Interest in the investigation of counseling and psychotherapy has both a history and evolution. It is unfortunate, however, that this common and long-standing concern in the constructive change of behavior and personality has fractionated rather than unified the interested parties. As Rogers (1963) has pointed out,
our differences as therapists do not lie simply in attaching different labels to the same phenomenon. The difference runs deeper. An experience which is seen by one therapist as healing, growth-promoting, helpful, is seen by another as none of these things. And the experience which to the second therapist is seen as possessing these qualities is not so perceived by the first. We differ at the most basic level of our personal experience.
Some people may feel that though we differ regarding specific incidents, . . . nevertheless in our goals and in our general directions there is much agreement and unity. I think not. To me it seems that therapists are equally divergent in these realms [p. 7].
One can best grasp the reasons for this current state of affairs by a more intense look at the evolution of psychotherapy research.
The era prior to the experimental investigation of psychotherapy can best be characterized as “academic tribalism.” The various schools of psychotherapy which existed consisted of more or less loosely organized theoretical formulations based on biased and unsystematic observations which could not be controlled and thus repeated in any reliable manner. Consequently, adherence to any of these theoretical views was based on faith, conviction, and personal satisfaction; and loyalties were maintained and perpetuated by identification with a particular set of esoteric rituals (Bandura & Walters, 1963). Blocker (1967) seems to have captured the essence of this evolutionary process when he states:
Much of the history of change in counseling and psychotherapeutic theory and practice contains elements which closely parallel those which tend to dominate the evolution of religious movements and political ideologies. In both cases a movement tends to be originated by a messianic figure, characterized by a kind of élan vital, who translates a deeply moving personal experience into universalistic terms. This leader quickly attracts a group of worshipful disciples who immediately begin to generalize the precepts promulgated by the master into the most widely applicable terms, [p. 4]
With little more than faith and the sheer force of opinion to back their untested propositions and doctrinaire assertions, it is not surprising that these so-called coteries or “schools” existed as factions, often diametrically opposed to one another in terms of their aims, methods, and goals (Arbuckle, 1967; Bandura & Walters, 1963; Rogers, 1963). However, they exhibited such a faith and hope in the efficacy of their respective techniques that failures in practice engendered only a minimum of what Festinger (1957) has termed “cognitive dissonance.”
Both Phillips (1956) and Blocher (1967) have shown how a number of convenient expressions arose to buttress and thus justify the already well developed and deeply held personal convictions of these “schools” : for example, “the client is unmotivated,” “he lacks ego-strength,” “he doesn’t have enough working-anxiety,” etc. “The convenience of this type of thinking . . . for counselors who readily lose a third to a half of their clients prematurely, is, of course, obvious” (Blocher 1967, p. 14).
The net effect of these newly-coined rationalizations was to effectively screen out any negative feedback by automatically attributing untoward consequences to defects in the client (Ellis, 1962). When coupled with the enormous resistance to extinction generated by relatively few “successes” delivered on an aperiodic schedule, these factors served only to reduce any dissonance which may have resulted from failures in practice and thus eliminate any reason for change or modification in techniques.
Finally, dissonance between “schools” was further avoided by limited contacts between members of the dissonant factions. Even when they did interact, the violent polemics which resulted often generated more heat than light. According to Hoskisson (1965): “they ... get together and wrangle and defame each other and have a wonderful time, . . . much of specialized scientific publication seems to consist in mutual condemnation of each other’s work” (p. 29).
With the introduction of scientific methods into this formerly sacrosanct domain, it was hoped that a common core of empirical knowledge would replace the theories based on tenacity, faith, and intuition (Campbell & Stanley, 1963). This laudable but peremptory wave of optimism soon envanesced, however, when the predicted rapprochement somehow never really materialized.
In a summary of the overall impact of the past 25 years of research in counseling and psychotherapy, Shlien (1966) has pointed out that, “Continued subscription [to an existing school of psychotherapy] is based upon personal conviction, investment, and observation rather than upon general evidence” (p. 125). In a similar vein, Eysenck’s (1952)1 first review of the outcome literature concluded with, “The figures fail to support the hypothesis that [existing forms of] psychotherapy facilitates recovery from neurotic disorders” (p. 323). His more recent reviews (Eysenck, 1955, 1960, 1961, 1965), as well as those of others (Baily, 1956; Bandura, 1964; Levitt, 1957), have led to essentially the same conclusions.2
The factors contributing to this current state of skepticism regarding psychotherapy are many and varied. As seen earlier, for a number of years this predicament was due to a lack of empirical research. However, as a number of authors have pointed out, the more recent causes stem from the fact that the existing empirical evidence is derived from such poorly organized and controlled research that the findings could be used to support almost anything, and thus, nothing (Blocher, 1967; Edwards & Cronbach, 1952; Kiesler, 1966; Paul, 1967; Sprinthall, 1967). In spite of this, researchers and practitioners alike have not tried to maintain a respectful tentativeness commensurate with their real ignorance of the problem. Rather, there is a tendency among partisans of each of the various positions to apply one standard of adequacy to inherently poor research when that research supports their theoretical position and an entirely different standard to the same type research when it is contrary to their position (Hunt, 1956).
As Goldstein et al. (1966) have pointed out, it is not uncommon for clinically minded researchers to disqualify and reject unfavorable results by pointing to methodological and control problems and then to cite at the same time favorably identical studies which support the position they advocate. More specifically, Bandura & Walters (1963) have shown how psychoanalysts frequently reject on the basis of inadequate translation negative findings when the research is based on a translation of psychoanalytic theory into learning terms. Yet the same psychoanalysts embrace positive findings with such enthusiasm that the purely psychic dividends which result unite to compel and further seduce their continued belief and increased entrenchment in psychoanalytic theory. What is the result? The era of “academic tribalism” still remains.
Schools of psychotherapy still exist with little more reliable, empirical foundations than before (Breger & McGaugh, 1965; London, 1964; Rogers, 1963). New schools continue to emerge here and there (Berne, 1961; Ellis, 1962; Eysenck, 1959, 1960,1965; Salter, 1961; Stamphl, 1967; Wolpe, 1958) again as factions radically opposed to the traditional ways of psychotherapy. Freshly minted ideas and glimmerings of understanding become so quickly encapsulated into the dogma of a “school” or coterie that they are seldom subjected to the scrutiny and “natural selection” of experimentation (Blocher, 1967; Breger & McGaugh, 1965; Dittmann, 1966). In fact, every effort is often made to protect and guard what each deems desirable, should it be “mistakenly confused” with the facts of research, for “nonconfirmation of a cherished hypothesis is acutely painful” (Campbell & Stanley, 1963, p. 3). ‘
Similarly, as before, these new schools maintain their autonomy and thus avoid any potential cognitive dissonance simply by building new jargons, creating new journals in which to publish their esoteric jargons (Behavior Research and Therapy, Voices, etc.), and generally divorcing themselves from the rest of the field. Successes by one school are either impugned as palliative, attributed to factors the critic’s school deems relevant, or blatantly disreputed (Rogers, 1963; Strupp, 1962; Wolpe, 1963; Wolpe & Lazarus, 1966). Finally, as earlier,. “Extremists on . . . [all] sides have not hesitated [in their expenditure of polemic words and ink] to discredit each other, even though well-controlled comparative studies are nonexistent” (Paul, 1966, p. 1).
Although collapse into a shapeless solipsism and feckless relativism, which are the death of science, is not eminent or pressing, this joint-catharsis-against-a-common-foe makes for little more than personal harmony and satisfaction within each faction; it contributes little in the way of constructive, cumulative knowledge (Campbell & Stanley, 1963). It is true that these precociously inspired theories often replace one another from time to time; however, since this displacement is typically not the result of well controlled, comparative research, the product rarely augments existing knowledge or serves as a sequel to what has gone before (Blocher, 1967). In short, it adds to the history of psychotherapy but not to its evolution.
The overall effect of this condition is forceably demonstrated in Colby’s (1964) analysis of the current predicament in psychotherapy—”Chaos prevails” (p. 347). Similarly, Rogers’s (1963) statement graphically portrays the net results of this current state of affairs—”The field of psychotherapy is a mess” (p. 8). Goldstein et al. (1966) feel that a “skein of confusion . . . represents our current level of understanding of the field of psychotherapy” (p. 10). Finally, London (1964) has conceded that “a detailed examination of the surfeit of schools and theories, of practices and practitioners that compete with each other conceptually and economically, shows vagaries which, taken all at once, make unclear what it is that psychotherapists do, or to whom, or why” (p. v).
As indicated above, a good portion of the confusion regarding psychotherapy is due to the fact that most existing schools of psychotherapy are based as much, if not more, on faith and dogma as on comparatively derived research findings. By worshiping their flimsy hypotheses into truths and then selecting “research” to bolster their already well developed personal convictions, these schools become implacable and categorically indestructable, that is, immune to dissonant empirical findings (Matarazzo, 1965). Evolution, if it can be called that, takes the form of “a fadish discard of old wisdom in favor of inferior novelties” (Campbell & Stanley, 1963, p. 2).
The foregoing analysis, although admittedly one-sided, is not intended to belie the importance of the immense problems and complexities which impede research in psychotherapy or allow, at best, for the most tenuous of controls
(Rubinstein & Parloff, 1959; Shlien, 1968; Strupp & Luborski, 1962). Indeed, much of the defection from experimentation to essay writing may be based on this very fact. As intimated above, a substantial portion of these difficulties can be attributed to very real and formidable methodological and control problems (Frank, 1959; Paul, 1967) and to the lack of a universal model for research in counseling and psychotherapy (Kiesler, 1966; Paul, 1966).
If counseling and psychotherapy are ever to mature beyond the level of “an undefined technique applied to unspecified problems with an unpredictable outcome” (Raimy, 1950, p. 93), then experimenters must abandon research models which perpetuate distinctions between existing schools and adopt those models which seek to define their techniques, circumscribe their limits of applicability, and demonstrate their efficacy in these delimited contexts (Gendlin, 1967; Gilbert, 1952; Kiesler, 1966; Sanford, 1953). In spite of the formidable obstacles created by methodological and control problems, and in the face of admonitions to the contrary (Hyman&Berger, 1965; Kiesler, 1966;Strupp, 1962), anum-ber of investigators feel that there is an adequate paradigm for research in counseling and psychotherapy which follows the foregoing prescription (Blocher, 1967; Edwards & Cronbach, 1952; Frank, 1959; Paul, 1967; Sprinthall, 1967). Such a design would consider simultaneously the following: input variables; process variables; outcome variables; follow-up variables. Let us consider each of these areas separately.
Input variables concern the ingredients that go into or make up the treatment process, that is, client variables and therapist variables. The former deals with the client’s presenting problem or distressing behavior as well as with the more stable personal-social characteristics that constitute his life style. The physical-social environment or the client’s life space completes this triad. Therapist variables incorporate primarily the techniques the therapist uses in treatment and the more stable personal-social characteristics that constitute his life style or personality.
CLIENT VARIABLES. In terms of research methodology, client variables can be reduced to the dual problem of sample selection and treatment focus. Borrowing his strategy from research on the effects of drugs, Frank (1959) has suggested a most promising approach to this problem. It involves the selection and description of clients in terms of what he calls “target behaviors.” In other words, clients who are going to be used as Ss for research are selected on the basis of common or similar distressing behaviors. The description or operational definition of this common presenting problem then becomes the dependent variable in the research design. Furthermore, with this approach the goal of treatment then becomes behavior change in a specified direction. The efficacy of this approach has been successfully demonstrated by Paul (1966) in his attempts to reduce interpersonal-performance anxiety in college freshmen.
According to Paul (1966), therapy research should “compare specific techniques in the treatment of an emotional problem that is delimited enough to allow rigorous experimental methodology, but significant enough to allow generalization from the findings and to have implications for further study in the broader field of counseling and psychotherapy” (p. 9). It is my belief that interpersonal anxiety is just such a problem.
Anxiety, generally, is a central explanatory concept in almost all contemporary theories of personality and is regarded as a principal causative agent for such diverse behavioral consequences as insomnia, immoral and sinful acts, instances of creative self-expression, debilitating psychological and psychosomatic symptoms, and idiosyncratic mannerisms of endless variety (Spielberger, 1966). It has been observed to have untoward consequences on motor behavior and coordination (Luria, 1932; Malmo, 1966), cognitive effectiveness (Saltz & Di Loreto, 1965; Spielberger, 1966), and physiological functioning (Grinker, 1966; Malmo & Shagass, 1949). Even more important, anxiety reduction is a cogent aspect of most, if not all, existing theories of psychopathology and is seen by most practitioners as a salient therapeutic goal.
Interpersonal anxiety, or anxiety that results from simple, routine interactions with other individuals, is particularly debilitating since it can disrupt normal, but required, daily routines, make otherwise healthy individuals unhappy and dissatisfied with themselves, and generally create discomforts that are unfortunate and otherwise avoidable. Consequently, the area of anxiety generally, and interpersonal anxiety in particular, seems fertile ground for research from both a theoretical and practical point of view.
In terms of a target focus for research, small interacting treatment groups provide a prototype stress situation for eliciting interpersonal anxiety, and thus they can be used as an objective criteria for assessing by means of behavioral ratings change in this target behavior from pre- to post-treatment. The importance of this type of criteria has been exhorted by Strupp (1962) : “if psychotherapy is effective, the benefits must be somehow demonstrable in the person’s behavior” (p. 457). In addition, since interpersonal anxiety undoubtedly restricts interpersonal activity, a periodic sample of interpersonal contacts can also serve as an external criteria for assessing both progress during the course of treatment, and outcome. The importance of using behavioral criteria external to the treatment setting has been noted by a number of researchers (Luborsky & Strupp, 1962; Paul, 1967; Zax & Klein, 1960).
If one views the most important test of the effectiveness of a particular therapeutic treatment as involving first a change in the client’s distressing behaviors and second a change in these behaviors outside of the treatment setting, then these two criteria, in conjunction with client self-reports, go a long way toward achieving these goals.
Even with selection on a homogeneous class of target behaviors, however, there is likely to be wide variation in the relatively stable personal-social characteristics of the client (Paul, 1967). However, if in addition to matching target behaviors, clients are further selected, described, and classified on the basis of these relatively stable life style characteristics, pre-treatment variability among clients can be reduced further. The importance of this type of research for purposes of assessing treatment effects unambiguously has been noted by many investigators (Blocher, 1967; Kiesler, 1966; Levinson, 1962; Paul, 1967; Sprinthall, 1967; Underwood, 1957). However, research of this nature is conspicuous by its absence.
Personality type, although an admittedly illusive and somewhat arbitrary choice in terms of the available empirical evidence on personal-social characteristics, looms as a salient prognostic construct in the minds of many experienced researchers and clinicians (Blocher, 1967; Gelder & Wolff, 1967; Kiesler, 1966; Lang et al., 1965; Lazarus, 1963; Sargent, 1961; Sprinthall, 1967). As Luborsky (1962) has noted, “Of all the influences which are thought to determine the change a patient can make through treatment, the patient’s personality is most often thought to be predominant” (p. 123).
There are at least two reasons for this belief. First and foremost is the very real fact that the founders of nearly all existing forms of psychotherapy derived their theoretical formulations and based their techniques on observations of radically different types of clients (Kiesler, 1966). Second, the evidence which does exist, while not entirely satisfactory in terms of either scope or consistency, suggests that counseling and psychotherapy, as traditionally conceived, are processes which have restricted applicability for only selected subgroups of the population (Blocher, 1967). Considered together, these two factors suggest that perhaps different forms of treatment are needed for differing combinations of personality type and presenting problem (target behaviors).
There is a growing body of extant evidence which suggests that therapists do in fact behave quite differently with different types of clients (Dittmann, 1966; Matarazzo, 1965; Strupp, 1962). However, the relationship between this and treatment outcome is unclear. In one study (Truax & Carhuff, 1965), it was found that “therapist transparency” was positively related to self-exploration in both hospitalized neurotics and delinquent adolescents. However, whereas self-exploration was positively related to client improvement among the neurotic group it was inversely related to positive personality change in the delinquent population.
It is this author’s belief that the dimension of introversion-extroversion has many attributes which warrant further interest and investigation. First, a number of investigators have identified these two continuous, bipolar qualities as accounting for the majority of variance on nearly all self-report personality questionnaires (Cattell, 1966; Eysenck, 1960; Guilford, 1959). Thus, we have here a major dimension of personality which is relatively easy to assess, has high empirical validity, and is relatively independent of any one, idiosyncratic operational definition. Personality attributes with such high construct validity are a rarity in this field and command further inquiry on that basis alone. Second, introverts and extroverts have been found to differ on a number of qualities which may be predictive of differential treatment outcome.
Extrapolating from Pavlov’s (1957) findings, Eysenck (1961) and Wolpe (1958, 1966) hypothesized that individuals differ with regard to conditionability (the speed and firmness with which conditioned responses are built-up or learned) and autonomic activity ( activity which affects the strength and intensity of a response once acquired). Eysenck (1961) further maintains that these qualities are related to the excitatory and inhibitory potential of an individual’s central nervous system, and that the latter two factors can serve as a basis for classifying individuals along one of two dimensions—normal-neuroticism or introversion-extroversion.
Mowrer’s (1966) concept of socialization further elaborates the relationship between nervous system functioning and the development of introvert-extrovert types. Since socialization is seen as dependent on conditioning, individuals with strong inhibitory potential and weak excitatory potential would be expected to form weak and unstable conditioned responses that extinguish easily. Such individuals would be weakly socialized. Those in whom the excitatory potential is strong and the inhibitory potential weak would be expected to form strong and stable conditioned responses that extinguish slowly. They would be strongly socialized. According to Eysenck (1961), “The former group would thus tend to develop introverted behavior traits . . . while the latter group would tend to develop extroverted behavior traits” (p. 27). If the foregoing analysis is correct, it would seem that a form of treatment based on counterconditioning would be more appropriate for introverts and one based on extinction more appropriate for extroverts.
Wolpin & Raines (1966) and Singer (1968) have suggested that introverts may have better visual imagery than extroverts, as they live in fantasy more and thus have more practice with visual imagery. Since he is cut off from his relationships with others (Hoskisson, 1965) an introvert’s imaginary or fantasy life may, ostensibly at least, constitute reality for him. As Franks (1961) has implied, the introvert is a “thinking,” “planning,” or “ideational” type whereas the extrovert is a “doing” or “acting” type.
Wolpin & Raines (1966) and Eysenck & Rachman (1965) have also noted that introverts are more dependent, suggestible, and conforming. In fact, Wolpin & Raines (1966) have found that “scores on the E scale [of the Maudsley Personality Inventory] may predict more generally willingness to comply or conform” (p. 35). In addition, Eysenck & Rachman (1965) and Eysenck (1967) have suggested that introverts may prefer a more mechanical and impersonal form of treatment.
Consistent with Franks’s (1961) description presented above, Eysenck & Rachman (1965) have characterized the extrovert as “sociable . . . needs to have people to talk to and does not like reading or studying by himself . . . he prefers to keep moving and doing things” (p. 19). He is “outgoing,” “talkative,” and “carefree” (p. 16). Experimental studies of introversion-extroversion (Eysenck & Rachman, 1965) indicate that extroverts, as compared to introverts, have a higher IQ-vocabulary ratio and higher sociability. Furthermore, unlike the introvert, the extrovert is cut off from himself and his feelings (Hoskisson, 1965) since, as Fenichel (1941) has pointed out, acting or doing often serves to circumvent awareness of one’s feelings or motives.
All of these considerations, when taken together, suggest a form of treatment for introverts which is based on a counter-conditioning model, is somewhat impersonal and mechanical in nature, and makes maximum use of the introvert’s suggestibility, dependency, and conformity as well as his excellent visual imagery. By the same token, the aforementioned also suggests a form of treatment for the extroverts that is based on an extinction model, is predominantly verbal in nature, and requires involvement in an intimate and very personal group interaction that emphasizes looking at and focusing on one’s “inner self” (i.e., thoughts, feelings, and fantasies).
In concluding this section on client variables, it appears that the selection, description, and classification of clients on the basis of pre-treatment individual differences provides a more adequate definition of the S sample, thus assuring both a greater replication of the sample across studies and a clearer focus for the assessment of outcome. In addition, utilization of a homogeneous problem area within personality type allows for more efficient assignment of Ss and thus greater equivalence of treatment and control groups, since Ss are classified along two dimensions simultaneously. All in all, it appears that the radical reduction in client pre-treatment variability permitted by these two operational distinctions undoubtedly will lead to more easily interpretable findings and more readily replicable results. Also, researchers should be able to better explain “inconsistent” or “contradictory” findings across studies.
THERAPIST VARIABLES. In discussing therapist variables, one is interested in both the therapist’s approach (both his philosophical orientation to treatment and his therapeutic techniques) and the more stable personal-social characteristics which constitute his life style or personality.
There are at least three ways in which one can treat the therapist variable (the effects of his personality vs. the affects of his approach). The first approach, which exists predominantly in fantasy, involves using each therapist as his own control. The advocates of this approach (Paul, 1966, 1967) aver that by having each therapist administer each of the treatments, one can hold the personal-social attributes of the therapist constant across groups, reducing placebo effects which may masquerade as treatment.
Not only is this goal impossible to achieve, but the strategy on which it is based is internally inconsistent. Having therapists objectively indicate their degree of commitment to certain techniques, as Paul (1966) has done, for example, excludes by definition an equal commitment to alternative approaches. The therapist’s personality already has entered into his choice and commitment. Once this is done, finding “therapists who are open-minded enough to learn to use contradictory methods without exhibiting attitudes that would greatly affect their approach” (Paul, 1966, p. 6) becomes an axiomatic impossibility.
Even if one were to grant the potential plausibility of the above approach, it does nothing to circumvent the original problem but rather creates a truly insoluable dilemma. If a therapist uses this approach and fails to achieve its supposed goal, then he is maximizing the differential influence of placebo effects in the direction of the techniques which he preferred originally. If one, in fact, succeeded in finding such “open-minded” therapists, they would most assuredly not be representative of other therapists of the same ilk. As Arbuckle (1967) has so appropriately pointed out, “Differences in counselors automatically become differences in counseling” (p. 224).
Perhaps the clearest statement regarding the inseparable nature of therapist and therapy variables has been offered by Frank (1959) who states :
It is obvious that the therapist and therapy variables cannot be completely separated. It is unlikely that a therapist can conduct different types of treatment with precisely equal skill or that his attitudes towards them will be identical. Therefore, differences in results obtained by two forms of therapy conducted by the same therapist may be due to therapist rather than treatment variables, especially since the faith of a therapist in a form of treatment may account for much of its efficacy (7). In our psychotherapy study the psychiatists disliked minimal treatment. They gave it reluctantly and felt that they were shortchanging the patients. The patients remained just as long in this type of treatment as in the other two, suggesting that they were not as lacking in confidence in it as the doctors, [p. 17]
An alternative and even more absurd approach is to attempt to get counselors to effectively role play confidence in techniques they regularly do not use or in which they do not believe (Snyder, 1962). This method, if effective, simply reintroduces any placebo effects in haphazard and uncontrollable amounts, mitigating the entire purpose of this circuitous approach. If ineffective, one has once again introduced placebo effects systematically. The real problem here, as above, is that one never really knows when and if one has or has not failed. In addition, in view of Rogers’s (1963) and Frank’s (1959) observations, it becomes difficult to imagine a Client-Centered therapist, for example, doing Rational Therapy or trying to role play confidence in such an approach, or vice versa! This is to say nothing of the inequity created by attempting to train three- to five-year veterans of Client-Centered therapy to do Rational Therapy in a week or less (Strupp, 1967, 1968).
Both of the above strategies lend an air of artificiality to the research treatments which is not present in the clinical treatment settings in which these techniques usually are administered. This latter fact serves only to reduce further the external validity or generality of the findings. Clearly, both of these approaches commit the error of misplaced precision.
A partial solution to the problem of therapist variables can be achieved by assuming, as Arbuckle (1967) does, that certain therapists choose certain techniques because they are certain kinds of people, that is, that the therapist’s personality and his treatment techniques are integrally and inseparably linked. Then, by securing therapists who are committed to techniques which one wishes to compare, by having each administer the techniques they respectively deem effective, and by comparing what they say they do with what they actually do (by means of audio or video tape), one is in a much better position to assess treatment conditions as they are most often administered with little or no loss in scientific rigor. Contrary to the beliefs of some (Paul, 1966, 1967), there is no reason to assume that this approach produces any greater variability in the administration of the treatments than the aforementioned strategies.
Using the above approach, one finds that the more therapists there are representing any one treatment, the more the results can be attributed to the effects of the treatment approach per se and not to any one therapist or his unique personal-social makeup. This can be achieved either by replication within and across studies or, in larger facilities, by a sampling of therapists within a “school” or treatment approach. The latter is similar to sampling IQ, introversion, anxiety, etc., from a more general population. Viewed in this vein, the absurdity of using each therapist as his own control becomes even clearer, since it would be tantamount to having each S serve as both introvert and extrovert or both high and low anxious, etc., within the same study.
Another distinct advantage of the treatment approach proposed above is that one can check on whether or not the assumptions of this model are met, a notable disadvantage of the previously discussed strategies. In addition to comparing what therapists say they do with what they actually do by means of tape recordings, objective assessment of the therapist’s personal-social characteristics can be made and similarities and differences noted. When client variables are spelled out in the manner described earlier, the data on therapist personal-social characteristics can be compared with the data on client personal-social characteristics, since, in many cases, assessment can be made by many of the same measuring devices. This is especially true of such personal-social characteristics as personality type, IQ, socio-economic status, age, sex, etc.
To the remaining dilatories, this does not mean that outcome results are simply a measure of therapist personal commitment. One can be committed strongly to walking to the moon, even though the efficacy of this technique (walking) can be proven virtually useless. This third approach attempts to hold personal commitment as constant as possible. Then, by using uniform outcome criteria for all treatment and control groups, the results can be interpreted as commitment to either effective or ineffective treatment, whatever the case may be.
It should be remembered that Paul’s (1966) conclusions, strictly speaking, are relevant only to insight-oriented therapists practicing systematic desensitization; and, at last count, there don’t seem to be too many of these individuals around.3 This, of course, assumes that one accepts Paul’s (1966) initial distinction between therapist personal-social characteristics and treatment techniques. If one discards this initial provision, then one is left with the conclusion that personal-social attributes are largely irrelevant to the issue of treatment techniques. The approach offered above, however, allows one to conclude that personal commitment is either too effective or ineffective counseling, whatever the results support.
In concluding this section on input variables, it becomes clear that selecting, describing, and classifying both clients and therapists on the basis of pre-treatment individual differences make for better controlled, more easily interpretable, and thus more legitimately generalizable research. The importance of defining clearly the variables within these two broad domains has been exhorted by Garfield & Affleck (1961) who maintain that the time to begin outcome studies is prior to intake. Similarly, Strupp (1962) has noted the potential value of this area when he states, “Research might make an important contribution by refining the selection of particular patients for particular therapists and for particular therapeutic methods” (p. 471).
In this manner, one may begin to bury the myths which, according to Kiesler (1966), have retarded progress in both the research and practice of counseling and psychotherapy, namely, the assumption that clients, therapists, and treatments are homogeneous entities. In so doing, one will simultaneously give birth to the area of “individual differences” within the fields of counseling and psychotherapy, an area which gave to psychology, generally, some of its earliest and most important discoveries (Sprinthall, 1967).
Process research, or the study of client-therapist interactions, owes its existence, reputation, and current fame to two historical events. The first of these has to do with the traditional role ascribed to insight and the second with the so-called criterion problem.
THE INSIGHT MODEL. Until relatively recently, it was felt that client self-exploration leading to insight was the modus operandi of successful treatment (Coons, 1957). Successful outcomes were believed to follow correct insight as surely as night followed day. Thus, the “natural” research focus was on the kinds of client-therapist interactions leading to “movement,” “improvement,” or “insight” during the course of treatment. Positive outcomes and generalization to “real life” settings could be expected to follow inevitably and thus need not be focused on directly. However, recent reviews of this area have demonstrated clearly that not only is this not the only way successful treatment comes about (Blocher, 1967; Matarazzo, 1965; Sprinthall, 1967), but that this entire model, in fact, may be in error (Hobbs, 1962; Paul, 1966; Szasz, 1961).
In view of the above, it seems clear that treating process rather than outcome as the end product and assessing the effectiveness of one or more treatment regimes by so-called process research is precarious business, to say the least. With the uncertainty of success following insight, it becomes immediately apparent that studies of the process of psychotherapy depend, for their validity, on the assumption that counseling and psychotherapy work (Frank, 1967; Greenhouse, 1964). However, in view of Eysenck’s (1952, 1961, 1965) and Shlien’s (1966) conclusions, the tenability of this assumption seems in grave doubt. If, as noted earlier, counseling and psychotherapy, as traditionally practiced, do work, they most certainly have restricted applicability for only a highly select subgroup of clients (Blocher, 1967; Sprinthall, 1967).
The problems of process research are even further exacerbated by the fact that the very nature of the research focus (i.e., client-therapist interaction) precludes the use of adequate control groups. Certainly, the use of a no-treatment control group does little or nothing to improve the design of process research. Similarly, the use of a placebo control provides for only the barest minimum beyond which the findings are un-interpretable. Utilizing a comparative approach, where a number of treatments are administered and thus a number of different types of client-therapist interactions are compared, does circumvent this problem somewhat. However, even with this approach, the absence of a control group makes it difficult to determine any cause-effect relationships. Finally, without at least outcome, if not both pre- and post-test measures, the results are of dubious value since one is, in effect, assuming at least a uniform and more often a positive outcome.
In regard to outcome, Goldstein et al. (1966) have offered very timely advice :
Research efforts should be directed toward the study of outcome(s) of psychotherapy. . . . Only after we have been able to demonstrate that we can consistently produce a particular change in behavior as a result of a particular manipulation does it seem advisable to expend effort in studying the “process” involved in the manipulation. We suggest that research be done on the changing of behavior, [p. 10]
Similarly, Edgar (1966) has made what seems an empirically reasonable demand when he states that, “ ‘Good’ counseling must be judged in relation to ‘good’ outcomes” (p. 1029).
THE CRITERION PROBLEM. A second and certainly not unrelated problem is the fact that interest in process variables has been a product of defection from interest in outcome. This “flight into process” (p. 127) as Zubin (1964) has termed it, was due largely to the so-called criterion problem. The problem of deciding what criteria to employ in evaluating outcomes owes much of its existence and is mute testimony to the fact that schools of psychotherapy still exist. In fact, the so-called criterion problem has virtually no operational definition outside the realm of “school” distinctions.4
It is interesting to note that although there are at least as many potential areas of focus in process as in outcome research, and thus as many, if not more, potential “criterion problems,” process research has never been strapped with similar problems of definition. This, it is believed, is due to the fact that process research was initiated and is being conducted currently by predominantly one “school” of psychotherapy, namely, the Client-Centered group; and, since their beliefs about the nature of the therapeutic process are relatively homogeneous, at least more so than between schools, the “criterion problem” never openly appeared in this area. But it is most assuredly there and cannot be used either as an excuse for avoiding outcome research or as a criticism of it.
As both Frank (1959) and Paul (1966, 1967, 1967a) have demonstrated, the selection, description, and classification of clients on the basis of relatively homogeneous pre-treatment variables, as described earlier, goes a long way in circumventing the problem of criteria selection. By selecting clients on common target behaviors, for example, one can begin to separate the criteria of successful treatment from school affiliation and in so doing construct a model for research which is more sensitive to client individual differences and less sensitive to the theoretical predilections of individual therapists. With this approach, one is in a much better position to assess the effects of counseling and psychotherapy in light of the only laudable criteria—whether or not the client gets better (Battle et al., 1966; Betz, 1962; Hoppock, 1953; Rickard, 1965).
The above approach does not require the abandonment of process research but rather involves a recasting of both process and outcome research into a unified model. Instead of treating the client-therapist interaction as the goal or focus of research, process variables could be incorporated into outcome research in such a manner that they aid in the assessment and understanding of how change occurs, during the course of treatment, in the pre-selected behaviors one is attempting to modify. This approach calls for the fusion of outcome and process variables within a single model and involves measuring sequential outcomes, as the client moves toward the ultimate goal of reduction in target behaviors and the kinds of client-therapist interactions which actually occur during each treatment session (Strupp, 1968).
By employing process data to assess how far along the client has progressed toward or away from a predefined or expected goal, one is in a much better position to offer plausible explanations rather than wild speculations regarding the outcome results (Gelfand & Hartmann, 1968; Reyna, 1964). As Bandura (1968) has pointed out, “By confining analyses solely to outcomes, investigators could readily affirm all sorts of causal conditions with immunity” (p. 247). Finally, the sequential measurement of change in pre-selected target behaviors, both within and outside of the treatment setting, by either self-report or behavioral observation, will not only allow one to observe how behavioral change proceeds but will undoubtedly make future directional predictions more probable and research more controllable.
Bandura (1968) has captured the essential features of this paradigm when he states :
In summary, one might reiterate the basic requirement that any study designed to elucidate change processes should include adequately measured outcomes that are systematically linked to their presumed controlling variables. When causal linkages are loose and ill-defined, spirited disputes flourish but little headway is made in delineating the conditions controlling behavioral changes and the mechanisms through which the effects are produced, [p. 249]
THE COMPARATIVE APPROACH. In addition to the aforementioned, there is a third and related problem which has hampered the evaluation of counseling and psychotherapy. This problem centers around the use of a research model that compares one treated group with one untreated or (no contact) control group. It seems of little or no value, in terms of adding to the general fund of knowledge concerning the efficacy of counseling and psychotherapy, to employ a simple treatment no-treatment model (Edwards & Cronbach, 1952;Paul, 1967). To demonstrate the effectiveness of a particular treatment by comparing it with no treatment at all, at least in psychotherapy research, is in effect stacking the cards in one’s own favor (Bergin, 1963; Blocher, 1967; Kiesler, 1966). All forms of intervention, from faith healing to aspirin to psychoanalysis, will undoubtedly produce some desirable effects, as will simply the passage of time (Eysenck, 1952).
As the above implies, and as a number of researchers have noted (Blocher, 1967; Kiesler, 1966; Paul, 1967), the simple treatment no-treatment model asks a meaningless question, that is, “Does psychotherapy work?” Furthermore, in terms of research methodology, this paradigm provides for only the barest minimum beyond which the research findings become uninterpretable (Campbell & Stanley, 1963; Goldstein et al., 1966; Underwood, 1957). Finally, since all forms of intervention undoubtedly “work,” this model, like the simple process model discussed earlier, serves only to perpetuate “school” distinctions by maximizing the occurrence of “positive” outcomes.
Research which purports to demonstrate the efficacy of a particular therapeutic approach must be based on a multivariate comparative model which either compares the outcomes of different techniques with the same problem and/or compares different problems with the same techniques, again on the basis of outcome and follow-up. Although an attention-placebo group may serve as an alternative treatment (Paul, 1967), considerably more information is gained by employing two or more “traditional” treatment approaches, in addition to a placebo and no-contact control group.
The comparative approach allows for the following : (a) a test of the “ relative effectiveness of each of the treatment techniques; (b) aids in pinpointing which approach (es) works best with which problem(s) and which type(s) of client; and (c) provision for the elimination of the maximum number of rival, yet plausible, explanations of the obtained results, which, of course, is the goal of empirical research (Campbell & Stanley, 1963). With regard to this latter asset, the comparative approach is potentially without equal. Moreover, it also has the potential of minimizing or even ultimately eliminating the “schools” approach, replacing it with a systematic eclectic approach founded on a common core of empirical knowledge derived from the commonalities and uniquenesses of each of the various competing approaches now in existence.
Grummon (1965) has captured the importance of this approach when he states that:
If we are going to rest our case about the effectiveness of counseling on whether or not it promotes this or that desirable behavior, we must also consider whether some other procedure might not produce the result more economically and to an even greater degree. We need much more of this kind of information before we can make intelligent value judgements about the usefulness of particular kinds of counseling, [pp. 63-64]
Grummon’s reasoning is especially important in an area such as counseling and psychotherapy where all approaches to treatment implicitly or explicitly claim to be effective with all problems. As London (1964) has pointed out:
Now if this plentitude of treatments involved much variety of techniques to apply to different persons under different circumstances by different specialists, there would be no embarrassment of therapeutic riches here, just as there is not within the many specialties of medicine or law or engineering. But this is not the case, .... One hardly goes to a psychoanalyst to be cured of anxiety and a nondirective therapist to be treated for homosexuality, as he might to a cardiologist for one condition and a radiologist for another. Nor does the same doctor use Freudian therapy for psychogenic ulcers and Rogerian treatment for functional headaches, as a physician might use medicine for one ailment and surgery for another, [p. 30]
The decision to utilize these three particular treatment approaches was based on a number of considerations. First and foremost is the fact that all three of these treatment approaches have dealt with the problem of interpersonal anxiety theoretically (Bandura & Walters, 1963; Ellis, 1962; Grum-mon, 1965; Rogers, 1955; Wolpe, 1958), and each claims to have treated it successfully with the techniques they respectively advocate (Bartlett, 1949; Dymond 1954; Ellis, 1962; Grummon, 1965; Grummon & John, 1954; Lazarus, 1960, 1963; Lazarus & Rachman, 1957; Rogers, 1954; Truax & Carkhuff, 1967; Wolpe, 1958; Wolpe, Salter & Reyna, 1964).
Second, there is reason to believe that these three treatment approaches may be differentially effective with the dependent variable of interpersonal anxiety. For example, Desensitization has been found to be quite effective with similar target behaviors (Land et al, 1965; Lazarus, 1963, 1966; Paul, 1966, 1967a). Similarly, if Ellis’s (1957, 1962) observations are correct, Rational-Emotive treatment may be superior to the more classical insight approaches such as Psychoanalytic and Client-Centered therapy. Although these approaches have never been compared in a single study, these preliminary observations suggest a tentative order of effectiveness as follows: Systematic Desensitization, a Rational-Emotive and Client-Centered treatment.
Furthermore, the characteristics of introverts and extroverts, presented earlier, may well make them differentially receptive to these three forms of treatment, thus producing different outcomes. For example, if the introvert was legitimately characterized as one who would profit most from a form of treatment based on counter-conditioning and one that makes maximum use of his suggestibility, dependency, conformity, and excellent visual imagery, then Desensitization is probably the most appropriate and Client-Centered the least appropriate form of treatment for him. Similarly, if the extrovert was legitimately characterized as one who would profit most from a form of treatment which is based on extinction but is predominantly verbal in nature and requires involvement in an intimate and personal group interaction which emphasizes self-exploration, then Client-Centered treatment seems the most appropriate and Desensitization the least appropriate form of treatment for him.
Finally, these three approaches represent radically different forms of therapy in terms of technique, treatment model, and level of organization addressed during treatment (affective, behavioral, cognitive). By maximizing the treatment differences, in this manner, one is in a much better position to control and check on the ongoing in-treatment procedures and thus make more definitive cause-effect conclusions, while capturing the essential features of most contemporary systems of psychotherapy currently in existence.
In summary, it is this author’s view that although “schools” were the inevitable consequence of theories based on tenacity, faith, and intuition, they have continued to exist largely because the bulk of contemporary research is of a process and treatment no-treatment nature and thus addresses itself to meaningless questions. Not only do these approaches yield very little useful information, they also tend to maintain and perpetuate school distinctions by allowing therapists to transfer their loyalties and affections from a set of highly esoteric rituals to a highly selective body of “empirical findings” without ever once taking a closer look at their underlying personal convictions (Hunt, 1956) to see whether these commitments are justified in terms of either the available evidence or competing approaches. It is this pseudo-scientific base which makes these schools currently so implacable and unyielding.
In order to overcome this predicament, it is believed that research in counseling and psychotherapy must move in two directions. First, it must move from research which emphasizes process to research which emphasizes outcome, with the inclusion of relevant process measures of client-therapist interactions and measures of sequential changes in the target behaviors under study to enhance understanding of the final results. Second, movement must occur from outcome studies which emphasize the use of a treatment no-treatment model to studies which emphasize research based on a multivariate comparative model.
With the inclusion of both process and outcome measures, in conjunction with assessment on relevant input variables, one can begin to ask the more appropriate question, “What form of treatment, administered by whom, is most effective for this particular client with that specific problem and under what set of circumstances?” (Blocher, 1967; Paul, 1967). Not until one begins to employ research designs which address themselves to this question will one be able to circumscribe the limits of applicability of various techniques and demonstrate their efficacy in these delimited contexts and thus be rid of the myths of which Kiesler (1966) speaks.
The need for follow-up data on clients receiving counseling and psychotherapy is of the utmost importance. In an area where placebo effects confound with and frequently masquerade as treatment, only the passage of time will reveal whether clients have truly changed or are simply placating their therapists by responding in a socially desirable manner. Although this problem is less severe when objective, behavioral criteria are employed, there are always immediate, short-term benefits which accrue from treatment but which never really become incorporated into the client’s permanent behavioral repetoire.
The other side of this issue is equally as important; namely, that what is learned or acquired as a consequence of treatment takes time to become effectively assimilated and implemented into action. The period immediately following treatment is undoubtedly the most crucial in terms of which behaviors will be discarded and which incorporated into the client’s way of life. Thus, without at least a brief follow-up period, treatment approaches may well get branded as either effective or ineffective on the basis of spurious information alone. This is to say nothing of the importance of checking on such currently controversial phenomenon as spontaneous remission, symptom substitution, or complete recitivism.
As Sargent (1960) has pointed out, however, “the importance of follow-up is equaled only by the magnitude of the methodological problems it presents” (p. 101). The most important of these problems is the very practical difficulty of sample maintenance and attrition (Paul, 1967). Since the rate of dropout seems to be directly related to the amount of time between post-testing and follow-up, and since dropout rates tend to be differential, selectively biasing the follow-up findings, tactical decisions favor the use of a short-term follow-up, since it improves the probability of total sample assessment. As Paul (1966, 1967a) has shown, when an adequate design with proper controls is evaluated by means of the same criteria, a six-week follow-up is as good as a two-year one.
Follow-up studies also suffer from “the uncontrollable nature of client experiences during the post-treatment period” (Paul, 1967, p. 333). This predicament may be eased somewhat by keeping the follow-up period brief and checking on such blatantly confounding factors as Ss receiving, additional treatment. Without some knowledge of whether or not Ss receive additional post-therapy treatment, cause-effect statements may be invalidly drawn.
Thus, the need for follow-up studies is obvious. The practical problems which surround such an effort, however, favor the use of short-term follow-ups which employ identical criterion measures and make every effort to secure data on crucial client behaviors during this post-treatment period.
The present study was organized in an attempt to compare the outcome of three distinct forms of counseling with two distinct personality types in the treatment of a specific, homogeneous problem. More succinctly, it is the purpose of this study to compare the relative effectiveness of Systematic Desensitization, Rational-Emotive, and Client-Centered group psychotherapy in the reduction of interpersonal anxiety in introverts and extroverts.
1For an explanation of these findings in terms of the therapist variable, see Truax & Carkhuff (1967). For an explanation in terms of pre-treatment individual differences in clients, see Blocher (1967) and Sprinthall (1967). For a general critique of the validity of Eysenck’s interpretations, see Kiesler (1966).
2Cross (1964) surveyed the literature since Eysenck’s 1952 review and found nine studies which used control groups. However, he felt they were so deficient in other respects that the findings still could not be interpreted unambiguously. More recent reviews (Dittman, 1966; Patterson, 1966) have led to essentially the same conclusions.
3Furthermore, no attempt was made to check on the nature or quality of the insight treatment as it was administered (Strupp, 1967). Consequently, his conclusions must be restricted to therapists who say they are insight oriented, and not to insight-oriented therapy per se.
4Although Strupp (1963) would have us believe that this problem is due solely to our lack of a theory of “normal” behavior, it seems he is saying the same thing since each “school” has an implicit or explicit theory of normal behavior and they are often in severe conflict with one another as Rogers (1963) has indicated.