Ritalin vs. response cost in the control of hyperactive children: a within-subject comparison

JOURNAL OF APPLIED BEHAVIOR ANALYSIS

RITALIN VS. RESPONSE COST IN THE CONTROL OFHYPERACTIVE CHILDREN: A WITHIN-SUBJECT

COMPARISON

MARK D. RAPPORT, H. ALLEN MURPHY, AND JON S. BAILEY

FLORIDA STATE UNIVERSITY

A within-subject comparison was made of the effects of methylphenidate (Ritalin) andresponse cost in reducing the off-task behavior of two boys, 7 and 8 years of age, whohad been diagnosed as having an attentional deficit disorder with hyperactivity. Severaldosages of Ritalin (5 to 20 mg/day) were evaluated with the results indicating varyingeffects of the drug for both children. Response cost (with free-time as the reinforcer)was superior to Ritalin in raising levels of on-task behavior and in improving academicperformance.DESCRIPTORS: hyperactivity, response cost, medication, Ritalin

Attentional deficit disorder with hyperactivityis a well-publicized and controversial childhoodproblem. Recent estimates have placed the inci-dence rate at 1.19% of the elementary schoolpopulation with relatively constant rates ob-served across grade levels (Lambert, Sandoval,& Sassone, 1978). These children are particu-larly noted for their impulsivity, attentionalproblems, and poor classroom performance.

By far, the most common treatment of hyper-activity is the prescription of medication, usuallymethylphenidate (Ritalin). However, for a vari-ety of reasons (cf. O'Leary, 1980) there has beenan increased interest in providing psychologicalalternatives to drug therapy (Pelham, Schnedler,Bologna, & Contreras, 1980).

The authors acknowledge the valuable assistanceof Mrs. Susan Holiday, master teacher at Sabal PalmElementary School, Tallahassee, Florida. Technicalassistance and advice regarding the apparatus used inthe experiment was provided by Ross Henderson,electronical engineer in the Psychology Department atFlorida State University. Mark Shamis providedvaluable assistance to this study by prescribing andmonitoring the children's drug dosages. Sincere ap-preciation is extended to Alan E. Kazdin, Steve Breun-ing, and Rolf Jacob, for providing valuable assistancein preparing this manuscript. Reprints may be ob-tained from the senior author who is now at TheUniversity of Rhode Island, Department of Psy-chology, Kingston, Rhode Island 02881.

Several previous studies have compared theeffectiveness of behavioral and medication tac-tics. The behavioral interventions which havebeen used fall into two broad categories. First,"behavior therapy" (Gittelman, Abikoff, Pol-lack, Klein, Katz, & Mattes, 1980; Gittelman-Klein, Klein, Abikoff, Katz, Gloisten, & Kates,1976; Loney, Weissenburger, Woolson, &Lichty, 1979; O'Leary, Pelham, Rosenbaum, &Price, 1976; Pelham et al., 1980) which consistsof an initial training of teachers in behavior man-agement techniques followed by (often weekly)consultation sessions. Second, "direct contin-gency management" which relies on immediatepoint reinforcement for appropriate behavior(Ayllon, Layman, & Kandel, 1975; Christensen,1975; Wulbert & Dries, 1977), point reductionfor inappropriate behavior (Rapport, Murphy,& Bailey, 1980), or continuous teacher attention(Shafto & Sulzbacher, 1977).With the exception of O'Leary et al. (1976)

the behavior therapy approach has been less ef-fective than medication. On the other hand, eachof the direct contingency management tactics hasequaled or surpassed the effects of drug therapy.The disparity of these results can be accountedfor in at least three ways. First, it could be thatthe types of dependent variables measured byresearchers using different strategies are differ-

205

NUMBER 2 (SUMMER 1982)1982, 15. 205-216

MARK D. RAPPORT et al.

entially affected by behavioral tactics and medi-

cation. Traditionally, studies using behaviortherapy have relied heavily on teacher ratings

with less emphasis on direct observation com-

pared to the contingency management proce-

dures. Second, none of the contingency manage-

ment programs has been compared to titrateddosages of medication. Conversely, titration hasbeen used by Gittelman-Klein, Klein, Abikoff,Gloisten, & Kates (1976), Loney et al. (1979),and Gittelman et al. (1980), all of whom foundgreater effects from drugs than behavior therapy.Third, it may be that direct contingency manage-

ment is a more powerful treatment than tradi-tional behavior therapy, as others have found thatreinforcers must provide sufficient cues to directthe child's attention toward specific task de-mands (Cohen, 1970; Parry, 1973). The partialreinforcement schedules typically used in behav-ior therapy protocols are usually not sufficient to

control hyperactive children's behaviors (Parry,1973).The obvious shortcomings of contingency

management interventions have been often un-

realistic time requirements, high adult-child ra-

tios, and extensive teacher training. While newer

psychostimulants have been developed (e.g.,pemoline), virtually no applications of alterna-tives to standard reinforcement programs havebeen forthcoming with hyperactive children.

Response cost may be a viable alternative to

standard reinforcement procedures for severalreasons. First, its effectiveness in treating disrup-tive children in classroom settings is well estab-lished (e.g., Hundert, 1976). Second, response

cost has been found to be more effective thaneither neuroleptic medication (Breuning, O'Neill,& Ferguson, 1980) or positive reinforcement(Worland, 1976). Third, undesirable side effectsfrequently associated with aversive proceduresare not typically reported with response cost

(Kaufman & O'Leary, 1972). Although response

cost appears to hold promise for treating hyper-active children, a direct comparison of response

cost and psychostimulant medication with hyper-

active children in classroom settings has not beenmade thus far.

There were several purposes of the presentstudy: (a) to develop a practical, easily used, al-ternative behavioral intervention to standard re-inforcement regimens for classroom use withhyperactive children; (b) to systematically eval-uate and compare response cost intervention andtitrated methylphenidate medication with base-line performance levels using a variety of depen-dent measures; and (c) to assess whether a func-tional relationship may be obtained for differenttarget behaviors using titrated mg/day doses ofstimulant medication in a classroom setting.

METHOD

Participants and SettingTwo hyperactive boys between the ages of

7 and 8 years participated in the study. Theywere given pseudonyms, Brian and Mitch, toprotect their identities. They were independentlydiagnosed as "hyperactive" by a physician andpsychologist using DSM III (American Psychi-atric Association, 1980) criteria for AttentionalDeficit Disorder with Hyperactivity. Additionalcriteria were (a) teacher ratings on the Abbrevi-ated Conners Teacher Rating Scale (ACTRS)above 15 (see Werry, Sprague, & Cohen, 1975);(b) no history of taking medication for hyperac-tivity; and (c) observed school behavior includ-ing low rates of academic completion, short at-tention span, disruptive behavior, impulsivity,and rates of on-task behavior below 609% whenengaged in academic seat work. The childrenwere of average intelligence and low to middlesocioeconomic status. The study was carried outduring two consecutive school years in a normal,second-grade classroom with one primary teacherpresent.

ApparatusThe apparatus used in the experiment for

Brian consisted of two wooden stands with num-bered cards attached to each. The cards for both

206

BEHAVIORAL AND STIMULANT TREATMENT OF HYPERACTIVITY

stands were made from poster board, numberedin a descending order from 20 to 0, and couldbe flipped down individually by the teacher andstudent.

The above apparatus was refined for Mitch inan effort to develop an easier to use response costdelivery system. The apparatus consisted of abattery-operated, electronic counter with a smalldigital display which was preset at zero prior toeach experimental session. The digital displayautomatically increased by one number eachminute when operative (FI: 1 min). The teacherused a hand-held apparatus to reduce Mitch'sdigital display by one number and illuminate ared light on the counter for 15 sec; signalingthat Mitch had been off task and consequentlylost one point. Thus, points could be both earnedand lost, depending on the boy's behavior.

AssessmentThe boys were observed twice daily for 20-

min periods. The first period began at 9:00 a.m.and the second at 9:40 a.m. During each obser-vation period, the class completed in-seat aca-demic work assigned by the teacher. Brian andMitch were observed by graduate psychologystudents for 80 intervals during each observationperiod throughout the study. Each interval wasdivided into 10 sec of observation followed by5 sec for recording. Observers were blind to bothwhen medication was administered and specificdosage levels, but not to response cost proce-dures.

Teacher ratings. Each Friday the classroomteacher completed the Abbreviated ConnersTeacher Rating Scale (Conners, 1973) on boththe experimental children and two control chil-dren, which reflected the children's behavior forthat week. The ACTRS has been demonstratedto be sensitive to the effects of both behaviortherapy and medication (O'Leary et al., 1976;O'Leary & Pelham, 1978; Sleator & von Neu-mann, 1974) in addition to correlating withclassroom behavior observations (Bolstad &Johnson, 1977).

Child behavior. The child's behavior was cate-gorized as either on task or off task in a mannersimilar to that used by Iwata and Bailey (1974)and Rapport et al. (1980). Off-task behavior wasdefined as visual nonattention to one's materialsfor more than 2 sec within the 10-sec recordinginterval, unless the student was talking to theteacher, had his hand raised above his head, orwas adjusting/glancing at his response cost ap-paratus.

Academic MeasuresPhonics. The phonics materials consisted of

the Scott Foresman Basics in Reading: DaisyDays Series (1978) and accompanying activities.There were typically between 30 and 50 prob-lems in each series depending on the level ofdifficulty and one series was assigned daily. Thechild's paper was graded after class by theteacher and correct responses taken from theaccompanying teacher's manual. Daily perfor-mance was recorded for both the percentage ofproblems completed and the percent correct.

Arithmetic. Arithmetic materials consisted ofthe Holt School Mathematics Series (1974),which was composed of arithmetic work prob-lems dealing with basic addition, subtraction,and word problems. Depending upon the levelof difficulty, 60 to 100 problems were assigneddaily. The child's paper was scored similarly tothe phonics assignment, i.e., daily percent com-pleted and correct.

Reliability. Reliability checks of child behav-ior were taken on 39% and 28% of the days forBrian and Mitch, respectively. Obtained andchance estimates (Hopkins & Hermann, 1977;Johnson & Bolstad, 1973) were computed foroccurrence, nonoccurrence, and overall agree-ment (Bijou, Peterson, & Ault, 1968). (Obtainedand chance estimates computed for occurrence,nonoccurrence, and overall agreement may beobtained from the senior author.) Overall reli-ability was consistently over 88%, with a meanof 96% for both children.

Reliability of academic measures were com-

207

MARK D. RAPPORT et al.

pleted on 40 (48%) and 48 (42%) of the occa-sions for Brian and Mitch, respectively. Checkswere made by the first observer and consisted ofscoring the academic assignments independentlyof the teacher. Agreement was defined as agree-ment on the number of problems completed andperformance accuracy. Reliability was computedby dividing agreement between observer andteacher by agreement plus disagreement andmultiplying by 100 to calculate the percentage.Observer-teacher agreement on problems com-pleted ranged from 91% to 100% (mean99.8%) and 98% to 100% (mean = 99.9%)for Brian and Mitch, respectively. Agreement onperformance accuracy was consistently 100%for both children across experimental conditions.

General ProcedureEach morning, the teacher wrote the two aca-

demic assignments on the chalkboard and de-livered specific instructions regarding their con-tent and requirements for completion. The classworked on the assignments for a 1-hr period,divided into three 20-min intervals. Each childspent one of these intervals in a small, teacher-held math group, while the remainder of theclass worked independently on their academicassignments. No data were recorded when thechildren were in their respective math groups.

Experimental ProceduresAn ABACBC within-subjects design was used

to compare the effects of methylphenidate andresponse cost on the children's on-task behavior,academic completion rate, academic accuracy,and social behavior.

Baseline L. During this condition the child'son-task behavior, academic performance, andacademic assignment accuracy were recorded fora period of several days to reflect preinterventionlevels of behavior.

Medication I. In this condition the parentsgave the boy a dosage of methylphenidate eachmorning, according to the prescribed dosageschedule outlined by White (1977), i.e., 5-mg

increments each week until symptomatic im-provement was noticeable and responding stable.A placebo condition was not included because itwas considered clinically unadvisable due to theseverity of the children's behavior. In an effortto control for possible expectation effects, par-ents told their child that he would be taking vita-min pills periodically to keep from catching acold. The parents were aware their children weretaking stimulant medication but were not awareof the specific dosages. Brian's medication pe-riods included dosages of 5 mg/day (.22 mg/kg), 10 mg/day (.44 mg/kg), and 15 mg/day(.63 mg/kg), lasting for a total of 31 days. Fur-ther increments were not administered due tothe occurrence of head jerks. Mitch was given5 mg/day (.23 mg/kg), 10 mg/day (.46 mg/kg), 15 mg/day (.64 mg/kg), and 20 mg/day(.92 mg/kg) during his medication trials, whichlasted 50 days. All medication changes (e.g.,switching from 5 mg/day to 10 mg/day, or to re-sponse cost) took place over weekends to ensureadequate adjustment. The teacher involved inthe study was blind to both when medication wasadministered and specific dosage levels.

Baseline II. The second baseline phase beganon a Monday morning (medication was termi-nated on the preceding Friday) to allow wash-outof the medication over the weekend. The wash-out period was consistent with the 3- to 4-hrhalf-life of methylphenidate following oral in-gestion. This phase was otherwise identical toBaseline I and lasted a total of 6 and 8 days forthe phonics assignment, and 12 days for thearithmetic assignment for Brian and Mitch, re-spectively.

Response cost 1. The response cost programwas initiated on days 44 and 73 in the phonicsassignment, and lasted for 17 and 13 consecutivedays for Brian and Mitch, respectively. On days50 (Brian) and 77 (Mitch), the response costprocedure was additionally implemented in thearithmetic assignment, and lasted for 11 and 9days, respectively. During this condition, thechild completed assignments at his desk while

208

BEHAVIORAL AND STIMULANT TREATMENT OF HYPERACTIVITY

the teacher conducted her small groups. Boththe teacher and the child had their cost apparatusadjacent to them. The child was told that hecould earn up to 20 min of free time for work-ing hard during each academic period in whichthe cost apparatus was used. If the child was notworking on his assignments, the teacher wouldflip a card down (Brian) or activate her appa-ratus (Mitch) and 1 min of the child's free timewould be lost. Brian was instructed to look at theteacher's apparatus and the number showingoccasionally, then to match his apparatus cardaccordingly. Mitch's apparatus automatically de-ducted 1 min of free time when the teacher acti-vated it and illuminated a red light to signal himthat he had been off task. The amount of freetime earned each day was equal to the numberon the apparatus showing at the end of the as-signment period and was provided approxi-mately 1 hr later.

Medication II. During this condition, themethylphenidate dosage determined most effec-tive in the previous medication trials was admin-istered daily. This decision was made jointly bythe attending physician and senior investigatorand was based upon the dependent measuresobtained. The 15 mg qAM dosage was judgedoptimal for both children due to the observedrates of on-task behavior, high academic com-pletion percentages, and teacher ratings. Thisphase was otherwise identical to the MedicationI condition.

Response cost IH. Following a medication-freeweekend for wash-out purposes, the responsecost intervention was simultaneously imple-mented in the phonics and arithmetic assign-ments on days 71 and 96, and lasted 12 and 19days for Brian and Mitch, respectively.

RESULTS

The effects of each intervention were evalu-ated on the percentage of intervals of on-taskbehavior, problems completed, and problemscompleted correctly.

Child Behavior

As depicted in Figures 1 and 2, prior to anyintervention both children were generally atten-tive for less than half the scored intervals withhigh variability observed across days. Upon in-troduction of the 5 mg/day dose of Ritalin, thepercentage of intervals scored as on-task in-creased appreciably for both children, but re-mained highly variable. Increasing the dosageto 10 mg/day resulted in no significant improve-ment of on-task performance for Brian and aslight increase for Mitch. A daily Ritalin dosageof 15 mg/day was administered on days 21 and44 for Brian and Mitch, respectively. Brian's on-task behavior during the phonics assignment in-creased slightly with no appreciable changenoted in mathematics. Mitch's mean on-task be-havior during this condition was essentially un-changed from the previous medication condi-tion; however, daily variability was restricted toa moderate degree. Medication was increased to20 mg/day Ritalin on days 55 through 64 withreduced variability and no change in mean on-task behavior noted. The datum indicated as"probe" in this condition represents a day inwhich Brian's parents forgot to administer themorning medication. It is interesting to notethat his on-task behavior dropped abruptly onthis day. During the no-treatment reversal phase,both children's on-task behavior decreased tonear original baseline levels.

Introduction of the response cost program ondays 44 and 73 in phonics and days 50 and 77in math, for Brian and Mitch, respectively, re-sulted in a high percentage of attentiveness forboth children across academic assignments areas.The multiple baseline lag on days 44 through49, and 73 through 76 in Figures 1 and 2 re-spectively, demonstrates a strong causal relation-ship between the implementation of the responsecost system and the students' increase in attend-ing behavior. Additionally, it allows two directcomparisons to be made between baseline andresponse cost conditions.

209

MARK D. RAPPORT et al.

5 MG RITALIN 10 MG RITALIN(.22 MG/KG) (.44 MG/KG)

13 MG RITALIN BMUME(.63 MG/KS)

Il

m cosT 15 MG RITALIN RESPOSS COSTI .65 MG/KG)

A A a

/

9-. I0 eel .*I- N~- :

I F~~~~~~~elSE I *le *

*

s (0 (s 20 25 30 35 40 45 s0 55 60

SESSIONS

a soo7 52

Fig. 1. Mean percentage of intervals of daily on-task behavior for Brian. Individual means for each condi-tion are indicated by dashed lines. The datum indicated as "probe" represents a no-medication day and wasnot computed in the mean.

A daily Ritalin dosage of 15 mg was reintro-duced on days 61 and 86 for Brian and Mitch,respectively, and resulted in a rate of attendingbehavior similar to the previous 15 mg/day con-dition. On day 89, Mitch's parents were in-structed to forego medication administration

SOU0 a

brow.|<9N* AL1 O, no1",z," is Us IPT&LI

(designated "probe" in Figure 2), which resultedin 'a dramatic decrease in on-task behavior.Neither of the data indicated "probe" was in-cluded in arriving at the 15-mg condition meansshown as dashed lines.

The response cost program was reinstituted on

( 23 WELKGCH (0000"61 0000100)""Van ) noolKS)

210

PHONICSBASELINE

(00-S-I

o0-so -

70-1so

MATH

IV

9/) 0

so

so -

To -

60 -

50-

40 -

30-_

20

60

50

II I I

I, . i

0 l Is to 0 so Is 40 00 0 0 00 a 0 To so a - M I00iv

SESSIONS

Fig. 2. Mean percentage of intervals of daily on-task behavior for Mitch. Individual means for each condi-tion are indicated by dashed lines. The datum indicated as "probe" represents a no-medication day and wasnot computed in the mean.

toWeSHIL's O*C^wtl Be'"StCST S Q"^T'

BEHAVIORAL AND STIMULANT TREATMENT OF HYPERACTIVITY

days 71 (Brian) and 96 (Mitch) and resulted inunprecedented levels of on-task behavior ob-served. Variability in the children's daily re-sponding was also noticeably reduced acrossacademic areas. The BCB portion of the experi-mental design, comparing response cost to medi-cation (Ritalin), demonstrates that both inter-ventions were effective in increasing the on-taskbehavior of hyperactive children, with responsecost resulting in higher rates in the two children.

Academic Assignment Completionand Accuracy

Figures 3 and 4 show the percentage of prob-lems completed for each of the two morning aca-demic assignments for Brian and Mitch, respec-tively. These graphs show that both childrenwere completing 60% or less of their daily as-signments on the average, with high variabilityacross days observed. Introduction of the 5 mg/day Ritalin condition resulted in marked in-creases in the numbers of problems completed byboth children, most notably in the mathematicsassignment. The 10 mg/day Ritalin condition

was initiated on days 16 (Brian) and 30 (Mitch),with a resultant increase in Mitch's phonics com-pletion rate. Introduction of the 15 mg/day do-sage continued to result in higher rates of re-sponding in Mitch's math assignment but didnot appreciably affect his phonics performance.Brian's phonics performance remained un-changed from the previous condition with lowerperformance levels observed in mathematics.Medication was not administered on the daymarked "probe" in Figure 3, with a resultantdecrease in math performance observed.

The 20 mg/day Ritalin condition was in ef-fect on days 55 through 64 for Mitch, with aresultant decrease in the percentage of problemscompleted in both phonics and math. This effectis remarkable, given the high rate of on-taskbehavior under the 20 mg/day Ritalin conditionobserved in Figure 2. It appears that being "ontask" does not guarantee high academic assign-ment completion rates.A return to baseline conditions on days 38

(Brian) and 65 (Mitch) resulted in assignmentcompletion rates similar to those obtained in

PHONICS

BASELINE S MG RITALIN 10 MG RITALIN 15 MG RITALIN BASELINE RESPONSE-COST IS MG RITALIN RESPONSE-COST(.22 MG/KG) (.44 MG/KG) (.63 MG/KG) (.65 MG/KG)

a toX MATH

o 100,at eo 4

each of the two morning assignments.SESSIONS

Fig. 3. Mean percentage of Brian's daily problems completed forIndividual means for each condition are indicated by dashed lines.

211

MARK D. RAPPORT et al.

so

w

SESSIONS

Fig. 4. Mean percentage of Mitch's daily problems completed for each of the two morning assignments.Individual means for each condition are indicated by dashed lines.

Baseline I. Overall, the ABA portion of the ex-perimental design, contrasting medication withbaseline performance levels, demonstrates thatRitalin may positively affect the assignmentcompletion rates of hyperactive children. Fur-ther, there appears to be a U-shaped dose-re-sponse curve with higher dosages (i.e., 20 mg/day) resulting in deteriorated responding.

The response cost program was implementedon days 44 and 50 in Brian's phonics and mathassignments, respectively, and resulted in un-precedented levels of academic performance. Asimilar result was observed in Mitch's academicperformance, with the percentage of assignmentscompleted increasing appreciably in both aca-demic areas. Reintroduction of the 15 mg/dayRitalin condition resulted in decreased rates ofresponding for both children across academicareas. As reported earlier, the "probe" indicatedin Figure 4 represents a no-medication day andwas not used in computing the mean in this con-dition. It is interesting to note that Mitch's per-formance was inconsistent on this day acrossvariables, remaining relatively stable in phonics,yet dramatically decreasing in mathematics. Re-sponse cost was reimplemented on days 71(Brian) and 96 (Mitch) and resulted in highacademic completion rates and reduced vari-ability across days. The CBC portion of the ex-perimental design, comparing response cost to

medication (methylphenidate), demonstrates thatboth interventions produce positive changes inacademic completion rates, with the responsecost program resulting in higher and more stableresponding.

Academic accuracy was also assessed for bothchildren across experimental conditions with noappreciable changes observed. (Individual graphsillustrating each boy's academic accuracy acrossexperimental conditions may be obtain from thesenior author.) Given the increased assignmentcompletion rates observed in both children, it isnoteworthy that both maintained rates of accu-racy averaging well above 80%.

Teacher-Rated Improvement

The ACTRS scores for Brian, Mitch, and twocontrol children are illustrated in Figure 5. Thisgraph allows comparisons to be made regardingthe differential effectiveness of medication andresponse cost interventions as viewed by theclassroom teacher. Control children were se-lected by the teacher prior to the experiment andrepresented "normal" children without inherentlearning or behavior problems. The 15 mg/dayRitalin and response cost conditions appear tobe the most effective in reducing teacher-viewedhyperactivity and concomitantly approachingnormality in these children.

212

BEHAVIORAL AND STIMULANT TREATMENT OF HYPERACTIVITY

TEACER RATINGS

a*v.

I~~~~~~~~E MOTl

Fig. 5. The Abbreviated Conners Teacher Rating Scale (ACTRS) scores for Brian, Mitch, and two classroomcontrol children across experimental conditions. Lower scores represent improvement in the child's behavior.

DISCUSSION

The results indicated that both the responsecost and psychostimulant medication (methyl-phenidate) interventions were effective in in-creasing on-task behavior and academic per-formance in children diagnosed as havingattentional deficit disorder with hyperactivity.The effect of individually titrated dosages of Rit-alin was contrasted with response cost to deter-mine the relative efficacy of each. The greatestimprovement in on-task behavior and academicperformance for both children occurred duringresponse cost.

The greater success of the response cost pro-gram in the present investigation may be attrib-uted to several factors. First, a necessary condi-tion of successful contingency managementprograms has been to direct the child's attentiontoward specific tasks using immediate positiveor negative consequences. Although positive re-inforcement is more desirable, its continuousapplication may actually distract a child fromongoing task demands by pushing him aboveoptimal arousal levels (Firestone & Douglas,1975). Mild, negative feedback coupled with

reinforcement (i.e., response cost) may be a rea-sonable alternative and has been shown to helphyperactive children modulate their respondingin laboratory tasks (Firestone & Douglas, 1975).Second, there was relatively little delay betweena child's off-task behavior and teacher-producedconsequences; an important consideration giventhe impulsivity of these children. Previous stud-ies have generally required the teacher to walkacross the room to provide reinforcement for on-task behavior, necessitating an undue time lagbetween behavior and consequences. Finally, be-havior therapy programs may be viewed as lesseffective and burdensome by teachers due to theirexcessive time demands and impracticality(Loney et al., 1979). The response cost proce-dure, however, required only brief teacher train-ing (i.e., approximately 15 min plus two feed-back sessions) and limited time involvementwhile operative (i.e., the teacher didn't have towalk across the classroom to check a card orprovide verbal praise).

Teacher ratings suggested that response costwas an effective intervention for improving theclassroom behavior of hyperactive children andnormalized them to an appreciable degree. Ques-

213

MARK D. RAPPORT et al.

tionnaires administered at the conclusion of thestudy indicated that both the teacher and studentsviewed the response cost procedure positively.

The 15 mg/day Ritalin dosage was the opti-mal dose for both children, resulting in levels ofon-task behavior and academic completion ratesclearly exceeding baseline measures. This is asurprising result given the findings reported byBarkley and Cunningham (1978) whose reviewof the literature reported little if any improve-ment in the academic achievement of hyperac-tive children while receiving stimulant medica-tion. However, the difference between a child'sdaily academic performance and overall achieve-ment may not be highly correlated.

Several similarities were found between thepresent results and past laboratory findings,providing external validity (albeit limited) to thelatter. For example, Sprague and Sleator (1977)found that hyperactive children performed betteron a short-term memory task under low dosages(.30 mg/kg Ritalin) and experienced deteriora-tion in learning under high dosages (1.0 mg/kg)of stimulant medication, i.e., U-shaped dose-re-sponse curve. The present results extend thesefinding in that academic performance continuedto improve with Ritalin dosages up to approxi-mately .65 mg/kg for both children. Walker(1980) reported a similar finding with hyperac-tive children on repeated acquisition tasks. Thechildren performed maximally at .70 mg/kg vs..30 mg/kg of methylphenidate. There are twopossible reasons for the differences between thepresent findings and Walker's (1980) findingsvs. those reported by Sprague and Sleator (1977).First, Sprague and Sleator's dose-response curveswere plotted between placebo, .30 mg/kg, and1.0 mg/kg Ritalin, with no intermediate dosagesadministered. Had intermediate dosages beenadministered (e.g., 70 mg/kg) their performancecurves may have been extended. Second, theyused laboratory tasks that may have been moresensitive to drug dose changes, i.e., laboratorytasks under highly controlled conditions willpresumably be more sensitive to subtle medica-

tion changes than will classroom measures. Thus,it may be that some hyperactive children willcontinue to improve with Ritalin dosages fallingbetween these levels (i.e., .30 and 1.0 mg/kg).

In the present study, deteriorated academiccompletion rates were observed at the .92 mg/kgRitalin dosage level, supporting the conclusionsdrawn by Sprague and Sleator (1977) regardinghigh (i.e., 1.0 mg/kg) dosages and associatedperformance deterioration. It appears that theoptimum dosage level for these children may beslightly higher that those previously reported(i.e., .30 mg/kg), with performance deteriora-tion occurring as the dosages approximate 1.0mg/kg.

Another similar conclusion between the pres-ent findings and those reported by Sprague andSleator (1977) and Walker (1980) is the obser-vation that different target behaviors improve atdifferent dosages. In the present investigation,the child's on-task behavior showed the greatestimprovement under the .92 mg/kg Ritalin dos-age, approximating the 1.0 mg/kg dosage levelreported by Sprague and Sleator. However, theclassroom teacher rated the children's social be-havior as maximally improved under the 15mg/day Ritalin condition (i.e., approximately.65 mg/kg) as opposed to the even higher 20mg/day dose. This may have been due to theteacher's sensitivity to the amount of academicwork completed each day and her rating the 20mg condition accordingly, i.e., the teacher wasrequired to score the child's academic work im-mediately after class and may have been influ-enced by the child's decreased assignment com-pletion rate despite the scale's loading on socialbehavior. This was confirmed by the teacher atthe study's conclusion.

Individually titrating medication dosageswith hyperactive children is a valuable clinicalmethod and should depend on feedback from avariety of dependent measures, especially giventhe idiosyncratic behavior of these children undersimilar medication dosages (Sprague and Sleator,1975). Drug dosage should represent a balance

214

BEHAVIORAL AND STIMULANT TREATMENT OF HYPERACTIVITY 215

between improved social behavior and learning.Without such considerations, erroneous conclu-sions may be drawn regarding treatment efficacy.A placebo control condition would have been

desirable and was planned at the inception ofthe study. As noted earlier, however, it was notclinically desirable, which may be a frequentproblem when working in naturalistic settings.The present study presents a practical alternativeto control for possible expectation effects whenplacebo administration is undesirable or not pos-sible. The results suggest that the "vitamin con-trol" was effective in limiting expectation effectsin several ways. First, on task and academic com-pletion rates improved gradually in both chil-dren across days, whereas little if any improve-ment is typically observed on ratings of specificbehaviors or global estimates with placebo treat-ment (Gittelman-Klein, Klein, Katz, Saraf, &Pollak, 1976; Henker, Whalen, & Collins,1979). Second, both children showed deterio-rated responding at higher stimulant doses, afinding consistent with the responder hypothesis.Third, follow-up student questionnaires indi-cated that both children believed they were tak-ing "vitamin pills" for cold prevention vs. medi-cation to help their behavior.

The inclusion of a response cost plus medica-tion condition was not included in the presentinvestigation due to time constraints. This typeof comparison has been lacking in the field thusfar and our future research will address suchissues.

Caution should be used in generalizing fromthe present results to hyperactive children ingeneral. Although precise measurement wasused, only two children were studied. Further, aspecialized response cost apparatus was used withboth children. Studies using dissimilar deliverysystems may not obtain identical results.

The response cost intervention may be usedin conjunction with or instead of psychostimu-lant medication. Clearly, medication is easier toadminister and does not require the extra effortthat contingency management interventions do.

Equally clear, however, is the fact that medicatedchildren continue to experience academic diffi-culty (Riddle & Rapoport, 1976; Weiss, Kruger,Danielson, & Elman, 1975) and that many donot respond positively to psychostimulant medi-cation. Thus, effective, easy to use behavioral in-terventions will continue to be required in thefuture.

REFERENCES

American Psychiatric Association. Diagnostic andStatistical Manual: DSM III (3rd ed.). Washing-ton, D.C.: American Psychiatric Association,1980.

Ayllon, T., Layman, D., & Kandel, H. J. A behav-ioral-educational alternative to drug control ofhyperactive children. Journal of Applied BehaviorAnalysis, 1975, 8, 137-146.

Barkley, R. A., & Cunningham, C. E. Do stimulantdrugs improve the academic performance of hy-perkinetic children? Clinical Pediatrics, 1978, 17,85-82.

Bijou, S. W., Peterson, R. F., & Ault, M. H. Amethod to integrate descriptive and experimentalfield studies at the level of data and empiricalconcepts. Journal of Applied Behavior Analysis,1968, 1, 175-191.

Bolstad, 0. D., & Johnson, S. M. The relationshipbetween teachers' assessment of students and thestudents' actual behavior in the classroom. ChildDevelopment, 1977, 48, 570-577.

Breuning, S. E., O'Neill, M. J., & Ferguson, D. G.A comparison of psychotropic drug, responsecost, and psychotropic drug plus response costprocedures for controlling institutionalized men-tally retarded persons. Applied Research in Men-tal Retardation, 1980, 1, 25 3-268.

Christensen, D. E. Effects of combining methyl-phenidate and a classroom token system in modi-fying hyperactive behavior. American Journal ofMental Deficiency, 1975, 80, 266-276.

Cohen, N. Psychophysiological concomitants of at-tention in hyperactive children. (Doctoral disser-tation, McGill University, 1970). DissertationAbstracts International, 1970, 32, 553B.

Conners, C. K. Rating scales for use in drug studieswith children. Psychopharmacology Bulletin,1973, 24-84.

Firestone, P., & Douglas, V. I. The effects of rewardand punishment on reaction times and autonomicactivity in hyperactive and normal children. Jour-nal of Abnormal Child Psychology, 1975, 3, 201-216.

216 MARK D. RAPPORT et al.

Gittelman, R., Abikoff, H., Pollack, E., Klein, D. F.,Katz, S., & Mattes, J. A controlled trial of be-havior modification and methylphenidate in hy-peractive children. In C. Whalen & B. Henker(Eds.), Hyperactive children. New York: Aca-demic Press, 1980.

Gittelman-Klein, R., Klein, D. F., Abikoff, H., Katz,S., Gloisten, A. C., & Kates, W. Relative efficacyof methylphenidate and behavior modification inhyperactive children: An interim report. Journalof Abnormal Child Psychology, 1976, 4, 361-379.

Gittelman-Klein, R., Klein, D. F., Katz, S., Saraf, K.,& Pollack, E. Comparative effects of methyl-phenidate and thioridazine in hyperkinetic chil-dren. I. Clinical results. Archives of General Psy-chitary, 1976, 33, 1217-1231.

Henker, B., Whalen, C. K., & Collins, B. E. Double-blind and triple-blind assessments of medicationand placebo responses in hyperactive children.Journal of Abnormal Child Psychology, 1979, 7,1-13.

Hopkins, B. L., & Hermann, J. A. Evaluating inter-observer reliability of interval data. journal ofApplied Behavior Analysis, 1977, 10, 121-126.

Hundert, J. The effectiveness of reinforcement, re-sponse cost, and mixed programs on classroombehaviors. Journal of Applied Behavior Analysis,1976,9, 107.

Iwata, B. A., & Bailey, J. S. Reward versus cost tokensystems: An analysis of the effects on student andteacher. Journal of Applied Behavior Analysis,1974, 1, 567-576.

Johnson, S. M., & Bolstad, 0. D. Methodologicalissues in naturalistic observation: Some problemsand solutions for field research. In L. A. Haner-lynch, L. C. Handy, & E. J. Mash (Eds.), Behaviorchange: Methodology, concepts, and practice.Champaign, Ill.: Research Press, 1973.

Kaufman, K. F., & O'Leary, K. D. Reward, cost, andself-evaluation procedures for disruptive adoles-cents in a psychiatric hospital school. Journal ofApplied Behavior Analysis, 1972, 5, 293-309.

Lambert, N. M., Sandoval, J. H., & Sassone, D. M.Prevalence estimates of hyperactivity in schoolchildren. Pediatric Annals, 1978, 7, 330-338.

Loney, J., Weissenburger, F., Woolson, R., & Lichty,E. Comparing psychological and pharmacologi-cal treatments for hyperkinetic boys and theirclassmates. Journal of Abnormal Child Psychol-ogy, 1979, 7, 133-143.

O'Leary, K. D. Pills or skills for hyperactive chil-dren. Journal of Applied Behavior Analysis, 1980,1, 191-204.

O'Leary, S. G., & Pelham, W. E. Behavior therapyand withdrawal of stimulant medication withhyperactive children. Pediatrics, 1978, 61, 211-217.

O'Leary, K. D., Pelham, W. E., Rosenbaum, A., &Price, G. Behavioral treatment of hyperkinetic

children: An experimental evaluation of its use-fulness. Clinical Pediatrics, 1976, 15, 511-515.

Parry, P. The effect of reward on the performanceof hyperactive children. Unpublished doctoral dis-sertation, McGill University, Montreal, 1973.

Pelham, W. E., Schnedler, R. W., Bologna, N. C., &Contreras, J. A. Behavioral and stimulant treat-ment of hyperactive children: A therapy studywith methylphenidate probes in a within-subjectdesign. Journal of Applied Behavior Analysis,1980, 13, 221-236.

Rapport, M. D., Murphy, A., & Bailey, J. S. Theeffects of a response cost treatment tactic onhyperactive children. Journal of School Psy-chology, 1980, 18, 98- 111.

Riddle, K. D., & Rapoport, J. L. A 2-year follow-upof 72 hyperactive boys. Classroom behavior andpeer acceptance. Journal of Nervous and MentalDisease, 1976, 162, 126-134.

Shafto, F., & Sulzbacher, S. Comparing treatmenttactics with a hyperactive preschool child: Stimu-lant medication and programmed teacher inter-vention. Journal of Applied Behavior Analysis,1977, 10, 13-20.

Sleator, E., & von Neumann, A. Methylphenidate inthe treatment of hyperkinetic children. ClinicalPediatrics, 1974, 13, 19-24.

Sprague, R. L., & Sleator, E. K. What is the properdose of stimulant drugs in children? InternationalJournal of Mental Health, 1975, 4, 75-104.

Sprague, R. L., & Sleator, E. K. Methylphenidatein hyperkinetic children: Differences in dose ef-fects on learning and social behavior. Science,1977, 198, 1274-1276.

Walker, M. K. Stimulant drugs. In S. E. Breuning &A. D. Poling (Eds.), Drugs and Mental Retarda-tion, Springfield, Ill.: Charles C Thomas, 1980.

Weiss, G., Kruger, E., Danielson, U., & Elman, M.Effect of long-term treatment of hyperactive chil-dren with methylphenidate. Canadian MedicalAssociation Journal, 1975, 112, 159-165.

Werry, J., Sprague, R., & Cohen, M. Conners'Teacher Rating Scale for use in drug studies withchildren-an empirical study. Journal of Abnor-mal Child Psychology, 1975, 3, 217-229.

White, J. H. The hyperactive child syndrome.American Family Practice, 1977, 15, 100-104.

Worland, J. Effects of positive and negative feed-back on behavioral control in hyperactive andnormal boys. Journal of Abnormal Child Psy-chology, 1976, 4, 315-326.

Wulbert, M., & Dries, R. The relative efficacy ofmethylphenidate (Ritalin) and behavior modifica-tion techniques in the treatment of a hyperactivechild. Journal of Applied Behavior Analysis,1977, 10, 21-31.

Received February 18, 1981Final acceptance November 24, 1981