multielement baseline design, a multiple schedule design, and a randomization design. The background of these terms is reviewed, and a new, more descriptive term, Alternating Treatments Design, is proposed. Critical differences between this design and a Simultaneous Treatment Design are outlined, and experimental questions answerable by each design are noted. Potential problems with multiple treatment interference in this procedure are divided into sequential confounding, carryover effects, and alternation effects and the importance of these issues vis-a-vis other single-case experimental designs is considered. Methods of minimizing multiple treatment interference as well as methods of studying these effects are outlined. Finally, appropriate uses of Alternating Treatments Designs are described and discussed in the context of recent examples. DESCRIPTORS: Single-subject design, methodology, comparison of two treatments To compare the effects of two or more treatments in applied research, each treatment is usually administered to a different group of subjects and differences are noted. Because considerable intersubject variability exists in each group (some subjects change and some do not), inferential statistics are often necessary to determine if an effect exists. This leads to problems in generalizing results from the group average to the individual subject or patient who should benefit from the research (Hersen & Barlow, 1976; Sidman, 1960). To avoid intersubject variability, an ideal solution would be to divide one subject in two and apply two different treatments simultaneously to each identical individual. This would eliminate intersubject variability and allow effects, if any, to be directly observed. Statements about other individuals could then be made through the usual process of replication and “logical generalization” (Edgington, 1966; Hersen & Barlow, 1976). Such a procedure exists in the family of singlecase experimental designs although it has been little used and often confused. It has been termed variously a multiple schedule design (Barlow & Hersen, 1973; Hersen & Barlow, 1976; Leitenberg, 1973), a multielement baseline design (Sidman, 1960; Ulman & Sulzer-Azaroff, Note 1), and a randomization design (Edgington, 1967). In addition, Kazdin and Hartmann (in press) use the term Simultaneous Treatment Design (see below). These terms were originated for somewhat different reasons, reflecting the multiple historical origins of single case research (Hersen & Barlow, 1976). Several proponents of the term multiple schedule (see Hersen & Barlow, 1976); Leitenberg, 1973) were associated in Vermont in the 1960s in an effort to apply operant procedures and methods to clinical problems (e.g., Agras, Leitenberg, Barlow, Reprint requests should be sent to David H. Barlow, Psychology Department, State University of & Thomson, 1969). These procedures and termiNew York, Albany, New York 12222. We would nology were derived directly from operant lablike to thank Harold Leitenberg, Mike Zeiler, and Warren Steinman for comments on an earlier draft oratories; but the term multiple schedule implies a distinct schedule associated with each stimulus of this manuscript. 199 200 DAVID H. BARLOW and STEVEN C. HAYES component, and this may not always obtain in makes this approach procedurally similar to the applied research, resulting in an unnecessary two approaches described above. narrowness in the term. Ulman and Sulzer-AzaThe basic feature of this design, under its variroff (1975) use Sidman’s term multielement ous names, is the fast alternation of two different baseline design to describe a procedure in which treatments or conditions, each associated with a different conditions or “treatments” are associ- distinct and discriminative stimulus. As Leitenated with different stimuli to establish experi- berg (1973) points out, this design “is based on mental control. In fact, multielement baselines discrimination learning principles; that is, if the as conceived by Sidman (1960) have little par- same behavior is treated differently in the presallel in applied research since the purpose is to ence of different physical or social stimuli, it “investigate relations between some single ex- will exhibit different characteristics in the presperimental operation and more than one be- ence of these stimuli” (p. 93). Thus in the typical havioral baseline” (p. 326). Of more direct design, after a baseline period, two treatments relevance is Sidman’s term multielement manip- (A and B) are administered, alternating with ulation in which the purpose is to study “the in- each other, and the effects on one behavior are teraction between a single behavioral baseline observed. For example, A may be administered and several qualitatively or quantitatively differ- in the morning and B in the afternoon, preceded ent experimental operations” (p. 323); in other by instructions such as, “This is treatment A” words, a comparison of the effects of two or more and “This is treatment B.” Conditions which treatments on one behavior. As in much basic might affect data other than treatments are counresearch, however, Sidman’s examples illustrate terbalanced as the experiment continues, such high-rate behavior brought to a point of stability as time of day, therapist administering the treatbefore introduction of experimental operation ment, or location of the treatment. For example, or “treatments.” Applied research, on the other B might be given in the morning one day and hand, is more often concerned with low-rate be- the afternoon the next. The data are plotted separately for each intervention to provide a haviors which are unstable. Edgington (1967, 1972), from a position out- ready visual representation of the effects of each side of operant psychology, originated the term treatment. Because confounding factors such as randomization design to describe his variation of time of administration have been neutralized a time series approach amenable to statistical (presumably) by counterbalancing, and because analysis. The design differs slightly from tradi- the two treatments are readily discriminable by tional operant application in that treatments are subjects through instructions or other discrimideliberately randomized across times of applica- native stimuli, differences in the individual plots tion: for example, ABBABAA rather than of behavior change corresponding with each ABAB. Treatments are repeated often enough treatment should be attributable to the treatment to allow statistical comparison of A and B phases itself, allowing a direct comparison between two (continuing a tradition begun by R. A. Fisher (or more) treatments. [Also see Kazdin and (0000) who explored the abilities of a lady to Hartmann (in press) for a discussion of the logic discriminate tea prepared in two different ways). of this design.] For example, McCullough et al. (1974) deThis frequent repetition of treatments usually requires fast alternation to obtain the necessary scribed treatment of disruptive behavior in a number of random observations, and the almost 6-year-old boy. Following a 5-day baseline peunavoidable discriminability of each condition riod in which cooperative behavior was meain applied research with human (e.g., McCul- sured, two treatments were introduced for a total lough, Cornell, McDaniel, & Mueller, 1974) of 4 days: (a) social reinforcement for coopera- 201 ALTERNATING TREATMENTS DESIGN A CASE STUDY PHASE ONE PHASE TWO PHASE THREE loo 0 100 cc 1
Purchase answer to see full attachment