BEHAVIOR THERAPY 7, 76--88 (1976)
Deviant Classroom Behavior as a Function of Combinations of Social and Token Reinforcement and Cost Contingency H I L L M . W A L K E R , H Y M A N HOPS, AND E D W A R D F I E G E N B A U M
University of Oregon The purpose of this study was to evaluate a series of program variables in modifying deviant classroom behavior within an experimental class setting. Experiment I evaluated combinations of one setting variable and three treatment variables (social reinforcement, token reinforcement, and cost contingency) in modifying the behavior of five subjects. Experiment II replicated the treatment procedures on a second group of subjects and evaluated combined effects of the above variables when applied to deviant behavior over an extended treatment period. Combinations of variables were less effective in controlling behavior than was simultaneous application of all treatment variables. Functional relationships were established between increases in appropriate behavior and both token reinforcement and cost contingency. Increases in appropriate behavior were associated with a change in setting and with manipulation of social reinforcement from the teachers. However, no attempt was made to establish functional relationships between these two variables and changes in appropriate behavior. Effects obtained with the combined replication of all treatment variables in Experiment II were virtually identical to those obtained in Experiment I.
Recently reviewed techniques for modifying classroom behavior include token reinforcement (O'Leary & Drabman, 1971 ; Kazdin & Bootzin, 1972); teacher attention and social reinforcement (Lipe & Jung, 1971); and cost contingency (Kazdin, 1972). Teacher attention and praise have been used to modify the study behavior of minimally disruptive children; token reinforcement and cost contingency have frequently been applied to more deviant classroom behavior. The effectiveness of these variables, either singly or in combination, has been well documented. Several studies have analyzed different combinations of treatment variables in the modification of minimally disruptive classroom behavior. Thomas, Nielsen, Kuypers, and Becker (1968) used social reinforcement This research was performed pursuant to contract No. OEG-0-72-0702, Bureau of the Handicapped, Office of Education, U.S. Department of Health, Education and Welfare through the Center at Oregon for Research in the Behavioral Education of the Handicapped. The authors thank Roberta Taussig for editing the manuscript, and Charles Greenwood and Joe Delquadri for critiquing it. Requests for reprints of this article should be sent to: Hill M. Walker, CORBEH, 1590 Willamette St., Eugene, OR 97401. The third author is now at the University of Utah. 76 Copyright© 1976by Associationfor Advancementof BehaviorTherapy. All rightsof reproductionin any form reserved.
COMBINATIONS OF PRAISE, TOKENS, AND COST
77
from the teacher and remedial instruction to eliminate a classroom behavior problem. The effectiveness of rules, praise, and ignoring as elements of elementary classroom control were examined by Madsen, Becker, and Thomas (1968). While rules alone had little influence upon classroom behavior, the combination of ignoring inappropriate behavior and showing approval for appropriate behavior was very effective. In contrast, O'Leary, Becker, Evans, and Saudargas (1969) found rules, structure, and the combination of praising appropriate behavior and ignoring inappropriate behavior to be generally ineffective in reducing disruptive behavior. Such apparent contradictions indicate a need for additional research in this area. Baer, Wolf, and Risley (1968) suggest that current behavioral procedures are complex and often diffuse in their application; when these approaches succeed, they clearly need to be analyzed into their effective components. The present study involved two separate experiments. Experiment I analyzed the relative effectiveness of various combinations of: (1) change in setting; (2) social reinforcement; (3) token reinforcement; and (4) cost contingency in modifying deviant classroom behavior. Experiment II evaluated effects of the total package condition on a second group of subjects. METHOD Subjects The first group studied consisted of four boys and one gift; the second, of five boys. The children, ranging in age from 6 to 9 years and enrolled in grades one, two, or three, were referred from elementary schools in the local school setting. All subjects were screened using behavior check-list ratings, achievement tests, standard auditory, visual, and general health tests, and observational data taken in the regular classroom. The subjects selected met the following criteria: (I) high scores on the acting-out subscale of the Walker Problem Behavior Identification Checklist (Walker, 1970); (2) high rates of disruptive classroom behaviors, such as noisy, aggressive movement around the room, and inappropriate peer interaction; (3) average or above average scores on intelligence tests; (4) academic performance deficits in the basic skills areas ranging from 3 months to 1.5+ years; (5) no gross physical or sensory deficit; and (6) low rates of appropriate behavior in the regular classroom setting relative to their peers.
Setting 1 The experimental classroom facilities adjoined and were affiliated with a public elementary school in the Eugene School District. The primary area for academic activities contained six double desks of approximately 20 x 45" (518 × 1143 mm) work surface, the teacher's desk, shelves, and tables for the display of materials for science and art projects, and a carpentry room with a variety of tools and wood. Adjacent rooms provided sink and table facilities and an observation room with a one-way mirror. Space was also available for individual testing, 1 Complete details of the experimental classroom setting can be found in Walker et al. (1971).
78
WALKER, HOPS AND FIEGENBAUM
tutoring, and remedial instruction. A small isolation (timeout) room containing a chair and desk adjoined the classroom. The children used the same playground and lunch facilities as the students regularly enrolled in the adjoining school. The children were in the experimental class daily for approximately 3 hr, 45 rain. Of that time, approximately 2 hr and 50 min were devoted to academic assignments, 45 min to physical education and recess, and 10 to point counting and exchanging tokens for back-up reinforcers at the end of the day.
Token R e i n f o r c e m e n t S y s t e m Token reinforcement was based on a point dispensing system. Subjects were able to earn points for appropriate social and academic behavior which could later be exchanged for such tangible back-up reinforcers as model cars and airplanes, games, books, chess sets, chemistry sets, toys, and athletic equipment. A large variety of back-up reinforcers was provided to increase the likelihood that at least one of them would be relevant to the deprivation conditions of any given subject (Ferster & DeMyer, 1962). Each child could earn a maximum of 35 points per day on the basis of two concurrent schedules (Catania, 1966, chap. 6). Subjects could receive points on both a variable interval schedule of reinforcement for appropriate classroom behavior and a fixed ratio schedule for correct academic responses and completion of assignments. At the end of each day, points could be exchanged for back-up reinforcers. These ranged in value from 25 to 200 points with occasional special items for 500 points. The point values, related to actual purchase price, were so arranged that sufficient points for the least expensive reinforcing stimulus could be earned with a very high proportion of appropriate behavior produced within a single day. However, the subjects were free to exchange their points for an inexpensive item or accumulate them for a more expensive one.
Classroom Observations Behavior coding system. An 11-category behavioral coding system was used to record the classroom behavior of the subjects. Five codes were designated a priori as appropriate (Appropriate Work, Group Activity, Appropriate Vocalization, Positive Physical Contact, Appropriate Movement), and six as inappropriate (Inappropriate Work, Nonattending, Noisy, Inappropriate Vocalization, Negative Physical Contact, Inappropriate Movement). Each subject's behaviors were recorded in 15-sec intervals for 6-min blocks. The 15-sec intervals were signaled via an earphone mounted in the observer's clipboard which contained a signal-emitting device. Observers were free to code more than one subject behavior during any interval. A minimum of 120 min of observation data was recorded for each subject over a 2-week baseline period in the regular classroom. During treatment in the experimental classroom, daily observations were made by having trained observers rotate among the subjects for 6-min blocks during all academic periods. Observer training and reliability. Observers were chosen from a volunteer population of graduate and undergraduate students enrolled in education and psychology courses. During the course of the study, a total of 10 observers plus a calibrating observer were used. Observer training began with mastery of the observation code definitions and manual of procedures. Next, videotapes of previous groups as well as the regular and experimental classroom settings were used for practice until five consecutive 6-rain observations with reliability coefficients of.90 or greater between each observer and the calibrating observer was reached. Weekly spot checks were used to monitor and maintain reliability. Observer reliability was calculated using the percentage agreement method. Agreements and disagreements were scored for each interval. Mean reliabilities during weekly spot checks ranged from 90 to 97%
COMBINATIONS OF PRAISE, TOKENS, AND COST
79
and averaged 93% across observers; by behavior category they ranged from 65 (negative physical contact) to 100% (positive physical contact).
EXPERIMENT
I
The design of this experiment allowed for evaluation of several combinat i o n s o f t h e t r e a t m e n t v a r i a b l e s . C o m p o n e n t s w e r e a d d e d in a c u m u l a t i v e fashion, with reversal designs used to evaluate the functional relationships between token reinforcement, cost contingency, and behavioral outcomes.
P r o c e d u res The study consisted of 10 phases over a period of approximately 4.5 months; (1) baseline1 (regular classroom), (2) baseline2 (experimental classroom), (3) social reinforcement, (4) social plus token reinforcement, (5) social reinforcement, (6) social plus token reinforcement, (7) social plus tokens plus cost contingency, (8) social plus tokens, (9) social plus tokens plus cost contingency, and (10) social plus cost contingency plus fading tokens. Phases 2, 5, 8, and 9 lasted 1 week (five school days); phases 1,3, 4, 7, and 10, 2 weeks (10 school days); and phase 6, 4 weeks (20 school days).
Phase 1, Baseline1 (Regular Classroom) Prior to entering the experimental classroom, 20 6-min observations were collected on each child in his regular classroom over a 2-week baseline period. Parents and teachers were asked not to inform the child of their selection for the experimental classroom until after the 2-week baseline period was over.
Phase 2. Baseline2 (Experimental Classroom) Observation data were also collected during a second, 1-week baseline period in the experimental classroom. The purpose of this phase was to test for behavior change(s) associated with a new setting, new materials, and new teachers. However, an attempt was made to eliminate effect(s) due to change in reinforcement schedules, by replicating the quality and quantity of positive and negative teacher attention each child had received in his regular classroom. The density per hour of teacher praise for appropriate behavior and teacher disapproval for inappropriate behavior that had been dispensed by each child's regular classroom teacher was calculated from data collected during baselinel, and the rates prorated for the 4-hr day of the experimental classroom. Teachers in the experimental classroom dispensed the praise and disapproval events uniformly over a 4-hr day. No attempt was made to equate total teacher attention received by the subjects in their regular classrooms. With a teacher-pupil ratio of 1:2.5 in the experimental classroom compared to 1:24 in the regular classroom, the children would have had to be on extinction for most of the day. However, the teachers were instructed to behave in a neutral manner during their other interactions with the children.
Phase 3. Social Reinforcement In this phase, the density of social reinforcement was markedly increased. Each child received approximately 30 praises per day for appropriate classroom (academic and social) behavior. The range of appropriate behavior, however, varied across the subjects. Inappropriate behavior was ignored during this phase. To insure that social reinforcement was distributed evenly, the class day was divided into four 1-hr periods, and approximately 7.5 praises were dispensed per hour. The teachers monitored their own behavior by recording each praise on a form on the child's desk. A short
80
WALKER, HOPS AND FIEGENBAUM
delay between the praise statement and the recording inhibited the subject's associating praise with the recording event.
Phase 4. Social Plus Token Reinforcement In this phase, the number of social reinforcers dispensed (approximately 30 per day per child) remained constant. In a parallel arrangement, teachers were instructed to dispense from 20 to 35 tokens as well, the 15-point range allowing the differential reinforcement of high and low quality performance. Social reinforcement preceded each token dispensed, but, since the latter was more variable, some praise was delivered alone.
Phases 5 and 6 Phases 5 and 6 were identical to Phases 3 and 4, respectively.
Phase 7. Social Plus Tokens Plus Cost Contingency Cost contingency involved the loss of earned tokens contingent upon the occurrence of specific deviant behaviors. Talking back to teacher, talking-out, not attending, out of seat, disturbing others, and playing with objects cost the child one point; teacher defiance cost two points; swearing, three points; and fighting and throwing objects, four points. When a deviant behavior occurred, the teacher placed a mark on a cost contingency record sheet beside the name of the specific deviant behavior located on the child's desk. Thus, the child knew the behavior to which it was applied, as well as the number of points lost. When a child had lost more points than he had earned, he was placed in the timeout room for 20 min. Thus, timeout was used as a back-up for children with initial high rates of deviant behavior. To insure that each child understood the procedure, the teacher discussed with the class the deviant behaviors to which cost contingency would be applied at the beginning of the phase.
Phases 8 and 9 Phases 8 and 9 were identical to Phases 6 and 7, respectively.
Phase 10. Social Plus Cost Contingency Plus Point Fade Conditions in this phase were identical to those in Phases 7 and 9, except that the number of points that could be earned per day was gradually reduced. During the first 3 days, the maximum number was reduced from 35 to 24 points, and then taken in three steps down to 11 points during the last 2 days of the fading point. Tokens were faded to prepare the children for return into their regular classrooms. No announcement was made, and all children continued receiving 30 social reinforcers per day. Cost contingency continued to be applied whenever deviant behavior occurred.
Results
Figure 1 shows the daily percentage of appropriate behavior for the group and for individual subjects during each phase of Experiment I. Each data point is the average daily proportion of observed time spent in the appropriate behavior categories. The five children averaged 34% appropriate behavior during baselinel in their regular classrooms. This increased during baseline2 in the experimental classroom to 47%. The introduction of social reinforcement during Phase 3 was associated with a further increase to a mean of 59% appropriate behavior. When tokens were introduced in Phase 4, the level of appropriate be-
81
COMBINATIONS OF PRAISE, TOKENS, A N D COST
u®_o=o~o=
uo
~o
o=
100
oo
5C • ~-• " ° " - ,
75
scz., /
"
~v'~'*"
o ~ o ~o~
=
,_.,.,.,.......,
......
. ......
o~oo~o
o o ~ o o o
_
~
.._...
<~g2~o ................... All Subjects
i"
\'~'1
i
$1
!\~i'\ ;.. • v' v i . a /
\.i" "\,.. ./"\!' "'- \.
""
:~' " ~-" ~
....
""
.....~.-'--.- v \ / ! oo / ,~
S2
@_,..
..\j
j
\ / \ I ~ . . ..,'-7\,'X"P-. ........ ~,\. "~'- " ~ - ~ " ~
==
S4
~= C
V "" 5 - Absent
\J/ t0
45
Ai\"
" ' 20
25
50
so 55
40
45
50
55
60
65
70
75
BO
D0ys
FiG. 1. Appropriate behavior as a function of change in setting, social reinforcement, token reinforcement, and cost contingency.
havior increased to 78%, but decreased to 66% when tokens were withdrawn in Phase 5. The reintroduction of tokens in Phase 6 returned the group mean to 79%. This replicated the effects of Phase 4 and showed a relationship between tokens and appropriate behavior, based on a constant level of social reinforcement. Cost contingency, introduced in Phase 7, produced a mean increase of 17% from 79 to 96%. This level dropped to 82% in Phase 8 with the removal of cost contingency, and returned to 96% when cost contingency was reintroduced in Phase 9. Thus, a causal relationship between cost and appropriate behavior was established on a base of social and token reinforcement. Fading points in Phase 10 had no effect upon the average amount of appropriate behavior shown by the five children. The group means in Phases 9 and 10 were 96 and 95%, respectively. Discussion
The purpose of Experiment I was to provide data about the differential effects of combinations of treatment variables in producing behavior
82
WALKER, HOPS AND FIEGENBAUM
change. The most powerful treatment effects were produced in those phases (7, 9, and 10) in which all three treatment variables and one setting variable were in operation (Fig. 1). The approximate proportion of appropriate behavior accounted for by the addition of tokens and cost was 19 and 17%, respectively. These results are consistent with prior research by Walker, Mattson, and Buckley (1971), who found that combining token reinforcement, social reinforcement, and aversive controls (timeout/suspension) was more effective than combinations of any two components. Greenwood, Hops, Delquadri, and Guild (1974) also found that a package condition containing praise and token reinforcement plus rules and feedback was superior to a combination of the latter two. In a study of teacher training techniques, adding social praise to feedback produced higher levels of teacher attention to student attending behavior than did either instructions or feedback alone (Cossairt, Hall, & Hopkins, 1973). Collectively, these data suggest that intervention procedures containing multiple treatment variables that are applied simultaneously may produce higher magnitude effects than do components of such interventions applied one or two at a time. This may be particularly true of studies in which positive reinforcement is used to increase appropriate behavior, while mild punishment procedures are simultaneously applied to inappropriate behavior (Bostow & Bailey, 1969). There was a 13% treatment effect associated with a simple change in setting from the regular classroom to the experimental classroom (Fig. 1). Although the present study did not attempt an experimental analysis of this effect, it has been replicated in studies of children with low study skills (Walker & Hops, Note 1) and socially withdrawn children (Walker, Hops & Greenwood, Note 2). If a setting change of the type described herein were functionally related to behavior change its effect may have been due to novel stimuli associated with the special class setting, such as reduced teacher-student ratios, new instructional materials, individualized instruction, and increased teacher attention. Another effect noted, but not analyzed, in this study was the 12% increase in appropriate behavior which followed introduction of social reinforcement. This effect has been well documented in studies by Hall, Lund, and Jackson (1968); Becker, Madsen, Arnold, and Thomas (1967); and Thomas, Becker, and Armstrong (1968). Overall, a treatment model consisting of change in setting, social reinforcement, tokens, and cost contingency appears to be very effective in modifying deviant classroom behavior. However, the procedure of adding variables cumulatively in Experiment I did not allow for the evaluation of token reinforcement or cost contingency in isolation. Nor was it possible to compare directly the effect of social reinforcement plus tokens with that of tokens plus cost contingency. Additional experiments in which such
COMBINATIONS OF PRAISE, TOKENS, AND COST
83
comparisons are made would further measure the relative effects of these variables. It would also be interesting to look at the effect of these variables in the absence of a setting change. A further limitation of the present study is the order in which the variables were introduced, which may have had an effect in determining the outcome. The results might have been different if token reinforcement had been introduced first, or social reinforcement last. A replication of the present study and studies which test for potential order effects are needed in order to describe precisely the parameters of each of these variables in modifying appropriate classroom behavior. E X P E R I M E N T II
The purpose of Experiment II was to evaluate the combined effects of change in setting, social reinforcement, token reinforcement, and cost contingency in modifying behavior over a longer period of treatment. During Experiment I these variables were applied singly and in combination for only brief periods; no treatment condition remained in effect for more than four successive weeks. Baseline data were recorded in each subject's regular classroom. All intervention data were recorded in the experimental classroom. Proced ures The package condition combining token reinforcement, social reinforcement, and cost contingency (as in Phases 7 and 9 in Experiment I) was implemented from the first day of intervention and remained in effect throughout the 45-day treatment period. Each day, individual subjects received 30 social reinforcers; possible points ranged from 20 to 35. Cost contingency was applied each time one of the specified deviant behaviors occurred. The instructional program for subjects was identical in Experiments I and II.
Results
There was a very substantial treatment effect associated with the combined application of social, tokens, and cost. The subjects averaged 38% appropriate behavior in baseline and 96% appropriate behavior during treatment in the experimental classroom, t (4) = 25.27, p < .001 (Fig. 2). Overall behavior changes from baseline to treatment are presented by behavior category in Fig. 3 to evaluate the relative responsiveness of each behavior category to treatment in a token economy (O'Leary & Drabman, 1971). Differences are masked when only broad categories such as percentage of study behavior, appropriate behavior, or disruptive behavior are used to evaluate the effects of intervention. During baseline, all five subjects had fairly high rates of nonproductive classroom behavior; Inappropriate Movement, Inappropriate Vocalization, and Nonattending. Inappropriate Work, a relatively low frequency category for four of the five subjects, was recorded approximately 17% of the time for subject 2. All of the subjects had correspondingly low percent-
84
WALKER,
HOPS AND FIEGENBAUM
INTERVEN
TION
75 subiect"
e~°--e..e L L
~
w.\,,, _,_,/o / . e ~ .
, p e ~ o . . . . . %e.a .~ . . . . .
.%,/-" ~ . r ~ , , ,
.........
°~" _
Nee
No..o 25 •
54
a
A
,~\j
.'0".
-
-
75
°\~e~o 1 O0
s
V 25 0
^
5 Z3 --Absenl
t0
,
t5
s
i
20
25
30
35
A
40
45
50
Doys
Fso. 2. Appropriatebehavioras a function of total package conditions. ages in the appropriate behavior categories, the highest being Normative at 13.29%. The greatest relative changes in appropriate behaviors occurred in the Work and Movement categories, which gained approximately five times and three times their own baseline levels, respectively. Four of the inappropriate categories showed almost a 100% decrease over that of baseline; Work, Vocalization, Movement, and Noisy. Inappropriate Physical, not very frequent during baseline but usually considered aversive to teachers, was virtually eliminated. Discussion The simultaneous application of social reinforcers, tokens, and cost in the experimental classroom setting produced an immediate rather than a
85
COMBINATIONS OF PRAISE, TOKENS, AND COST
WORK
~_
--
NORMATIVE
.~
~T,
NON AT TENDING
NOISY
VOCALIZATION PHYSICAL
MOVEMENT
._~
• o
~
p,,. O
,~
so
"I.i
0 I~i
20
g z
~rJ
ALl- SUB,JECTS
g
L
APPROPRIATE
,N~0~ATE
FIG. 3. Behavior category changes from baseline to intervention.
gradual change in behavior. Each subject substantially increased his rate of appropriate behavior from the first day of treatment (Fig. 2). The results of Experiment II replicated the findings of Phases 7, 9, and I0 in Experiment I, in which the combined application of the same package conditions produced high and stable rates of appropriate behavior. However, in spite of the intersubject replication effects, it should be noted that a causal relationship between the package condition and the observed changes in behavior was not established vis-/~-vis an experimental analysis or experimental/ control group comparison in Experiment II. The effect of the treatment model was replicated across behavior categories to varying degrees. All categories of inappropriate behavior showed a decrease from baseline to treatment. All appropriate behavior categories, with the exception of normative, showed a corresponding increase. Differing amounts of cost contingency were required to suppress deviant behavior in the two experiments. The per subject daily frequency applied in Experiment I (mean = 4.48; range = 1.2-8.00) was higher and more variable than it was in Experiment II (mean = 1.55; range = 0.69--2.83).
86
WALKER, HOPS AND FIEGENBAUM
This effect may have been due to the differences in the length of time cost was in operation. In Experiment I, cost was used for a total of 5 weeks, interrupted in the middle by a 1-week withdrawal of the procedure. In contrast, cost was in effect continuously for 3.5 months with the subjects in Experiment II. Thus, cost could have had a greater suppression effect in Experiment II, simply because it was applied consistently for a longer period of time. In summary, the results of Experiment II replicated the findings obtained in Phases 7, 9, and 10 of Experiment I. Evaluation of the combined application of social reinforcers, tokens, and cost suggested these variables were effective in reducing deviant behavior and accelerating appropriate behavior within the experimental class setting.
GENERAL DISCUSSION O'Leary and Drabman (1971) have pointed out that token systems usually consist of a collection of treatment techniques which are applied simultaneously. In most token systems, token reinforcement techniques are not applied in isolation. Praise, timeout, cost contingency, rules, structure, and ignoring (extinction) are often used to supplement the token procedures. While such token programs are generally effective, it is often difficult to determine which component variables, or combination(s) of variables, are responsible for the observed treatment effects. Components analysis techniques can be used to generate data on this question. Components analysis attempts to demonstrate the functional relationships between various treatment variables and behavioral outcomes. The application of these techniques to studies of token systems can contribute to the identification of effective, economic combinations of treatment variables for modifying classroom behavior. This methodology can also be used to help determine the relative contribution of each treatment variable or combination of variables in producing an overall treatment effect. Additional studies of program components are needed so that the most economical, yet powerful, treatment programs can be developed for classroom use. Token-based intervention procedures are needed that are maximally effective, but that contain the smallest possible number of treatment variables. It is entirely possible that many token programs include treatment techniques that are unnecessary, in that they contribute little to an overall treatment effect, ff these techniques could be identified through components analysis, token programs might be greatly simplified while retaining their effectiveness. This would, in turn, reduce staff training, monitoring, and supervisory requirements associated with the token economy (Walker & Buckley, 1974).
COMBINATIONS OF PRAISE, TOKENS, AND COST
87
REFERENCE NOTES 1. Walker, H. M., & Hops, H. Use of normative peer data as a standard for evaluating classroom treatment effects. Journal of Applied Behavior Analysis, in press. 2. Walker, H. M., Hops, H., & Greenwood, C. R. Modification of social withdrawal within an experimental class setting. Paper presented at the annual meeting of the Council for Exceptional Children Convention, New York, April 1974. Center at Oregon for Research in the Behavioral Education of the Handicapped, University of Oregon, 1590 Willamette St., Eugene, OR, 97401.
REFERENCES Baer, D. M., Wolf, M. J., & Risley, T. R. Some current dimensions of applied behavior analysis. Journal of Applied Behavior Analysis, 1968, 1, 91-97. Becker, W. C., Madsen, C. H., Arnold, C. R. & Thomas, B. A. The contingent use of teacher attention and praise in reducing classroom behavior problems. Journal o f SpecialEducation, 1967, 1, 287-307. Bostow, D. E., & Bailey, J. B. Modification of severe disruptive and aggressive behavior using brief timeout and reinforcement procedures. Journal of App6ed Behavior Analysis, 1969, 2, 31-37. Catania, A. C. Concurrent opcrants. In W. Honig (Ed.), Operant behavior: Areas of research and application. New York: Appleton-Century-Crofts, 1966. Cossairt, A., Hall, R. V., & Hopkins, B. L. The effects of experimenter's instructions, feedback, and praise on teacher praise and student attending behavior. Journal of Applied Behavior Analysis, 1973, 6, 8%100. Ferster, C. B., & DeMyer, M. K. A method for the experimental analysis of the behavior of autistic children. The American Journal o f Orthopsychiatry, 1962, 32, 8%98. Greenwood, C. R., Hops, H., Delquadri, J., & Guild, J. Group contingencies for group consequences in classroom management: A further analysis.JournalofAppliedBehavior Analysis, 1974, 7, 413-425. Hall, R. V., Lund, D., & Jackson, D. Effects of teacher attention on study behavior. Journal of Applied Behavior Analysis, 1968, 1, 1-12. Kazdin, A. E. Response cost: The removal of conditioned reinforcers for therapeutic change. Behavior Therapy, 1972, 3, 533-546. Kazdin, A. E., & Bootzin, R. The token economy: An evaluative review. Journal of Applied Behavior Analysis, 1972, 5, 343-373. Lipe, D., & Jung, S. M. Manipulating incentives to enhance school learning. Review o f Educational Research, 1971, 41, 249-280. Madsen, C. H., Jr., Becker, W. C., & Thomas, D. R. Rules, praise, and ignoring: Elements of elementary classroom control. Journal of Applied Behavior Analysis, 1968,1, ! 39-150. O'Leary, K. D., Becker, W. C., Evans, M. B., & Saudargas, R. A. A token reinforcement program in a public school: A replication and systematic analysis. Journal of Applied Behavior Analysis, 1969, 2, 3-13. O'Leary, K. D., & Drabman, R. Token reinforcement programs in the classroom. Psychological Bulletin, 1971, 75, 37%398. Thomas, D. R., Becker, W. C., & Armstrong, M. Production and elimination of disruptive classroom behavior by systematically varying teacher's behavior. Journal of Applied Behavior Analysis, 1968, 1, 35-45. Thomas, A., Nielson, L. J., Kuypers, D. S., & Becker, W. C. Social reinforcement and remedial instruction in the elimination of a classroom behavior problem. Journal o f Special Education, 1968, 2, 291-306.
88
WALKER, HOPS AND FIEGENBAUM
Walker, H. M. The Walker problem behavior identification checklist. Test and manual. Los Angeles: Western Psychological Services, 1970. Walker, H. M., & Buckley, N. K. Token reinforcement techniques. Eugene, OR: E-B Press, 1974. Walker, H. M., Mattson, R. H., & Buckley, N. K. The functional analysis of behavior within an experimental class setting. In W. C. Becker (Ed.), An empirical basis for change in education. Chicago: Science Research Associates, 1971.