clinical innovation and evaluation: integrating practice ...gdaviso/library/davison's...

12
Clinical Innovation and Evaluation: Integrating Practice with Inquiry Gerald C. Davison, University of Southern California Arnold A. Lazarus, Rutgers University We explore the complex interplay of clinical discovery and controlled evaluation, demonstrating how experi- ence in the applied arena provides invaluable insights and ideas about the complexity of the human condition and of ways to intervene effectively. Case studies have features that earn them a firm place in psychological research, and to ignore their potential contributions is to limit severely the kind of knowledge that can be generated by more systematic modes of inquiry. Some li mitations of group designs in comparative therapy re- search are also reviewed, again highlighting the impor- tance of idiographic analyses of single cases. Innovation and creative advancement are most readily nurtured via immersion in clinical/applied work, but at the same time the nature of that work is inevitably shaped by theories and hypotheses that clinicians bring into the applied setting. These abstractions are themselves in- fluenced by the clinician's interpretations of data, which interpretations are molded by theoretical and metatheoretical preconceptions. In this complex and i nteractive fashion, clinical innovation is part of a non- linear network of forces that includes personal biases, professional allegiances, epistemological assumptions, theoretical preferences, and familiarity with and use of certain bodies of data. Key words: case studies, psychotherapy outcome research. [Clin Psycho) Sci Prac 1:157-168, 1994] At a recent professional meeting, a colleague was dis- cussing a new and unusual clinical procedure that its pro- Address correspondence to Gerald C. Davison, Department of Psychology, University of Southern California, Los Angeles, CA 90089-1061. ponents had reported the previous year in a clinical case study-a technique that, it was claimed, could eliminate long-standing fears that had proven recalcitrant to other interventions. "How could such a thing be possible?" we asked with the utmost skepticism and even derision, all the while harboring the hope that the clinical results were valid and replicable. The response from our trusted friend: "Look, who knows if it works or not, or how it works if it does work? But I do know two things: First, we can't afford to ignore a promising finding; and second, the people making the claims have their heads screwed on straight." Continued our friend: "So I've begun doing some controlled laboratory research on the procedure. I'll call you next year when my results are in." Our own reply to him: "Fine, Tom, we're impressed that you're im- pressed enough to spend some time on it. If your experi- ments tell us that the clinical reports are probably true or at least believable, we'll be after you to teach us how to do it." There are a couple of points we wish to make with this little exchange. First, innovations by clinicians are the lifeblood of advances in the development of new thera- peutic interventions. Second, whether or not attention is paid to a discovery-especially if that discovery borders on the unbelievable-depends in large measure on a prior pro hominem judgment we have made about the integrity and standing of the creative clinician. And third, clinical innovations are often (though definitely not invariably or even primarily) investigated by more exper- i mentally minded workers whose subsequent findings may persuade others that the previously unbelievable technique is worth a closer look. In what follows we try to make explicit the nature of the complex interplay between clinical innovation and controlled evaluation. The importance of attending to the idiographics of individual cases is also explored in a © 1994 AMERICAN PSYCHOLOGICAL ASSOCIATION D12 0969-5893/94/$5.00 157

Upload: duonghanh

Post on 13-May-2018

234 views

Category:

Documents


1 download

TRANSCRIPT

Clinical Innovation and Evaluation: Integrating Practicewith InquiryGerald C. Davison, University of Southern CaliforniaArnold A. Lazarus, Rutgers University

We explore the complex interplay of clinical discoveryand controlled evaluation, demonstrating how experi-ence in the applied arena provides invaluable insightsand ideas about the complexity of the human conditionand of ways to intervene effectively. Case studies havefeatures that earn them a firm place in psychologicalresearch, and to ignore their potential contributions isto limit severely the kind of knowledge that can begenerated by more systematic modes of inquiry. Somelimitations of group designs in comparative therapy re-search are also reviewed, again highlighting the impor-tance of idiographic analyses of single cases. Innovationand creative advancement are most readily nurturedvia immersion in clinical/applied work, but at the sametime the nature of that work is inevitably shaped bytheories and hypotheses that clinicians bring into theapplied setting. These abstractions are themselves in-fluenced by the clinician's interpretations of data,which interpretations are molded by theoretical andmetatheoretical preconceptions. In this complex andi nteractive fashion, clinical innovation is part of a non-linear network of forces that includes personal biases,professional allegiances, epistemological assumptions,theoretical preferences, and familiarity with and use ofcertain bodies of data.

Key words: case studies, psychotherapy outcomeresearch. [Clin Psycho) Sci Prac 1:157-168, 1994]

At a recent professional meeting, a colleague was dis-cussing a new and unusual clinical procedure that its pro-

Address correspondence to Gerald C. Davison, Department ofPsychology, University of Southern California, Los Angeles,CA 90089-1061.

ponents had reported the previous year in a clinical casestudy-a technique that, it was claimed, could eliminatelong-standing fears that had proven recalcitrant to otherinterventions. "How could such a thing be possible?" weasked with the utmost skepticism and even derision, allthe while harboring the hope that the clinical results werevalid and replicable. The response from our trustedfriend: "Look, who knows if it works or not, or how itworks if it does work? But I do know two things: First,we can't afford to ignore a promising finding; and second,the people making the claims have their heads screwedon straight." Continued our friend: "So I've begun doingsome controlled laboratory research on the procedure.I'll call you next year when my results are in." Our ownreply to him: "Fine, Tom, we're impressed that you're im-pressed enough to spend some time on it. If your experi-ments tell us that the clinical reports are probably true orat least believable, we'll be after you to teach us how todo it."

There are a couple of points we wish to make withthis little exchange. First, innovations by clinicians are thelifeblood of advances in the development of new thera-peutic interventions. Second, whether or not attention ispaid to a discovery-especially if that discovery borderson the unbelievable-depends in large measure on aprior pro hominem judgment we have made about theintegrity and standing of the creative clinician. And third,clinical innovations are often (though definitely notinvariably or even primarily) investigated by more exper-imentally minded workers whose subsequent findingsmay persuade others that the previously unbelievabletechnique is worth a closer look.

In what follows we try to make explicit the nature ofthe complex interplay between clinical innovation andcontrolled evaluation. The importance of attending tothe idiographics of individual cases is also explored in a

© 1994 AMERICAN PSYCHOLOGICAL ASSOCIATION D12 • 0969-5893/94/$5.00

157

discussion of the inherent limitations of group designs inshedding light on how particular patients change in ther-apy research. Our sometimes abstract discussions shouldbe considered in the light of the highly personal judg-ments that clinical workers and scientists make aboutwhat is worth a closer look. This article is not a reviewof comparative outcome studies leading to conclusionsabout the superiority of one theory or technique overanother. Like an earlier effort (Lazarus & Davison, 1971),it is written from the perspectives of both authors andis therefore biased in the choice of examples and in thepreference for a cognitive-behavioral approach.

CLINICAL INNOVATION AND CONTROLLEDRESEARCHMany regard the laboratory and the clinic as oppositeends of a continuum. Research is said to be precise, con-trolled, and uncontaminated. The ideas that flow fromapplied settings are often regarded as woolly, riddled withbias, purely anecdotal, and even useless. Our abidingbelief is that the path between the laboratory and theclinic is a two-way street (Woolfolk & Lazarus, 1979). Weaver that most new methods have come from the workof creative clinicians. Furthermore, as we hope to show,the process of discovery that is carried on within the clini-cal practices of some therapists is the equivalent ofresearch.

Scientists and practicing clinicians can each offerunique contributions in their own right and can conceiv-ably open hitherto new and unsuspected clinical-experimental dimensions for research and practice. Ideastested in the laboratory may be applied by the practitionerwho, in turn, may discover important individual nuancesthat remain hidden from the laboratory scientist simplybecause the tight environment of the experimental test-ing ground makes it impossible for certain behaviors tooccur or for certain observations to be made. Conversely,ideas formulated in the clinic, provided that they areamenable to disproof, can send scientists scurrying offinto laboratories and other research settings to subject theclaims of efficacy to controlled tests. Cases in point arecited further on.

While it is proper to guard against ex cathedra state-ments based upon flimsy and subjective evidence, it is aserious mistake to discount the importance of clinicalexperience per se. There is nothing mysterious about thefact that repeated exposure to any given set of conditions

CLINICAL PSYCHOLOGY: SCIENCE AND PRACTICE • VI N 2, WINTER 1994

makes the recipient aware of subtle cues and contingen-cies in that setting which elude the scrutiny of those lessfamiliar with the situation. Clinical experience enables atherapist to recognize problems and identify trends thatare usually beyond the perceptions of novices, regardlessof their general expertise. It is at this level that new ideascome to the practitioner and often constitute break-throughs that could not be derived from animal analoguesor tightly controlled investigations. Different kinds ofdata and differing levels of information are obtained inthe laboratory and the clinic. Each is necessary, useful,and desirable.

CLINICAL INNOVATION AND EXPERIMENTATIONA valuable perspective upon the clinical research enter-prise should follow from a concerted effort to make theprocess of clinical experimentation and innovation moreexplicit. Innovation is the outcome of experimentation,for it is when we try new things that true innovators havethe capacity to appreciate relationships that may go unno-ticed by less resourceful, less observant workers.

Most practitioners, for instance, have discovered howdifficult it is to console many individuals immediately fol-lowing an important loss in their lives (such as death of afriend, dismissal from a job, rejection by a lover, or a simi-lar deprivation). The ensuing depression often remainsunaltered by reassurance. Supportive therapy over severalweeks, notwithstanding the ever-present risk of suicide inthese cases, often heralds the diminution of the patient'sreported misery. In the present context, the clinical inno-vator is the person who addresses the problem of whatcan be done to facilitate rapid recovery from such adepression. This attitude demands some form of experi-mentation. The actual experimental operations usuallyare determined in part by the therapist's own theoreticalorientation. For instance, a therapist with a proclivity fororganic notions will be more inclined to search for aneffective combination of drugs or some other biologicalmode of intervention. The psychologically oriented ther-apist will search for effective psychosocial procedures.The cognitive theorist might look for newer and deepermediating belief systems that mitigate subjective miseryrather than for novel means of psychomotor expressive-ness, which those who espouse various experiential theo-ries might be inclined to develop. Occasionally, a senseof desperation may lead a clinician to make a responsethat fits neither his or her theoretical preconceptions nor

158

his or her more usual empirical resources. Many clinicaladvances are preceded by what we might term a frustra-tion-observation sequence.

Let us consider the practitioner who has expendedenergy, time, and effort to alleviate the suffering of asomewhat depressed but extremely demanding individ-ual. The therapist's fund of methods and techniques hasbeen exhausted to no avail. Despite attempts to interveneat the level of family relationships, to tap underlying guiltsand hostilities, to assess for and alter biased thinking anddysfunctional beliefs, and to ply the patient with appro-priate medication and inspiration, the net result is ademanding and threateningly dependent person whosedesperation evokes anxiety in the therapist. At this stage,the harassed and perplexed practitioner may advocate acourse of action dictated solely by pragmatic conveniencerather than by theoretical confidence. Out of keepingwith his or her usual practices, the therapist may confinethe patient to bed for 10 days and forbid any patient-therapist communication during this period. In all can-dor, the therapist's principal motive might simply be to"get the patient off my back" for a while. Ten days later,the patient is seen again and, quite remarkably, reportsfeeling much better.

Unplanned or unexpected clinical improvements areoften dismissed as "fortuitous events" or "spontaneousremissions," but the clinical innovator is the one whocarefully notes a variety of possible cause-effect sequencesand thus discovers therapeutic levers that less inquisitivecolleagues are apt to overlook. A propitious clinical out-come might stimulate innumerable questions. In thehypothetical case already mentioned, one might simplypose the obvious question: "Of what value mightenforced bed rest be for certain cases of depression?"The clinician might then look for an additional casepresenting with similar problems and try the bed-restintervention, this time not out of frustration but in anincreasingly systematic effort to evaluate its potentialeffectiveness. If favorable changes are again observed, dis-cussion at professional meetings and publication of casereports can set the stage for comparing experiences withother practitioners and conceivably stimulate experimen-tally minded clinical researchers to conduct clinical trials,where experimental design, in its most rigorous sense,becomes essential. The independent variable-confiningX category of patients to Y units of enforced bed rest-calls for all the salient safeguards against experimenter

bias, contamination, and the like. An empirical rulemight even evolve. "A period of 7 to 10 days of enforcedbed rest can be expected to ameliorate depressive reac-tions in persons between the ages of 15 to 60 years inwhom the depressive reaction has been clearly precipi-tated by an event no more than 3 months previously, andin persons who have no history of chronic withdrawalreactions." Further experimentation will unearth excep-tions to the rule and often call for its modification andrevision. Eventually, practitioners may acquire a preciseand scientific clinical rule or procedure with a predictablesuccess rate and clear lines of contraindication. This routeis quite familiar in the practice of medicine but is some-what neglected in the practice of psychotherapy, where,until recently, clinicians seemed bound to use only thosemethods that were justified by their theories.

Another important kind of innovation occurs whenone borrows a technique from another orientation andincorporates it into one's own (different) conceptualframework and mode of practice. The complex episte-mological issues surrounding this practice, for which Laz-arus has been a strong proponent for many years, wererecently the subject of a lively debate (Lazarus & Messer,1991) and will not be repeated here. What is relevant tous is Messer's contention that, for example, when a non-Gestalt therapist like Lazarus uses the empty chair tech-nique, the importation of that procedure into a differentframework necessarily changes it enough so that itbecomes something else. The technique may well effectimprovement unanticipated by the creator and also lenditself to beneficial modifications and extensions.

The down side of this creative step is that any eviden-tiary justification of the original technique is probably notapplicable to the new conceptual and applied setting.Simply stated, Perls's empty chair is not Lazarus's emptychair. But the advantage of this eclectic maneuver is thatone is necessarily creating a new technique both by virtueof how one thinks about it and by virtue of operationalchanges that are almost certainly going to be introduced(and we assume that the latter follow from the former orare at least correlated with it).

THEORY AND PRACTICEMany difficulties arise when different theorists endeavorto reconcile identical empirical facts within divergenttheoretical models. The efficacy of the aforementioned"bed-rest hypothesis," if empirically established, will be

CLINICAL INNOVATION AND EVALUATION: INTEGRATING PRACTICE WITH INQUIRY • DAVISON AND LAZARUS 15 9

explained organically by organicists, psychoanalyticallyby psychoanalysts, behaviorally by behaviorists, and soforth. All too often a useful method will be employed bypractitioners of different theoretical persuasions only if itcan be "explained" according to their own favorite the-ories.

A common avenue of clinical experimentation con-sists of the development of techniques arising out of thetherapist's predilections. This was the route followed bymost of the psychoanalytic offshoots. Very often,although departing from his teachings and generatingindependent hypotheses of their own, Freud's formerpupils did not deviate very widely in matters of tech-nique-free association, dream interpretation, and analy-sis of transference retained their preeminence. Thedifferences revolved around points of emphasis, timing,and content of interpretations. The respective deviationsin technique were usually dictated by the different theo-retical views that the Freudian revisionists espoused(although none of them systematically evaluated theeffects of their innovations).

It stands to reason that a theorist who believes thatemotional disturbances arise out of feelings of inferioritymight develop and use different methods and techniquesthan a therapist who holds to a theory of unconscioussexual repression. The grave error is then to assume thatif a technique proves successful in achieving its desiredresults, the process that gave rise to it is thereby necessar-ily strengthened or confirmed. For example, a Rankianmight have reasoned that a depressed individual is activelyreliving the birth trauma and craving an intrauterinerespite. Employing enforced bed rest as a symbolic returnto the womb, and then discovering a clinical improve-ment in X number of patients, the committed Rankian ismost likely to resist the notion that the clinical outcomemight be unrelated to Rankian theories about the basictherapeutic process. Techniques may, in fact, prove effective forreasons that do not remotely relate to the theoretical ideas thatgave birth to them.

There is another side to the theory/practice issue,however, that we feel is sometimes dismissed. Whenselecting therapeutic techniques it matters very muchwhich theoretical notions a clinician espouses during theconduct of all clinical activities. For example, if oneassumes that a given phobic reaction is best conceptual-ized as an anxiety-avoidance gradient, and furthermore is

CLINICAL PSYCHOLOGY: SCIENCE AND PRACTICE • V1 N2, WINTER 1994

not secondary to a basic underlying condition that is theproper focus for treatment, one is more likely to employ,with confidence, a technique like desensitization or otherexposure-based interventions (Wolpe, 1958). Conversely,if one holds to a view that all phobias are adaptive tothe extent that they protect the individual from libidinalimpulses that would be devastating were they allowedexpression, it would seem that the clinician might chooseto dwell upon the presumed unconscious conflicts andignore the manifest phobia. This is not to say that onlyone particular theoretical stance will lead to a particularintervention; rather, it is to say that the "set" with whicha clinician approaches a problem determines the clini-cian's own clinical behavior and view of what occurs. Thisis one reason why we advocate caution, tentativeness, andempirical testing when adopting any theoretical position.Often such positions harden commitments rather thanfacilitate discovery, which is the real purpose of theories.

Once one has assimilated certain theoretical con-structs, it is necessary to apply these nomothetic prin-ciples to an idiographic case (Levine, Sandeen, &Murphy, 1992). Gordon Allport (1937) was identifiedwith the so-called "nomothetic-idiographic contro-versy," but Maher (1966) made a convincing argumentagainst a necessary incompatibility between these twoapproaches. The application of a general principle in aparticular case depends not only on a familiarity with theprinciple but also on an accurate assessment of the givencase. The example Maher uses is that of an engineer whomust build a bridge across a particular river: "In order tobuild a bridge over a certain river, we must know thedetails of the soil mechanics, water flow, prevailingwinds, topography, traffic usage, availability of labor andmaterials, and so on. When we consider all these, thetotal picture might not be like any other bridge that hasever been built. Nevertheless, none of the principles orassumptions that go into the final decisions could bemade in contradiction to the laws of physics, economics,and the like" (p. 112). We return to this important issuelater.

SOME CHARACTERISTICS OF CASE STUDIES ASRELATED TO CONTROLLED RESEARCHAs mentioned at the outset, explicating the processes ofclinical innovation is necessarily personal and idiosyn-cratic. We therefore focus on our own experiences and

160

use the resultant methods and findings to communicatewhat we mean by clinical experimentation. Otherendeavors (e.g., Kazdin, 1994) concentrate on a widerange of research and design issues in clinical research andin many ways complement our own efforts.

When creative clinicians learn new things frompatients and invent new procedures to resolve difficultproblems, they are conducting a form of research. Thereis usually a series of clinical trials or experiments in whichreactions of a patient or several patients to the procedureare observed. How can clinicians make these discoveries,and how can they learn about therapeutic change andinvent techniques for promoting it? A complete aware-ness of what information can and cannot be derived fromcase studies constitutes the first essential step toward theachievement of any meaningful clinical innovations thatdeserve to be labeled "research."

As we articulated in an earlier publication (Lazarus &Davison, 1971), there seem to us to be several characteris-tics unique to case studies that earn for them a firm placein psychological research. We outline them below andthen elaborate on each one:

1. A case study may cast doubt upon a general theory.2. A case study may provide a valuable heuristic to

subsequent and better controlled research.3. A case study may permit the investigation, although

poorly controlled, of rare but important phenomena.4. A case study can provide the opportunity to apply

principles and notions in entirely new ways.5. A case study can, under certain circumstances, pro-

vide enough experimenter control over a phenomenonto furnish "scientifically acceptable" information.

6. A case study can assist in placing "meat" on the"theoretical skeleton."

1. A Case Study May Cast Doubt Upon a General Theory.The successful handling of a particular case may under-score an important exception to a theory. For example, agiven theory may hold that a certain kind of problem isuntreatable. If a therapist succeeded in making an impactupon the recalcitrant problem, this would cast doubtupon the tenets of the theoretical viewpoint under con-sideration.

A particular theory may also predict that certain meth-ods will prove antitherapeutic. For example, when a fairlyelaborate case history was presented to two different

audiences-one made up mainly of psychodynamicpractitioners, the other comprising clinicians whoespoused a family systems perspective-both groups pre-dicted rapid relapse for different reasons (Lazarus, 1989a).The case was that of a 32-year-old man with multiplecomplaints, including generalized anxiety, depression,obsessive-compulsive problems, somatization, agora-phobia, and social isolation. The treatment followed abroad-based cognitive-behavioral approach (multimodaltherapy; Lazarus, 1989b). Because certain psychodynamicissues were not addressed or resolved, the practitionerswho identified themselves with psychodynamic theoryasserted that the unresolved conflicts would render thetreatment gains impermanent and ephemeral. Onespokesperson made the dire prediction that the clientmight decompensate and end up in a mental hospitalwithin 3 to 5 years. The family systems therapists cor-rectly pointed to the powerful mother-son enmeshmentthat was a fundamental part of the client's problems, andthey contended that the improper management of thisdyadic system (from their theoretical standpoint), plusother triangulating and scapegoating issues that werebypassed, heralded only a temporary remission. The factthat a 7-year follow-up revealed that the client had main-tained and further extended his therapeutic gains bringssome of their respective theoretical notions into seriousquestion. Let it be remembered that only one clearlynegative instance is sufficient to cast doubt on any gen-eral hypothesis.

2. A Case Study May Provide a Valuable Heuristic to Subse-quent and Better Controlled Research.Case studies in clinical psychology are probably bestknown for suggesting new directions that can be pursuedsystematically by laboratory investigators. Examples arelegion. The research in systematic desensitization thatvirtually exploded in the late 1960s into the 1970s proba-bly would not have been undertaken without the clinicalsuccesses being reported by several pioneering prac-titioners (e.g., Wolpe & Lazarus, 1966). The cognitive-behavioral therapy movement of the later 1970s thatextends into the present is derived largely from the clini-cal reports and theoretical propositions first propoundedby Ellis (1962) and Beck (1967). The role of small-N orsingle-case studies is thoroughly described in a volumededicated exclusively to case study methodology (Yin,

CLINICAL INNOVATION AND EVALUATION: INTEGRATING PRACTICE WITH INQUIRY • DAVISON AND LAZARUS 16 1

1989). (The limitations of comparative group designs arediscussed in a later section.) It has been argued that if weare ever to discover what aspects of a particular techniqueresult in therapeutic change, and exactly how this isachieved, global outcomes must be broken down into aseries of small interrelated changes. Single-case method-ology and designs are ideal for this purpose.

Let us turn to an example of a case study that led tosubsequent and well-controlled research. In a book ofLazarus's collected papers, Dryden (1991) included theaccount of a 19-year-old woman with a severe hand-washing compulsion. She had been treated in the early1960s, and Dryden comments: "It is interesting to notehow Lazarus struggled to find a means whereby the cli-ent's compulsive washing could be eliminated. In retro-spect, he finally administered a form of `responseprevention', which is now considered an indispensablecondition in the amelioration of most compulsive habits"(p. 36).

Initially, some of the best researched and well-documented methods of the 1960s had been applied inthis case with great diligence but without success-pro-gressive relaxation, systematic desensitization, and covertsensitization. Finally, the use of a portable faradic shockunit that delivered an unpleasant but entirely safe andnondamaging electrical impulse to her upper arms, wasbrought into play whenever she overdid any hand-washing. This resulted merely in a temporary hand-washing avoidance, but did not lessen her desire or urgeto wash. It was only when the procedure dramaticallyincreased the latencies between her hand-washing urgesand the act of self-cleansing (i.e., protracted response pre-vention), that significant clinical gains accrued. Subse-quently, "avoidance of the reinforcing habit" wasidentified as the active treatment ingredient. Thereafter,in similar cases, the "aversion therapy" was discarded, andresponse prevention became the mainstay (see Spiegler &Guevremont, 1993). More recent therapies for obsessive-compulsive disorders point to the synergistic impact ofadding medication (e.g., clomipramine) and cognitiverestructuring to exposure and response prevention(Franks, Wilson, Kendall, & Foreyt, 1990).

Davison's research on Articulated Thoughts in Simu-lated Situations (ATSS) (Davison, Robins, & Johnson,1983) is another example of the way in which case studiescan provide a valuable heuristic to subsequent and better

CLINICAL PSYCHOLOGY: SCIENCE AND PRACTICE • V1 N2, WINTER 1994

controlled research (see Davison, Navarre, & Vogel, inpress; Davison & Neale, 1994).

3. A Case Study May Permit the Investigation, Although PoorlyControlled, of Rare but Important Phenomena.Human beings are capable of harming themselves, andothers, in the most unusual ways. It is the practicing clini-cian who is most likely to encounter the vagaries ofhuman conduct. Unusual case reports from "field observ-ers" can add to clinical and experimental knowledge. Forexample, one of us was consulted some years ago by ayoung married couple who practiced a peculiar sexual rit-ual. As a prelude to sexual intercourse, the young manwould draw blood by cutting a small incision on the palmof his wife's right hand. She would then stimulate hispenis, using the blood of her right palm as a lubricant.Normal intercourse would then ensue, and the momentthe wife felt her husband ejaculating, she was required todig her nails deep into the small of his back or buttocks.Each sadomasochistic act was followed by guilt andremorse. But the most serious consequence of theseabnormal practices was the fact that repeated septicwounds tended to develop.' The couple in question wereboth intelligent college graduates who had indulged inthese gory activities intermittently for 3'/z years. Theywere vague and uncertain about the beginnings of theirstrange behavior. Signs of delusions, hallucinations, con-crete or overinclusive thinking, and other psychoticbehaviors were not present.

Knowledge, in cases such as this, can be advanced intwo ways. First, a detailed study could conceivably throwlight on the genesis of the problem and thereby add newinsights into human aberrations. Secondly, a successfultreatment strategy applied in this case may have relevancefor overcoming other deviant behaviors. In regard to thefirst consideration, only vague and very tentative infer-ences could be drawn. At best, there appeared to be someassociations with menstruation (probably safe vis-a-vispossible pregnancy), but the husband's connectionbetween blood per se and sexual excitement remainedobscure. It is possible that an extended analysis of thecouple may have proved enlightening, but a remarkablysimple remedy proved so successful that they were nolonger motivated to undergo further therapy. When ittranspired that they had desisted from engaging in theirsadomasochistic acts since commencing treatment, but

162

that the husband had been completely impotent, the wifewas advised to obtain a harmless red dye and add it to anytransparent lubricant. With the use of this mixture, thehusband's potency was restored and the need for actualbloodletting was obviated. The same mixture applied tothe husband's buttocks at the commencement of thepractice was recommended in place of a painful andharmful routine. For reasons unknown, the birth of theirfirst child put an end to their unusual practices altogether.Here again there is tremendous potential for clinical dis-covery.

4. A Case Study Can Provide the Opportunity to Apply Prin-ciples and Notions in Entirely New Ways.The clinical setting affords the opportunity and challengeto develop new procedures based on techniques andprinciples already in use. It is a truism that one will lookin vain for the "textbook case." Clinicians are often facedwith problems for which existing procedures seemunsuitable or insufficient. At the same time, certainaspects of a particular clinical problem may call for a newway of relating old principles and procedures to the reso-lution of the problem. This issue is related to Point 6below, but nevertheless seems worthy of separate illustra-tion here.

In one of our early case reports (Davison, 1966), some"tried and true" procedures were employed in a novelcontext. The use of deep-muscle relaxation has an exten-sive history in medicine, clinical psychology, and psychia-try. The many and varied applications of relaxationprobably share the implicit or explicit purpose of reduc-ing subjective feelings of anxiety. In the case describedbelow, it was possible to use relaxation in a different wayto handle a problem that was hitherto considered unap-proachable by relaxation training. Clinical innovationimplies the discovery that "old" methods can be appliedto new problems, as well as the discovery of new methodsfor overcoming common but seemingly intractable syn-dromes.

The case was that of a middle-aged, male hospitalpatient diagnosed as "paranoid schizophrenic," primarilyon the basis of his complaints about "pressure points" onhis forehead and in other parts of his body. These so-called pressure points were believed by him to be signalsfrom outside forces impelling him toward certain deci-sions. The man had received treatment for 2 months

without any change in these pressure points. In fact, hehad even managed to have the medical staff approve theremoval of a cyst over his right eye in the hope that thismight remove the pressure points. Unfortunately, this hadno effect upon his paranoid delusions. Because of theirtheoretical orientation, the psychiatrists and residents hadbeen restricting their clinical investigations to his pasthistory and, not surprisingly, were finding events in hispast to which they assigned considerable etiological sig-nificance. Nonetheless, the pressure points remainedunabated. The therapist met this man in a Grand WardRound in a psychiatric hospital, during which heinquired of the patient whether he would describe him-self as a "tense" or "anxious" individual. This aspect ofthe clinical picture had been largely ignored by the pres-enting physician. When the patient reported that he wasindeed very anxious, the therapist agreed to attempt ther-apy with him as a demonstration case.

During the first session, the therapist concentrated onclearly delineating those situations in which the manbecame particularly aware of his pressure points. Thepatient was able to identify several such situations thatwere, at the same time, clearly anxiety-provoking. Forexample, being a truck driver, he would often get thepressure points when he was lost and late with a truckloadof goods. He then saw them as helpful in deciding howto reach his destination. This led the therapist to inquirewhether decision-making situations of any importancewere, in general, anxiety-provoking. Indeed they were,and indeed they set the occasions for the most frequentoccurrence of his pressure points.

Having satisfied himself that there was a close relation-ship between anxiety and the pressure points, the thera-pist asked the patient to extend his right arm, clench hisfist, and slowly bend the wrist downwards so as to bringthe closed hand toward the inside of his forearm. Theintent was to produce a feeling of severe muscle tensionin the forearm, and this is precisely what the patientreported. He reported, also, however, that it felt verymuch like a pressure point. The therapist, believing thathe had a good enough relationship with the man to avoidbeing thrust into the patient's delusional system by dis-agreeing with him, suggested an alternative interpretationof the pressure points: perhaps they were simply a conse-quence of his becoming tense and anxious in particularkinds of situations. It was suggested to the man that, in

CLINICAL INNOVATION AND EVALUATION: INTEGRATING PRACTICE WITH INQUIRY • DAVISON AND LAZARUS 16 3

the absence of a naturalistic scientific explanation, he, likeother people, tended to explain strange occurrences insomewhat supernatural or mystical terms. The patientagreed that the merit of the therapist's hypothesis was thatit seemed amenable to an empirical test. The meanswould be to train him in deep muscle relaxation and thento determine whether the relaxation could control theoccurrence of the pressure points. The man consented tothis, and relaxation training was undertaken. Conven-tional relaxation therapy extended over several weeks.Outside of therapy, the man was instructed to pay carefulattention to the occurrence of the pressure points and toconfirm or weaken the assumed connection betweenanxiety, especially in decision-making situations, and theemergence of troublesome pressure points. The mancited enough occurrences to confirm the hypothesis,and as he was becoming more and more proficient inrelaxation, he also reported some degree of control overthe intensity and even the persistence of the pressurepoints by means of differential relaxation. After eightadditional sessions over a 9-week period, the man wasbeginning to refer to the pressure points as "sensations,"and his conversation was generally losing its "paranoidflavor."

What we have here is the application of differentialrelaxation as a means of testing a nonparanoid hypothesisabout bodily sensations. Clearly, there is much more tothe case than can be explained by relaxation principlesalone. For instance, it is likely that new cognitions wereinduced simply via persuasion. Nevertheless, a functionalanalysis of the man's clinical picture led to the hypothesisthat the pressure points were part of a general anxietyreaction to specific kinds of situations. While it is possiblethat the pressure points had complex symbolic meaningsfor the patient, relaxation was effective in controlling thesensations. This helped the patient to account for the sen-sations in terms of a tension reaction rather than as aproduct of external forces. That the man became lessparanoid as therapy proceeded does suggest that the useof differential relaxation in conjunction with what wascalled "cognitive restructuring" was indeed an importantelement in the therapy. Furthermore, having the patientcreate his own pressure points and then apply learnedrelaxation skills to reduce them as a way to alter theirmeaning is similar to an important component of Bar-low's empirically validated therapy for panic disorder. Byspinning in a chair or repeatedly climbing up and down

a step, the patient learns that sensations hitherto interpre-ted as an impending panic attack are actually controllableby relaxation or other coping skills and therefore nothingto fear (e.g., Craske & Barlow, 1993). There is a growingbody of research attesting to the clinical efficacy of Bar-low's treatment.

5. A Case Study Can, Under Certain Circumstances, ProvideEnough Experimenter Control Over a Phenomenon to Furnish"Scientifically Acceptable" Information.Thus far we have at least implicitly accepted the com-monly held view that case reports are intrinsically uncon-trolled. However, one can look to the work of theSkinnerians in both laboratories and clinical settings fordisproofs of this point of view. As has been documentedin many places, one can establish a reliable baseline forthe occurrence of a given behavior in an individual caseand then demonstrate changes that follow the alterationof a particular contingency. Then we may return thebehavior to its original level by changing the contingencyonce again. This is the familiar A-B-A design; numerousand ingenious variations on the basic reversal design havebeen described elsewhere (e.g., Barlow & Hersen, 1984;Hayes & Leonhard, 1991; Kazdin, 1982).

Much of the earliest controlled work in behavior ther-apy was conducted within the Skinnerian framework.Especially noteworthy were the efforts of Wolf, Bijou,and Baer at the University of Washington, and of Ayllonand Azrin (1968) at Anna State Hospital in Illinois in the1960s. To illustrate the work of the former group, wedescribe a case in which it became necessary to reinstatewalking in a 6-year-old autistic child who had regressedto the point where he crawled around on his hands andknees more than 80% of the time. This was achieved byinstructing his teachers to offer him candy and social rein-forcement (attention and praise) intermittently for walk-ing, while completely ignoring him when he wascrawling. Within 2 weeks the child walked normally andseldom crawled. One of the teachers questioned the rele-vance of the reinforcement contingencies and maintainedthat it was merely noncontingent love and approval thataltered the child's behavior. To test this hypothesis, theteachers were again directed to offer "love and approval,"only this time to make it coincide with crawling whileignoring the child when he was walking. In less than aweek the child had reverted to pretreatment levels ofcrawling. Finally, by reversing the contingencies once

CLINICAL PSYCHOLOGY: SCIENCE AND PRACTICE • V1 N2, WINTER 1994

164

more, he stopped crawling and resumed normal walking.The control of the child's crawling by reinforcement con-tingencies constitutes reliable scientific data on the natureof this behavior (Harris, Johnston, Kelley, & Wolf, 1964).(Under many circumstances it would be unethical orimpractical to reverse contingencies in order to reinstateproblem behaviors. There are alternatives in single-subject research, such as multiple baseline designs, thateliminate the need for reversals.) Present-day corrobora-tions and extensions of these paradigms have been pre-sented by Butterfield and Cobb (1994).

6. A Case Study Can Assist in Placing "Meat" on the "Theoreti-cal Skeleton."The reader will recall our earlier suggestion that the theo-retical notions to which clinicians subscribe bearimportantly on the specific decisions they make in a par-ticular case. Clinicians in fact approach their work with agiven set, a framework for ordering the complex data thatare their domain. But frameworks are insufficient. Theclinician, like any other applied scientist, must fill out thetheoretical skeleton. Individual cases present problemsthat always call for knowledge beyond basic psychologi-cal principles.

Illustration of this point can be underscored by refer-ring to desensitization procedures in general and phobicreactions in particular. The general technique of desensi-tization has been detailed quite specifically (e.g., Wolpe,1990). In the management of less simple and straightfor-ward cases, however, the mechanistic sequences may nothold up. In these instances, the "meaty" issues involvedecisions about precisely what idiosyncratic variations toplace on the hierarchy, whether desensitization is evenappropriate to the case, and if so, whether crucial dimen-sions of anxiety have been properly spelled out. Apartfrom relaxation and positive imagery, what other easilyapplicable antianxiety responses can be employed withina desensitization framework? This question became fairlyurgent when it was found that several people who wereresistant to relaxation procedures were also havingdifficulty in conjuring up vivid images (Lazarus & Mayne,1990). In placing further "meat" on the "theoretical skel-eton," various cognitive procedures have been added tothe usual desensitization sequence (Davison & Neale,1994, pp. 556-557). These additions probably would nothave arisen if difficult individual cases had not called forrevisions, refinements, and extensions of existing meth-

ods. Here again it is likely to be the practitioner who iscompelled to amplify theories and techniques in order toaccommodate individual variations that expose deficien-cies in our existing areas of knowledge.

LIMITATIONS OF GROUP DESIGNS AND THE NEEDFOR OBJECTIFIED SINGLE-CASE STUDIESClinicians are usually concerned with particular cases.Since group designs, such as those used in the usual com-parative outcome studies, provide information on aver-ages, therapy researchers have long appreciated theirlimitations in informing the practitioner about how toproceed in the individual case. As alluded to earlier, thisdialectical tension between the nomothetic and the idio-graphic has been a theme in psychology at least sinceGordon Allport's classic writings on personality (e.g., All-port, 1937). The pros and cons of single-case methodol-ogy have been thoroughly analyzed by Hilliard (1993).

There is, however, an important limitation of groupresearch that is seldom if ever discussed. Consider thesimplest of all therapy studies, involving an experimentalgroup and a placebo control group. We have becomeaccustomed over the years to expect some degree ofimprovement in placebo groups, sometimes even to thedegree that within-condition changes are significant.The researcher, of course, hopes that any such improve-ment will be exceeded by positive changes in the experi-mental condition. But consider the following frequentlyencountered situation: Subject A in the experimentalgroup improves significantly, and Subject B in the pla-cebo control group improves to the same degree. Can weattribute the improvement of Subject A to a particularfeature of the experimental condition? Another way toput the question is as follows: Given that Subject Aimproved in the experimental condition, can we say hewould not have improved to the same degree if he hadbeen assigned to the control condition (for Subject Bshowed the same improvement, and it is common to findsome degree of improvement even in placebo condi-tions)? Furthermore, since placebo elements are admit-tedly a part of the experimental condition-hence theinclusion of a placebo control group-can we say withconfidence that Subject A's improvement was not due tothe placebo elements inherent in the experimental condi-tion? We suggest that the answers to these questions is no.

Reports of comparative outcome research at leastimply that improvements in experimental subjects are

CLINICAL INNOVATION AND EVALUATION: INTEGRATING PRACTICE WITH INQUIRY • DAVISON AND LAZARUS 165

due to something particular about that condition vis-a-vis a control group, even though there is always variancein change scores in both groups. But consider this: AsBergin (1966, 1970) alerted us long ago, there is usuallysome deterioration among some subjects in experimentalconditions, even when the group on average improvessignificantly with respect to pretreatment status and morethan control conditions. How frequently do authors attri-bute this worsening to something special about theexperimental condition? Our answer: Never. With theselimitations of group designs in mind, it is important toconsider the legitimacy and importance of a researchmodel that is based not upon variation between patients,but upon different averages or responses within a patient( Hilliard, 1993).

Individual patients may be studied in two ways. First,they may be used as "their own control." In this connec-tion, individual patients are studied more carefully than isusual when group comparisons are under investigation,but the findings can be added to hypotheses that still cen-ter around group norms. Second, in the truly intensiveindividual clinical design, each subject becomes his or herown laboratory, and hypotheses that arise are tested solelywith reference to that particular individual. In the latterinstance, the patient's variability and reaction patternsmay be studied minute to minute, hour to hour, day today, session to session, and so on. Statistical probabilitiescan be computed, and experimental design in its mostrigorous sense can be applied. The patient's behavior canbe described in terms of a multidimensional or multivari-ate probability distribution, and therapeutic progress canthen be assessed in relation to these probability distribu-tions. Symptom frequency and symptom intensity can bewoven into the measures obtained and form part of theoverall evaluation of treatment effects.

Much greater precision in these studies has followedthe use of recordings, films, and videotapes. Since anygiven clinical observer is subject to a personal within-rater variability, this factor seems less likely to distort andinfluence results when cases can be evaluated by differentraters (notwithstanding problems of inter-rater reliabil-ity). Advances in telemetry and other electronicrecording devices have added further impetus to objectiv-i ty and quantitative accuracy.

The general trend in clinical research is in the direc-tion of greater specificity. Broad questions such as "Ispsychotherapy effective?" are now considered meaning-less and have been replaced by the standard scientific

question: What specific treatment is most effective for this indi-vidual with that particular problem working with this therapist ofthat orientation, and under which set of circumstances? (SeePaul, 1967; Strupp and Bergin, 1969). Yet, when aimingfor specificity, the major drawback of extensive statisticaldesigns is, as just shown, the fact that they yield onlygroup norms and probabilities, and do not tell us verymuch about a given individual in the group. Only fine-grained study of individual cases permits us to relate ther-apeutic effect to specific patient characteristics.

When an individual therapeutic effect follows asequence of treatment methods within an appropriatelycontrolled framework, numerous patient-therapist char-acteristics in whose context the effect took place can bespecified. One can thus narrow down the particularpatient and technique variables involved. Strictly speak-ing, specific inferences are valid only with respect to theindividual case itself, but if one relates the particular indi-vidual's most relevant characteristics to similar attributesin other people, general theories can be formulated interms of these common characteristics. One does notfocus upon identical cases (since everyone is unique,there are no completely identical cases), but there areoften sufficient similarities and obvious dissimilarities topermit the evaluation of treatment effects on the basis ofthese related and unrelated features. The basic emphasis isupon the documentation of clinical research, with specialreference to objective ratings and the statistical study ofthe course of a given patient's treatment, in relativelyconcrete and operational terms.

CONCLUSIONBecause of our therapeutic bias, we have emphasizedcognitive-behavioral experiments and clinical trials thatare directly related to treatment and that are intended topartially alleviate emotional suffering in a field where somany people seek help and so few ever really find it.Nevertheless, the field of clinical experimentation asresearch need not be involved with treatment per se orrestricted to any particular theoretical orientation. Studieson intermittent reinforcement, for example, were notconducted with treatment applications in mind, but theresults are certainly relevant to clinical experimentalinvestigations. Many areas of investigation are carried outwith normal subjects. The basic aim of clinical experi-mentation is to be able to predict changes following spe-cific experimental operations. Our own interests favorthose kinds of experiments that endeavor to provide a

CLINICAL PSYCHOLOGY: SCIENCE AND PRACTICE • V1 N2, WINTER 1994

16 6

framework of scientifically based knowledge for the treat-

ment of specific abnormalities of behavior, or that are

intended to explore particular techniques in order to

either examine their validity or improve them and specify

limiting conditions for their application. We believe that

the kinds of continuing interactions detailed in this article

between innovations in the applied arena and controlled

inquiry in research settings represent promising strategies

for enhancing conceptual and procedural knowledge in

what might properly one day become the clinical sci-

ences.

NOTES1. This couple was seen more than 25 years ago, well

before the AIDS crisis. Clearly this kind of bloodletting

sexual ritual would be regarded as more dangerous today

than it was then.

2. We wish to acknowledge that there are problems in

transferring the medical concept of placebo to psycho-

therapy. The medical referent denotes belief and expec-

tancies that patients can bring to a drug treatment

situation and that have been shown to lead to therapeutic

change even when the medication provided is biochemi-

cally inert. There is no analogous situation in the field of

psychotherapy, because patients' ideas and expectations

are an integral part of any psychotherapy. Some have

argued that nonspecificfactors is a preferable term, viz., ".. .

to what extent does a given treatment produce therapeu-

tic changes over and above the changes that would result

from the presence of these nonspecific factors alone (Kaz-

din, 1986, p. 50). But, as Kazdin cautions, this reconcep-

tualization is also fraught with difficulties. Our use of the

term placebo control, therefore, may bother some readers,

but our argument goes beyond the controversies of pla-

cebo conditions in psychotherapy research.

ACKNOWLEDGMENTThis is a revised and updated rendition of a handbook

chapter by Lazarus and Davison (1971).

REFERENCESAllport, G. W (1937). Personality: A psychological interpretation.

New York: Holt, Rinehart, & Winston.

Ayllon, T., & Azrin, N. H. (1968). The token economy: A motiva-tional system for therapy and rehabilitation. New York:

Appleton-Century-Crofts.Barlow, D. H., & Hersen, M. (1984). Single-case experimental

designs: Strategies for studying behavior change (2nd ed.). New

York: Pergamon Press.

Beck, A. T. (1967). Depression: Clinical, experimental, and theoreti-cal aspects. New York: Harper & Row.

Bergin, A. E. (1966). Some implications of psychotherapyresearch for therapeutic practice. Journal of Abnormal Psychol-ogy, 71, 235-246.

Bergin, A. E. (1970). The deterioration effect: A reply to

Braucht. Journal of Abnormal Psychology, 75, 300-302.

Butterfield, W H., & Cobb, N. H. (1994). Cognitive-

behavioral treatment of children and adolescents. In D. K.

Granvold (Ed.), Cognitive and behavioral treatments (pp. 65-

89). Belmont, CA: Brooks/Cole.

Craske, M., & Barlow, D. H. (1993). Panic disorder and agora-

phobia. In D. H. Barlow (Ed.), Clinical handbook ofpsychologi-cal disorders (pp. 1-47). New York: Guilford.

Davison, G. C. (1966). Differential relaxation and cognitive

restructuring in therapy with a "paranoid schizophrenic" or

"paranoid state." Proceedings of the 74th annual convention of theAmerican Psychological Association. Washington, DC: Ameri-

can Psychological Association.Davison, G. C., & Neale, J. M. (1994). Abnormal psychology (6th

ed.). New York: Wiley.

Davison, G. C., Navarre, S. L., & Vogel, R. S. (in press). Artic-ulated thoughts in simulated situations: A think-aloudapproach to cognitive assessment. Current Directions in Psy-chological Science.

Davison, G. C., Robins, C., &Johnson, M. (1983). Articulated

thoughts during simulated situations: A paradigm for study-

ing cognition in emotion and behavior, Cognitive Therapy andResearch, 7, 17-40.

Dryden, W (1991). The essential Arnold Lazarus. London:

Whurr Publishers.

Ellis, A. (1962). Reason and emotion in psychotherapy. New York:

Lyle Stuart.

Franks, C. M., Wilson, G. T., Kendall, P C., & Foreyt, J. P.

(Eds.). (1990). Annual review of behavior therapy: Theory andpractice (Vol. 12). New York: Guilford.

Harris, F. R., Johnston, M. K., Kelley, C. S., & Wolf, M. M.

(1964). Effects of positive reinforcement on regressed crawl-

ing of a nursery school child. Journal of Educational Psychology,55, 35-41.

Hayes, S. C., & Leonhard, C. (1991). The role of the individualcase in clinical science and practice. In M. Hersen, A. E.

Kazdin, & A. S. Bellack (Eds.), The clinical psychology hand-book (2nd ed., pp. 223-238). New York: Pergamon Press.

Hilliard, R. B. (1993). Single-case methodology in psychother-

apy process and outcome research. Journal ofConsulting andClinical Psychology, 61, 373-380.

Kazdin, A. E. (1982). Single-case research designs: Methods for clini-cal and applied settings. New York: Oxford University Press.

Kazdin, A. E. (1986). The evaluation of psychotherapy:

CLINICAL INNOVATION AND EVALUATION: INTEGRATING PRACTICE WITH INQUIRY • DAVISON AND LAZARUS 16 7

Research design and methodology. In S. L. Garfield & A. E.Bergin (Eds.), Handbook of psychotherapy and behavior change(3rd ed.) (pp. 23-68). New York: Wiley.

Kazdin, A. E. (1994). Methodology, design, and evaluation inpsychotherapy research. In A. E. Bergin & S. L. Garfield(Eds.), Handbook of psychotherapy and behavior change (4th ed.,pp. 19-71). New York: Wiley.

Lazarus, A. A. (1989a). The case of George. In D. Wedding &R. J. Corsini (Eds.), Case studies in psychotherapy (pp. 227-238). Itasca, IL: Peacock.

Lazarus, A. A. (1989b). The practice of multimodal therapy. Balti-more, MD: Johns Hopkins University Press.

Lazarus, A. A., & Davison, G. C. (1971). Clinical innovation inresearch and practice. In A. E. Bergin & S. L. Garfield(Eds.), Handbook of psychotherapy and behavior change (pp. 196-213). New York: Wiley.

Lazarus, A. A., & Mayne, T. J. (1990). Relaxation: Some limita-tions, side-effects, and proposed solutions. Psychotherapy,27, 261-266.

Lazarus, A. A., & Messer, S. B. (1991). Does chaos prevail? Anexchange on technical eclecticism and assimilative integra-tion. Journal of Psychotherapy Integration, 1, 143-158.

Levine, F. M., Sandeen, E., & Murphy, C. M. (1992). The ther-apist's dilemma: Using nomothetic information to answeridiographic questions. Psychotherapy, 29, 410-415.

CLINICAL PSYCHOLOGY: SCIENCE AND PRACTICE • VI N2, WINTER 1994

Maher, B. A. (1966). Principles ofpsychopathology: An experimentalapproach. New York: McGraw-Hill.

Paul, G. L. (1967). Strategy of outcome research in psychother-apy. Journal of Consulting Psychology, 31, 109-118.

Spiegler, M. D. & Guevremont, D. C. (1993). Contemporarybehavior therapy (2nd ed.). Belmont, CA: Brooks/Cole.

Strupp, H. H., & Bergin, A. E. (1969). Some empirical andconceptual bases for coordinated research in psychotherapy.InternationalJournal of Psychiatry, 7, 18-90.

Wolpe, J. (1958). Psychotherapy by reciprocal inhibition. Stanford,CA: Stanford University Press.

Wolpe, J. (1990). The practice of behavior therapy. New York: Per-gamon Press.

Wolpe, J., & Lazarus, A. A. (1966). Behavior therapy techniques.New York: Pergamon Press.

Woolfolk, R. L., & Lazarus, A. A. (1979). Between laboratoryand clinic: Paving the two-way street. Cognitive Therapy andResearch, 3, 239-244.

Yin, R. K. (1989). Case study research: Design and methods (2nded.). Newbury Park, CA: Sage.

Received August 23, 1993; revised April 22, 1994; acceptedMay 16, 1994.

168