Psychotherapy Volume 28/Fall 1991/Number 3

COUNTERTRANSFERENCE AND THE REPUTEDLYEXCELLENT THERAPIST

STEVEN L. VAN WAGONER CHARLES J. GELSOJEFFREY A. HAYES ROBERTA A. DIEMER

University of Maryland

The extent to which therapists whowere reputed to be excellent differ fromthe therapists in general is examined,with respect to five qualities theorizedto be important in the management ofcountertransference feelings—self-insight, integration, empathy, anxietymanagement, and conceptualizingability. A sample of experiencedpsychotherapists (n = 122) evaluatedtherapists they selected as excellentsignificantly more positively thantherapists in general on all five of thetheorized qualities. Contrary toexpectation, reputedly excellentpsychodynamic, learning, andhumanistic therapists were generallyrated as equivalent on these theorizedqualities. Reputedly excellentpsychodynamic therapists, however,were rated more favorably thanexcellent humanistic therapists onconceptualizing ability.

The concept of countertransference has been amatter of great attention, debate, and revisionsince the concept was first used by Freud (Peabody& Gelso, 1982; Robbins & Jolkovski, 1987). ForFreud and other classical psychoanalytic theorists,

The authors wish to thank Drs. James Mahalik, Judith Marx,and Terence Tracey for their helpful comments on an earlierdraft of this manuscript.

Correspondence regarding this article should be addressedto Charles J. Gelso, Department of Psychology, Universityof Maryland, College Park, MD 20742.

countertransference includes only those reactionsto the client that are triggered by the analyst'sown unconscious needs and neurotic conflicts,and are viewed as inappropriate and potentiallydamaging to the therapeutic endeavor (Arlow,1985; Reich, 1951, 1960). A more contemporarydefinition posits countertransference as all of thetherapist's feelings and attitudes toward the client(Fromm-Reichmann, 1950; Heimann, 1950;Giovacchini, 1975; Kernberg, 1975). The mostpertinent criticism of this "totalist" definition isthat if countertransference is to include both un-conscious, conflict-based feelings toward the client,as well as reality-based feelings, it becomes overlyinclusive and too broad to be useful (Gelso &Carter, 1985; Watkins, 1985). While these defi-nitional disputes have yet to be resolved, a commonelement of all of the definitions is that the therapist'sinternal reactions need to be attended to, under-stood, and in one way or another, managed. Animperative for the therapist, therefore, is to bringthese reactions into awareness, examine them,and use them in the service of the work, ratherthan permitting them to impede effective treatment(Fromm-Reichmann, 1950; Gelso & Carter, 1985;Heimann, 1950; Reich, 1960).

While there is generally a dearth of empiricalwork on countertransference itself, a few studieshave examined ways in which therapists behavecountertransferentially with their clients. Coun-tertransference behavior in the form of therapistwithdrawal, antagonism, or hostility has beenshown to emerge in the presence of client materialthat is in an area of unresolved conflict for therapists(Cutler, 1958), client transference (Luborsky &Singer, 1974; Mueller, 1969), and intensely neg-ative client emotion (Beery, 1970; Gamsky &Farwell, 1966; Haccoun & Lavigueur, 1979). Suchbehavior has been shown to contribute to unsuc-cessful counseling outcomes (Singer & Luborsky,1977). While these studies demonstrate the po-

411

S. L. Van Wagoner et al.

tentially negative impact the client can have ontherapist behavior, not all therapists respondcountertransferentially to client material, trans-ference, or emotion. Might some therapists beespecially good at managing their countertrans-ference? Do therapists who manage countertrans-ference well possess special qualities that helpthem do this, or is their ability simply a functionof their general therapeutic skill?

We wondered whether a therapist who is con-sidered excellent by colleagues would differ fromthe average therapist in the management of coun-tertransference and, if so, in what ways. Doesthe reputedly excellent therapist possess qualitiesthat allow him or her to moderate countertrans-ference feelings so they do not contaminate theprocess? Do reputedly excellent therapists differfrom therapists in general in terms of these qualitiesor characteristics? Additionally, do reputedly ex-cellent male and female therapists differ on thesecharacteristics? Finally, do reputedly excellenttherapists whose orientations are primarily psy-chodynamic, humanistic, and behavioral in factdiffer from one another, in terms of qualities thatmoderate countertransference?

The present study sought to answer the abovequestions. In order to do so, however, we firstneeded to develop conceptualization of the factorsthat we thought mediate how a therapist managescountertransference feelings. Based on a reviewof the theoretical and empirical literature, wetheorized the existence of five interrelated factors:(a) self-insight, (b) empathic ability, (c) self-in-tegration, (d) anxiety management, and (e) con-ceptualizing ability.

Therapist self-insight refers to the extent to whicha therapist is aware of one's own feelings andunderstands their basis. For Freud (1910/1959),the analyst's ability to help the analysand is impededby one's own resistances and unconscious conflicts.The analyst must, therefore, recognize counter-transference, and then begin a process of analyzingthe basis for these reactions. Robbins and Jolkovski(1987) found that awareness of countertransferencefeelings was inversely related to withdrawal ofcounselor involvement with the client, a measureof countertransference behavior. They concludedthat therapists who are aware of their feelings arein a better position to do something about thembefore they are manifested behaviorally. Thesefindings suggest that therapist insight into coun-tertransference feelings allows one to remain en-gaged effectively in the therapeutic interaction.

Empathic ability, our second ingredient, consistsof both affective empathy, that is, the ability toput oneself into another's shoes and temporarilypartake of that one's feelings; and diagnostic em-pathy, the intellectual understanding of the other'sexperience. Empathy, in psychoanalytic theory,consists of a partial or trial identification with theclient's emotional experience, and this partialidentification serves as an avenue for understandingthe client's inner world (Arlow, 1985; Greenson,1960; Reich, 1951; 1960). Countertransferenceoccurs, however, when the therapist is unable toextricate him or herself from this identification.A study by Peabody and Gelso (1982) providespartial empirical support for this conception. Theyfound empathic ability to be positively related toawareness of countertransference feelings, whichin turn, was inversely related to countertransferencebehavior. In addition, Peabody and Gelso (1982)found a trend toward a negative relationship be-tween empathic ability and countertransferencebehavior. These findings suggest that empathicability might be a moderator of countertransferencein the sense that the empathic therapist is typicallymore aware of one's own countertransferencefeelings, putting one in a better position to dosomething about them, before they interfere withthe therapeutic endeavor.

A third factor theorized to be important to themanagement of countertransference, self-integra-tion, broadly refers to the therapist's psychologicalhealth and possession of a stable identity, andincludes the ability to differentiate oneself fromothers, as well as the ability to put aside one'sown needs in the service of the client. Cutler(1958) found that therapists were more likely torespond in an ego-oriented, countertransferentialmanner to client material that was in an area ofunresolved conflict for the therapist, suggestingthat the more psychologically sound a therapistis, the less likely countertransference will impedethe work. Therapist's must possess the ability topartake of the client's emotional experience, butalso the capacity to stand back from this trialidentification and observe it from an objective,analytic stance (Greenson, 1960; Reich, 1960).Greenson (1960) emphasizes that the critical re-quirement is to "partake of the quality and notthe degree of the feelings, the kind, and not thequantity" (p. 418). Self-integration is closely relatedto this empathic process. Greenson (1960) discussesthe "pathology of empathy" as a condition whereempathy leads to countertransference in therapists

412

Countertransference and the Reputedly Excellent Therapist

who are experiencing either a "temporary neuroticdisturbance within [themselves]" or "in those whohave a chronically precarious mental equilibrium"(p. 419). These countertransferences can eitherinhibit empathy on the part of the therapist, orresult in the "loss of control of empathy" wherethe therapist is unable to differentiate his or herreactions from those of the client. For the therapistto engage in this process of merging and separatingwithout the interference of countertransferencereactions appears to require a stable sense of selfwith firm ego boundaries (Gorkin, 1987). Thisalso requires that the therapist can put aside one'sown needs in the service of the client. The needfor approval (Bandura, Lipsher & Miller, 1960)and the need to nurture (Mills & Abeles, 1965)have been shown to moderate therapist's coun-tertransference reactions. Therapists possessingthese needs were more likely to display counter-transference behavior toward their clients, sug-gesting that counselors who had yet to resolvethese areas of personal conflict for themselveswere less likely to manage their countertransferencereactions productively.

The anxiety management factor refers to theextent to which the therapist is anxious in general(trait), as well as the tendency to experience anxietyin the therapy setting (state). That therapist anxietyoccurs in the presence of strong client emotion,and that it is related to countertransference behavior,has been suggested by several studies (Beery,1970; Gamsky & Farwell, 1966; Hayes & Gelso,1990; Haccoun & Lavigueur, 1979; Russell &Snyder, 1963; Yulis & Kiesler, 1968). Freud (1926/1959) thought that the ego will be called upon todefend itself against the danger signaled by thepresence of anxiety, and findings from the abovestudies highlight some of the countertransferencebehaviors therapists employ to manage their ownanxiety, such as withdrawing from personal in-volvement with the client, avoiding negative clientemotion, or responding to negative emotion in ahostile or antagonistic manner. The better abletherapists are in managing their anxiety adaptivelyin the presence of clients' strong emotion, theless likely they will manifest countertransferencebehavior.

The last factor theorized by us to be importantto countertransference management is concep-tualizing ability, which refers to the extent towhich the therapist can conceptualize the clientdynamics in terms of the therapeutic relationshipwith the client, as well as in the context of the

client's past. Application of a theoretical frameworkto case conceptualization is considered a part ofthis factor. While the ability to conceptualize clientmaterial and behavior well is invaluable to thetherapist in terms of guiding the work, it has alsobeen shown to be a useful moderator of counter-transference feelings toward clients. Robbins andJolkovski (1987) found that reliance on a theoreticalframework interacted with awareness of counter-transference feelings to moderate therapist in-volvement, such that under conditions of lowawareness of feelings, theoretical understandingresulted in emotional dis'incing from the client.When theoretical understanding combined witheven moderate levels of awareness of one's feel-ings, however, use of a theoretical frameworkfacilitated involvement. The implication of Robbinsand Jolkovski's finding is that therapists who relyon theory in the presence of their own heightenedawareness of countertransference feelings mightbe cognitively interpreting their feelings in a waythat helps them remain therapeutically involvedwith the client. This process has been describedas critically necessary (Reich, 1960) and potentiallybeneficial (Fromm-Reichmann, 1950; Gelso &Carter, 1985; Heimann, 1950) to the therapeuticendeavor.

As stated, our primary purpose was to determineif reputedly excellent therapists differed fromtherapists in general on these five ingredientstheorized to be important to the management ofcountertransference feelings. A second generalpurpose was to assess if the particular theory oftherapy relied upon by the reputedly excellenttherapist relates significantly to qualities theorizedto be important in countertransference manage-ment. Specifically, although psychodynamic, hu-manistic, and behavioral therapies are essentiallyequal in their global effectiveness (Lambert,Shapiro & Bergin, 1986), would therapists adheringto those orientations actually differ with respectto the possession of qualities thought to moderatecountertransference?

Method

Participants

Two groups were combined to form the presentsample. The first group consisted of 93 participantssampled randomly from the membership of Di-vision 29 (Psychotherapy) of the American Psy-chological Association. The second group consisted

413

S. L. Van Wagoner et al.

of 29 doctoral level professionals in the Departmentof Psychology, Department Counseling and Per-sonnel Services, and the University CounselingCenter of a large mid-Atlantic university. Thisgroup contained only subjects who had practiced,supervised, and/or empirically investigated personalcounseling or psychotherapy within the last twoyears. The university sample was included so thatpersonal follow-ups of their responses could bepursued for a later study. Nonsignificant differencesbetween the university sample and the Division29 sample on the dependent variables under in-vestigation permitted combining both samples inthe subsequent analyses.

Of the 122 subjects, 78 (63.9%) were femaleand 44 (36.1%) were male, with a mean age of47.9 years. Clinical psychologists accounted for64.8% of the sample, followed by counseling psy-chologists (23%), and other (11.4%)(e.g., psy-chiatrists, social workers). Ninety-six percent ofthe sample held a doctorate. The mean postgraduateexperience conducting psychotherapy was 15.4years (SD = 9.4), ranging from 1 to 51 years.On the average, participants spent slightly morethan half of their professional activity conductingpsychotherapy (mean = 51.8%, SD = 32.1), withthe rest of their time devoted to teaching (mean =7.9%, SD = 13.5), supervision (mean = 9.4%,SD = 9.0), empirical study (mean = 4.3%, SD =8.4), and scholarly writing pertaining to counselingor psychotherapy (mean = 3.8%, SD = 8.1).Other professional activities accounted for a littleless than one fourth of the subjects' professionaltime (mean = 23.4%, SD = 26.3).

We asked participants to rank order their the-oretical orientation in terms of their belief in, andadherence to, Psychodynamic (e.g., Freudian,Sullivanian, Ego Psychology, Self Psychology,etc.), Humanistic (e.g., Rogerian, Existential,Gestalt, etc.), or Learning (e.g., Behavioral,Cognitive—Behavioral, Social Learning, etc.)perspectives. Subjects were not to tie any rankings.Of the total sample, 50% ranked psychodynamictheory as their primary theoretical orientation,followed by 23% learning theory, and 22% hu-manistic. The remaining 5% either tied their pri-mary ranking or did not rank their orientation.

Instruments

Countertransference Factors Inventory (CFI).The CFI is a 50-item, Likert type inventory de-signed for the present study to measure the extentto which therapists possess certain interrelatedqualities or characteristics theorized to be important

in the management of countertransference. Thesecharacteristics were conceptualized in terms offive rationally derived therapist factors (self-insight(e.g., "usually comprehends how his/her feelingsinfluence him/her in the therapy", "understandsthe background factors in his/her life that haveshaped his/her personality"); self-integration (e.g.,"effectively distinguishes between clients' needsand his/her own needs", "is psychologically bal-anced") empathy (e.g., "is perceptive in his/herunderstanding of clients", "intuitively understandsthe clients"); anxiety management (e.g., "is com-fortable in the presence of strong feelings fromothers", "does not experience a great deal of anxietywhile conducting therapy"); and conceptualizingability (e.g., "often conceptualizes his/her role inwhat transpires in the counseling relationship","conceptualizes relationship dynamics in terms ofthe client's past"). The CFI was subjected to aseries of reviews and revisions by the authors,resulting in the fifty items selected for the finaldraft. Care was taken to ensure that each subscalecontained a representative number of items sam-pling the domain of characteristics thought to beimportant in countertransference management. Thefive factors made up the subscales of the inventory.The number of items per subscale broke down asfollows: insight, integration, and empathy (11 itemseach), anxiety management (8 items), and con-ceptualizing ability (9 items). It should be reiteratedthat the "factors" of the CFI were rationally derivedrather than empirically derived. They were basedon the authors' theoretical deductions, which inturn, were derived from the empirical and theo-retical literature, as well as clinical experience.Content and face validity data, along with internalconsistency reliability, are presented below.

Eleven experts in the areas of transference andcountertransference were then selected to assessthe content and face validity of the items. Weselected these experts based on their extensiveknowledge, scholarly writing, and/or empiricalinvestigation in the areas of transference andcountertransference. Eight of the eleven expertswere male. Experts had mean age of 54.3 years(range 35-74), and as a group, averaged 22.9years experience conducting postgraduate psy-chotherapy and 7 scholarly publications in thearea of transference and countertransference. Onthe average they spent 35.3% of their professionaltime conducting psychotherapy.

The CFI was mailed to the experts who wereasked to rate the importance of each item as itrelated to the management of countertransference.

414

Countertransference and the Reputedly Excellent Therapist

The expert form of the CFI was in a 5-choice,Likert-type format, with alternatives ranging from"Not Important" to "Very Important." In the di-rections, experts were told that each item repre-sented a characteristic "that a therapist might po-tentially possess that may or may not be criticalin the management of his or her own counter-transference reactions," and "assuming that alltherapists experience countertransference feelingswith their clients at one time or another," wereasked to rate each characteristic as to how importantthey thought they were in the management ofcountertransference (1 = Not Important, 2 =Slightly Important, 3 = Somewhat Important, 4 =Important, 5 = Very Important). Return rate forthe experts was 100% with one follow-up lettera month after the initial mailing.

Measures of central tendency were computedon the experts' rating of the CFI. All but 6 ofthe 50 items had a mean, median, and/or modeof 4 ("Important") or greater. Of the remainingsix items, all had a mean, median, and modeof 3 ("Somewhat Important") or greater. Thesesix items were distributed across all five sub-scales. Mean item scores and standard deviations,computed for each of the five subscales, wereas follows: insight (mean = 4.09, SD — .45),empathy (mean = 3.89, SD = .59), concep-tualizing ability (mean = 4.02, SD = .52), anx-iety management (mean = 3.92, SD = .64),and self-integration (mean = 4.05, SD = .50).The mean item score for the total CFI was 4.01(SD = .52). The mean item scores within eachof the subscales and for the total scale suggestthat the experts as a group found that the itemsreflect characteristics important to countertrans-ference management, and that they accuratelyrepresented the domain for each of the five salientfactors.

In addition to the CFI for experts, three additionalforms of the CFI, varying only in the characteristicsof the therapist that participants were asked torate, were developed for subjects. Two of theseforms asked subjects to "think of a normal dis-tribution of therapists, with [their] concept oftherapist-in-general as encompassing the middleof that distribution." They were then to identifyeither a male or a female therapist with whoseclinical work they were familiar, and who they"would consider to be in the excellent range ofthis distribution," and then rate that person on theCFI. Ratings were on a 5-point Likert scale interms of the extent to which participants agreedthat the rated therapist possessed the given char-

acteristic (1 = strongly disagree, 5 = stronglyagree). The university participants were asked torate either an excellent male or excellent femaletherapist. Two thirds of the Division 29 samplewas also asked to rate either an excellent male orexcellent female therapist. The remaining thirdof the Division 29 participants, however, wasasked to "imagine the "therapist-in-general" con-ducting therapy, and rate him/her" on the CFI.(The Cm for subjects differed from the CFI forexperts only with respect to the Likert type anchorsand stems of the items, otherwise, the items wereidentical).

To assess relability in terms of internal con-sistency, Cronbach's alpha was computed for thetotal CFI and each of the five subscales on theresponses of the 122 subjects. For the total CFI,alpha was .97. For the subscales, alphas were:insight (.91), empathy (.92), conceptualizing ability(.88), anxiety (.91), and self-integration (.91).

Before concluding our description of the CFI,two additional points need to be clarified. First,although we use the term "countertransferencefactors" in the inventory's title, it must be keptin mind that the CFI is not a direct measure ofcountertransference. Rather, it is a measure of anindividual or group's status on five interrelatedand rationally-derived factors theorized to be im-portant (by the authors and expert judges) in themanagement of countertransference reactions.Might the five factors of the CFI be important totherapist operations other than countertransferencemanagement, e.g., overall therapist functioning?Undoubtedly the factors of integration, insight,anxiety management, empathy, and conceptual-izing ability are related to a host of therapist be-haviors. Yet the term countertransference factorsis justified because the CFI is based on a theoreticalformulation about countertransference, and thecontent and face validity data described abovepertain directly to countertransference manage-ment.

The second point needing clarification pertainsto the expected and actual interrelatedness of theCFI's five subscales. To begin with, in our theo-rizing of the five factors, we fully expected arelationship among them, with each factor generallycontributing to the other in clinical practice, e.g.,anxiety management, insight, and empathy arepart of, contribute to, and are affected by self-integration. At the same time, we expected eachfactor to have sufficient non-overlapping varianceto justify it as a separate construct. In fact, thedata do support this conception of interrelatedness

415

S. L. Van Wagoner et al.

and separateness. The correlation coefficients forthe largest subsample, those rating the excellenttherapist (n = 91), were as follows for each pairof subscales: Conceptualizing Ability (CA) withIntegration (Int) = .58; CA with Insight (Ins) =.54; CA with Anxiety Management (AM) = .31;CA with Empathy (EM) = .60; Int with Ins =.68; Int with AM = .71; Int with EM = .65; Inswith AM = .55; Ins with EM = .69; AM withEM = .51.

ProcedureFollowing several revisions of the CFI, subject

packets were prepared for mailing, each containinga cover letter, demographic questionnaire, the CFI,a stamped, self-addressed envelope for returns, apostcard that could be used for requesting theresults of the study, and a sealed envelope con-taining a brief explanation of the study. Participantswere asked to first complete the CFI followed bythe demographic questionnaire. Directions for theCFI were given on the first page of the inventory.Subjects were asked to rate on the CFI either: (a)their concept of a therapist in general; (b) a par-ticular male therapist considered by the subjectto be excellent in the field; or (c) a particularfemale therapist considered to be excellent in thefield. Therapists from the Division 29 sampleswere randomly assigned to one of the above threeconditions, while therapists from the universitysample were randomly assigned to only the ex-cellent male and excellent female conditions.Subjects who rated excellent therapists were in-structed to select only a therapist with whom theyhave worked or observed. At the beginning ofthe CFI, they were to rank that therapist accordingwhat they perceived the therapist's theoretical ori-entation to be. They were asked not to tie anyrankings. In addition, they were to write the ther-apist's name on a postcard provided for referencewhile rating that person on the CFI, so that raterswould not inadvertently substitute other therapiststhroughout the task, and as a result, increase thelikelihood of providing a more generic rating oftheir image of the excellent therapist. By askingsubjects to visualize the actual therapist and imaginehim or her conducting therapy, we hoped to removesome of the abstraction and tendency for subjectsto fall into a perceptual "set" of some generictherapist. Following the rating they were askedto destroy the postcard with the therapist's nameon it to ensure the anonymity of the person beingrated, place the demographic questionnaire and

the CFI into the provided stamped, self-addressedenvelope, and mail the packet back to us. Theywere then invited to open the provided envelopemarked "Debrief," containing a description of thestudy.

Two weeks after the initial mailing, a postcardreminder was mailed to nonrespondents, and fourweeks following the initial mailing we sent anothercomplete packet of materials to nonrespondents.Several steps were taken to ensure the maximumpossible return rate. Cover letters were carefullyworded expressing the value we placed upon par-ticipants' time and input, and assuring that theanonymity of the participants was preserved, aswell as the anonymity of the person they rated.The researchers personally signed all cover letters.In addition, study results were made readily avail-able to participants. Finally, the "Debrief" wasprovided as an added incentive for completing thesurvey.

One participant returned the materials completed,but had indicated that he did not conduct psy-chotherapy. Of the remaining 189 participantssurveyed, 122 returned the completed materials,resulting in a 65 percent rate of return.

DesignThe overall design of the study was a levels-

by-treatment, randomized blocks design. Therewere two primary independent variables. The mainindependent variable of the study, to which par-ticipants were randomly assigned, was degree oftherapist excellence. This was divided into threelevels: (a) excellent male therapist, (b) excellentfemale therapist, and (c) therapist in general. Thesecond primary independent variable was the the-oretical orientation of the excellent therapists se-lected for rating by the subjects, and this variablealso contained three levels: (a) psychodynamic,(b) learning, and (c) humanistic. Gender of theparticipant was also analyzed to control for anypossible systematic error due to this variable. Be-cause theoretical orientation applied only to theexcellent male and female therapists and not thetherapist in general, the design was not completelycrossed.

Results

Differences between the Samples

Because participants rating excellent therapistscame from two sources, Hotelling T2 tests forindependent samples were performed comparing

416

Countertransference and the Reputedly Excellent Therapist

the Division 29 and university samples' ratingsof excellent therapists on the five subscales of theCFI. The Division 29 raters did not approachdiffering from the university sample on either theirratings of excellent male therapists (T2 =.17, p = .34) or excellent female therapists (T2 =.11, p = .46). As a result, the two samples werecombined for all subsequent analyses.

Gender, Excellence, and CountertransferenceManagement

Mean item scores, standard deviations, and asummary of post hoc univariate F tests and Scheffet tests for the total CFI and the five subscales arepresented in Table 1 based on the ratings of theexcellent male and excellent female therapists,and the therapists in general. A two-way MANOVAwas performed for participant's gender (male ratervs. female rater) by excellence (excellent malevs. excellent female vs. therapist in general) onthe five subscales of the CFI. A significant andvery large multivariate effect was found for ther-apist excellence using Wilk's criterion (F [2, 116]= 9.77, p < .001). No significant effects forparticipant gender or the participant gender byexcellence interaction were detected. Follow-upunivariate F tests revealed that all of the subscalessignificantly contributed to the significant MAN-OVA effect (see Table 1). Scheffe's Multiple RangeTest, the most conservative of the univariateposthoc procedures (Haase & Ellis, 1987), wasemployed to examine pairwise comparisons among

the means of the three levels of the excellencefactor on the total scores of the CFI, as well asthe subscale scores, to determine where the dif-ferences between the three levels of excellenceexisted. The Scheffe procedure with the criticalvalue set at p = .01 was used to minimize thebuild-up of the Type I error rate that occurs whenmultiple dependent variables are examined (Haase& Ellis, 1987). Post hoc comparisons revealedsignificant differences between ratings of excellentmale therapists compared to therapists in general,and ratings of excellent female therapists comparedto therapist in general; however, no differencesexisted between excellent male and excellent femaleratings. Examination of the mean item scores inTable 1 highlights the direction of the differencesbetween the groups. Clearly the excellent maleand excellent female therapists were rated sig-nificantly higher on all of the subscales of theCFI, as well as the total CFI as revealed by atwo-way ANOVA (F [1, 116] = 49.33, p <.001), in full support of our first hypothesis. Therewas no significant main effect for participant gen-der, nor was there a significant interaction effectbetween gender and excellence.

Theoretical Orientation andCountertransference Management

Mean item scores, standard deviations, and asummary of post hoc ANOVAs and Scheffe /Tests for the total CFI and the five subscales arereported in Table 2 for the three levels of theoretical

TABLE 1. Ratings of Countertransference Factors According to Therapist Excellence. Means, StandardDeviations, and Post Hoc Univariate F and Scheffe t Tests of Excellence Differences

Countertransference Factors

Conceptualizing AbilityIntegrationInsightAnxietyEmpathyTotal CFI

ExcellentMale

(n =

M

4.294.304.084.324.254.24

40)

SD

.39

.40

.48

.46

.43

.43

Therapist Excellence

ExcellentFemale

(« =

M

4.384.244.114.084.384.24

= 51)

SD

.46

.47

.51

.66

.42

.50

Therapist inGeneral(n =

M

3.523.393.203.273.473.37

31)

SD

.53

.58

.54

.64

.59

.57

F"(2, 116)

37.42*37.28*35.68*28.49*38.93*49.33*

GroupDifferencesb

M,F > TIGM,F > TIGM,F > TIGM,F > TIGM,F > TIGM,F > TIG

'F is based on post hoc ANOVAs for all of the CFI subscales.bGroup differences are based on post hoc Scheffe t Tests. M = Male; F = Female; and TIG = Therapist

in General.*p< .001.

417

S. L. Van Wagoner et al.

TABLE 2. Ratings of Countertransference Factors According to Theoretical Orientation. Mean Item Scores,Standard Deviations, and Post Hoc Univariate F and Scheffe / Tests of Theoretical Differences

Countertransference Factors

Conceptualizing AbilityIntegrationInsightAnxietyEmpathyTotal CFI

Theoretical '

Psychodynamk(« =

M

4.454.304.184.144.374.29

= 60)

SD

.39

.45

.53

.61

.45

.49

Orientation

: Learning(» =

M

4.194.213.914.144.214.13

12)

SD

.43

.45

.42

.55

.41

.45

Humanistic(» =

M

4.094.193.974.334.254.16

17)

SD

.46

.44

.41

.51

.34

.43

F'(2, 116)

6.02**.51

2.31.72

1.001.37

GroupDifferences'1

P > H

'F is based on post hoc ANOVAs for all CFI subscales.bGroup differences are based on post hoc Scheffe t Tests. P = Psychodynamic; and H = Humanistic.* p < .05** p < .01

orientation. Of the excellent therapists who wereselected by participants, 60 were perceived ashaving a psychodynamic orientation, 12 as learn-ing, and 17 as humanistic. To examine the secondmajor hypothesis, that psychodynamic and hu-manistic therapists would be viewed more posi-tively on the five salient factors than the learningtherapists, a one-way MANOVA was performedon the subscales of the CFI, with the independentvariable, theoretical orientation, consisting of 3levels (psychodynamic vs. learning vs. humanistic).MANOVA revealed a significant multivariate ef-fect, F [2, 116] = 2.11, p' < .05). Univariate Ftests revealed significant differences between thegroups on the conceptualization subscale. Sig-nificant differences existed between ratings ofpsychodynamic therapists and humanistic therapistsas revealed by Scheffe's post hoc test (p = .01).Examination of the means in Table 2 indicatesthat psychodynamic therapists were rated morepositively on conceptualizing ability than human-istic therapists. A one-way ANOVA on the totalCFI score, however, revealed no significant dif-ferences between the theoretical orientations onan overall measurement of countertransferencemanagement.

DiscussionExcellent male and female therapists, when

compared to therapists in general, were viewedas having more insight into their feelings and thebasis for those feelings; as having a greater capacity

for empathy in the sense of being able to partakeof the client's emotional experience, as well ashaving an intellectual understanding of clientemotions; as being more highly integrated, andtherefore, more able to differentiate client needsfrom their own needs; as possessing less anxietyin general and with clients in the session; and asbeing more adept at conceptualizing client dy-namics, in both the context of the therapeuticrelationship, and the client's past.

While some studies have suggested that coun-tertransference is inevitable, and that therapistsrespond to clients in predictable ways (Beery,1970; Cutler, 1958; Gamsky & Farwell, 1966;Haccoun & Lavigueur, 1979; Luborsky & Singer,1974; Mueller, 1969), these findings taken at facevalue might be misleading. Excellent therapists,by virtue of their possession of more characteristicsimportant to the management of countertransfer-ence than therapists in general, might defy thenorm in that they are less likely to respond toclient material, transference, or intense emotionin the predictable ways suggested by the research.Moreover, some studies support the notion thattherapists do differ in their ability to managecountertransference. Therapists possessing a strongneed for approval (Bandura et al., 1960) and needto nurture (Mills & Abeles, 1965), low or ex-cessively high levels of empathy (Peabody & Gelso,1982), low level of awareness of countertrans-ference feelings (Robbins & Jolkovski, 1979),and heightened anxiety (Beery, 1970; Gamsky &

418

Countertransferen.ee and the Reputedly Excellent Therapist

Farwell, 1966; Haccoun & Lavigueur, 1979; Rus-sell & Snyder, 1963; Yulis & Kiesler, 1968),have been shown to exhibit greater amounts ofcountertransference behavior. Similarly, findingsfrom the present study combined with the aboveresearch suggest that the generally excellent ther-apist, being more highly integrated, more empathic,and better able to manage anxiety than the typicaltherapist, in addition to being more insightful abouthis or her own feelings and conflicts, and moreskilled in conceptualizing client dynamics, wouldbe less likely to manifest countertransference be-havior, thus maximizing his or her capacity forresponding to clients constructively. Given an un-derstanding of how excellent therapists differ fromthe norm with respect to the above characteristics,therapists in training, as well as experienced ther-apists, might strive to model what excellent ther-apists do in therapy, and not assume that the normis what everybody does or even should do.

Another implication from the findings is thatit may be wise for therapists to seek therapy forthemselves if they truly are to minimize the impactof countertransference on their work. Such personaltherapy would appear to make sense, especiallysince the five ingredients we have theorized asimportant in countertransference management maywell be affected by personal therapy. For example,Strupp (1973) found that experienced therapistswho had undergone personal therapy were moreresponsive and empathic to the needs of theirclients. We would likewise expect that factorssuch as self-insight, anxiety management, andself-integration to be positively affected by personaltherapy.

Although excellent therapists from the threetheoretical orientations did appear to differ overallon the five subscales of the CFI (see the MANOVAresults), subsequent univariate analysis revealedthat the only significant effect by theoretical ori-entation was between the excellent psychodynamictherapists and the excellent humanistic therapistson the subscale, Conceptualizing Ability. Excellentpsychodynamic therapists received higher ratingson the subscale than excellent humanistic therapists.Neither the psychodynamic nor the humanistictherapists differed as predicted from the excellentlearning therapists on any measure.

The difference between excellent psychodynamicand humanistic therapists on conceptualizing abilitydoes make sense when one considers the differentialvalue placed upon conceptualizing in these twoapproaches. Psychodynamic theory places a pre-

mium on conceptualizing client dynamics and thetherapeutic relationship, especially with regard totransference and countertransference (Gelso &Carter, 1985). Humanistic theory, on the otherhand, places a premium on experiencing in themoment, and on living within the experience ofthe therapeutic moment. The therapist is to en-counter the client subjectively, rather than thinkabout the client during the hour (Meador & Rogers,1984; Polster & Polster, 1973; Rogers, 1980).

It will be recalled that the large majority of theexcellent therapists that participants selected tobe rated in the study were of a psychodynamicorientation (i.e., 60 of 89, or 67%.). Althoughthis imbalance does not cause serious problemsin terms of the validity of our quantitative analysis,it does appear to add an important twist to howthe findings may be interpreted. Thus, althoughexcellent psychodynamic, learning, and humanistictherapists differ little in terms of ingredients thatare seen as important in countertransference man-agement (except for conceptualizing ability), whena sizeable sample of psychotherapists is asked torate an excellent therapist, the therapist who ischosen is quite typically a psychodynamically ori-ented therapist. The 67% figure is well beyondthe percentage of the participants who rate them-selves as psychodynamic (50%). Why so manyof the excellent therapists chosen for rating in thisstudy were dynamic remains an interesting questionawaiting future research.

The above discussion notwithstanding, it appearsthat in general, excellence overrides theoreticalorientation in terms of the ingredients that facilitatecountertransference management. While learningtheorists have not traditionally focused a greatdeal of attention on the therapeutic relationship,and in particular, their own reactions to clients(Messer, 1986), they (at least those reputed to beexcellent) do appear to possess characteristics thatwould theoretically help them manage their coun-tertransference reactions. Moreover, there appearsto be emerging importance placed on the therapeuticrelationship in behavior therapy (Goldfried, 1982;Lazarus, 1981;Meichenbaum&Gilmore, 1982),even resulting in some behavior therapists ex-amining their reactions to clients (Goldfried &Davison, 1976). Given this trend, we may expectto see learning therapists attending to the therapeuticrelationship more and more, and possibly to theirown reactions to their clients within mat rela-tionship. While excellent learning therapists mightnot emphasize countertransference awareness as

419

S. L. Van Wagoner et al.

important in the delivery of therapy, they equalpsychodynamic and humanistic therapists in theirperceived possession of factors that might helpthem manage countertransference feelings.

A potential limitation of the present study isthat raters might have had a perceptual "set" forexcellence dictating that excellent therapists wouldbe rated highly on everything. This would beconsistent with the social psychology literaturedemonstrating that people will be perceived in away that is congruent with a particular schema orstereotype (Hamilton, 1979). On the other hand,excellent therapists might, in fact, possess moreof the characteristics measured. The fact that wehad the raters visualize the therapist they rateddoing therapy should have removed some of theabstraction of the task, and attenuated the perceptualset problem. Future research might explore theinfluence of cognitive sets on ratings of excellenttherapists.

A second limitation pertains to the fact thatdata were gathered by asking experienced therapistsfor their perceptions of the imagined excellenttherapist and the therapists in general. Such per-ceptions are important in and of themselves, butmuch caution must be exercised in generalizingfrom perceptions to the actual behaviors of excellenttherapists. Finally, although we believe that thereare therapists who generally excel in their work,the master therapists so to speak, it is also likelythat no therapist is highly effective with all typesof clients. A therapist who is excellent with onegroup may not be effective with another. Thus,within the concept of overall excellence, we wouldexpect considerable therapist by client variabilityin therapist effectiveness.

In conclusion, the results do suggest that ex-cellent therapists are perceived to possess morequalities theoretically related to the managementof countertransference than therapists in general.That this is the case regardless of the theoreticalorientation of the excellent therapists requires fur-ther investigation. It should be stressed that theresults do not indicate directly that excellent ther-apists are in fact better than therapists in generalat managing countertransference, although thatcertainly would be a reasonable inference. Thenext steps in the research would be to determineif excellent therapists actually do more effectivelymanage their countertransference reactions, as wellas to assess directly if our five theorized factorsrelated to actual countertransference feelings aswe have theorized and as our expert judges support?

Moreover, is the management of these feelingsrelated to treatment outcome? There is considerablesupport to suggest that therapy, overall, is equallyeffective for all the major theoretical schools(Lamber, Shapiro & Bergen, 1986). Perhaps thekey to their success lies in the existence of non-specific factors common to all techniques thatpositively influence outcome (Kazdin, 1986). Re-search does suggest that relationship factors arepowerful determinants of therapeutic success(Lambert, Shapiro & Bergin, 1986). That coun-tertransference management might be a by-productof therapist possession of general qualities im-portant in all therapeutic endeavors, independentof theoretical orientation, requires further inves-tigation.

ReferencesARKOWITZ, H. & MESSER, S. B. (EDS.) (1984). Psychoanalytic

theory and behavior therapy: Is integration possible? NewYork: Plenum.

ARLOW, J. A. (1985). Some technical problems of counter-transference. Psychoanalytic Quarterly, 54(2), 164-174.

BANDURA, A., LIPSHER, D. H. & MILLER, P. E. (1960). Psy-chotherapist's approach-avoidance reactions to patient'sexpression of hostility. Journal of Consulting Psychology,24, 1-8.

BEERY, J. W. (1970). Therapists' responses as a function oflevel of therapist experience and attitude of the patient.Journal of Consulting and Clinical Psychology, 34, 239-243.

BIRD, K. D. (1975). Simultaneous contrast testing proceduresfor multivariate experiments. Multivariate Behavioral Re-search, 10, 434-352.

BLANCK, G. & BLANCK, R. (1979). Ego Psychology II: Psy-choalalytic developmental psychology. New York: ColumbiaUniversity Press.

CUTLER, R. L. (1958). Countertransference effects in psy-chotherapy. Journal of Consulting Psychology, 22, 349-356.

FREUD, S. (1910/1958). Future prospects of psychoanalyticpsychotherapy. In J. Strachey (ed. and trans.), The standardedition of the complete psychological works of SigmundFreud (Vol. 11). London: Hogarth.

FROMM-REICHMANN, F. (1950). Principles of intensive psy-chotherapy. Chicago: University of Chicago Press.

GAMSKY, N. R. & FARWELL, G. F. (1966). Counselor verbalbehavior as a function of client hostility. Journal of Coun-seling Psychology, 13, 184-190.

GELSO, C. J. & CARTER, J. (1985). The relationship in coun-seling and psychotherapy. The Counseling Psychologist,13, 155-244.

GIOVACCHINI, P. L. (ED.). (1975). Tactics and techniques inpsychoanalytic therapy (Vol. 2): Countertransference. NewYork: Jason Aronson.

GOLDFRIED, M. R. (1982). Resistance and clinical behaviortherapy. In P. L. Wachtel (ed.), Resistance: Psychodynamicand behavioral approaches. New York: Plenum.

GOLDFRIED, M. R. & DAVISON, G. C. (1976). Clinical behaviortherapy. New York: Holt, Rinehart, & Winston.

420

Countertransference and the Reputedly Excellent Therapist

HAASE, R. F. & ELLIS, M. V. (1987). Multivariate analysisof variance. Journal of Counseling Psychology, 34(4), 404-413.

HACCOUN, C. M. & LAVIGUEUR, H. (1979). Effects of clinicalexperience and client emotion on therapists' responses.Journal of Consulting and Clinical Psychology, 47, 416-418.

HAMILTON, D. L. (1979). A cognitive-attributional analysisof stereotyping. Advances in Experimental Social Psychology.

HAYES, J. (1987). A phenemenological view of the relationshipbetween anxiety, empathy, and countertransference. Un-published masters thesis. University of Maryland.

HAYES, I. & G E L S O , C . I. (1991). Effects of therapist-trainees'anxiety and empathy on countertransference behavior.Journal of Clinical Psychology, 147, 284-290.

HEIMANN, P. (1950). On countertransference. InternationalJournal of Psychoanalysis, 31, 81-84.

HOWARD, K. I., ORLINSKY, D. E. & HILL, J. A. (1969). Thetherapist's feelings in the therapeutic process. Journal ofClinical Psychology, 25, 83-93.

JOHNSON, M. (1978). Influence of counselor gender on reactivityto clients. Journal of Counseling Psychology, 25, 359-365.

KAZDIN, A. E. (1986). Research designs and methodology.In S. L. Garfield and A. E. Bergin (eds.), Handbook ofpsychotherapy and behavior change. New York: John Wiley.

KERNBERG, O. F. (1975). Borderline conditions and patho-logical narcissism. New York: Jason Aronson.

LAMBERT, M. J., SHAPIRO, D. A. & BERGIN, A. E. (1986).The effectiveness of psychotherapy. In S. L. Garfleld andA. E. Bergin (eds.), Handbook of psychotherapy and be-havior change. New York: John Wiley.

LAZARUS, A. (1981). The practice of multi-modal therapy.New York: McGraw-Hill.

LUBORSKY, L. B. & SINGER, B. (1974). The fit of therapist'sbehavior into patient negative expectations: A study oftransference-countertransference contagion. Unpublishedmanuscript, University of Pennsylvania School of Medicine.

MEICHENBAUM, D. & GILMORE, J. B. (1982). Resistance froma cognitive-behavioral perspective. In P. L. Wachtel (ed.),Resistance: Psychodynamic and behavioral approaches. NewYork: Plenum.

MESSER, S. B. (1986). Behavioral and psychoanalytic per-spectives at therapeutic choice points. American Psychologist,41(11), 1261-1272.

MILLS, D. H. & ABELES, N. (1965). Counselor needs foraffiliation and nurturance as related to liking for clients andcounseling process. Journal of Counseling Psychology, 12,353-359.

MUELLER, W. J. (1969). Patterns of behavior and their reciprocalimpact in the family and in psychotherapy. Journal ofCounseling Psychology, 16, (2, Pt. 2).

NORCROSS, J. C. (ED.) (1986). Handbook of eclectic psycho-therapy. New York: Brunner/Mazel.

PEABODY, S. A. & GELSO, C. J. (1982). Countertransferenceand empathy: The complex relationship between two di-vergent concepts in counseling. Journal of Counseling Psy-chology, 29(3), 240-245.

REICH, A. (1951). On countertransference. International Journalof Psychoanalysis, 32, 25-31.

REICH, A. (1960). Further remarks on countertransference.International Journal of Psychoanalysis, 41, 389-395.

ROBBINS, S. B. & JOLKOVSKI, M. P. (1987). Managing coun-tertransference feelings: An interactional model usingawareness of feeling and theoretical framework. Journal ofCounseling Psychology, 34, 276-282.

RUSSELL, P. & SNYDER, W. (1963). Counselor anxiety inrelation to amount of clinical experience and quality ofaffect demonstrated by clients. Journal of Consulting Psy-chology, 22, 358-363.

SINGER, B. A. & LUBORSKY, L. (1977). Countertransference:The status of clinical versus quantitative research. In A.Gurman and A. Razin (eds.), Effective psychotherapy:Handbook of research. New York: Pergamon.

STRUPP, H. H. (1973). Psychotherapy: Clinical, research,and theoretical issues. New York: Jason Aronson.

WATKINS, C. E. (1985). Countertransference: Its impact onthe counseling situation. Journal of Counseling and De-velopment, 63(6), 356-361.

YULIS, S. & KIESLER, D. J. (1968). Countertransference re-sponse as a function of therapist anxiety and content ofpatient talk. Journal of Consulting and Clinical Psychology,32, 414-419.

421