JOURNAL OF APPLIED BEHAVIOR ANALYSIS

THE TOKEN ECONOMY: A DECADE LATERALAN E. KAZDIN

WESTERN PSYCHIATRIC INSTITUTE AND CLINICUNIVERSITY OF PITTSBURGH SCHOOL OF MEDICINE

In the last decade, the token economy has been extended widely across populations andbehaviors in treatment, rehabilitation, educational, and community settings. Outcomeresearch has expanded as well to include large-scale program evaluations and comparativeand combined treatment studies of the token economy. In a previous review (Kazdin &Bootzin, 1972), several obstacles were identified for the effective application of the tokeneconomy. These included identifying procedures to enhance program efficacy, to trainstaff, to overcome client resistance, and to promote long-term maintenance and transferof training. The present paper discusses recent advances in research and reviews progresson the major issues identified previously. New issues have become salient in the lastdecade that pertain to the extension of the token economy to institutional settings. Thedemands for maintaining the integrity of treatment, the ability to integrate token econ-omies within existing institutional constraints, and the disseminability of the procedureson a large scale are major issues that may dictate the future of the token economy.DESCRIPTORS: token economies, reinforcement programs, dissemination

In the middle and late 1960's, the token econ-omy emerged as a promising intervention intreatment, rehabilitation, and educational set-tings. Major impetus for the development of thetoken economy was the pioneering work of Ayl-lon and Azrin (1965, 1968b) who developedand evaluated a program for chronic psychiatricpatients. Several other programs emerged forpsychiatric patients, delinquents, the mentallyretarded, children in classroom settings, andother populations (see Kazdin & Bootzin, 1972;O'Leary & Drabman, 1971). Since the late1960's and early 1970's, the breadth of applica-tions has increased greatly. Within areas that al-ready received attention, programs extended todiverse subpopulations. For example, within psy-chiatric populations, the token economy has beenextended beyond chronic psychiatric patients topatients with acute disorders (Gershone, Errick-

Completion of this paper was supported by a Re-search Scientist Development Award (MH00353)from the National Institute of Mental Health. Re-quests for reprints should be sent to the author, West-ern Psychiatric Institute and Clinic, 3811 O'HaraStreet, University of Pittsburgh School of Medicine,Pittsburgh, Pennsylvania 15213.

son, Mitchell, & Paulson, 1977), organic brainsyndromes (Murphy, 1976), psychosomatic dis-orders (Wooley, Blackwell, & Winget, 1978),and autistic children (Hung, 1977), to mentiona few. The token economy has also been ex-tended to new areas of research such as behav-ioral medicine (Ferguson & Taylor, 1980), be-havioral ecology (Lloyd, 1980) communitypsychology (Glenwick & Jason, 1980), and geri-atric psychology (Hussian, 1981).

Within the last decade, extension of the tokeneconomy has also been evident in the numberof studies comparing the token economy withother interventions (e.g., Bushell, 1978; Paul &Lentz, 1977; Stoffelmayr, Faulkner, & Mitchell,1979) and examining the potential benefits ofcombining a token economy with other proce-dures (e.g., Greenberg, Scott, Friesen, & Pisa,1975; McCreadie, Main, & Dunlop, 1978). Themost dramatic extension has been the applica-tion of token economies on a much larger scalethan had been the case previously. Perhaps thelargest extension has been the Behavior AnalysisFollow Through program for disadvantaged ele-mentary school children (Bushell, 1978). The

431

1982, 153,431-445 NUMBER 3 (FALL 1982)

ALAN E. KAZDIN

Follow Through program was developed to fol-low up the gains provided by an earlier interven-tion program, Head Start. However, the FollowThrough project was designed specifically to testdifferent models of intervention including Be-havior Analysis. The program has been imple-mented in kindergarten through third gradesand has included over 7,000 children in approxi-mately 300 classrooms and 15 separate citiesthroughout the United States. Other programseven though smaller in scale still represent ma-jor extensions of the token economy in educa-tional settings by encompassing entire schools(e.g., Boegli & Wasik, 1978) and classes fromseveral different schools (Rollins, McCandless,Thompson, & Brassell, 1974; Thompson, Bras-sell, Persons, Tucker, & Rollins, 1974).A decade ago, Bootzin and I evaluated the

token economy research and identified severalissues and obstacles (Kazdin & Bootzin, 1972).These included: (a) maintaining behavior andensuring generalization, (b) training staff to im-plement the token economy, (c) increasing clientresponsiveness to the contingencies, and (d)overcoming client resistance to the program.Since the earlier review, the literature has pro-liferated and considerable progress has beenmade. An evaluative review of research is nowwell beyond the confines of a chapter or journalarticle (see Kazdin, 1977; O'Leary, 1978). Thepresent article discusses issues raised by tokeneconomies in light of applications over the lastdecade. The purpose is to examine progress onsalient issues identified previously as in need ofresearch and significant issues that have recentlyemerged.

PROGRESS ON SALIENT ISSUESEnhancing Effects of Token Economies

From the inception of the token economy, re-ports indicated that some number of participantsmay not respond to the contingencies. For exam-ple, Ayllon and Azrin (1965), reported that18% (n = 8) of chronic psychiatric patients

were generally unaffected by the procedures.Since this report, other programs with psychi-atric patients, the mentally retarded, delinquents,and children in school settings have continued toreport a small but consistent percentage of par-ticipants who fail to respond (see Hemsley,1978; Kazdin, in press).

Research has examined whether particularclient variables contribute to responsiveness tothe token economy, but the evidence has beeninconsistent. For example, in token economiesfor psychiatric patients, degree of patient with-drawal, social isolation and length of hospitaliza-tion are negatively correlated with improvementin some studies (Ayllon & Azrin, 1968b; Ful-lerton, Cayner, & McLaughlin-Reidel, 1978) butunrelated or even positively correlated with re-sponsiveness in other studies (Allen & Magaro,1971; Mishara, 1978). Similarly, conflicting evi-dence within and across target populations hasbeen provided for the relations between age, IQ,and gender and responsiveness to the contingen-cies (e.g., Fullerton et al., 1978; Mishara, 1978;Moran, Kass, & Munz, 1977).A significant development over the last dec-

ade is recognition that lack of responsiveness tothe token economy may reflect more on the pro-gram than on clients who fail to respond. Lackof responsiveness usually refers to the failure ofsome clients to respond to a set of contingenciesthat is standardized across all clients. The modelon which the token economy is based does notpropose that identical contingencies will be uni-versally effective. Considerable evidence existsthat persons who do not respond initially to aprogram may readily respond when some altera-tions are made in the contingencies.

Several procedures can be used to improve re-sponsiveness. Perhaps the most obvious one isto vary the magnitude of reinforcement. In-creases in the number of tokens or the value ofback-up events enhance client responsiveness(e.g., Ayllon, Milan, Roberts, & McKee, 1979;Bassett, Blanchard, & Koshland, 1975; Rickard,Melvin, Creel, & Creel, 1973). Another proce-

432

TOKEN ECONOMY

dure for improving responsiveness is reinforcersampling (Ayllon & Azrin, 1968b) which con-sists of exposing the client to a portion of thebackup reinforcer such as a part of a meal or afew minutes of special social activity. By sam-pling a portion of the reinforcers on a noncon-tingent basis, the client is more likely to purchasethe event with tokens and consequently to en-gage in token-earning behaviors (e.g., Ayllon &Azrin, 1968a, 1968b; Curran, Lentz, & Paul,1973). Allowing clients to preselect the backupreinforcers for which they will be working (Kaz-din & Geesey, 1980) or to earn their way off thetoken system for meeting high criterion levels ofperformance (Kazdin & Mascitelli, 1980) canalso enhance responsiveness.

Viewing a token program from the standpointof an economic system has also identified vari-ables that can be manipulated to enhance perfor-mance (see Fisher, Winkler, Krasner, Kagel,Battalio, & Basmann, 1978). Essentially, token-earning behaviors represent work output; thetokens represent income or wages; backupevents represent expenditures; and accumulatedtokens can be viewed as savings. Altering theamount of savings clients are permitted to ac-crue, increasing the costs of backup events (in-flation), increasing the consumption of backupevents by expanding the range and attractivenessof these events, stimulating spending throughoccasional sales, or placing expiration dates onthe tokens to promote spending, have increasedresponsiveness (e.g., Hung, 1977; Milby, Clarke,Charles, & Willcutt, 1977; Winkler, 1973).

Responsiveness to token reinforcement canalso be enhanced by involving peers in the pro-gram. For example, permitting peers to sharethe consequences earned by a particular individ-ual has increased responsiveness of that individ-ual to the contingencies (Feingold & Migler,1972; Kazdin & Geesey, 1977). Using a peer-manager system of reinforcement in which amember of the peer group administers and with-draws tokens can enhance performance as well(Phillips, Phillips, Wolf, & Fixsen, 1973). Fi-

nally, the addition of response cost or fines canimprove client performance in a token economy(e.g., McLaughlin & Malaby, 1977; Walker,Hops, & Fiegenbaum, 1976).When persons initially fail to respond, perfor-

mance can be improved by different programvariations. The fact that many variables mayovercome initial unresponsiveness to the pro-gram does not mean that token economies donot "fail" to produce the desired changes or ulti-mate treatment goals. However, the lack of re-sponsiveness often can be readily controvertedwith changes in the contingencies.

Staff TrainingTraining the staff who administer a token

economy has been recognized as an importantissue throughout the history of reinforcementprograms. In recent years, the importance hasbeen substantiated by demonstrations showingthat the extent to which staff administer the con-tingencies as intended (e.g., Jackson, 1976; Mc-Laughlin et al., in press) is directly related toclient behavior change. In the last decade a greatdeal has been learned about alternative trainingprocedures and their effects.

Several methods have been evaluated and in-clude variations of instructions, modeling androle playing, informative feedback, and directreinforcement using approval and attention, spe-cial privileges or "tokens" (e.g., money, tradingstamps). Procedures that combine several tech-niques and include direct reinforcement of staffperformance have been especially effective, asdemonstrated in school, hospital, and commu-nity settings (see Bernstein, 1982; Kazdin,1980b).An important issue is integrating staff train-

ing within the constraints of existing settings.For example, many potent reinforcers that mightbe used to develop staff performance are usuallyunavailable for contingent application. Theproblem has been circumvented in many pro-grams by creative use of such reinforcers as cer-tificates of recognition and approval from super-

433

ALAN E. KAZDIN

visors (Burg, Reid, & Lattimore, 1979; Bushell,1978; Montegar, Reid, Madsen, & Ewell, 1977).A remaining issue is determining whether train-ing can be effectively carried out on a large scale.Promising results along these lines have alreadyemerged in the context of training teachersto administer the Behavior Analysis FollowThrough program (see Bushell, 1978; Jackson,1976).Over the last decade, additional information

has emerged related to staff training and admin-istration of token economies. Several studieshave shown that staff behaviors often change asa function of administering a program. Acrossdifferent populations and settings, the adminis-tration of a token economy has been associatedwith staff increases in nonverbal and verbal ap-proval for appropriate client behavior and de-creases in disapproval and withdrawal of priv-ileges for inappropriate behavior (Boegli &Wasik, 1978; Breyer & Allen, 1975; Trudel,Boisvert, Maruca, & Leroux, 1974). Also, staffworking on a token economy ward show morepositive attitudes toward patients and towardtreatment than staff working on wards with con-ventional treatments (McReynolds & Coleman,1972; Milby, Pendergrass, & Clarke, 1975). Thesocial climate or ward atmosphere of token econ-omy wards has also been found to be more posi-tive on several dimensions (e.g., spontaneity,affiliation) than on conventional wards (Wilkin-son & Reppucci, 1973). Thus, the token econ-omy appears to produce changes in several facetsof the social climate in which staff and clientsfunction.

Client Resistance to the ProgramClient resistance refers to expressions of an-

ger, complaints, and rule breaking in response tothe token economy. Although only a few reportsindicate client resistance, adverse client reactionsare important to discuss for different reasons.First, a token economy restructures much of thereward system in most settings. Consequently,the potential for coercion is great. Conceivably,basic amenities previously provided noncontin-

gently might be withheld until they are earned.Second, legal issues raised by token economiesin institutional settings have received increasedattention in the United States (Martin, 1975).In specifying patient rights and basic conditionsof institutional care, the courts have influencedthe types of events that can be used as reinforc-ers (see Kazdin, 1977).

Within legal guidelines, it is still possible todesign programs that clients find aversive. Forexample, Biklen (1976) reported a token econ-omy in a psychiatric hospital that led to patientanger at the system and rejection of the tokens.Objections were based in part on the contingentdelivery of many rewards that were given freelybefore the program. Also, many of the rein-forced activities seem puerile (e.g., games, crafts,childlike parties). Similarly, Zeldow (1976)noted adverse reactions of psychiatric patients toa system that seemed to consist of inflexible rulesthat staff rigidly imposed and the lack of patientrecourse for complaints about the system.

Consideration of client reactions to tokeneconomies is critical because, in institutional set-tings, residents usually have the legal right towithdraw from the program. Because the courtshave been involved increasingly in the rights ofinvoluntarily confined persons, programs havechanged. The onus has fallen on investigators toidentify and to provide reinforcers that are ordi-narily unavailable in the setting rather than touse basic amenities to which persons are entitledby right (Wexler, 1973). With creative selectionof incentives, client resistance is less likely toresult because existing reinforcers from the set-ting are not lost. Indeed, because of the diverserewards that are added to the program, there isan incentive for participants to remain willinglyin the program. Also, in some programs, clientsare explicitly given the option of leaving theprogram without penalty (e.g., Ayllon et al.,1979). Other procedures can be used to over-come client resistance and to help protect clientrights. For example, providing opportunities forclients to have input into the system such as se-lecting rewards or negotiating the contingencies

434

TOKEN ECONOMY

may increase reactions to the program (see Kar-raker, 1977; Kazdin, 1980a). In short, severaloptions are available to overcome client resis-tance.To operationalize client resistance, an impor-

tant point of departure would be to assess clientreactions to the program (see Wolf, 1978). Therelationship between client reactions and respon-siveness to the contingencies could then be moresystematically evaluated than has been the casein current research. Moreover, interventionscould be designed to have impact on client eval-uations of and participation in the program.

Long-Term Effects: Response Maintenanceand Transfer to TrainingAn issue of obvious importance in any inter-

vention program is whether the effects are main-tained after the program is terminated andcontinue outside of the treatment setting. Theseissues, referred to as response maintenance andtransfer, respectively, or as generalization collec-tively, have been of special concern in tokeneconomy research for different reasons. First, to-ken economies and other contingency-based pro-grams are often evaluated in experimental de-signs (e.g., ABAB) where the intervention istemporarily suspended and behavior returns toor near baseline levels. Thus, even during evalu-ation of the program, therapeutic gains may belost or partially lost. Second, token economiesoften represent environmental arrangementsthat depart dramatically from the usual environ-ments to which individuals will return. Thus,little transfer of training might be expected fromtreatment to extratreatment settings.

In the last decade, several developments canbe identified regarding maintenance and transferof intervention effects. To begin with, manymore programs are available that report follow-up data than in previous years. Also, for manyprograms, the follow-up data are reported afterthe program has been in effect for protractedperiods (e.g., several months or a few years).Thus, the long-term effects of treatment have

been given a better test than in demonstrationprojects with short intervention phases.

The results from different studies show thatgains produced by token economies are notinevitably lost. For example, Paul and Lentz(1977) compared social learning, milieu ther-apy, and routine hospital care for the treatmentof chronic psychiatric patients. The social learn-ing program was based primarily on a tokeneconomy where patients received incentives fora variety of adaptive behaviors on the ward suchas attending activities and engaging in self-careactivities or social interaction. Although patientsin both social learning and milieu programsimproved, the social learning program was con-sistently more effective on measures in the hospi-tal, discharge of patients and status in the com-munity from 1.5 up to 5 years after terminationof the program.

In school settings, follow-up data have alsoindicated that intervention effects are at leastpartially maintained. For example, the changesin academic achievement obtained from the Be-havior Analysis Follow Through program werestill evident 2 years after the program had beenterminated and the children had entered class-rooms in which token programs were not in ef-fect (Bushell, 1978). Similarly, in a junior highschool program for serious behavior problemadolescents, token economies led to reductions inexpulsions, suspensions, and grade failure. Atfollow-up 3 to 4 years later, gains were still evi-dent (Heaton & Safer, 1982; Safer, Heaton, &Parker, 1981). Adolescents who participated inthe program showed higher rates of entranceinto high school and school attendance, betterclassroom conduct, and lower rates of with-drawal from school relative to control subjects.However, by the end of senior high school, tokeneconomy and control groups showed comparablerates of school enrollment and high school grad-uation.With delinquent youths, follow-up results

have been obtained for the teaching-familymodel (based on Achievement Place) (Wolf,Phillips, Fixsen, Braukmann, Kirigin, Willner,

435

ALAN E. KAZDIN

& Schumaker, 1976). The model has been ex-tended to approximately 150 different grouphomes throughout the United States and a fewforeign countries (Jones, Weinrott, & Howard,Note 1). (Actually, the token economy is onlypart of a much larger program in the teaching-family model and relies on self-government, askills training curriculum, a relationship withthe teaching parents, and procedures to reinte-grate youths into the community [Kirigin,Braukmann, Atwater, & Wolf, 1982).) Evalua-tions of large-scale extensions of the model haveshown that measures of offenses and reinstitu-tionalization from 1 to 3 years after the programare no different for youths who complete theprogram and those who participate in more tra-ditional programs (Kirigin et al., 1982; Joneset al., Note 1). In contrast, an extensive evalua-tion of the Achievement Place home where theprocedures have been especially well developedand monitored has shown that youths who par-ticipate in the program have much lower rates(approximately one half) of reinstitutionaliza-tion, and a much higher rate of school atten-dance in the community than youths in a moretraditional detention setting (Kirigin, Wolf,Braukmann, Fixsen, & Phillips 1979). However,in the follow-up period, contact with the policeand the courts was not different between groups.The absence of differences on police and courtcontacts is difficult to interpret. Non-Achieve-ment Place youths were more likely to be rein-stitutionalized at follow-up and hence were nolonger candidates for police and court contacts.

In general, the above programs show that theeffects of participation in a token economy maystill be evident up to a few years after the pro-gram has been terminated. However, importantqualifiers need to be highlighted to place the fol-low-up data in perspective. First, in many in-stances a token economy is only one componentof the program. For example, in school pro-grams the token economy has been associatedwith smaller classroom size, individualized in-struction, parent involvement in classroom pro-cedures, home-based reinforcement, and other

procedures that may contribute to maintenanceand transfer in their own right (e.g., Bushell,1978; Safer et al., 1981). Similarly, in theAchievement Place program, the token economyis only part of a much more comprehensive pro-gram including multiple procedures noted ear-lier. Thus follow-up results cannot be attributedspecifically to the token reinforcement contin-gencies. However, the primary applied concernis whether after participation in a token econ-omy, the gains in behavior are necessarily lost;several programs indicated that they are not.

Second, in many cases, follow-up data showthat gains are sustained in some areas of perfor-mance but lost in others. Thus, the long-termeffects of a program are not simply evaluated bywhether the gains are retained or not. For exam-ple, in the junior high school program of Saferet al. (1981), some measures reflected mainte-nance of intervention effects (school attendance)and others did not (graduation from highschool). The same was true for the comparisonof Achievement Place and other facilities wherefollow-up gains for delinquents were differenton some measures but not on others (Kiriginet al., 1979, 1982; Jones et al., Note 1).

Third, with long-term follow-up, interveningexperiences (e.g., hospital aftercare) can obfus-cate the effects of the original program. Inter-vening and current environmental contingenciesmay exert more immediate impact on perfor-mance than a program a few years earlier. Thus,it is no surprise that several investigators havecautioned that token programs may have im-mediate impact but perhaps should not be ex-pected to alter future performance unless theenvironments to which persons return promotecontinuation of the gains (Bushell, 1978; Kiri-gin et al., 1982; Paul & Lentz, 1977).

Although many programs have shown thatthe effects of token economies are partially main-tained, maintenance and transfer continue to besalient issues. Whether intervention effects willbe maintained is still not entirely predictable.Several programs have shown that behavioralgains are lost when the program is terminated.

436

TOKEN ECONOMY

Why behaviors are maintained after some pro-grams but not others is not obvious. Simple hy-potheses such as those stating that particular tar-get behaviors are likely to be maintained are noteasily supported. Responses that might be ex-pected to be maintained by the natural environ-ment (e.g., social interaction) are lost in somestudies and maintained in others (see Kazdin,1980b). Programs that have shown long-termchanges often have been in effect for relativelylong periods. Protracted participation in a pro-gram may develop greater stability in the targetbehaviors so that they are less likely to be lost orto depend on immediate changes in the environ-ment once the client leaves the program.

As a general rule, it is still prudent to assumethat behavioral gains are likely to be lost in vary-ing degrees once the client leaves the program.Thus, special efforts are required to ensure thatthe gains are maintained, a point cogently madeby Baer, Wolf, and Risley in 1968 and fre-quently (and deservedly) cited ever since. In thelast decade, considerable progress has been madein identifying strategies that increase the likeli-hood that behaviors are maintained and extendto new settings (see Kazdin, 1980b; Stokes &Baer, 1977). The proposed strategies include:removing the token economy gradually so thatbehaviors are maintained with less direct rein-forcement; reinforcing behaviors under a varietyof situations so that the behaviors are not re-stricted to a limited range of cues; substitutingnaturally occurring reinforcers such as praiseand activities in place of tokens; altering theschedule and delay of reinforcement to prolongextinction; and using peers and clients them-selves as reinforcing agents to sustain long-termperformance across a variety of situations.

EMERGENT ISSUESAdvances over the last decade have not en-

tirely resolved the questions about the tokeneconomy and variables that contribute to be-havior change. However, over the years otherissues have emerged that directly pertain to the

limitations of token economies. The issues per-tain to the feasibility of implementing effectivetoken economies outside the domain of researchand demonstration projects.Integrity of TreatmentThe essential ingredients for beginning a to-

ken economy typically include identifying thetarget behaviors, the medium of exchange (to-kens) and backup reinforcers and specifying therelations among performance, token earnings,and expenditures. The success of token econo-mies is largely attributed to what the program is,i.e., the specific contingencies, rather than howthe program is conducted. Too little attentionhas been accorded the manner in which the pro-gram is monitored and implemented. Severalprocedures are often included in the programto help monitor the treatment to ensure that theprogram is carried out correctly. These proce-dures may be critical to the successful implemen-tation of token economies.To appreciate the point, reconsider the pro-

gram of Paul and Lentz (1977) which producedmarked changes in chronic psychiatric patients.Several features of the Paul and Lentz studyprobably contributed to the success of the tokenprogram. First, training of staff to implementthe treatments was extensive. Clinical staff re-ceived academic training that consisted of care-fully planned instruction in the different pro-cedures, using a detailed treatment manual as aguide. Training included opportunities for roleplaying, modeling, rehearsal, and feedback. Theacademic training was followed with on-the-jobtraining and supervised practice.

Second, monitoring of treatment was exten-sive to ensure that the programs were adminis-tered as planned. (The assessment proceduresused to monitor treatment are detailed in a seriesof papers published in a special issue of the Jour-nal of Behavioral Assessment, 1979, 1(3).) Su-pervisory staff monitored data on staff-patient in-teraction daily and provided positive feedback tostaff for flawless performance or corrective feed-back for departures from the desired procedures.

437

438 ~~~~~ALANE. KAZDIN

Professional observers monitored staff and pa-tients over the entire course of the programwhich provided a further check on execution ofthe program.

Third, the program included several person-nel in roles that depart from the usual staffing ofinpatient programs. Among the positions wereseveral interns who helped implement and eval-uate the program, persons to monitor staff-pa-tient interaction, professional observers, alreadymentioned, Ph.D. level staff to supervise the re-search, and so on. The mere presence of separateresearch and clinical staff and Ph.D. level re-search supervisors to monitor the day-to-day pro-gram, added a special feature to ensure properimplementation of the treatments.

Finally, the treatment procedures were rela-tively complex, as would be expected with tech-niques that focus on difficult clinical problems.The treatment was described in manual form.Constant updating of procedures was handledthrough memoranda to clarify implementationof practices, to answer questions, and so on. Boththe subtle day-to-day details and the resources tohave knowledgeable personnel to address suchquestions are very special program features thatmay have helped ensure that treatment was con-ducted as intended.

Characteristics such as those mentioned abovemay have had major bearing on the clinical im-pact of treatment and the generality of resultsto other clinical settings. The results, viewed su-perficially, suggest that a token economy canproduce dramatic inhospital and extrahospitalchanges and return chronic patients to the com-munity. However, the Paul and Lentz programwas implemented with multiple procedures toevaluate and monitor the execution of treatment.Programs without these latter procedures mayfall quite short of the mark in producing simi-lar changes.One of the major problems of treatment and

program evaluation is ensuring the integrity oftreatment, i.e., that treatment is carried out asintended (Rossi, 1978; Scheirer, 1981; Sechrest',West, Phillips, Redner, & Yeaton, 1979).-Moni-

toring the integrity of treatment is essential toensure that the program is being conducted cor-rectly. Different reports have indicated that to-ken economies deteriorate when supervision overexecution of the program is withdrawn or is notin place from the beginning. For example, Bas-sett and Blanchard (1977) reported a tokeneconomy in a prison setting for male adult of-fenders. When the director took a leave of ab-sence and provided supervision only on a con-sulting basis, the program deteriorated rapidly.Specifically, staff withheld tokens for appropri-ate behaviors, increased the use of fines, and be-came inconsistent in the magnitude of fines thatwere invoked. The eventual return of the pro-grams director and careful monitoring of thecontingencies, returned the program to its origi-nal state.

Similarly, Scheirer (1981) described a tokeneconomy for female chronic psychiatric patientsthat failed from the beginning in part becauseof the absence of personnel within the programto supervise staff directly and to ensure theirproper execution of the contingencies. No singleperson was in a position of authority to coordi-nate and supervise the day-to-day details of run-ning the program as part of the ward routine.

Finally, Rollins, Thompson, and their col-leagues developed token economies in severalelementary school classrooms (Rollins et al.,1974; Thompson et al., 1974). The programswere quite effective in altering student deport-ment and academic achievement. When the in-vestigators left the setting, the resources forsupervision and data collection were also with-drawn. When they returned 1 year later for fol-low-up assessment, the programs had been dis-continued and teacher and student behavior hadreturned to preprogram rates.

In general, the potent effects of token econo-mies may result in part from procedures in-cluded to ensure treatment integrity. Perhaps, aminimal condition to monitor treatment execu-tion is the continuous data collection on clientor staff behavior. Assessment procedures havebeen carefully developed and tested by Paul and

438

TOKEN ECONOMY

his colleagues to monitor patient and staff be-havior and program execution in institutionalsettings (see Engel & Paul, 1979; Licht, 1979;Mariotto, 1979; Power, 1979; Redfield, 1979).Of course, in many institutional settings, continu-ous data collection is often difficult to implementunless special consultants with outside resourcesare available (see Scheirer, 1981). With little orno feedback about direct execution of the pro-gram or its effects on client behavior, the integ-rity of treatment and the efficacy of the programare likely to be sacrificed.

Administrative and Organizational IssuesToken economies are frequently implemented

in institutional settings such as schools, psychi-atric facilities, and institutions for the mentallyretarded. Programs must work within the con-fines of organizational structures, administrativehierarchies, and external regulatory procedures.Organizational and administrative issues fre-quently dictate the extent to which implement-ing an effective token economy is feasible (seeScheirer, 1981). Constraints in permissible prac-tices within the institution, lack of authority orpower to follow through on program decisions,limited resources, and a variety of other sociopo-litical issues may interfere with beginning theprogram and maintaining the integrity of treat-ment once the program is initiated. Althoughorganizational obstacles have long been recog-nized in applied research (e.g., Tharp & Wetzel,1969), their significance in delimiting the ef-fects of token economies has only been fullyappreciated relatively recently (Ayllon et al.,1979; Bushell, 1978; Liberman, 1979; Rep-pucci & Saunders, 1974; Scheirer, 1981).

Professionals who are responsible for design-ing, implementing, and evaluating token econo-mies may not initially recognize organizationaland administrative issues as central to the pro-gram. However, the issues come into sharp focusas they affect more familiar variables that areknown to influence program effectiveness. Forexample, limited resources within an institutionmay be translated into a small budget for backup

reinforcers, insufficient staff to reward clients asfrequently as might be required, and absence ofpersonnel to assess behavior of clients or staff.In one instance, a limited budget resulted in theabsence of backup reinforcers in the store whenpsychiatric patients were to spend their tokens(Scheirer, 1981).

Administrative and institutional obstacles of-ten become obvious when beginning staff train-ing. As noted earlier, the administration of directreinforcement to staff has been an important in-gredient in many staff training programs. Yet,potent reinforcers (e.g., money, vacations, shiftpreferences, work breaks) usually are unavail-able for contingent application. Thus, staff train-ing programs often rely on in-service training,workshops, and feedback which by themselvesmay be less effective training procedures thanwhen used in conjunction with potent reinforc-ers. Even if incentives (e.g., certificates, recogni-tion in a institutional newsletter) are provided,staff behavior is not always easily assessed to en-sure that incentives are applied contingently.Whether the program variations known to be

effective can be implemented or implementedroutinely in applied settings is a major questionfacing the field at the present time. Questionsabout the feasibility of extending the token econ-omy have become more prominent over the lastdecade because they have direct implications forthe dissemination of the token economy, as dis-cussed below.

Dissemination of the Token EconomyMerely demonstrating that the token economy

is effective is not enough for it to be widelyadopted. A dramatic illustration of the processesand obstacles associated with program adoptionwas provided by Fairweather and his colleagueswho developed an effective aftercare program(community lodge) for psychiatric patients. Aresearch project was designed to disseminate theprogram to hospitals throughout the UnitedStates and to evaluate alternative methods ofpromoting program adoption. The results re-vealed a very small proportion (approximately

439

ALAN E. KAZDIN

10%) of the settings adopted the program (Fair-weather, Sanders, & Tornatzky, 1974) butprovided important information about the pro-gression from research and development to dis-semination and social policy change (Fair-weather & Tornatzky, 1977).

The characteristics of programs that may con-tribute to their widespread adoption have beenelaborated by Fawcett, Mathews, and Fletcher(1980) in the context of behavioral communitypsychology. Drawing from work on dissemina-tion in other fields (e.g., extension of technologyto underdeveloped countries), Fawcett et al.(1980) discussed the notion of an "appropriatetechnology" which refers to procedures that arecompatible with the context, resources, philoso-phy, and values of the settings in which they willbe used. Interventions that are contextually ap-propriate and likely to be adopted should be(a) effective, (b) relatively inexpensive, (c) de-centralized and controlled by local participants,(d) flexible enough to permit local input, (e)sustainable with local rather than outside re-sources, (f) relatively simple and comprehensi-ble, and (g) compatible with existing values,goals, and perceived needs of the setting.A few of the conditions of an appropriate

technology have begun to be addressed in re-lation to token economics. For example, theexpense (cost) of token economies has been ex-amined in different ways. Operating costs, cost-ef-fectiveness, and cost-benefit analyses have shownthat major benefits accrue from adopting tokeneconomies in relation to existing alternativessuch as routine psychiatric care or detention cen-ters for juvenile offenders (e.g., Foreyt, Rock-wood, Davis, Desvousges, & Hollingsworth,1975; Kirigin et al., 1979; Paul & Lentz, 1977).A major issue that has not been addressed is

the extent to which successful programs can besustained by local resources. Token economiesreported in the research literature often havefederal, state, or local funding to provide manyof the resources that promote effective programimplementation (e.g., money for staff who col-lect data, special reinforcers). Also, consultants

with interests in research and affiliations withacademic departments often play an importantrole in program development and implementa-tion. Resources not otherwise available in theinstitution (e.g., undergraduates) are oftendrawn upon to serve critical functions to main-tain and evaluate the program.

For example, the Behavior Analysis FollowThrough program was part of a specially fundeddemonstration project (Bushell, 1978). Execu-tion of the program was monitored centrally andlocally and teacher and student performancewere regularly evaluated. The favorable resultssuggest that the token economy could be ex-tended on a large scale provided the mechanismsfor monitoring and supervision are in place aswell. Whether the program could obtain theeffects without outside resources included as partof a special research project remains to be seen.Even with the program closely monitored andsupervised, treatment integrity and program ef-fectiveness varied across different sites.

If effective token economies are to be dissemi-nated, they may need to be designed to dependless on extraneous resources. Some attemptshave been reported to reduce outside monitoringand input. For example, token programs in theschools developed by Rollins et al. (1974) de-pended on extensive training and supervision byoutside consultants. To make the program lessdependent on outside resources, principals weretrained in behavioral techniques so they couldreturn to their schools, train teachers, and moni-tor the programs (Rollins & Thompson, 1978).Extrainstitutional resources continued to be usedto monitor program effectiveness. Yet, the at-tempt to develop programs that can functionautonomously with minimal outside resources isan important step.

Flexibility of the token economy warrantsspecial comment insofar as it may relate to treat-ment integrity. The willingness of administra-tors, staff, and others to adopt a particular pro-gram may be a function of the perceived oractual flexibility of the procedures. Yet, theremay be clear limitations in allowable flexibility

440

TOKEN ECONOMY

to ensure that changes in behavior are achieved.Procedural flexibility to promote local adoptionand treatment integrity to ensure effective appli-cations may need to be balanced. It may be criti-cal to delineate program characteristics that areknown or are likely to be essential for behaviorchange and those that are not. For example, ina token economy, reinforcers generally need tobe delivered contingently and behaviors thatcompete with the target behavior should not beinadvertently or directly reinforced. On theother hand, many of the details of the programsuch as the selection of target behaviors, tokens,or backup reinforcers, and the use of staff orpatients to administer consequences, individualor group contingencies, and other features canvary widely (Kazdin, 1977). Dissemination ofthe token economy may be facilitated by delin-eating the principles that need to be followedand the alternative ways in which they can betranslated into effective procedures. In this way,flexibility can be delineated even for essential in-gredients of the program.

The attention that token economies have re-ceived in research, claims made about their ef-fects, and their seeming ease of implementation,may already have fostered their dissemination.The extent to which token economies are is useis not known; published reports of individualprograms, of course, greatly underestimate theiractual prevalence. Programs conducted in mostapplied settings may not be evaluated or, if eval-uated, their effects may not be reported. Whethersuch programs conducted routinely in appliedsettings vary procedurally from those reportedin research and whether the outcomes approachthose reported in well monitored programs re-main to be determined.General Comments

Substantive questions about the token econ-omy and variables that contribute to its efficacyare by no means resolved. The array of programvariations and their relative effectiveness war-rant continued research. However, within thelast decade other issues that have become salient

pertain to the requirements of implementing to-ken economies effectively in applied settings aspart of routine care. The demands for maintain-ing the integrity of treatment, the ability to inte-grate token economies within administrative andorganizational constraints, and the ease of dis-seminating the token economy effectively on alarge scale appear to be especially relevant issuesat this time in dictating the future of the tokeneconomy.

SUMMARYThe token economy has been extended greatly

in recent years in terms of the range of popula-tions, settings, and target problems to which ithas been applied. A decade ago, salient issuespertained to how to improve client responsive-ness to the program and how to promote mainte-nance and transfer. Since then, several advanceshave been made in these areas. The paucity offollow-up data, a perennial issue in outcome re-search, still characterizes the token economy.However, in the last decade, several studies haveindicated that the effects of token economies areat least partially maintained and extend to extra-treatment settings.

Salient issues have emerged related to the ex-tension of the technique as part of routine prac-tice in treatment, rehabilitation, and educationalsettings. Many issues for effectively implement-ing the token economy have been identified re-lated to the requirements for maintaining theintegrity of treatment and integrating the pro-gram within administrative constraints of institu-tional settings in which such programs are likelyto be useful. The primary question is whetherthe token economy can be implemented effec-tively outside the context of demonstration orresearch projects which include special featuresto sustain the integrity of treatment and to over-come institutional obstacles. Apart from contin-uing to refine the technique and understandingthe variables that may contribute to its efficacy,the next step for research is to explore and evalu-ate procedures to integrate token economies rou-

441

442 ALAN E. KAZDIN

tinely into settings where programs are likely tobe of use.

REFERENCE NOTE1. Jones, R. R., Weinrott, M. R., & Howard, J. R.

The national evaluation of the teaching-familymodel. Final report to the National Institute ofMental Health, Center for Studies in Crime andDelinquency, June 1981.

REFERENCESAllen, D. J., & Magaro, P. A. Measures of change

in token-economy programs. Behaviour Researchand Therapy, 1971, 9, 311-318.

Ayllon, T., & Azrin, N. H. The measurement andreinforcement of behavior of psychotics. Journalof the Experimental Analysis of Behavior, 1965,8, 356-383.

Ayllon, T., & Azrin, N. H. Reinforcer sampling: Atechnique for increasing the behavior of mentalpatients. Journal of Applied Behavior Analysis,1968, 1, 13-20. (a)

Ayllon, T., & Azrin, N. H. The token economy: Amotivational system for therapy and rehabilita-tion. New York: Appleton-Century-Crofts, 1968.(b)

Ayllon, T., Milan, M. A., Robert, M. D., & McKee,J. M. Correctional rehabilitation and manage-ment: A psychological approach. New York:Wiley, 1979.

Baer, D. M., Wolf, M. M., & Risley, T. R. Somecurrent dimensions of applied behavior analysis.Journal of Applied Behavior Analysis, 1968, 1,91-97.

Bassett, J. E., & Blanchard, E. B. The effect of theabsence of close supervision on the use of responsecost in a prison token economy. Journal of Ap-plied Behavior Analysis, 1977, 10, 375-379.

Bassett, J. E., Blanchard, E. B., & Koshland, E. Ap-plied behavior analysis in a penal setting: Target-ing "free world" behaviors. Behavior Therapy,1975, 6, 639-648.

Bernstein, G. S. Training behavior change agents:A conceptual review. Behavior Therapy, 1982,13, 1-23.

Biklen, D. P. Behavior modification in a state men-tal hospital: A participant-observer's critique.American Journal of Orthopsychiatry, 1976, 46,53-61.

Boegli, R. G., & Wasik, B. H. Use of the tokeneconomy system to intervene on a school-widelevel. Psychology in the Schools, 1978, 15, 72-78.

Breyer, N. L., & Allen, G. J. Effects of implement-ing a token economy on teacher attending behav-

ior. Journal of Applied Behavior Analysis, 1975,8, 373-380.

Burg, M. M., Reid, D. H., & Lattimore, J. Use of aself-recording and supervision program to changeinstitutional staff behavior. Journal of AppliedBehavior Analysis, 1979, 12, 363-375.

Bushell, D., Jr. An engineering approach to theelementary classroom: The Behavior AnalysisFollow Through project. In A. C. Catania & T. A.Brigham (Eds.), Handbook of applied behavioranalysis: Social and instructional processes. NewYork: Irvington, 1978.

Curran, J. P., Lentz, R. J., & Paul, G. L. Effective-ness of sampling-exposure procedures on facilitiesutilization by psychiatric hard-core chronic pa-tients. Journal of Behavior Therapy and Experi-mental Psychiatry, 1973, 4, 201-207.

Engel, K. L., & Paul, G. L. Systems use to objectifyprogram evaluation, clinical, and managementdecisions. Journal of Behavioral Assessment,1979, 1, 221-238.

Fairweather, G. W., Sanders, D. H., & Tornatsky, L.G. Creating change in mental health organiza-tions. New York: Pergamon, 1974.

Fairweather, G. W., & Tornatsky, L. G. Experimen-tal methods for social policy research. Oxford:Pergamon, 1977.

Fawcett, S. B., Mathews, R. M., & Fletcher, R. K.Some promising dimensions for behavioral com-munity technology. Journal of Applied BehaviorAnalysis, 1980, 13, 505-518.

Feingold, L., & Migler, B. The use of experimentaldependency relationships as a motivating pro-cedure on a token economy ward. In R. B. Rubin,H. Fensterheim, J. D. Henderson, & L. P. Ullmann(Eds.), Advances in behavior therapy. New York:Academic Press, 1972.

Ferguson, J. M., & Taylor, C. G. (Eds.). The com-prehensive handbook of behavior medicine, Vol-ume 3: Extended applications and issues. NewYork: Spectrum, 1980.

Fisher, E. F., Jr., Winkler, R. C., Krasner, L., Kagel,J., Battalio, R. C., & Basmann, R. L. Economicperspectives in behavior therapy: Complex inter-dependencies in token economies. Behavior Ther-apy, 1978, 9, 391-403.

Foreyt, J. P., Rockwood, C. E., Davis, J. C., Des-vousges, W. H., & Hollingsworth, R. Benefit-cost analysis of a token economy program. Pro-fessional Psychology, 1975, 6, 26-33.

Fullerton, D. T., Cayner J. J., & McLaughlin-Reidel,T. Results of a token economy. Archives ofGeneral Psychiatry, 1978, 35, 1451-1453.

Gershone, J. R., Errickson, E. A., Mitchell, J. E., &Paulson, D. A. Behavioral comparison of atoken economy and a standard psychiatric treat-ment ward. Journal of Behavior Therapy andExperimental Psychiatry, 1977, 8, 381-385.

TOKEN ECONOMY 443

Glenwick, D., & Jason, L. (Eds.). Behavioral com-munity psychology: Progress and prospects. NewYork: Praeger, 1980.

Greenberg, D. J., Scott, S. B., Pisa, A., & Friesen, D.D. Beyond the token economy: A comparisonof two contingency programs. Journal of Consult-ing and Clinical Psychology, 1975, 43, 498-503.

Heaton, R. C., & Safer, D. J. Secondary school out-come following a junior high school behavioralprogram. Behavior Therapy, 1982, 13, 226-231.

Hemsley, D. R. Limitations of operant proceduresin the modification of schizophrenic functioning:The possible relevance of studies of cognitive dis-turbance. Behavioural Analysis and Modification,1978, 2, 165-173.

Hung, D. W. Generalization of "curiosity" ques-tioning behavior in autistic children. Journal ofBehavior Therapy and Experimental Psychiatry,1977, 8, 237-245.

Hussian, R. A. Geriatric psychology: A behavioralperspective. New York: Van Nostrand Reinhold,1981.

Jackson, D. A. Behavior analysis certification: Aplan for quality control. In T. A. Brigham, R.Hawkins, J. Scott, & T. F. McLaughlin (Eds.), Be-havior analysis in education: Self-control andreading. Dubuque, Iowa: Kendall/Hunt, 1976.

Karraker, R. J. Self versus teacher selected reinforc-ers in a token economy. Exceptional Children,1977, 43, 454-455.

Kazdin, A. E. The token economy: A review andevaluation. New York: Plenum, 1977.

Kazdin, A. E. Acceptability of time out from rein-forcement procedures for disruptive child behav-ior. Behavior Therapy, 1980, 11, 329-344. (a)

Kazdin, A. E. Behavior modification in applied set-tings (2nd ed.). Homewood: Ill.: Dorsey, 1980.(b)

Kazdin, A. E. Failure of persons to respond to thetoken economy. In E. B. Foa & M. G. Emmelkamp(Ed.), Failures in behavior therapy. New York:Wiley, in press.

Kazdin, A. E., & Bootzin, R. R. The token economy:An evaluative review. Journal of Applied Behav-ior Analysis, 1972, 5, 343-372.

Kazdin, A. E., & Geesey, S. Simultaneous-treatmentdesign comparisons of the effects of earning rein-forcers for one's peers versus for oneself. BehaviorTherapy, 1977, 8, 682-693.

Kazdin, A. E., & Geesey, S. Enhancing classroomattentiveness by preselection of back-up reinforc-ers in a token economy. Behavior Modification,1980, 4, 98-114.

Kazdin, A. E., & Mascitelli, S. The opportunity toearn oneself off a token system as a reinforcerfor attentive behavior. Behavior Therapy, 1980,11, 68-78.

Kirigin, K. A., Braukmann, C. J., Atwater, J. D., &Wolf, M. M. An evaluation of teaching-family(Achievement Place) group homes for juvenileoffenders. Journal of Applied Behavior Analysis,1982, 15, 1-16.

Kirigin, K. A., Wolf, M. M., Braukmann, C. J., Fix-sen, D. L., & Phillips, E. L. Achievement Place:A preliminary outcome evaluation. In J. S. Stum-phauzer (Ed.), Progress in behavior therapy withdelinquents. Springfield, Ill.: Charles C Thomas,1979.

Liberman, R. P. Social and political challenges tothe development of behavioral programs in orga-nizations. In P. Sjoden, S. Bates, & W. S. Dockens,III (Eds.), Trends in behavior therapy. New York:Academic Press, 1979.

Licht, M. H. The Staff-Resident Interaction Chrono-graph: Observational assessment of staff perfor-mance. Journal of Behavior Assessment, 1979, 1,185-197.

Lloyd, K. E. Reactions to a forthcoming energyshortage: A topic in behavioral ecology. In G. L.Martin & J. G. Osborne (Eds.), Helping in thecommunity: Behavioral applications. New York:Plenum, 1980.

Mariotto, M. J. Observational assessment systemsuse for basic and applied research. Journal ofBehavioral Assessment, 1979, 1, 239-250.

Martin, R. Legal challenges to behavior modifica-tion: Trends in schools, corrections and mentalhealth. Champaign, Ill.: Research Press, 1975.

McCreadie, R. G., Main, C. J., & Dunlop, R. A.Token economy, pimozide and chronic schizo-phrenia. British Journal of Psychiatry, 1978, 133,179-181.

McLaughlin, T. F., & Malaby, J. E. The comparativeeffects of token-reinforcement with and withouta response cost contingency with special educa-tion children. Educational Research Quarterly,1977, 2, 34-41.

McLaughlin, T. F., Williams, R. L., Truhlicka, M.,Cady, M., Ripple, B. J., & Eakins, D. Modelimplementation and classroom achievement inthe Northern Cheyenne Behavior Analysis Fol-low Through project. Child and Family BehaviorTherapy, in press.

McReynolds, W. T., & Coleman, J. Token economy:Patient and staff changes. Behaviour Research andTherapy, 1972, 10, 29-34.

Milby, J. B., Clarke, C., Charles, E., & Willcutt, H. C.Token economy process variables: Effects of in-creasing and decreasing the critical range of sav-ings. Behavior Therapy, 1977, 8, 137-145.

Milby, J. B., Pendergrass, P. E., & Clarke, C. J. To-ken economy versus control ward: A comparisonof staff and patient attitudes toward ward environ-ment. Behavior Therapy, 1975, 6, 22-29.

444 ALAN E. KAZDIN

Mishara, B. L. Geriatric patients who improve intoken economy and general milieu treatment pro-grams: A multivariate analysis. Journal of Con-sulting and Clinical Psychology, 1978, 46, 1340-1348.

Montegar, C. A., Reid, D. H., Madsen, C. H., & Ewell,M. D. Increasing institutional staff to residentinteractions through in-service training and super-visor approval. Behavior Therapy, 1977, 8, 533-540.

Moran, E. L., Kass, W. A., & Munz, D. C. In-pro-gram evaluation of a community correctionalagency of high-risk offenders. Corrective and So-cial Psychiatry, 1977, 23, 48-52.

Murphy, S. T. The effects of a token economy pro-gram on self-care behaviors of neurologically im-paired inpatients. Journal of Behavior Therapyand Experimental Psychiatry, 1976, 7, 145-147.

O'Leary, K. D. The operant and social psychologyof token systems. In A. C. Catania & T. A. Brig-ham (Eds.), Handbook of applied behavior anal-ysis: Social and instructional processes. NewYork: Irvington, 1978.

O'Leary, K. D., & Drabman, R. Token reinforce-ment programs in the classroom: A review. Psy-chological Bulletin, 1971, 75, 379-398.

Paul, G. L., & Lentz, R. J. Psychosocial treatment ofchronic mental patients: Milieu versus social-learning programs. Cambridge, Mass.: HarvardUniversity Press, 1977.

Phillips, E. L., Phillips, E. A., Wolf, M. M., & Fixsen,D. L. Achievement Place: Development of theelected manager system. Journal of Applied Be-havior Analysis, 1973, 6, 541-561.

Power, C. T. The Time-Sample Behavioral Check-list: Observational assessment of patient function-ing. Journal of Behavioral Assessment, 1979, 1,199-210.

Redfield, J. Clinical Frequencies Recording System:Standardizing staff observations by event record-ing. Journal of Behavioral Assessment, 1979, 1,211-219.

Reppucci, N. D., & Saunders, J. T. Social psychol-ogy of behavior modification: Problems of imple-mentation in natural settings. American Psycholo-gist, 1974, 29, 649-660.

Rickard, H. C., Melvin, K. B., Creel, J., & Creel, L.The effects of bonus tokens upon productivity ina remedial classroom for behaviorally disturbedchildren. Behavior Therapy, 1973, 4, 378-385.

Rollins, H. A., McCandless, B. R., Thompson, M., &Brassell, W. R. Project Success Environment:An extended application of contingency manage-ment in inner-city schools. Journal of EducationalPsychology, 1974, 66, 167-178.

Rollins, H. A., & Thompson, M. Implementationand operation of a contingency management pro-gram by the elementary school principal. Ameri-

can Educational Research Journal, 1978, 15, 325-330.

Rossi, P. H. Issues in the evaluation of humanservices delivery. Evaluation Quarterly, 1978, 2,573-599.

Safer, D. J., Heaton, R. C., & Parker, F. C. A be-havioral program for disruptive junior highschool students: Results and follow-up. Journalof Abnormal Child Psychology, 1981, 9, 483-494.

Scheirer, M. A. Program implementation: The or-ganizational context. Beverly Hills, Calif.: Sage,1981.

Sechrest, L., West, S. G., Phillips, M. A., Redner, R.,& Yeaton, W. Some neglected problems inevaluation research: Strength and integrity oftreatments. In L. Sechrest, S. G., West, M. A.Phillips, R. Redner, & W. Yeaton (Eds.), Evalua-tion studies: Review annual (Vol. 4). BeverlyHills, Calif.: Sage, 1979.

Stoffelmayr, B. E., Faulkner, G. E., & Mitchell, W. S.The comparison of token economy and socialtherapy in the treatment of hard-core schizo-phrenic patients. Behavioural Analysis and Modi-fication, 1979, 3, 3-17.

Stokes, T. F., & Baer, D. M. An implicit technologyof generalization. Journal of Applied BehaviorAnalysis, 1977, 10, 349-367.

Tharp, R. G., & Wetzel., R. J. Behavior modifica-tion in the natural environment. New York: Aca-demic Press, 1969.

Thompson, M., Brassell, W. R., Persons, S., Tucker,R., & Rollins, H. Contingency management inthe schools: How often and how well does itwork? American Educational Research Journal,1974, 11, 19-28.

Trudel, G., Boisvert, J., Maruca, F., & Leroux, P.Unprogrammed reinforcement of patients' behav-iors in wards with and without token economy.Journal of Behavior Therapy and ExperimentalPsychiatry, 1974, 5, 147-149.

Walker, H. M., Hops, H., & Fiegenbaum, E. Devi-ant classroom behavior as a function of combina-tions of social and token reinforcement and costcontingency. Behavior Therapy, 1976, 7, 76-88.

Wexler, D. B. Token and taboo: Behavior modifica-tion, token economies, and the law. CaliforniaLaw Review, 1973, 61, 81-109.

Wilkinson, L., & Reppucci, N. D. Perceptions ofsocial climate among participants in token econ-omy and non-token economy cottages in a juve-nile correctional institution. American Journal ofCommunity Psychology, 1973, 1, 36-43.

Winkler, R. C. An experimental analysis of eco-nomic balance, savings and wages in a token econ-omy. Behavior Therapy, 1973, 4, 22-40.

Wolf, M. M. Social validity: The case of subjectivemeasurement or how applied behavior analysis is

TOKEN ECONOMY 445

finding its heart. Journal of Applied BehaviorAnalysis, 1978, 11, 203-214.

Wolf, M. M., Phillips, E. L., Fixsen, D. G., Brauk-mann, C. J., Kirigin, K. A., Willner, A. G., &Schumaker, J. B. Achievement Place: The teach-ing-family model. Child Care Quarterly, 1976, 5,92-103.

Wooley, S. C., Blackwell, G., & Winget, C. A learn-ing theory model of chronic illness behavior: The-

ory treatment and research. Psychosomatic Medi-cine, 1978, 40, 379-401.

Zeldow, P. B. Some antitherapeutic effects of thetoken economy: A case in point. Psychiatry, 1976,39, 318-324.

Received October 26, 1981Final acceptance February 2, 1982