Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Bias and clinical judgment in counselling and psychotherapy: extending theory and research design Grigg, Donald G 1993

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1993_fall_grigg_donald.pdf [ 5.68MB ]
JSON: 831-1.0098837.json
JSON-LD: 831-1.0098837-ld.json
RDF/XML (Pretty): 831-1.0098837-rdf.xml
RDF/JSON: 831-1.0098837-rdf.json
Turtle: 831-1.0098837-turtle.txt
N-Triples: 831-1.0098837-rdf-ntriples.txt
Original Record: 831-1.0098837-source.json
Full Text

Full Text

BIAS AND CLINICAL JUDGMENT IN COUNSELLING AND PSYCHOTHERAPY:EXTENDING THEORY AND RESEARCH DESIGNbyDONALD GLEN GRIGGB.A., University of British Columbia, 1977A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF ARTSinTHE FACULTY OF GRADUATE STUDIESDepartment of Counselling PsychologyWe accept this thesis as conformingTHE UNIVERSITY OF BRITISH COLUMBIAAugust, 1993© Donald Glen Grigg, 1993In presenting this thesis in partial fulfillment of therequirements for an advanced degree at The University ofBritish Columbia, I agree that the Library shall make itfreely available for reference and study. I further agreethat permission for extensive copying of the thesis forscholarly purposes may be granted by the Head of my Departmentor by his or her representatives. It is understood thatcopying or publication of this thesis for financial gain shallnot be allowed without my written permission.Department of Counselling PsychologyThe University of British Columbia2075 Westbrook PlaceVancouver, B.C., CanadaV6T 1W5Date: August 1993 iiABSTRACTIn an investigation of how counsellors make decisionsabout how to intervene on behalf of their clients twelvetrainees (six family counsellors and six person-centeredcounsellors) sorted a series of 32 possible counsellingprocedures from most to least desirable into a quasi-normal Q-distribution. Each subject completed the sort under fourconditions, once for each hypothetical counselling client.The sex, age, gender, type of disorder, and severity ofdisorder were held constant across all conditions. Conditionsvaried as to socioeconomic status (poor versus middle class)and ethnicity (Caucasian versus Native) of the hypotheticalclient. The counselling alternatives ranked by the subjectshad been pre-selected to represent two dimensions of cognitiveattribution: internality versus externality and stabilityversus instability. The sorted cards yielded preferencescores for each attributional category for each experimentalcondition. Analysis of variance was applied to the scores foreach subject across all four conditions as a test for ethnicor socioeconomic status differences as a function ofattributional pattern. The scores for all subjects were thencorrelated with each other, and this matrix ofintercorrelations submitted to factor analysis yielding aprincipal axis solution. Varimax rotation was used to produceorthogonal factors. The factor loadings on each orthogonalfactor were then used to calculate weighted standard scoresindicative of the preference shown for each item within theiiipattern indicated by the factor. The items were than rankedby their standard scores, scored by attribution pattern, andanalysis of variance applied.Analysis of variance for individual counsellors revealedonly one subject for whom, across all conditions, ethnicityand socio-economic status were significant variables.Factor analysis revealed three distinct patterns ofresponse, one pattern for family counsellors and two distinctpatterns of response for the person-centered counsellors.Analysis of variance indicated that the family counsellorspreferred items classified in the attributional schema asexternal or external-stable. The first group of person-centered therapists showed a preference for items classifiedas stable. The second group of person-centered therapistsalso preferred items classified as stable, but alsosystematically ranked unstable and external items lowest.Patterns were highly correlated across all experimentalconditions except the "poor Caucasian" condition. In thiscondition a fourth pattern emerged representing only twoperson-centered therapists and showing no significantpreferences on the attributional schema. The second person-centered pattern essentially disappeared, while the familytherapists split into two less differentiated patterns. Inthe "poor Caucasian" condition there were four subjects whoseresponses correlated about equally with more than one pattern,something which did not occur in the other conditions.ivData confirm that attribution theory and Q-methodologycan be used to distinguish patterns of clinical decision-making as a function of type of therapist and clientcharacteristics. Results also tend to confirm that clinicaldecision-making, as a pattern of attribution, is influenced bysocio-economic status of the client.VTABLE OF CONTENTSABSTRACT^ iiLIST OF TABLES^ viiiLIST OF FIGURES ixACKNOWLEDGMENTS^ xCHAPTER I - INTRODUCTION^ 1Nature of the Problem 1Purpose of the Study^ 3Definition of Terms 4A Note About Methodology^ 5A Cognitive Attributional Modelof Therapist Judgment  6Research Questions^  7CHAPTER II - REVIEW OF THE LITERATURE^ 10Introduction^ 10Defining and Operationalizing Bias in Psychotherapy^ 10Client Characteristics That Evoke Bias^ 13Methodology^ 13Attribution Theory^ 23Innovation In Research Design^ 33CHAPTER III - METHODOLOGY^ 35Introduction^ 35Purpose of the Study^ 35Q-Methodology^ 36Q-Methodology: History and Rationale^ 38Subjects^ 39Data Collection^ 40Q-Sort Materials and Demographic Sheet^ 42viExperimental Materials^ 43Introduction to Scoringand Statistical Procedures^ 46Scoring the Q-Sorts^ 46Analysis of Variance for Individual Cases^ 47Factor Analysis of Pooled Cases^ 47Analysis of Variance for Hypothetical Types^ 48Determining and Comparing Pattern Content 49Descriptive Statistics and Secondary Analyses^ 50CHAPTER IV - RESULTS^ 52Introduction 52Q-Sorts^ 52Analysis of Variance for Individual Cases^ 55Factor Analysis of Pooled Cases^ 56Analysis of Variance for Hypothetical Types^ 60Determining and Comparing Pattern Content 64Severity of Depression^ 66CHAPTER V - DISCUSSION 68Introduction^ 68Persons, Traits, and Occasions^ 68The Discrimination of Types 69Cognitive Attribution as a Decisional Model^ 71Situations as Evocative of Difference^ 71Severity of Depression^ 72Some Critique of the Study 73Two Kinds of Person-Centered Counsellors^ 78Individual Differences^ 79viiClinical Decision-Making and Bias^ 81Implications for Theory and Research 84REFERENCES^ 87Appendix A: Recruitment of Subjects^ 94Appendix B:Appendix C:Appendix D:Appendix E:Instruction Sets and Experimental Materials ^ .97Q-Sort Items and Their Rater Reliabilities ^ 108Analysis and Content of Patterns^ 112Analysis of Variance for Q-Scores ofIndividual Subjects^ 140viiiLIST OF TABLESTable 1 - Ages and level of training for subjects^ 40Table 2^- Means and standard deviations by attributionalvariables for family therapists^ 54Table 3^- Means and standard deviations by attributionalvariables for person-centered therapists^ 54Table 4 - Summary of significant effects found in ANOVA doneon each subject's Q-score^ 56Table 5 - Rotated factor matrix for Poor Native condition.. .58Table 6 - Rotated factor matrix for Middle Class Nativecondition^ 56Table 7^- Rotated factor matrix for Poor Caucasiancondition 58Table 8 - Rotated factor matrix for Middle Class Caucasiancondition^ 59Table 9^- Significant effects and trends in ANOVA's performedon patterns formed by hypothetical types^ 61Table 10 - Correlations of Q-scores of hypotheticaltypes^ 63Table 11 - Items preferred by FC over PC2^ 64Table 12 - Items preferred by PC2 over FC 65Table 13 - Items preferred by PC1 over PC2^ 65Table 14 - Items preferred by PC2 over PC1 66Table 15 - Analysis of variance for subject's estimation ofthe severity of depression for hypotheticalclients^ 67ixLIST OF FIGURESFigure 1 - Simplified attributional model^ 29Figure 2 - Attributional model including nomothetic vs.ideographic dimensions^ 32Figure 3 - Distribution of preferences in Q-sort task^ 45xACKNOWLEDGEMENTSThe writer expresses his appreciation to the members ofhis committee, Dr. M. J. Westwood, chairman, Dr. F. I.Ishiyama, and Dr. A. J. More. Special thanks are owed to Dr.Westwood, not only for his scholarly consultation, but alsofor his ability to restore motivation and focus in times ofdiscouragement or confusion.Deep gratitude goes to Professor Walter Boldt whoprovided extensive methodological consultation with patience,generosity, and a brilliant and creative command of hissubject.The subjects who participated in this study gave of theirtime, their serious attention, and their professionalcommitment. They are owed a large debt of thanks.The author owes personal gratitude to his wife, AndreaGauthier, who endured much, and was still willing to proof-read a very esoteric document. Ron Peters provided invaluableassistance both through his support as a friend and throughhis knowledge and talent with statistical analysis. RichardCopley provided encouragement from the beginning, and hissupport was there right through to the end.1CHAPTER 1INTRODUCTIONNature of the ProblemIn the literature pertaining to the study of bias inpsychotherapy, clinicians are enjoined not to base theirdecisions on such client characteristics as race, gender,economic status or social position. Including these factorsin decision-making is considered to be bias. Commenting onstudies that find that women and minorities are judged to bemore disturbed or requiring more treatment, Lopez (1989)asserts that a "major assumption underlying this bias is thatthese differences are not due to actual differences betweengenders, age groups, or racial/ethnic groups but to error onthe part of practitioners" (p. 184).The literature on cross-cultural counselling andpsychotherapy enjoins clinicians to take cultural factors intoaccount in deciding how and when to intervene (Marsella &Pedersen, 1981). Encouraging practitioners to see culture asa dimension of relevant difference, Sue and Sue (1990) notethat when "minority group experiences are discussed they aregenerally analyzed from the 'White middle-class perspective'"(p. 8).While the two literatures may seem to be at odds, theyare easily untangled at a philosophical level. The biasliterature asks clinicians not to be influenced by irrelevantfactors such as race or status, and the cross-culturalliterature asks clinicians to view culture as a relevant2dimension of difference. Having recourse to a hierarchy oflogical types, any difference in the messages of theliteratures can be resolved by saying that consistentlyintervening on behalf of the client's best interest may alsomean intervening in different ways depending on the culturalcontext.However, in the pragmatic world of clinical practice,cultural variables are often confounded with issues ofracial/ethnic difference and differences in socioeconomicstatus--the only client variable effect found to consistentlyengender bias (Abramowitz & Dockecki, 1977). Acknowledgingthis problem, Draguns (1981) states that "a more thoroughresearch effort on the effects of ethnicity, race, minoritystatus, and judgment class, all of these variables beinginterrelated, is necessary in order to resolve this issue" (p.19) [italics added].This study does not purport to untangle Draguns entirelist of variables and their interrelationships. However, itdoes provide a demonstration of a method for empirically andqualitatively studying variables related to culture and bias,while at the same time relating these to types of clinicalinterventions and differences in orientations to counsellingand psychotherapy.3Purpose of the StudyThe intent of this thesis is refinement in theconceptualization and methodology in the study of bias incounselling and psychotherapy.The typical study of bias has little theoreticalunderpinning. Characteristics of clients, which are usuallytaken to be the objects of prejudice (gender, ethnicity, race,socioeconomic status, and so on), are presumed to be factorsthat ought to be irrelevant to clinical decision-making. Theethical principle (as opposed to theoretical construct)invoked is usually distributive justice, the notion ofproviding similar treatments for similar problems.Consequently, the independent variable in a typical study ofbias is usually some factor such as race or gender, and thedependent variable is the outcome of some clinical judgment.In the classic experiment, if all other factors are heldconstant, the difference in clinical judgment is attributed tothe operation of bias.However, it is also argued in the cross-culturalcounselling literature that client variables such as gender,race, age, ethnicity and socioeconomic status representdifferences that ought to be treated with sensitivity, thatought to be taken into account in therapist decision-making.Theoretically, this study is a step toward disentangling thesearguments by explicating processes underlying therapistdecision-making and separating categorical decisions (such asclassification and diagnosis) from continuum decisions (such4as severity of disorder, or type and amount of treatmentindicated).The experimental paradigm typical of this area ofresearch is generally operationalized by sending casedescriptions to large numbers of counsellors andpsychotherapists. The therapists are asked to make somejudgements about how to proceed with the process of helpingthe client described. Some putative source of bias issystematically manipulated. In a study of gender bias, forexample, half the sample would receive material describing theclient as female, the other half would be presented with anidentical client profile, except that the client would bedescribed as male. The typical experiment involves largesample sizes, multiple dependent measures, large variances,small mean differences, high levels of statisticalreliability, and little demonstration of clinical relevance.Most typically, there is no study of "within-subjects"variance or of the psychological phenomena underlying theprocess of judgment.Definition of TermsLater in this document, important precedents in the useand definition of terminology will be discussed. However, forthe present discussion, more precise meanings for basic termsare important for clear understanding. Bias is taken to meanthe application of extraneous or irrelevant factors into theprocess of clinical decision-making. Counselling and5psychotherapy are taken to mean consultations provided byprofessionals with the goal of changing the behavior,cognition, or affect of the client. Cognitive attributiontheory is a set of constructs fundamentally associated withthe work of Heider (1958) explaining how observers attributecauses to the behavior of others.A Note About MethodologyThis study uses a very different experimental design thanhas been employed previously in this area of research, and soa few early comments are in order to explain the rationale forproceeding with an old problem in a new way. First, thedesign is experimental so as to produce "cause and effectresults". The research question was made specific: Is therean ethnic bias or a judgment bias demonstrated whencounselling trainees plan counselling interventions in termsof options structured according to cognitive attributiontheory. All subjects respond to all four experimentalconditions so that individual patterns and within-subjectsdifferences can be examined. Experimental controls areachieved statistically or through standardizing theexperimental materials. The experimental task is an analogyto actual counselling practice, specifically the review of aclinic intake report in preparation for a counsellinginterview. The response set in each experimental condition isa Q-sort, making use of the special properties of Q-methodology for exploring decisional processes.6The design theory underlying the study can be describedas "replication logic". Each subject's responses form asingle universe of content, which is to say an experiment inand of itself, with data points which can be statisticallyanalyzed. The next step in classical science is to replicatethis experiment. The hypothesis is that therapists of similartheoretical orientation will give similar responses, thusforming a "family" of subjects or, more abstractly, a group ofreplicating experiments, with therapist theoreticalorientation being the linking subject variable.The two therapist groups to be studied are person-centered therapists with their emphasis on endogenous growth(internal attribution) and systemically oriented familytherapists with their emphasis on interactions between people(external attribution).A Cognitive Attributional Model of Therapist JudgmentIn this study cognitive attribution theory is taken as aconceptual basis for structuring and analyzing the responsesof the subjects. Within this framework, the variables ofinterest are the cognitive dimensions that decision-makersapply to making judgements, especially about other people.The attributions made about the behavior of others can beconstrued as lying somewhere on a dimension of internal versusexternal (manifestation of a personal character trait asopposed to response to the environment) and on anotherdimension of stable versus unstable. In a classic study of7gender prejudice, men accounted for high scores by women on amathematics task as a function of ideal working conditions andgood luck: external, unstable attributions. In anotherexperimental condition, when the men were reported to havescored well, the men attributed this performance tomathematical aptitude and predilection to diligent effort:internal, stable attributions.In at least some cases, the interventions of counsellorsand therapists can be taken to be indicative of theirattributions about clients and their problems. The communitypsychologist engaged in social action programs could beconstrued to be taking an external unstable position about thedifficulties in question, looking for change in the malleablepolitical world outside the person. The psychoanalyst using apsychodynamic frame of reference to interpret the structure ofpersonality is using an internal stable attribution: change issought through modifying a stable entity (the personality)within the person. In this study, the possible responses ofthe subjects to the experimental materials is structuredaccording to the possible dimensions of the two-dimensionalattribution theory described above.Research QuestionsThe general hypothesis is that clinicians will differ intheir response patterns according to their theoreticalorientations. A further hypothesis is that they willdifferentially respond to white, middle-class clients with8attributions that are unstable and external, and that theywill respond to non-white poor clients with attributions thatare stable and internal. These latter attributions shouldcorrelate with higher estimates of severity of disorder,since, from a clinical perspective, change would be construedto be much more difficult. These predictions are consistentwith general attribution theory (as explicated below) (Weary,Stanley, & Harvey, 1989) as well as Lopez's (1989) findingthat three quarters of bias studies that includedsocioeconomic status as a variable found an overpathologizingbias.It is also hypothesized that family-oriented counsellorswill differ from person-centered counsellors in theirattributional patterns. Family therapists are concerned lesswith internal, intra-personal events and more concerned withexternal, unstable inter-personal events (Goldenberg &Goldenberg, 1985). By contrast, person-centered counsellorscan be expected to emphasize internal, stable aspects ofclients (Meador & Rogers, 1984).This study is intended to show within-subject differencesin clinical judgment, as well as cross-sectional differences.It is also designed to yield data about how clinicians maketheir judgements. The experimental design allows for thecontrol of, and manipulation of, several sources of bias atthe same time. This allows for the study of interactionsamong variables. Similarly, client characteristics can easilybe altered from experiment to experiment. At a theoretical9level, the outcomes indicate the workings of an attributionalprocess, without necessarily defining response to theindependent variables as a bias (i.e. extraneous and/orirrelevant).1 0CHAPTER IILITERATURE REVIEWIntroductionIn reviewing the literature on bias in psychotherapy forthe purposes of this study, there are a number of points ofemphasis. First, the notion of what bias is, and how theconcept of bias is operationalized, is crucial tounderstanding this literature. Within this area, ideas fromthe literature on cross-cultural counselling form points ofcomparison and alternatives for understanding differences inclinical judgment. Second, the methodology of research isimportant to this study since it is in the area of researchstrategy and methodology that this writer sees the most roomfor critical judgment and innovation. And, third, theoryunderlying the study of bias is explored, with particularemphasis on the possibilities of cognitive attribution theoryas a frame of reference.Defining and Opera tionalizing Bias in PsychotherapyThe problem of bias in psychotherapy could be put as aconfrontation with the questions: Do therapists, eitherindividually or as a group, serve their clients objectivelyand sensitively, without the operation of bias or prejudice?This question implies that the study of bias is a question ofconsistent application of professional values. These valuesinclude the specific codes of ethical conduct published,endorsed, and enforced by most professional human service1 1organizations. The principles underlying these codes can beclassified and summarized as welfare (action taken is in theinterest of the client), beneficence (doing good by not doingharm), distributive justice (similar treatment for similarproblems) and autonomy (the rights of clients to makedecisions in their own best interests) (Cayleff, 1986).Framed as an ethical issue, bias would most obviously take theform of violating one of the principles above as a function ofsome client characteristic that ought to be irrelevant to theclinical considerations of the case. From a comparativestandpoint, the notion of distributive justice, the idea ofproviding similar treatment for similar problems, provides anoperational principle. It has been typical of research inthis area to operationalize bias as difference in therapistprofessional actions as a function of client characteristicssuch as race, ethnicity, gender, age, or handicap.Lopez (1989), in a major review, states that "the termbias implies a prejudgment or prejudice" and goes on toassert that "a major that differences [inclinical judgment] are not due to actual difference.. .but toerror on the part of practitioners." Bias, then, refers toerror or inaccurate therapist judgment following from theapplication of irrelevant or extraneous information toclinical decision-making, usually by a process that is outsidethe awareness of the clinician (Lopez, 1983; Schlossberg,1977). Clearly, the operation of bias in psychotherapy hasgreat potential to lead to violation of the ethical principles12above, as well as to put into question the objective stance ofthe therapist. Even if we allow, as Fromm (1947) does, thatethics and values can be objectively studied from the point ofview of scientific psychology, the question of bias inpsychotherapy remains philosophically subtle.The question is also operationally complicated by thenotion that bias is also operationalized as a problem ofclinical judgment.More complication arises when social scientists seek tooperationalize the concept of bias in psychotherapy and planempirical studies of this problem. Researchers must generatedefinitions of bias, prejudice, objectivity, therapy, and soon that are consonant with the broad connotations ofprofessional practice and still allow for quantification or,at least, reliable classification. Beyond this, theresearcher interested in bias must generate a research design,an empirical paradigm, that yields results that are valid andreliable while at the same time clinically significant. Giventhat a major goal of this study is to address experimentalmethodology in this area of research, the emphasis below willbe on controlled experiments as opposed to archival orquestionnaire research. Again, acknowledging the limits ofthe present study, the review below emphasizes continuumjudgements (such as severity of client problem or therapeuticintervention indicated) in clinical decision making as opposedto classification judgments (such as diagnosis, all of whichis experimentally controlled in the present case).13Client Characteristics That Evoke BiasGender, age, ethnicity, race, age, and socioeconomicstatus have all been studied as possible client variables thatimpact on the clinical judgment of therapists. Lopez (1989),in a major review, concludes that the only client variablewith consistent empirical support is social class. Othervariables seem not to consistently "show up" as significant inspite of several decades of research and hundreds of studies.Many of the difficulties associated with the apparentinconsistency seem to be conceptual and methodological.MethodologyThe literature on racial bias seems to epitomize thedifficulties encountered in research on bias. A review bySattler (1977), while failing to document systematic bias,pointed to the limitations of both the analogue and archivalresearch methods. Voluntary participation by therapists inmost analogue designs limited reliability, as did the lack ofcontrol for social desirability effects. Designs almostinvariably lacked control for "same race" versus "mixed race"therapist/client dyads, and placebo controls were almostalways absent. In a further review of this literatureAbramowitz and Murray (1983) noted that the confounds inherentin research designs to date so slant results toward the nullhypothesis that analogue methodology is, in their opinion,"well nigh irredeemable". These authors assert that "a14wholesale moratorium on the continued wholesale adoption thusseems entirely justified" (p. 247). This extreme position,while reflecting a lack of results, fails to take into accountthe source of the problem and to fully evaluate thealternatives. The alternatives are to either design researchthat transcends the problems of control and confounds, or torely solely on naturalistic studies that are, by the logic oftheir designs, descriptive and correlative at best. If biasis a problem, then it is a problem to be solved. Whilearchival and naturalistic studies may have proved the abilityto identify and predict the operation of bias, without datathat indicates cause and effect relationships, professionalcounsellors and those concerned with their behavior will knowfrom correlational study only that bias exists and ispredictable. While these methods may have heuristic value,they cannot fully explain how bias comes about or how tochange it.Murray and Abramson (1983b) offer a compendium ofresearch and theory on the subject of bias in psychotherapy.While acknowledging that some important results have beenfound, the general consensus of the many authors contributingto this work is that empirical data has, to date, beenconfusing at best and equivocal at worst. The editorsconclude that "it seems we are just learning how to study biasin psychotherapy" (p. 351). The study of bias lackstheoretical clarity and a reliable data base of valid researchresults.15Rather than despair of any hope for analogue research onbias, Davidson (1983) takes aim at the theoreticalunderpinnings of bias research. The often-cited study byBroverman, Broverman, Clarkson, Rosencrantz, and Vogel (1970)(in which a large sample of therapists were asked to ratebipolar adjective pairs descriptive of general mental health)is a first target. Davidson notes that Broverman-likeresearch fails to show how results of paper and pencil teststranslate into the emotion-laden arena of thepsychotherapeutic encounter. Davidson also notes thatattempts to replicate Broverman have been disappointing.Pointing out that few research designs test articulated theoryor contain compelling analogies to "real" therapist behavior,Davidson recommends that countertransference, thepsychoanalytic concept which accounts for the therapist'sfeelings about and expectations for the client, form thetheoretical basis for bias research. Direct study ofclinicians' behavior and attitudes with clients isrecommended. Presumably, such research would involveassessment of the therapist's emotional and cognitivefunctioning, linking this to clients' immediate responses andlater behaviors. Davidson stops short of describing inpractical, operational terms just how this is to beaccomplished.Davidson's principle that research, to be meaningful,must come to terms with the client/therapist encounter iscertainly fundamental. However, the criticism of "paper and16pencil tests" as operational measures is equally basic. Alarge body of well-replicated literature documents the lack ofcorrelations between actual behavior and subjects' (orclients') assertions about what they do (Mischel, 1968).Thus, experiments that operationalize dependent measures interms of paper and pencil responses to hypothetical situationsare really measuring a correlate of an unreliable correlationrather than the phenomenon of interest. This problem isreflected in a study by Murray and Abramson (1983a). Adescription of a client presenting problem was sent to a largesample of therapists with the request that the case beevaluated. Subjects were told that their responses would beused as a data base for a computer screening system, but infact, the sex and attractiveness of the client description wassystematically varied while the case description was heldconstant. Results of questionnaires returned were offered bythe authors as evidence of an attractiveness bias. In thiscase, the implicit operational definition of bias is that, allother things held constant, differences based on clientcharacteristics represent bias. In order to find compellingevidence in this study that this is a bias operative in thetherapists' behavior with clients, one would have to believethat answering a questionnaire concerned with differentialscreening of client problems is a good analogy to counsellingprocess. Furthermore, the authors report levels ofstatistical significance, but not absolute magnitudes of meandifferences or measures of variance. In a large sample study17(N = 971), small differences that are clinically trivial maywell show up more reliably, in a statistical sense, than largedifferences (associated with large variances between subjects)that are clinically crucial (Sidman, 1960). Inherent,fundamental design flaws render results of this studyclinically ambiguous at best.Even if the limitation of statistical design, multipledependent measures, and weak analogy could be transcended,this study would still fail to address the fundamental concernunderlying the study of bias. In the opinion of this author,counsellors are concerned with bias because they are concernedwith offering a complete service to clients that is consistentwith the ethics of professionalism. If the Murray andAbramson study does, indeed, tell us that therapists, inconducting therapy, behave differently when they encounter anattractive client, have we really encountered a bias? Is theclinician aware of the source of the judgment, is the judgmentaccurate, and is the information relevant to the decision? Ifthe therapist quite consciously takes client attractivenessinto account in understanding the uniqueness of theindividual, while taking care not to transgress any of theethical guidelines above, then bias is not the issue. Carefulcontrol of client variables and presenting problem minimizethis issue in Murray and Abramson (1983a), but thisdistinction between bias and sensitive discrimination isvitally important. What should surprise us in doing researchis finding that a salient variable is unrelated to behavior,18just as in observing therapy we should be surprised to see apractitioner insensitive to a dimension of individualuniqueness. Meaningful research about bias tells the readerwhat dimensions or aspects of the therapist or client arecausally related to undermining the egalitarian practice ofpsychotherapy.Still the Murray & Abramson (1983a) study represents akind of classic paradigm for studying difference. The centraltactic is to have therapists review case descriptions, holdingall factors constant, varying only one client characteristicbetween subject groups. Typically, subjects are therapistsrandomly assigned to groups, and the results are, necessarily,normative descriptions of population samples since repeatedmeasures of individuals would jeopardize fundamental controls.This tactic was used by Saxon & Spitznagel (1992) to expose anage bias among vocational counsellors. Hardy and Johnson(1992) showed case material to 185 graduate students,systematically varying gender, alcoholic state, and thesocioeconomic status of the "client". They found that thepresence of alcoholism negatively affected judgments ofprognosis--but, again, is this evidence of bias? Isalcoholism really irrelevant to prognosis? Burk and Sher(1990) used the same research strategy to demonstrate thatcounsellors have negative perceptions of children ofalcoholics. But is being the child of an alcoholicunimportant, something that should be left out of clinicaljudgment? Gartner & Harmatz (1990) kept their case19descriptions constant among all 363 counsellors who reviewedthe material, varying only the religious and political viewsof the hypothetical clients. These authors described thedifferential response of groups of therapists to thesevariables as "ideological transference". Huitt & Elston(1991) used this method to explore attitudes to the disabled,while Maroney & Golub (1992) used this technique to quantifynurses' bias against obese patients. Hansen and Reekie (1990)used the "standard case description/vary only thedemographics" method to assess how social workers differed intheir perceptions of male and female clients. Male socialworkers saw male clients as more disturbed, but, again, themethodology stops cold at identifying differences. Whichparticular therapists perceive this difference? How do theyperceive it? What does it mean to their clinical judgmentprocess? And under what conditions, and on what occasions, isthe independent variable relevant or irrelevant? Thisresearch strategy seems to say that clinical judgment is ablack box, that aggregating the responses of many therapistsis a generalizable result, and that client variables arealways irrelevant to clinical decision-making.These tacit assumptions are at odds with notions basic toanother literature, the literature of cross-culturalcounselling. Here, client variables such as race, socialposition, ethnicity, and so on are explicitly seen as factorsthat therapists should take into account insofar as they arecorrelates of or reflections of cultural differences (Sue and20Sue, 1990). It would be naive to argue that the literature onbias is insensitive to cultural issues which are oftenconfounded with race, ethnicity, and socioeconomic status.However, the acceptance of simple difference as evidence ofbias ignores the importance of culture and the dynamics ofclinical judgment.In spite of its short-comings and difficulties, Lopez(1983) defends the importance of analogue research, but callsfor new designs. He deplores the lack of theoretical basisfor research design and recommends cognitive attributiontheory as a frame of reference to guide research: "Empiricalstudies that could identify specific attributional patternswould greatly assist those attempting to ameliorate andprevent clinical bias" (Lopez, 1983 p. 362). Lopez suggeststhat specific presenting problems be systematicallyinvestigated, relating referral problems to therapist andclient variables. Lopez asserts that research designs need toaddress themselves to the process of clinical judgment (notsimply its manifestations) and to the question of when in thetherapeutic process bias occurs.When therapists form clinical judgements they require adata base, a decisional strategy, and a set of priorities.Falvey (1992) found that therapists rely most heavily oninterview data to inform themselves about clients, but alsoattend to reports from other professionals, giving little.Korchin (1976) notes that clinicians tend, in their decision-making to use less information than is actually available to21them. O'Donohue & Fisher (1990), using survey and interviews,could find no systematic decision-making procedure in thecounsellors they studied. However, there is evidence thattherapists consistently form causal hypotheses, and that theytend to remember information that conforms to these hypotheses(Strohmer & Shivey, 1990), often maintaining these notions inthe face of new and contradictory evidence (Ellis & Robbins,1990). Presumably, therapists have some theoretical "roadmap" by which they organize information into priorities.Clearly, a research strategy that related a theoreticalposition to a decisional strategy and a set of clinicalpriorities and then related these to specific clinicalproblems would be extremely useful in understanding clinicaldecision-making.Introducing the notion of dynamics of clinical decision-making into the literature on bias in psychotherapy is anotherdimension that defines counselling and therapy as a process,not an event. Few studies of bias take this reality intoaccount (Abramowitz and Murray, 1983). Rather, the typicaldesign is a cross-sectional "snapshot" of intake, assessment,disengagement, or outcome. Populations are studied at onepoint in time rather than individuals across time. Definingtherapy as a process implies higher validity (i.e. closeranalogy) for longitudinal experimental designs. Processresearch, empirical single-case studies, repeated measuresdesigns, and replication logic (Yin, 1989) seem to be obviousimprovements over the current state of affairs. Research22designs that contain reliable repeated measures carry thepotential to study processes and particular events, as well asindividuals and populations. Such designs have potential tobring to light a wholistic, but empirically documented,picture.Taken together, the discussion above forms a set ofcriteria for the design of research in psychotherapy. Thedesign should be experimental in nature so as to produce"cause and effect" results rather than correlations orpredictions. The design must be based on a clearlyarticulated theory such that ameliorative action is clearlyindicated when causes and effects are discovered. Resultsshould be clinically meaningful as well as statisticallysignificant. To be useful, data must be directed atattributes that can be identified with individual therapistsand counsellors rather than statistically defined populations.The analogy represented by the experiment must be a compellingparallel with the actual role-consistent behavior oftherapists with clients and, furthermore, the results must bereplicable in real clinical situations. The ability tocontrol for the nature of the presenting problem and itsseverity, while meeting other design demands, is crucial.Client variables not intrinsic to the hypotheses of the study,and especially socioeconomic differences, must be controlledfor. Theory and design must take into account the dynamics ofdyadic therapy, the change in relationship over time, that isthe hallmark of the therapeutic encounter. These principles23represent a criterion list for evaluating research design andoutcome.While at a practical level, the present study does notaddress itself to changes over time in the therapist/clientrelationship, ideas for how to apply the basic design andtheory to longitudinal research are presented below. However,it remains to take up Lopez' challenge to apply cognitiveattribution theory to clinical judgment.Attribution TheoryAttribution theory as a formal psychological system hasits roots in the pragmatic psychology of Heider (1958).Heider took as his premise that all people have a basic needto believe that they control their environment. It followsfrom this premise that a basic human quality would be to seekto understand why people do things. Attribution theory positsthat, armed with this understanding of why others behave asthey do, we can take action to predict and control whathappens to us. In an attribution theory frame of reference,we are all "scientists" using theory to attribute causes tothe interpersonal events in our lives. Attribution theoryposits that all people make causal attributions about thebehavior of others.Insofar as it is concerned with the need to predict andcontrol, Heider's theoretical premises are very similar tothose of Rotter's work on locus of control (Rotter, 1966) andSeligman's theory of learned helplessness which, in recent24years, has taken an explicitly attributional tack (Alloy,Peterson, Abramson, & Seligman, 1984). In the present study,this premise of inferred causation on the part of theperceiver is undertaken as well. The assumption is made thattherapists seek to understand the client in order to make theencounter predictable and controllable.Heider proposes that there are two kinds of basicattribution: internal (or, as some refer to it, dispositional)and external (or situational). Again, these constructs arevery similar to the internal/external dichotomies posited inthe locus of control literature.Internal attribution is in play when the cause ofbehavior is thought to lie within the person as opposed tothat person's environment or circumstances. An internalattribution makes the causal assumption that behavior observedreflects a unique property of the person. A personalattitude, a religious belief, a character trait or apersonality dimension given as a cause of behavior are allexamples of internal attribution. When internal attributionsare made, the cause is said to be inside the person: Joe workshard because he has within him the characteristic ofpersistence.External attribution is in play when factors in thesocial or physical environment are said to be causingbehavior. If we say that Joe works hard to earn money, to getpraise, or to achieve grades, we are making externalattributions about the causes of Joe's behavior. Here, the25fundamental point is that the cause of behavior is seen to beoutside the person doing the behaving.Skinner (1974) has noted that people routinely selectinternal or external attribution according to the desirabilityof the behavior to be explained. When a student performs wellin preparing a paper, it is typical for this performance to beattributed to internal causation: intelligence, diligence,careful attention, writing skills, and the like. But when awritten assignment is poorly done, the same student is likelyto favor external attributions about causes of behavior:sickness, fatigue, stress, and a host of extenuatingcircumstances are invoked to account for the poor showing. Acrucial point is that when we make a purely situational orexternal attribution we are implicitly saying that this personwould not engage in this behavior without these situationalfactors.The selection of attributional dimension can also beindicative of biases or prejudices (Zimbardo & Lieppe, 1991).The common and near-classic example is that of commentary onathletes. It is not unusual to hear that athletes of colorhave an abundance of natural talent--internal attribution--while white players perform well because of hard work:external attribution. The unstated but logical corollary isthat colored athletes, as persons, are just plain lucky to beborn with talent, while the white athletes are hard workerswith lots of "character". Choosing another common-placeexample, sex-role stereotypes also demonstrate the workings of26the attributional dimension of internal versus external(Hamilton, 1983). Deaux (1976) and Deaux & Emswiller (1974)reported on a series of studies in which the dependentvariable is the attributions made about differential behaviorof men and women on tasks where traditional stereotypes wouldimpute some advantage or disadvantage to one gender oranother. For example, when women outperformed men on amathematical task the subjects accepted the outcome as factualand valid--but accounted for it with external attributionsabout the success of the women. The test was unfair, the menwere fatigued, the women got lucky: these were theexplanations offered. When the poles were reversed, and themen were seen to have done better, the attributions werealmost all internal, having to do with the skills, talents andintelligence of the men or the internal deficits of the women."What is skill for the male is luck for the female," is thephrase that summarizes the findings.What factors influence the formulation of internal versusexternal attributions? Kelley (1972) asserts that there arethree perceived properties of behavior that are criticaldeterminants: nonnormative versus normative, consistent versusinconsistent, and distinctive versus nondistinctive.Nonnormative behavior leads to internal attributions, as whenwe assume that we are dealing with a rude person when we seethat person behave in a rude way to someone known to be kindand generous. Internal attributions are also more likely whenbehavior is seen to be consistent. In a simple example, being27consistently on time for appointments may lead to anattribution of compulsiveness (negative trait) or punctuality(positive trait) but in either instance the attribution isabout some internal factor influencing behavior. Indeed, wecould say that a great deal of psychology is about internalattributions since consistency of behavior over time andacross situations is a defining feature of personality theory.Distinctiveness is a slightly more subtle determinant ofinternal attribution. When a behavior is particular to agiven set of circumstances, i.e. distinctive, then we areinclined to make an external attribution, an attribution thatthe circumstances account for the behavior. Conversely, whenbehavior is non-distinctive we make an internal attribution.For example, someone is seen to pray not only in church but atother times as well--the praying is nondistinctive--then theattribution typically made is internal, having to do with suchqualities of person-hood as piety or spirituality. Externalattribution about the demand characteristics of churchservices is reserved for the "Sunday Christian" who prays onlyat church.The logic of the attributional theory is that we assumeinternal causation of behavior when no one stimulus seems tobe causing it and yet it occurs consistently (Weary, Stanley,& Harvey, 1989). It follows that attributions will varyaccording to how much information is available concerningnormativeness, consistency and distinctiveness. It isimportant to note that the causal attributions of abnormal28psychology have traditionally been internal. Abnormalbehavior is nonnormative, inconsistent, and distinctive.Internal versus external attribution accounts for thespatial dimension. Where is the cause? Is it inside theperson or outside the person? A second dimension ofattribution is temporal. Is the causative factor stable--which is to say, temporally ongoing--or is it unstable orshort-lived?Weiner, et. al. (1972) has proposed the following model,refined by Hamilton (1983), of attributional process whichincorporates the "spatial" (internal versus external) and the"temporal" (stable versus unstable). This model posits thefour categories of causal attribution. This representation ofattribution theory reflects Heider's original thinking aboutthe crucial distinction between what he called "can" versus"trying", the distinction between ability and effort. Itseems to make intuitive good sense that skills and abilities,as well as effort and motivation, do not, by themselvesproduce outcomes. Outcomes are also determined by situationaldemand characteristics (here called "task difficulty") and byunpredictable and uncontrollable events, or, put simply, luck.This, then, forms a simple model of attributed causation.29Figure 1.Simplified attributional model.Internal^ExternalStable Ability Task difficultyUnstable Effort LuckAny scholar who has struggled to provide a theoreticallycoherent and empirically rigorous account of any phenomenonwill attest to the notion that everyday causal thinking isusually subject to oversimplification. Taylor and Fiske(1975) point out that internal attributions require lessinformation and less keen observation. But whatever thecause, Ross (1977) has described attributing cause todispositions over situations as the fundamental attributionerror. It is, therefore, predictable from all that has beensaid above that therapists will tend to make internalattributions about clients. Mainstream theories of therapy,personality, and change are replete with internal attribution.Furthermore, self-report and observations taken outside ofday-to-day contexts all tend toward external attribution.Relating this notion to the model above, it seems a fairgeneralization that psychological theory is differentiallyoriented more toward notions about skill and motivation thanto task difficulty and the role of chance. The former seemavailable to influence, while the latter seem not to be.30Still, this question of internal attribution is a matterof more or less. In social psychology radical situationalistshave taken an opposite tack (Ross, 1977). Radicalbehaviorists have also rejected reliance on inner causes asexplanations for behavior (Skinner, 1974; Zuriff, 1985). And,more recently, general family therapy theory has emphasizedbehavior as a function of relational interactions as opposedto manifest traits of persons (Goldenberg & Goldenberg, 1989).Going even further in moving attributions outside the person,strategic family therapists such as Erikson and Haley (Haley,1973, 1987) and Madanes (1981) emphasize the role of systemichomeostasis, but also developmental issues as determinants ofbehavior. Even more explicitly external in theirattributions, solution-focussed theorist/clinicians such asEpston & White and (1991) talk about "situating" problemsoutside of persons and describe "externalizing"--talking aboutproblems, such as anxiety, as being outside of and separatefrom the client--as potent clinical techniques.Attribution theory forms, then, what we might call a"meta-theory" or cognitive model (Ivey, Ivey, & Simek-Downing,1987), which accounts for some dimensions of distinction inthe judgment and choice process of persons, and also accountsfor some of the fundamental premises of psychological theories(Schneider, 1973). It also provides a model for stereotypingof person-perception. Hamilton (1983) puts it succinctly:"The term stereotype is in essence a cognitive structuralconcept, referring to a set of expectations held by the31perceiver regarding members of a social group. It is, then,similar to an implicit personality theory, in this case one'sgroup membership being the stimulus cue on which a number ofinferences about the person are based." (p. 105). The modelabove, with its two dimensions of internal versus external andstable versus unstable, is taken to be sufficient to form thetheoretical underpinning of the present study.This model is sufficient only insofar as the currentstudy is concerned with a "cross-sectional snapshot" of thetherapeutic encounter. The study is not concerned withchanges in the therapeutic relationship over time. If it wereto account for those changes, a further dimension might wellbe needed. Other theorists have suggested expanding thecognitive attributional theory to include a third dimensionsuch as global versus specific or controllable versusuncontrollable (Alloy et al., 1984; Lopez, 1983). GordonAllport (1937, 1961), in context of grappling with the problemof individuality, suggests that psychotherapy andpersonological psychology makes a distinction between what hecalls nomothetic science and ideographic science. Nomotheticapproaches are concerned with the study of generalized humanbehavior, while ideographic approaches focus on theunderstanding of particular persons (Korchin, 1976). Sincethe concept of nomothetic understanding subsumes the notion ofcomparative, dimensional, and global attributions whileideography is addressed to the unique, personal, and specific32in human behavior, this author suggests these attributionalconcepts as a third dimension of the model.The complete model is illustrated in the followingfigure.Figure 2.Attributional model including nomothetic vs. ideographicdimensions.Nomothetic^ Ideographic(comparative, dimensional)^(unique, personal)Stable^Unstable^Stable^UnstableIn-ternalCharactertraitsDevelopmentalStagePersonalitydynamicsMotivationlevelEx-ternalSocioculturalEmployment/financialFamilybackgroundPersonalcrisisIn order to operationalize the conceptual model for thestudy of therapist behavior, each of the eight dimensions ofthe attributional model has been assigned a psychologicalconstruct that is both an explanatory construct and apotential focus of therapeutic intervention. Thus, forexample, personality dynamics is classified as an internal,stable, ideographic dimension of causal attribution whilesociocultural milieu is assigned to the external stablenomothetic category.The dimension of nomothetic versus ideographic alsorelates to the process of psychotherapy. As Milgram (1974)points out, a vital dimension of person perception is thetension between information input and the capacity to processthis information. In everyday encounters, especially with33strangers, people formulate nomothetic (general) attributionsabout others which serve to guide behavior through briefencounters or casual relationships. However, as relationshipsdeepen, the attributions made by the perceiver become moreideographic as the individuality of the other person becomespart of the information base of the encounter. Ideally, then,attributions made by therapists at the beginning stages ofpsychotherapeutic relationships would tend to be nomothetic(since more personal information is lacking), but shouldchange as a function of time and successful interaction to amore ideographic dimension concerned with the particulars ofthe client as a unique person. When this does not happen, thetheory above would suggest a lack of functional communication,a specific nonfunctional disorder, or a therapist bias isindicated. Clinicians, in making decisions, tend to use lessinformation than is actually available to them (Korchin,1976). Where this tendency is operative, theory would predicta trend to nomothetic attribution since these dimensions areusually more salient. This, too, can be indicative oftherapist bias when therapist judgment is shown to be based onirrelevant information or a failure to take relevantinformation into account.Innovation in Research DesignThe following sections detail the formulation of aquantitative study of bias in clinical judgment with severalunique features. The design is experimental, such that cause34and effect inferences are possible. As shown above, cognitiveattribution theory is used as a basis such that thedifferences shown can be accounted for in conceptual terms.The design involves repeated measures of the same subject sothat within-subject differences--in contrast to the between-groups comparisons found above--can be explicated. A case-study design employing replication (as opposed to populationsampling logic) is used so as to improve the clinical, asopposed to statistical, reliability of the method. Thepresenting problem and its severity is controlled to eliminatethe possibility of classification error as a confound. Theexperiment is designed to explicate processes in thedecisional behavior of counsellors. Socioeconomic status, theclient variable most reliably known to evoke bias, is used asan independent variable so that its differential effect can beevaluated. And, finally, it is intended that the analogy onwhich the experiment is based is more consistent with in-rolebehavior than the typical case evaluation task so often usedin this area of research.35CHAPTER IIIMETHODOLOGYIntroductionIn this chapter the specific content of the study isgiven. The research question is put in operational terms andthe Q-methodology is described in terms of process, history,and rationale. Sections beyond this describe the subjects andtheir recruitment, the experimental materials, scoring of Q-sorts, and statistical analyses applied to the scores.Purpose of the StudyThis study was developed to demonstrate that clinicaldecision-making, as a possible expression of bias incounselling and psychotherapy, could be studied experimentallyusing attribution theory as a frame of reference. Q-methodology was used to operationalize the theory, inform thescoring and measurement, and as a template for the statisticalanalysis. The specific question was to determine whetherthere was a significant ethnic bias and/or socioeconomic biasin counselling trainees in deciding how to clinicallyintervene in terms of options structured according tocognitive attribution theory. The experimental analogy isthat the trainees were asked to consider descriptions ofclients, imagining all the while that this is done inpreparation for conducting a counselling session. Thetrainees then sort cards labelled with counselling proceduresaccording to a pre-determined quasi-normal distribution from36most preferred to least preferred. Q-methodology was selectedas being appropriate for investigating choice and decisionbehavior, and as a paradigm for studying several interactingdimensions of difference sufficiently subtle that they wouldnot be revealed through direct questioning or observation.Q-MethodologyThe following paragraphs offer an intuitive descriptionof Q-methodology, linking this method to the notions ofmultiple single-case studies and replication logic. Theproperties of the data produced by Q-sorts are discussed withrespect to advantages and limitations in the analyses that canbe applied. The particular application to this study isdescribed.Q-methodology, at a pragmatic level, asks subjects toindicate choice behavior by sorting a pre-determined andlimited range of options according to some pre-determineddimension (usually most preferred to least preferred). Q-methodology is a research tactic particularly apt for thestudy of decision-making (McKeoun, Bruce, & Thomas, 1988).The options presented to the decider can be structuredaccording to some theoretical premise. As well, the method isapplicable to single case studies and small groups since eachsubject's responses, or each pre-selected group of subjects'responses (Rinn, 1961), yield a complete "universe" of scores(Yin, 1989). The sampling logic of using more that onesubject is not to sample a population, as in the traditional37statistical model. Rather, the strategy is to replicate thesame experiment represented by the first test-case or criticalcase.This notion of "replication logic" is critical not onlyto the broad research strategy, it is also crucial to thetechnique. In a typical study in psychology involving aseries of responses to stimuli--such as the items on apsychological inventory--the response given by a given subjectto a given item is assumed to be independent and takes onmeaning in a statistical comparison to the mean of all otherpersons responding to that item. The method is normative. InQ-methodology the items are not independent (only one item canbe ranked first!) and the measurement that gives meaning tothe item is comparison with the mean for all the other itemsfor that person. The technique is ipsative as opposed tonormative.This notion of single-case ipsative study with each itemresponse being non-independent has important implications foranalysis as well. Statistics based on interval data withcentral limits and normal distribution can be applied to theresponses of a given subject. However, the scores of thesubjects on a Q-sort cannot, technically, be pooled directlyfor a valid comparison of groups because the item responsesare not independent. The assumption of normal distributionapplies within the subject's data--the allowable responses arestrictly structured this way--but not across the group ofsubjects. Data can, however, be aggregated on the basis of38pattern analysis, and the technicalities of this procedure areexplicated below.In the present study, each of six family counsellors andeach of six person-centered counsellors completed anexperiment with four conditions, making a total of twelvequantitative ipsative case studies. It was hypothesized thatthe family counsellors would replicate each other, and sowould the person-centered counsellors. Factor analysis is thestatistical tool used to determine whether there are, in fact,replicating patterns of response in the combined scores of allthe subjects.Q-Methodology: History and RationaleIn the review above it was suggested that improvement inresearch in bias might well include research designs that candeal with several independent variables, experimentalconditions, and subject variables where the fundamentalresearch question has to do with choice behavior. Thefollowing section is an explanation of how Q-methodology canbe applied to a research question with many dimensions."Q-methodology" is a label original with WilliamStephenson (1953) who used it to describe this particularapproach to factor analysis. Stephenson was a student ofCyril Burt who conceptualized the possibilities of factoranalytic research as involving comparisons and interactionsbetween and among three general dimensions of psychologicaldifference: types of persons (usually operationalized as39subject variables), traits (usually operationalized as testscores), and occasions (usually operationalized asobservational or experimental conditions). Traditional factoranalysis looks at two dimensions: persons and traits. The twodimensions considered in a longitudinal study of a given traitare persons and occasions. Biographical studies of a givenperson look at traits and occasions; again, two dimensions.In short, most tactics of educational and psychologicalresearch consider two dimensions. Q-methodology allows forthe examination of three dimensions. In the present case thepersons of interest were the different "families" ofcounsellors, the trait of interest was the pattern ofattribution, and the occasions of interest were the fourexperimental conditions. As noted above, other researchtactics used in the area of bias in counselling andpsychotherapy have generally looked at only two dimensions.SubjectsSubjects were selected based on only three criteria: (1)being female (control for gender), (2) being in the final yearof a Master's Program in Counselling Psychology (control fortype and level of training), and (3) being self-described as a"person-centered counsellor" or a "family counsellor".^Theauthor attended several graduate classes in the Department ofCounselling Psychology, Faculty of Education, University ofBritish Columbia using a standard request for volunteers(shown in Appendix A). Subjects were asked to give their name40and telephone number, as well as their counsellingorientation. The final sample consisted of twelve returns.All subjects were female. Important controls were level oftraining and age. These data are summarized in Table 1.Table 1Ages and level of training for subjectsType of Counsellor^N^Median Age Group^MedianNumberof CoursesFamily^ 6^35 - 39^ 6Person-centered^6^30 - 34 5Overall, the subjects ranged from the 25-29 year age group tothe 45-49 year old group. The fewest number of coursescompleted was five; the greatest was nine.Data CollectionOf the thirty-four who gave their names as volunteers,nine subjects describing themselves as family counsellors andnine describing themselves as person-centered counsellorsagreed to complete the experimental task when contacted bytelephone. Each subject was mailed a package of materials tocomplete at home. The package consisted of four clientprofiles (detailed below and shown in Appendix B), the cardsand envelopes for four Q-sorts (one set for each clientprofile), instruction sets, a demographic sheet, and a returnenvelope. The package also contained some coffee, tea, hot41chocolate mix, and a granola bar. The intention behind the"snack" was to thank people who had been generous with theirtime, but also to encourage them to use the time to completethe task seriously.Initially, ten of the eighteen subjects responded withinfour weeks. A letter prompting the task and return of thematerials (see Appendix A) was sent out to those who had notyet returned the items. Two more packages of materials werereturned. The final return rate was a predictable two-thirds.The principal investigator's telephone number was givenin anticipation of problems with a lengthy cognitive taskinvolving a complicated written instruction set and noexemplar (to prevent modelling a particular type of sortingresult). However, only one person called with any difficultywith the instruction set, and that problem was relativelyminor. The subject thought that there might be a distinct setof cards for each client profile. She left the question on ananswering machine, and by the time her call was returned a fewhours later she had reread the instructions and completed thetask. In all cases the subjects had sorted the cards andplaced them in the envelopes in the correct frequencies.Although the logic of the design is such that self-endorsement as a particular type of counsellor should besufficient condition for inclusion in the study, subjects werealso asked basic information about courses taken (as aconfirmation of approximately equal level of training) andage.42Q-Sort Materials and Demographic SheetIn order to create the material for the Q-sort, just over100 specific counselling and psychotherapeutic procedures andinterventions were listed using the tables and indices oftextbooks in counselling and psychology (Corsini, 1984; Egan,1986; Ivey, Ivey, & Simek-Downing, 1987, Kanfer & Goldstein,1986, Korchin, 1976). The author then attempted to sort themas to whether they represented an internal or an externalattribution about the client, and whether they represented astable or unstable attribution about the client. Repeatingthis sort several times revealed that about half the itemsseemed to not be amenable to this sort of procedure. Theremaining 58 items were listed and sent out to ten of theauthor's colleagues with a request that they sort the itemsaccording to the attributional schema and return the results.The sorted lists were then tabulated to determine how reliablyobservers could assign a given procedure or technique to agiven category. This tabulation, along with the completelist, is shown in Appendix C.Forty items for the entire sort would have been ideal.However, the primary consideration was to have enough items togenerate a range of options and variance in response, whilenot including items of poor reliability. Given the necessarytrade-off between "sample size" and item reliability, eightitems per category, thirty-two in total, was taken to be aminimal requirement for the size of the sorting task. In both43the Internal Stable category and the External Unstablecategories eight items of at least 80% observer reliabilitywere available. In the Internal Unstable categorization fouritems of only 60% reliability and three of 70% had to beincluded to make up the necessary eight items. In theExternal Stable category there were only six items altogether,one of which was only 60% reliably sorted. Two more itemswere generated on a rational basis to complete the eight itemsfor the category.Subjects were also asked to complete a brief demographicsheet (shown in Appendix C), checking off their age category,courses in counselling completed, and counselling orientation.This last item was included as a check against possible mis-classification of counselling orientation.Experimental MaterialsEight intake assessment summaries, each representing asummary of a screening interview in a clinic for depression,were generated from the items of the Beck Depression Inventory(Beck, Ward, Mendelson, Mock, Erbaugh, 1961; Beck, 1974).Symptoms within the inventory are classified as indicators ofdepression by appearance, thought content, vegetative signsand psycho-social performance. These factors emerge reliablyin factor analysis of item endorsement. Each sign or symptomwithin the inventory is categorized for depth of depressionfrom none through mild, moderate, and severe. Items ratedmild were listed under their associated factor headings and44each was assigned a number. Following this, one item from eachof the four lists was randomly assigned to one of the fourcase descriptions. Thus, each of eight case descriptions wasconstructed to contain four indicators of depression, one eachfrom each symptom factor group.Eight such profiles were generated and used for the pilotphase. In the study proper, only four fictitious clientprofiles were used. These were selected randomly.The descriptions were completed by assigning eachhypothetical client a name and indicating that each lives inVancouver. A concluding sentence was added to eachdescription indicating that the client had responded toempathic statements from the interviewer indicative ofdepressed affect. The affective adjectives were taken fromthe Lubin Adjective Checklist (DACL forms A & B) (Zuckerman &Lubin, 1965) and each is associated with the experience ofdepression.For use in a validity check, one version of thedescriptions, with no information other than that describedabove, was generated. For use in the experimental conditions,information was added to the "summary" section of a fictitiousdepression clinic intake form. Sections of the form were usedto indicate age, geographic area, ethnicity, and socioeconomicstatus of the client. The experimental version of thedescriptions is presented in Appendix B.In order to validate control for type and severity ofpresenting problem, the stimulus material nominally45representing eight mildly depressed clients, but containing nodemographic information, was presented to five experiencedcounsellors who were asked to rate the descriptions for typeand severity of disorder. Rating was done on a five-pointscale, 1 being mild, 3 was moderate and 5 severe. Responsesranged from 2 to 4 with a mean of 3.64, s.d. = .34. All weredescribed as depressed.The instruction set preceding the experimental materialspresented to the subjects indicated that the summaries ofscreening interviews were taken at a clinic for depressedclients. Subjects were asked to imagine themselves preparingto counsel the client described using the possible activities,techniques and actions described on the Q-sort cards.Subjects were instructed to sort the cards according to thedistribution shown in figure 3.Figure 3.Distribution of preferences in Q-sort task.xXXX^x^XX^X^XX^X^X^X^XX^X^X^X^XX^X^X^X^x^X^XX^X^X^X^X^X^X 1^2^3^4^5^6^7Final Groups for SortingMost Preferred Least PreferredThe cards to be sorted were shuffled before being mailedout and the client profiles were presented in random order.46Subjects also rated each client problem for severity using ascale from one (mild) to three (moderate) to five (severe).The complete instruction set is shown in Appendix B.Introduction to Scoring and Statistical ProceduresThe data were treated in four broad steps. First, thecard sorts were scored and categorized by attribution.Second, analysis of variance was applied to each subject'sscores (to each experiment, as it were), to determine whether,within the linear schema of experimental conditions versusattributional categories, there were significant main orinteractional effects. Third, the results were submitted tofactor analysis (of subjects correlated with every othersubject, not items on items as is more typical) in order todetermine whether there were discernable patterns of responseindicative of replication of the experiments. Fourth, theloadings on the factors were used to generate weightedstandard scores for all the items within the patterns, andthen the items, ranked by standard score, were given Q-scoreswhich were submitted to ANOVA. This last step essentiallytreats the orthogonal patterns as hypothetical subjects in theexperiment and scores and analyzes the responses accordingly,showing how a subject almost perfectly correlated with thepattern would have responded to the experimental condition.47Scoring the Q-SortsQ-scores for the card sorts were arrived at by assigningeach item a score according to the pile in which it wasplaced. Cards in the first pile got a seven, those in thesecond were scored as six and so on down to the seventh pilewhere cards in that category were scored as one each.^Thequasi-normal distribution of the sort, combined with thenumerical scoring, is the chief rationale for using thestatistics of interval scores. Since the items had been pre-categorized according to the attribution represented, thecategories of attribution each contained eight scores.Preference for that particular attribution was operationalizedas the higher the score, the greater the preference.Analysis of Variance for Individual CasesEach subject's scores were analyzed by analysis ofvariance (ANOVA) involving four variables: Ethnicity (Nativeor Caucasian) (ET), Socioeconomic Status (poor or middleclass) (SS), Internal/External attribution (IE), andStable/Unstable attribution (SU). Since each of the foursorts involved a complete Q-sort, there could be no differencein mean or variance from one experimental condition toanother. All effects of the experimental variables (i.e. thedemographic characteristics of the hypothetical clients) inthe ANOVA would be found in the interactions with theattribution variables.48Factor Analysis of Pooled CasesFactor analysis was completed in three stages. First,each subject's responses were correlated with every othersubject's responses. This was done once for each of the fourexperimental conditions, once for each Q-sort. Then, second,the matrix of intercorrelations was submitted to factoranalysis such that, relative to the typical way of doing theprocedure, the subjects were treated as variables and theitems were treated as observations. Third, a principal axissolution was obtained which was submitted to Varimax rotationto achieve the simplest orthogonal structure criterion. Basedon the data from the actual sorts, this procedure essentiallyconstructs hypothetical types of persons responding to theexperimental task. This procedure also indicatesquantitatively, through the factor loadings, how the responsesof the subjects are patterned, which subjects adhered to whichpattern, and to what degree.Analysis of Variance for Hypothetical TypesEach hypothetical type represents a pattern of sorting.To determine the content of this pattern, a weighted score foreach item was calculated based on the following algorithm.For each subject whose pattern of scoring correlatedpositively (.5 or higher) with the factor, each of theoriginal raw item scores was multiplied by a weighting factor.The weighting factor was calculated as the factor loading for49the subject divided by one minus the square of the loading.The formulae are as follows:r = loadingweight = r/(1 - r2)weighted score = (raw item score) x (weight)(Conceptually, the procedure is essentially the reverse of aleast squares fit.) The weighted item scores for all thesubjects correlated with the factor were added together.These sums were then transformed into standard scores, or Z-scores. A positive Z-score indicates to what degree a givenitem was preferred, relative to the average item, within thatpattern. A negative Z-score indicates the degree to which anitem was not preferred relative to the average item in thatpattern. Z-scores allowed the items to be ranked from most toleast preferred and the Q-scoring and ANOVA procedure abovewas applied to the response pattern of the hypothetical type.The choice to treat loadings of .5 or higher asrepresentative of the pattern associated with a given factor,and to disregard loadings less than .5, requires someexplanation. As Rinn (1961) points out, there are no "hardand fast rules" in this area, but the choice of a cut-offpoint has some analogy to the choice of levels of statisticalsignificance. If small factor loadings (representing Q-scoresets with relatively more variance from the hypothetical typerepresented by the factor) are included in the calculation,then greater and greater amounts of ramdom variance areintroduced into the definition of the pattern. This, in turn,50increases the probability of a Type II error, which is to say,failing to discriminate one pattern from another.The cut-off is also related to the discrimination powerof the Q-sort categories. In the present study, thecategories, taken overall, are about 75% reliable asdemonstrated in the observer reliability study (see appendixC). This leaves about 25% or variance attributable to randomor unspecified factors. Since loadings of .5 can also beinterpreted as indicating that the factor accounts for about25% of the variance in the subject's scores, then .5 can alsobe taken to be the outer limit of the discriminating power ofthe measurement instrument. Still, it is important to notethat in other applications of Q-sorting, where the items arenot organized into categories of limitied reliability, thediscriminating power of the method is not so limited and muchsmaller loadings can be taken into account in definingpatterns.Determining and Comparing Pattern ContentFor purposes of determining which specific counsellinginterventions were preferred and not preferred within a givenpattern, Z-scores of greater than 1 were taken to indicatevery positive preferences; less than -1 were taken to indicatethose least preferred. In this way the content of most andleast preferred items for each hypothetical type wasspecified. The difference between the Z-score associated withan item on one pattern or factor and the Z-score for the same51item on another pattern also indicates the degree anddirection of difference in the preference shown for that itemin comparing any two patterns.Descriptive Statistics and Secondary AnalysesMeans and standard deviations for the subjects'attribution scores and their estimates of the severity of thedepressions for the hypothetical clients were calculated.Very much as a secondary or adjunct analysis, severity scoreswere submitted to ANOVA by type of therapist, age category,and level of training. Similarly, ANOVA was performed on theentire sample as per the procedure used with each subjectexcept that type of therapist (TH) was included as a variable.52CHAPTER IVRESULTSIntroductionThis chapter gives the results. It is organized fromsimplest description to most complicated results, which is tosay from the least transformation of the data to the greatest.Means and standard deviations for each cell in the Q-scoringmatrix are presented first. Then ANOVAs for individual casesare summarized. Following this, the factor analytic resultsand the analyses of the hypothetical types are presented.Tables showing the counselling interventions associated withthe hypothetical types are then presented and the significantdifferences between the content of the patterns is presented.A final section reports on the subjects' estimates of theseverity of the depression for the hypothetical clients.Q-SortsThe means and standard deviations of the Q-scores byvariable and by subject are shown below. Because the task isa forced-choice forced-distribution model, the mean for anysubject's results on the four variables, Stable (ST), Unstable(UN), Internal (IN), and External (EX), or on the fourpossible combinations is always the same, in this case 4.Therefore, scores over 4 indicate a preference, while scoresless than 4 indicate that this category is not preferred.These results are shown in the following tables.53Table 2Means and standard deviations by attributional variablesfor family therapists. Variables are Internal (IN), External(EX), Stable^(ST) and Unstable (UN) and their combinations.Subj.^#Variable1Mean S.D.2Mean S.D.3Mean S.D.InSt 3.25 2.44 4.28 2.61 3.25 2.74InUn 3.75 2.78 3.28 2.86 3.84 3.02ExSt 5.34 3.10 5.25 2.86 5.72 2.06ExUn 3.66 2.32 3.22 2.27 3.13 1.21UN 3.70 3.62 3.25 3.65 3.48 3.26ST 4.30 3.95 4.77 3.87 4.48 3.43IN 3.50 3.70 3.78 3.87 3.55 4.08EX 4.50 3.87 4.23 3.65 4.42 2.39Subj. # 4 5 6GrandVariab Mean S.D. Mean S.D. Mean S.D. MeanInSt 3.53 2.29 3.72 2.81 3.38 3.96 3.57InUn 3.03 2.47 3.47 2.45 3.47 1.84 3.47ExSt 5.13 2.22 5.34 2.32 5.28 2.42 5.34ExUn 4.31 2.20 3.38 3.10 3.75 1.91 3.57UN 3.67 3.31 3.42 3.95 3.61 2.65 3.52ST 4.33 3.19 4.53 3.64 4.33 4.64 4.46IN 3.28 3.37 3.59 3.72 3.42 4.37 3.52EX 4.72 3.12 4.36 3.87 4.52 3.08 4.4654Table 3Means and standard deviations by attributional variablesfor person-centered therapists. Variables are Internal (IN),External^(EX),combinations.Stable (ST) and Unstable (UN) and theirSubj.Var.#^7Mean S.D.8Mean S.D.9Mean S.D.InSt 4.63 3.11 5.25 2.40 4.63 2.36InUn 4.28 3.17 4.25 2.40 3.84 3.66ExSt 4.25 2.87 4.25 2.40 4.69 2.28ExUn 2.75 2.16 2.25 1.94 2.84 2.43UN 3.52 3.84 3.25 3.08 3.34 4.39ST 4.44 4.23 4.75 3.39 4.66 3.28IN 4.45 4.44 4.75 3.39 4.23 4.35EX 3.50 3.59 3.25 3.08 3.77 3.34Subj. 10 11 12GrandVar. Mean S.D. Mean S.D. Mean S.D. MeanInSt 3.75 3.93 4.16 2.53 4.88 2.11 4.55InUn 3.19 2.33 3.34 2.65 3.63 1.98 3.76ExSt 4.91 2.74 5.16 3.19 5.25 2.40 4.75ExUn 4.16 2.08 3.34 2.65 2.25 1.94 2.93UN 3.67 3.13 3.34 3.75 2.94 2.77 3.34ST 4.33 4.79 4.66 4.07 5.06 3.19 4.65IN 3.47 4.57 3.75 3.66 4.25 2.89 4.15EX 4.53 3.44 4.25 4.15 3.75 3.08 3.84Putting aside the issue of statistical significance for amoment and simply inspecting the means seems to indicate a55Putting aside the issue of statistical significance for amoment and simply inspecting the means seems to indicate apattern. The family therapists consistently score over fiveon the external-stable category, showing consistentpreferences for the stable and external dimensions of themodel. The case of the person-centered therapists is notquite so clear. A high score is found consistently on thestable dimension, with low scores consistently found on theunstable dimension and in the external-unstable category.Since item scores are not independent, any data aggregated bysimple addition, as in the case of the grand means shown inthe table, should be interpreted with extreme caution, usefulonly as very rough descriptions. However, quantitative testsof statistical significance can be applied to each subject'sdata, as detailed in the next section.Analysis of Variance for Individual CasesAn analysis of variance was performed on the data matrixfor each subject. The main effects and significantinteractions are summarized in Table 4. Where the maineffects were significant, the mean differences on theattributional dimensions were also significant (p < .05).Both groups of counsellors preferred stable attributions tounstable attributions.^The family counsellors preferredexternal attributions and the person-centered counsellorspreferred internal attributions. Similarly, where theinteractions of the two attributional dimensions were56significant, so were the mean differences (P < .05), with thefamily-oriented trainees preferring external-stable choicesand the person-centered counsellors preferring internal-stableoptions and showing a negative preference for the external-unstable category.Only one subject showed significant effects attributableto a client variable (socioeconomic status), and this subjectwas the only family counsellor for whom the IE by SUinteraction was not significant.Table 4Summary of significant effects found in ANOVA done on eachsubject's Q-scores.Effect^ Family^Person-Counsellors^CenteredCounsellors(N= 6)^(N = 6) Main Effects (p < .05)Internal/External (IE)^5 4Stable/Unstable (SU) 6^6Two-Way Interactions (p < .05)IE by SU^ 5^4Socioeconomic Status by IE^1 0Socioeconomic Status by SU 1 0No significant interaction with EthnicityNo significant 3-way or 4-way interactionsANOVA tables showing the effects and interactions foreach subject are given in Appendix E.57Factor Analysis of Pooled CasesThe rotated factor matrices for the factor analyses ofeach subject correlated with all other subjects are shown inTables 5 to 8, one table for each experimental condition. Allthe tables show the varimax rotation with Kaisernormalization.Table 5Rotated factor matrix for Poor Native condition.Loadings greater than .5 are shown by **.FACTOR^1 FACTOR^2 FACTOR^3Eigenvalue 5.75926 1.57807 1.05677Pct of Var 48.0% 13.2% 8.8%Cum Pct 48.0% 61.1% 70.0%Sub. #1 .71237** .18837 .22832Family 2 .71674** .20348 .364603 .69470** .13074 .26354Counsellors 4 .78985** .18106 -.167615 .58284** .53431** .030106 .79661** .30810 .202967 .02674 .80648** .33032Person- 8 -.00079 .08807 .88647**Centered 9 .21614 .44331 .61458**10 .39031 .75877** -.00197Counsellors 11 .37405 .70002** .2578512 .41914 .18155 .80769**58Table 6Rotated factor matrix for Middle Class NativeLoadings greater than .5 are shown by **.condition.FACTOR^1 FACTOR^2 FACTOR^3Eigenvalue 5.87422 1.74050 1.20238Pct of Var 49.0% 14.5% 10.0%Cum Pct 49.0% 63.5% 73.5%Sub. #1 .78670** .05861 .21067Family 2 .79619** .27488 .337113 .66594** .36214 .31432Counsellors 4 .77905** -.02283 -.028645 .78063** .28451 .125396 .58214** .17837 .563507 .09286 .41106 .77546**Person- 8 -.03665 .95036** .02321Centered 9 .30854 .80314** .2339110 .23574 -.06399 .76955**Counsellors 11 .50116** .17907 .60842**12 .46582 .78745** .1732859Table 7Rotated factor matrix for Poor Caucasian condition.Loadings greater than .5 are shown by **.* marks loadings greater than .45, but less than .5.FACTOR 1Eigenvalue^5.87422Pct of Var 49.0%Cum Pct 49.0%Sub. #1^-.27343Family^2 .47840*3^.27385Counsellors^4 .114735^.340196 .23727FACTOR 2 FACTOR 3 FACTOR 41.74050^1.20238^1.0392614.5% 10.0% 8.7%63.5%^73.5%^76.1%.81560**^.09270^.14994.15208^.46123*^.36693.82921**^.09592 .08793.06300^-.83674** -.05759.69654** -.01987^.12916.56308**^.64718** -.015927Person-^8Centered 910Counsellors 1112.20886.94601**.81526**.29845.08950.79706**.02485^-.00567^-.11441^-.08463.19517 .03252.46494*^.51351**.46578*^.27150.35131 .20345.92036**.09876.18713.38473.68807**.1076560Table 8Rotated Factor Matrix for Middle Class Caucasian condition.Loadings greater than .5 are shown by **FACTOR^1 FACTOR^2 FACTOR^3Eigenvalue 5.39961 1.68236 1.33054Pct of Var 45.0% 14.0% 11.1%Cum Pct 45.0% 59.0% 70.1%Sub. #1 .84306** .16496 .05594Family 2 .38492 .58003** .239223 .72699** .35915 .25369Counsellors 4 .62332** .36233 -.151935 .85909** -.03550 .281366 .64816** .27221 .295097 -.12959 .82157** .29005Person- 8 -.07933 .23132 .90449**Centered 9 .27054 .02907 .80424**10 .33372 .70572** .14085Counsellors 11 .43822 .73505** -.0296512 .44332 .30053 .66419**The tables show that the factor analysis generallyproduces three groups, one composed of family counsellors andthe others made up of two kinds of person-centered counsellor.In three of the four scenarios, all of the family counsellors,save for subject #2 in the Middle Class Caucasian condition,load onto a single factor. In the case of the Poor Caucasiancondition, four or the six maintain a pattern, one of these,subject #6, becoming a mixed type, while subject #4 loadednegatively onto another factor and subject #2 wasundifferentiated. This pattern will be referred to as PatternFC, with Factor 2 in the Poor Caucasian condition taken torepresent this pattern, while Factor 3 of the Poor Caucasiancondition will be referred to as Pattern FC2.61Where the person-centered counsellors are concerned,subjects #8, #9, and #12 were consistently loaded onto thesame factor with comparatively high loadings. In the case ofthe Poor Caucasian condition, where the coherence of thefamily counsellor pattern seemed to break down, these threecounsellors loaded onto a factor accounting for 49% of thevariance. This pattern will be referred to as PC1. Subjects#7, #10, and #11 also tend to be loaded onto the same factor--except in the case of the Poor Caucasian condition whereSubject #10 seems to load onto both factors favoured by thefamily counsellors. This pattern will be referred to as PC2.Overall, it can be said that there are three patterns,FC, PC1, and PC2, with the family counsellor pattern splittinginto two under the Poor Caucasian condition.Analysis of Variance for Hypothetical TypesThe factor loadings were used to calculate weightedscores for each item for each of the thirteen factors. Theweighted scores were then converted to standard or Z-scores,sorted from most to least preferred, and then each factor orpattern was scored as if it was a sort done by a hypotheticalindividual. Analysis of variance was then performed on thesedata. The ranked items and associated Z-scores, along withcell means and ANOVA tables, are presented in Appendix D.Table 9 lists the main effects and interactions that werefound to be significant across all the ANOVAs.62Table 9Significant effects and trends in ANOVA's performed onpatterns formed by hypothetical types.Experimental Main Effect^PATTERNCondition or Interaction FC PC1 PC2Poor Native Internal/External^(IE) .003 .010Stable/Unstable (SU) Trend .000 .051(.063)IE by SU .019 Trend(.077)Middle Internal/External^(IE) .008 .004ClassNative Stable/Unstable (SU) .008 .001 Trend(.083)IE by SU .000Poor Internal/External^(IE) .005 .007 TrendCaucasian (.069)Stable/Unstable (SU) Trend .000 Trend(.106) (.109)IE by SU .005Middle Class Internal/External^(IE) Trend .008Caucasian (.078)Stable/Unstable (SU) .026 .001 .016IE by SU .014FC - Family Counsellor PatternPC1 - Person-Centered Counsellor Pattern #1PC2 - Person-Centered Counsellor Pattern #2In summary, the FC group shows significant effects forall dimensions of the attributions. PC1 pattern shows maineffects but not interactions. PC2 responds primarily to thestable/unstable dimension. The FC2 factor, found only underthe Poor Caucasian condition had significant main effects for63IE (P < .014) and SU (p < .026) but the interaction was atrend (p < .078). The FC2 pattern was similar to the PC1pattern. Complete results of the ANOVAs for the hypotheticaltypes are found in Appendix E.To this point in the analysis the identification of threepatterns operating across all four experimental conditions (asopposed to twelve separate patterns) was based on similarityof main effects as well as the same subjects consistentlyloading onto the factors. As a formal test of the consistencyof the patterns across experimental conditions, the Q-scoresassociated with each pattern were correlated with all theother Q-scores for all the other patterns. The statisticalanalogy here is multimethod-multitrait analysis, except thatit is, in the situation in hand, a multipattern-multiconditionanalysis (Campbell & Fiske, 1959). The complete table isshown in Table 10. If the patterns are consistent across theexperimental conditions, then the "multicondition-unipatterntriangles" should contain large, statistically significantcorrelations. All of the "triangles" contain correlationssignificant at the p < .001 level. Although consistentlysignificant at this high level, correlations involving eitheror both the FC2 pattern or the Poor Caucasian condition weresmaller in magnitude.If the patterns are distinct from each other, then thecorrelations outside the triangles should be small, and notstatistically significant. Of the forty-nine correlationsoutside the "triangles", nine reached significance, all lessTable 10Correlations of Q-scores of each hypothetical type with theQ-scores of each other type, experimental condition byexperimental condition.PATTERN:^FCCONDITION: PNFCMNFCPCFC2PCFCMCPC1PNPC1MNPC1PCPC1MCPC2PNPC2MNPC2PCPC2MCPNFCMNFCPCFCPCFC2MCFCPNPC1MNPC1PCPC1MCPC1PNPC2MNPC2PCPC2MCPC21.00 .89**1.00.90**.90**1.00.72**.71**.69**1.00.81**.90**.87**.60**1.00.58**.56**.56**.63**•45*1.00.65**.58**.61**.65**.51*.90****.68**1.00.52*.50*.48*.57**.38.85**.85**.82******.93**.97***.95**.95**1.00**N of cases: 32^1-tailed significance: * - .01** - .001Conditions: PN - Poor Native^MN - Middle Class NativePC - Poor Caucasian MC - Middle Class CaucasianPatterns:^FC - Family Counsellors PC - Person-CenteredCounsellors65than .65 in magnitude, and all but two were correlationsbetween the FC and the PC1 patterns.Determining and Comparing Pattern ContentAppendix D contains the lists of clinical options mostand least preferred by each hypothetical type listed factor byfactor, experimental condition by experimental condition. Asshown above it is also possible to compare the patterns bylooking at the differences between the standard scores for agiven item. The following table lists differences of at leastone standard score. When the standard scores for the PC1 andthe FC patterns were compared, there were no items for whichthe difference in the scores exceeded one.Table 11Items preferred by FC over PC2Item^Z-scoreItem^ Type DifferenceTracking interactional patterns^EU^1.89Constructing a "Life Line" EU 1.45Mediating communication between familymembers^ EU^1.25EU - External/Unstable AttributionES - External/Stable AttributionIS - Interal/Stable AttributionIU - Internal/Unstable Attribution66Table 12Items preferred by PC2 over FCItem^Z-scoreItem^ Type DifferenceAnalyze cognitive style and cognitivedistortions^ IS^2.02^Discover and correct irrational ideation IS 1.71Thought stopping IU^1.40Client self-monitoring of thoughts andassociated emotions^ IU^1.04EU - External/Unstable AttributionES - External/Stable AttributionIS - Interal/Stable AttributionIU - Internal/Unstable AttributionTable 13Items preferred by PC1 over PC2Item^Z-scoreItem^ Type DifferenceAnalysis of characterological type^IS^2.17Directing client to change questionsto statements^ IS^1.59Values clarification IS 1.20Free association IS^1.06Analysis of developmental stage^IU 1.65Analysis of resistance^ IU^1.10EU - External/Unstable AttributionES - External/Stable AttributionIS - Interal/Stable AttributionIU - Internal/Unstable Attribution67Table 14Items preferred by PC2 over PC1Item^Z-scoreItem^ Type DifferenceGenerating a specific behavior problemdefinition^ EUSetting specific, observable, measurablegoals for counselling outcome^EUCollect a family history^ ESMultigenerational pattern analysis^ESAnalyze cognitive style and cognitivedistortions^ IS1.971.661.351.221.07EU - External/Unstable AttributionES - External/Stable AttributionIS - Interal/Stable AttributionIU - Internal/Unstable AttributionSeverity of DepressionThe estimates of the severity of depression given by eachsubject for each hypothetical client were examined by analysisof variance to determine whether these estimates were afunction of the type of client or whether they were a functionof the type of counsellor trainee or a function of otherdemographic variables. There is a main effect for age, aninteraction between level of education and type of therapist,with person-centered therapists who had taken fewer graduateclasses nominating higher levels of depression. The three-wayinteraction refers to the higher estimates of depression fromthe younger, person-centered therapists who had taken fewercourses.68Table 15Analysis of variance for subject's estimation of the severityof depression for hypothetical clients.Severity of Depression EstimatesBY Hypothetical client (CL)Level of Education (ED)Family versus Person-Centered counsellor (FP)Age (AG)Source of Variation^Sum of^MeanSquares^DF^Square FSignifof FMain Effects 5.385^6^.898 1.366 .283CL .494^3 .165 .251 .860ED .214^1^.214 .326 .576FP .043^1 .043 .065 .802AG 4.510^1^4.510 6.866 .018*2-way Interactions 5.656^12^.471 .718 .717CL^ED .319^3 .106 .162 .920CL FP .339^3^.113 .172 .914CL^AG .964^3 .321 .489 .694ED FP 3.257^1^3.257 4.958 .040*ED^AG .077^1 .077 .117 .737FP AG .001^1^.001 .001 .9733-way Interactions 4.608^10^.461 .702 .711CL^ED^FP .506^3 .169 .257 .856CL ED AG .122^3^.041 .062 .979CL^FP^AG .689^3 .230 .350 .790ED FP AG 3.200^1^3.200 4.872 .041*4-way Interactions .433^2^.217 .330 .724CL^ED^FP .433^2 .217 .330 .724AGExplained 16.083^30^.536 .816 .696Residual 11.167^17^.657Total 27.250^47^.580Effects significant at p < .05 are shown by a *69CHAPTER VDISCUSSIONIntroductionThe chapter is organized as a commentary, first, on thedata produced. Then the shortcomings and limitations of thestudy are discussed. Following this, interpretive commentsare given with respect to the major findings. Finally,implications for theory and research are discussed.The fundamental research question was: Is there an ethnicor judgment bias in the selection of counselling interventionsfor counsellor trainees given a choice of actions structuredaccording to cognitive attribution theory? If we take bias tobe simple difference attributable to these variables, then theanswer is yes, there is evidence of an interaction of theethnic and socioeconomic variables to produce a discernabledifference in how the clients are perceived in terms ofattributional dimensions. The attributional pattern for theexperimental condition in which the hypothetical client waspoor and Caucasian was different from the other conditions.However, if we take a more subtle and complex view of bias,then it is not at all clear that the results suggest a bias inthe sense of ethical violation. The results do, however,define dimensions of difference.Persons, Traits, and OccasionsUnder the design of this study, data are produceddescribing different types of counsellors (persons), defining70their traits (attributional preferences) and differentiatingthese by their response to different experimental conditions(occasions).^Data have been presented to indicate thatdifferent types of counsellors demonstrate different traitsunder different conditions. This result, it seems to thiswriter, not only supports this type of research design, butalso provides commentary on other research strategies asbeing, literally, two-dimensional. As Ivey, Ivey, & Simek-Downing (1989) so succinctly put it, the clinical question forthe practicing therapist is almost always: Which interventionfor which client when?The Discrimination of TypesAs was suggested in the literature review, there has beena tendency in designing bias research to assume that alltherapists are members of the same normally distributedpopulation and that individual differences among therapistscan be controlled statistically usually by resort to verylarge sample sizes. In the present study, there was anexplicit hypothesis that there were at least two "types" oftherapists, and that these therapists would differ in theirchoices of intervention, and that the differences could beaccounted for abstractly through the constructs of cognitiveattribution theory. The data support this contention, exceptthat it is clear that while there is only one "type" of familycounsellor, there are two discernable types of person-centeredcounsellors. The data indicate that, in general, the family71counsellors select interventions that are, in theattributional classification, External, External-Stable andnot Internal. One group of person-centered counsellors, whohave been labelled PC1, prefer interventions with the a prioriclassification of Stable and indicate that they would notemploy interventions classified as Unstable or External. Thesecond group of person-centered counsellors are, within theattributional framework, systematic in their choices onlyinsofar as they choose interventions classified as Stable.These findings are important at least two ways. First,they demonstrate that there are identifiable subgroups withinthe population, and that the differences between these groupsis nonlinear. This suggests that the statistical design ofstudies comparing groups of therapists treats differences inpatterns as variability and error rather than real difference.This, then, will tend to mask differences, to produce Type IIerrors. A second important dimension of this finding is that,in discriminating the groups, the negative choices, thenomination of counselling options least preferred, was crucialto discriminating the types. This, then, supports Lopez's(1989) argument that many sources of bias may well goundetected because we look at only commissions and not atomissions in choice and selection behavior.Cognitive Attribution as a Decisional ModelWhile it did not account for all the data, cognitiveattributional theory as a framework for accounting for72clinical decision-making received good support in this study.In the four factor analyses performed, the scores based onattributional dimensions resulted in patterns--which is to sayrelatively simple factor solutions--that accounted for over70% of the variance. However, this leaves just under 30% ofthe variance to some other source. And, of course, thesorting of counselling procedures into attributionalcategories will bear other interpretations as will bediscussed below. Still, the general result is that cognitiveattribution theory, as operationalized in this study, wasuseful in accounting for a large proportion of the variance.Situations as Evocative of DifferenceThe experimental condition that produced patterns ofresponse different from all the others was the Q-sort done forthe hypothetical Poor Caucasian client. Under this condition,the trainees representing the PC1 pattern became much moreclear in their choice-making, accounting for 41% of thevariance, whereas in the other conditions they consistentlyaccounted for less than 12% of the variance. The PC2 patternwas reduced to only two of the twelve subjects, and thisseemed to be more of a choice of cognitive interventions thanany single attributional theme: five of the six positivechoices with a standard score greater than one contained adirect reference to client thought process. The familycounsellors retained their positive emphasis on External andExternal-Stable attributions, but broke into two discernable73patterns, with one group distinguishing itself by clearlychoosing against using techniques classified as Internal. Aswell, in the Poor Caucasian condition, one-third of thesubjects became "mixed types", having correlations of greaterthan .5 with more than one pattern. One subject's pattern wasdistinguishable only as being the opposite of (loading factorof -.84) the FC2 pattern. To try to put it all more simply,the PC1 group stood their ground, a few family counsellorsremained consistent, but the rest of the subjects seemed toeither "jump ship", change their decisional criteria, orbecome undifferentiated. This condition contrasts clearlywith the other three experimental conditions where, with minorvariations, the same subjects accounted for the same patternswhich, in turn, accounted for about the same amount ofvariance.Clearly, something happens to decisional process when thePoor Caucasian client is encountered. Subjects tighten uptheir decisional strategy, loosen up, or change. No one, itseems, was unaffected. Hypotheses that may account for thisresult are discussed below.Severity of DepressionIt is interesting to note that the severity of depressiondata seemed to be unrelated to the fictional client scenarios,to the point where the F ratios for the Ethnicity andSocioeconomic status were almost equal to one. The age of thetherapist seemed to matter here, with the younger group seeing74the hypothetical clients' difficulties as more severe, andthis was exacerbated if the therapist was not only young, butalso a person-centered counsellor, and also had takenrelatively fewer advanced courses in counselling psychology.This finding is consistent with the common-sense notion thatbeing more mature, being better educated, and differentiallyattending more to situational factors (as opposed todispositional factors) tends toward more conservativejudgements of levels of pathology.Some Critique of the StudyThere are a number of dimensions of critical examinationthat bear on the validity of the study, its results, and theirinterpretation. They are presented below in what amounts to alist in which the objection is stated and a response follows.Note that the objections are not "straw men" to be disposedof, nor is there a comprehensive or inclusive rebuttal to allthe possible short-comings of this study.The fundamental logic of the study, that of replicationas opposed to population sampling, is flawed. This is reallya question in philosophy of science, and more particularly indefining the necessary and sufficient conditions for validinference. A principal objection to single-case research isthat the results are not generalizable, which is to say,inductive logic cannot be applied to extrapolation from theresults. To begin to answer to this objection, it isimportant to note that all generalizations in science proceed75from the specific to the epistemic based on the notion ofsamples being probabilistically representative of theirpopulations. What is different about this study is that theextrapolation, or generalization, is done a priori, "up front"as it were, with the prediction of replication based on theselection of particular subject variables, in this caseorientation to counselling and psychotherapy. The design ofthe study is, in this sense, an explicit empirical test of thegeneralizability of a theoretical proposition.The experimental analogy, that of reviewing patentlyfictitious and very sparse descriptions of clients, and thenranking pre-determined options for interventions, is unrelatedto actual in-role therapist behavior. This criticism strikesat the core of Abramowitz & Murray's (1983) objection toanalogue studies in this area. They argue that the contrivedconditions produce contrived results, especially resultscontaminated by social desirability effects. In the presentstudy, the only safeguards against social desirability effectswere the confidentiality of the task, done in the privacy ofthe subject's home, and the fact that the criteria by whichthe interventions were included in the sort were notdisclosed, nor were they obvious. The experimenter was ableto speak with about half of those returning the completedtasks, and none of them had any notion as to the inclusioncriteria for the Q-sort. The task was intended to be verysimilar to reviews of "intake summaries", a task routinelydone by counsellors in agency practice. The task was clearly76in the nature of a "role play", which might invite a frivolousresponse, yet the data indicate that the sorting task wasanything but random. It seems a fair inference that,contrived or not, the subjects took the task seriously. Themore or less obvious difference in the experimental conditionsrelated to Ethnicity and Socioeconomic Status invites socialdesirability bias, but also serves as a very conservativecontrol for Type I errors.It is certainly possible to argue that a Q-sort justisn't a good analogy to "real life" choice or decisionalbehavior. This criticism accepted, then the study becomes alaboratory artifact. It is arguable, on the other hand, that,like a projective technique, the sorting task is an externalscreen on which to project internal processes.The Poor Caucasian condition may have produceddifferential results, but that is because the condition wassimply different with a difference that cannot be attributedto the experimental variables. This possibility cannot becategorically ruled out. However, whatever the differencewas, it seemed to affect all of the subjects. As has beenshown, both a priori and post hoc controls for level ofperceived pathology do not account for the difference.Neither is there a question of order effects. However, it ispossible that this particular client description somehowevoked the different response. The only truly adequate testwould have been to run the experiment on a large sample ofsubjects, but without altering any of the independent77variables from condition to condition. This check was onlypartially done (the larger study being quite beyond the scopeof this investigation), with experienced counsellors asked toverify type of disorder and level of severity, and this testdemonstrated more than adequate control for these factors.Another tactic would be to follow the lead of Abramowitz &Herrera (1981) and statistically control for type and severityof disorder.The assertion that the results demonstrate cognitiveattribution theory rests on very poor reliability of assigningitems to attributional categories; consequently, results areof dubious reliability and may not reflect the theory at all.This criticism is certainly well-founded. However, it doesnot imply a categorically negative problem for reliability.The reliability of the categorizations was lower than had beenhoped for. This meant that there was more heterogeneity tothe sort task, making the task more life-like and theunderlying structure even less obvious. This low reliabilityof item classification would also tend to add to randomvariance, thus decreasing the statistical power of theanalysis. In spite of these factors, the attributionaldimensions seemed to account for the preponderance of thevariance both in the ANOVAs and factor analyses. Equallytrue, however, is the assertion that the theory did notaccount for all of the results, and in the case of someindividuals on some occasions, it accounted for very little oftheir behavior.78The whole study is an elaborate tautology because thefamily-oriented therapists chose interventions particular totheir theory, while the person-centered therapists chosepreferences particular to their theory, and the attributionalpatterns of the two are obviously different. The study simplydemonstrates a difference that was obvious from the start.This criticism is important, but it has only a relativelynarrow relevance. To begin with, it is not altogether clearthat a therapist trainee's statement of adherence to aparticular body of theory will translate into actual choicebehavior. For instance, a study by Sloan et al. (1975) (citedin Ivey, Ivey, & Simek-Downing, 1987) demonstrated thecounter-intuitive result that behaviorally oriented therapistsdemonstrated higher levels of empathy, self-congruence andinterpersonal contact relative to humanistically orientedtherapists. The design of this study closes some of the gapbetween stated intention and actual behavior. It was also notan expectable result that there would be two kinds of person-centered therapists, and that these two patterns would be asuncorrelated as the data shows. Furthermore, tracking theattributional pattern of individuals from experimentalcondition to experimental condition seems to show that forsome individuals their pattern of clinical decision-making issituationally specific rather than being closely tied to atheoretical position.None of these data has anything to say about therapistinteraction with clients, or about the client's response to79the interactions, or even whether the therapist trainees, onceengaged in the emotion-laden arena of counselling and therapy,would do anything like what they said they would do in thedetached, highly cognitive situation of doing a Q-sort. Theresponse to this statement is, simply, to endorse it. It wasbeyond the scope of this study and its experimental analogy todeal with these very important issues. Rather, this study isaimed at explicating processes happening in the therapistprior to interaction with the client.Two Kinds of Person-Centered CounsellorsIt was hypothesized that two kinds of counsellors wouldemerge from the factor analysis. There were, instead, three.The family counsellors formed one group, but the person-centered counsellors formed two groups. The person-centeredcounsellors had this much in common. Both groups preferredattributions that were classified as Stable. However, the Pd1group also seemed to know what they did not like, and that wasUnstable or External items. Knowing not only what you like,but also what you do not like, seemed to be an advantage fromthe point of view of treating all the hypothetical clients thesame. It was this group that did not change or disperse theirpattern of response under the Poor Caucasian condition.It is clear that greater methodological rigor, or aparticular kind, would have produced the hypothesized two-factor solution. Selecting subjects based on the rather loosecriterion specified above--female, in second year of graduate80training, and self-described as "person-centered"--did notproduce a unitary pattern of clinical choice-making. Inretrospect, it is more than possible that specifying selectionbased on such labels as "Traditionally Rogerian" to capturewhat has been called the PC1 group or "Person-Centered/Cognitive" to recruit the PC2 group would haveproduced homogenous results. However, it is not necessarilytrue that the therapist trainees would have found such adefinition meaningful, that they could reliably and with self-awareness self-select into such categories. The test implicitin the methodology of the study was to determine whether self-nomination as a person-centered therapist would indicatemembership in a homogenous group with respect to patterns ofclinical decision-making. The answer is, quite emphatically,that the group of trainees who call themselves person-centeredare not a homogenous group by the cognitive attributionalcriteria of this study. This is an important finding insofaras it tells consumers of counselling services that this label,person-centered therapist, is not necessarily descriptive of aunitary body of clinical practice.Individual DifferencesThis study makes it possible to track the pattern ofresponding for individual subjects from one experimentalcondition to another. As noted in the results section, mostof the subjects adhered to a pattern that was highly81correlated across experimental conditions. However, therewere some marked exceptions.Subject #2 responded as part of the PC2 pattern for theMiddle Class Caucasian condition (r = .58), was a "mixed type"under the Poor Caucasian condition (factor loadings of almost.5 on both the PC1 and FC2 patterns), clearly adhered to theFC pattern where the Middle Class Native was concerned (r =.80), and was also with the FC group for the Poor Nativecondition (r = .72). Analysis of variance for this subject'sscores found no significant effects for the Ethnicity orSocioeconomic Status variables. However, "tracking" thesubject from one pattern to another as experimental conditionschange seems to indicate that the subject is respondingdifferentially to the experimental conditions, and that thelinear assumptions of the ANOVA technique treats these changesas error variance. Put in other terms: While examination oflinear categories indicates no significant quantitativedifference, the Q-method points out qualitative changes inresponse.Subject #5 stayed with the FC pattern under allconditions except the Poor Native client description. Here,#5's responses loaded .58 with the FC pattern and .53 with thePC1 pattern.Subject #4 correlated highly with the FC pattern, exceptwhen doing the Q-sort for the Poor Caucasian. On thisoccasion #4 was distinguished by responses negatively loading82onto the FC2 pattern (r= -.84) and by not correlatingpositively with any other factor (all other loadings < .11).Under the Poor Caucasian condition, Subject #10 left thePC2 pattern shown in all other conditions and became a "mixedtype" loading modestly (.46 and .51) onto the two FC patterns.These examples seem to have a lot to say about the studyof bias and clinical decision-making. It appears that, evenin a small N study like the one in question, clinicaldecision-making is highly idiosyncratic and explication ofthese patterns at an individual level relies on ipsativetechniques. At an individual level, patterns of clinicaldecision-making are not necessarily stable or generalizablefrom one situation to another.Clinical Decision-Making and BiasThe patterns discussed above are at odds with findingstypified by O'Donohue & Fisher (1990), who found throughsurvey and interview techniques, that clinical decision-makingfollows no systematic procedure. The difference, in thisinstance, seems to be methodological. The experimentalparadigm, coupled with Q-sort measurement, seems to revealhighly systematic procedures. However, the exceptions to theclear-cut patterns also support Mattingly's (1991) assertionthat clinical reasoning is highly idiosyncratic.How, then, do weaccount for the results? First thesubjects seemed to respond differentially to the Poor83Caucasian with the family therapists becoming lessdifferentiated, some person-centered counsellors adhering evenmore closely to their patterns, and other person-centeredcounsellors taking a cognitive tack. Yet, these samecounsellors responded about the same to the Native clients,rich and poor, and to the Middle Class Caucasian client. Whywere there no differences here? Why does socioeconomic statusmake a difference only if you are Caucasian? Park & Rothbart(1982) account for such results in terms of the "outgrouphomogeneity" hypothesis. This hypothesis suggests that in-group members--in this case Caucasians--are seen as morecomplex and variegated while members of outgroups--in thiscase Natives--are seen as relatively uncomplicated and similarto each other. Thus, socioeconomic status would notdifferentiate native clients; they're all pretty much thesame--but it would differentiate Caucasian clients. Theobserved results are consistent with this hypothesis.There are other possibilities for describing thedifferential treatment of the Poor Caucasian client. Soder(1990) suggests that a distinction be made between prejudiceand ambivalence. In the case of the factor analytic patternfor the Poor Caucasian there is ample evidence that could beinterpreted as the operation of ambivalence. Note that thefactor loadings are generally lower, that there are four, notthree, patterns, and that there are four "mixed type"responses. Higgins & Borgh (1987) add another dimension tothe notion of ambivalence, hypothesizing that such judgements84arise out of competing motivations. Ambivalence is thecondition of close competition between two motivations.Ambivalence as an explanatory concept is also consistentwith attribution theory. If one accepts as fact that Nativepeople are subject to economic discrimination, and,consequently, economic distress is a part of the experience ofnative people in general, then the coincidence of being poorand native is more than understandable, it's expectable ornormative. However, it is possible to think of poverty for aCaucasian person as quite separate from the issue of race.Given the information provided, the best that can be saidabout the cause of the Caucasian's poverty is that he is downon his luck or suffering the secondary results of depression.Bad luck falls into the quadrant of the attributional modellabelled External-Unstable--and the unstable dimension was theone least preferred by all counsellors in all conditions. Inshort, the Poor Caucasian condition may be evokingattributions involving the least control, least clarity, andlowest area of preference for all the counsellors.Returning to the competing motivation hypothesis, It isinteresting to speculate that the competing motivations couldhave to do with, first, a desire to provide an effectiveefficient professional service, and at the same time remainculturally sensitive to out-group members. Pedersen andMarsella (1982) have pointed out that it is ethicallyincumbent upon counsellors and psychotherapists to provide aninformed culturally sensitive service. It has been further85emphasized that providing such a service necessitates becominginformed about the particulars of a client's culturalbackground and milieu (Sue & Sue, 1990). If we assume thatsocioeconomic status is also a cultural difference, then itmakes sense that therapists would respond to the poor clientmore tentatively and in a less organized way; they are awareof wanting to help, but at the same time needing moreinformation. The "outgroup homogeneity" hypothesis wouldexplain why no such competing emotions arise in the case ofthe native clients, i.e. the natives are all pretty much thesame.There are other possibilities. Freudian theory suggestscounter-transference as a possible explanation (Davidson,1983). The Poor Caucasian client could be seen as,differentially among the four clients, the most likelycandidate for counter-transference since this client isethnically similar (supporting the necessary identification)but also experiencing financial and status distress which is amarker for anxiety in a culture that values affluence andposition. The anxiety and the identification are the rightingredients for counter-transference.Implications for Theory and ResearchThis study demonstrates that an alternative methodologyin the study of bias and clinical decision-making can yieldresults that inform not only the relationship between clientvariables and clinical judgment, but also how therapist86orientation fits into this process. Further, the usefulnessof cognitive attribution theory as a frame of reference hasbeen demonstrated. As Pedersen (1991) has shown, a dimensionof differential treatment of culturally different clients isthe attributional basis--internal versus external--of thecounselling theory being applied.Further research in this area will benefit from havingsubjects respond to several experimental conditions andresponding with options structured according to some testabletheory--as in the present study--but with the addition of morelife-like analogy such as video-tape. A further point ofinvestigation is to determine whether Q-sort choices arecorrelated with actual counsellor behavior, a point that issimply assumed in the present investigation. The presentstudy examines only one point in the therapeutic process,specifically preparation for a first interview. Studies thatexamine other points in therapy may find that anotherdimension of attribution, global versus specific or nomotheticversus ideographic, may be useful in accounting for the shiftin understanding of the client from the general to thespecific, from the normative to the ipsative.The data produced by this study could be taken asevidence for bias in counselling and psychotherapy, but onlyif bias is defined as simple difference based on clientcharacteristics, and only if the assumption is made that thesedifferences are irrelevant to therapy. An alternative, andthis author believes, better interpretation runs something87like the following. Counselling trainees are generallyconsistent in their decisional strategy concerning plans tointervene in a first encounter with a client. However, whenthe client characteristic is low socioeconomic status, and theclient is a member of an "in-group", decisional strategieschange, some demonstrating less attributional coherence, somemore. This finding is congruent with the notion thatsocioeconomic status is the most salient and powerful clientvariable affecting therapist behavior, and that this variableis mediated by in-group versus outgroup membership. It seemscongruent with its theoretical orientation that the group ofperson-centered counsellors who seemed to demonstrate mostclearly not only what they would do in counselling, but alsowhat they would not do, changed the least in response to thesocioeconomic variable. However, it is also congruent thatfamily counsellors, focussing more on external factors, wouldchange their decisional pattern when the client's context isless familiar and more obviously distressing. The "outgrouphomogeneity" hypothesis seems to account for the lack ofdifferential response when the client variable is ethnicity.This finding, in turn, supports the notion of specifictraining in cross-cultural training and the ethicalresponsibility of counsellors to change "outgroups" to"ingroups", at least within the context of counselling andtherapy, by becoming knowledgeable about the cultural contextof the client.88Does this study demonstrate the operation of bias? Theanswer depends, ultimately, on where one stands in terms ofdefinitions. Two dimensions of biased decision-making havetraditionally been emphasized: (1) lack of self-awareness onthe part of the decider, and (2) basing a decision onirrelevant factors. This study does not address the firstdimension, self-awareness. However, if the reader seessocioeconomic status as irrelevant to treatment of depression,then a socioeconomic bias, for Caucasian clients seems to beindicated. However, if socioeconomic status is a relevantfactor in clinical decision-making for the type of clientpresented in this study, then we have evidence of an ethnicbias: An important factor, socioeconomic status, is completelyignored for the Native clients! If we include, in thedefinition of bias (as this writer does) not only takingirrelevant factors into account, but also failing to respondto relevant factors, then the results point to an ethnic bias,not a socioeconomic bias. Even more importantly, theseresults suggest that an understanding of bias and clinicaldecision-making is not likely to come through the examinationof isolated variables. The phenomenon of interest is aninteraction of persons, traits, and occasions.89REFERENCESAbramowitz, S. I. & Murray, J. (1983). Race effects ofpsychotherapy. In J Murray & P. R. Abramson (Eds.), Bias in psychotherapy (pp. 215-255). New York: Praeger.Abramowiz, S. I. & Dockecki, P. R. (1977). The politics ofclinical judgment: Early empirical returns. Psychological Bulletin, 84, 460-476Abramowitz, S. I. & Herrera, H. R. (1981). On controlling forpatient psychopathology in naturalistic studies of sexbias: A methodological demonstraction. Journal of Clinical and Consulting Psychology, 49, 597-603.Alloy, L. B., Peterson, C., Abramson, L. Y., & Seligman, M. E.P. (1984). Attributional style and the generality oflearned helplessness. Journal of Personality and Social Psychology, 46, 681-687.Allport, G.W. (1937). Personality: A psychological interpretation. New York: Holt.Allport, G.W. (1961). Comments on earlier chapters. In RolloMay (Ed.), Existential psychology. New York: Random House.Beck, A.T. (1974). The development of depression: A cognitivemodel. In J. R. Friedman and M. M. Katz (Eds.), Thepsychology of depression: Contemporary theory and research(pp. 186-207). New York: Wiley.Beck, A.T., Ward, C.H., Mendelson, M., Mock, J., & Erbaugh, J.(1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561-571.Broverman, I.K., Broverman, D. M., Clarkson, F. E.,Rosenkrantz, P. S., & Vogel, S. R. (1970). Sex rolestereotypes and clinical judgements of mental health.Journal of Consulting and Clinical Psychology, 34, 1-7.Burk, J. P. & Sher, K. J. (1990). Labeling the child of analcoholic: Negative stereotyping by mental healthprofessionals and peers. Journal of Studies on Alcohol,51, 156-163.Campbell, D. T. & Fiske, D. W. (1959). Convergent anddiscriminant validation by the multitrait-multimethodmatrix. Psychological Bulletin, 56, 81-105.Cayleff, S. E. (1986). Ethical issues in counseling gender,race, and culturally distinct groups. Journal of Counseling and Development, 64, 345-347.90Corsini, R. J. (1984). Current psychotherapies (3rd ed.).Itasca, IL: Peacock Publishers.Crocker, J. & Voelkl, K. (1991). Social stigma: The affectiveconsequences of attributional ambiguity. Journal of Personality and Social Psychology, 60, 218-228.Davidson, C. V. (1983). Making the conceptual leap from sexbias to countertransference: A closer look at the patient-therapist dyad. In J. Murray & P. R. Abramson (Eds.), Bias in psychotherapy (pp. 215-255). New York: Praeger.Deaux, K. (1976). Sex: A perspective on the attributionprocess. In J. H. Harvey, W. J. Ickles, & R. F. Kidd(Eds.), New directions in attribution research (Vol. 1).Hillsdale, NJ: Erlbaum.Deaux, K. & Emswiller, T. (1974). Explanation of successfulperformance on sex-linked tasks: What is skill for the maleis luck for the female. Journal of Personality and Social Psychology, 29, 80-85.Draguns, J. G. (1981). Cross-cultural counseling andpsychotherapy: History, issues, current status. In A. J.Marsella & P. B. Pedersen (Eds.), Cross-cultural counselingand psychotherapy (pp. 3-27). New York: Pergamon Press.Egan, G. (1986). The skilled helper: A systematic approach toeffective helping (3rd ed.). Monterey, CA: Brooks/Cole.Ellis, M. V. & Robbins, E. S. (1990). Anchoring errors inclinical judgments: Type I error, adjustment, ormitigation? Journal of Counseling Psychology, 37, 343-351.Epston, D. & White, M. (1990). Narrative means to therapeutic ends. New York: W. W. Norton & Co.Falvey, J. E. (1992). From intake to intervention: Inter-disciplinary perspectives on mental health treatmentplanning. Journal of Mental Health Counseling, 14, 471-489.Fromm, E. (1947). Man for himself: An inauiry into the psycholoay of ethics. New York: Fawcett Publications.Gartner, J. & Harmatz, M. (1990). The effect of patient andclinician ideology on clinical judgment: A study ofideological countertransference. Psychotherapy, 27, 98-106Goldenberg I. & Goldenberg H. (1985). Family therapy: Anoverview. Monterey, CA: Brooks/Cole91Higgins, E. T. & Bargh, J. A. (1987). Social cognition andsocial perception. Annual Review of Psychology, 38, 369-425Haley, J. (1973). Uncommon therapy: The psychiatric techniques of Milton H. Erickson. New York: Norton.Haley, J. (1987). Problem-solving therapy. San Francisco,CA; Jossey-Bass.Hamilton, D. L. (1983). Effects of bias on attribution andinformation processing: A cognitive-attributional analysisof stereotyping. In J. Murray & P. R. Abramson (Eds.),Bias in psychotherapy (pp. 103-123). New York: Praeger.Hansen, F. J. & Reekie, L. (1990). Sex differences inclinical judgments of male and female therapists. SexRoles, 23, 51-64.Hardy, D. M. & Johnson, M. E. (1992). Influence of therapistgender and client gender, socioeconomic status, andalcoholic status on clinical judgments. Journal of Alcohol and Drug Education, 37(2), 94-102.Heider, F. (1958). The psychology of interpersonal relations.New York: Wiley.Huitt, K. & Elston, R. R. (1991). Attitudes toward personswith disabilities expressed by professional counselors.Journal of Applied Rehabilitation Counseling, 22(2), 42-43.Ivey, A. E., Ivey, M. B. & Simek-Downing, L. (1987)Counselling and psychotherapy: Integrating skills, theory, and practice (2nd ed.). Englewood Cliffs, NJ: Prentice-Hall.Kanfer, F. H. & Goldstein, A. P. (1986). Helping peoplechange (3rd ed.). New York: Pergamon.Kelley, H. H. (1972). Attribution theory in socialpsychology. In D. Levine (Ed.), Nebraska symposium onmotivation (vol. 51). Lincoln: University of NebraskaPress.Korchin, S. J. (1976). Modern clinical psychology: Principles of intervention in the clinic and community. New York:Basis Books.Lopez, S. R. (1989). Patient variable biases in clinicaljudgement: Conceptual overview and methodologicalconsiderations. Psychological Bulletin, 106, 184-203.92Lopez, S. R. (1983). The study of psychotherapy bias: Someconceptual issues and some concluding remarks. In J.Murray & P. R. Abramson (Eds.), Bias in psychotherapy (pp.353-365). New York: Praeger.Madanes. C. (1981). Strategic family therapy. San Francisco,CA: Jossey-Bass.Maroney, D. & Golub, S. (1992). Nurses' attitudes towardobese persons and certain ethnic groups. Perceptual andMotor Skills, 75, 387-391.Marsella, A. J. & Pedersen P. B. (Eds.). (1981). Cross-cultural counseling and psychotherapy. New York: PergamonPress.Mattingly, C. (1991). What is clinical reasoning? AmericanJournal of Occupational Therapy, 45, 979-986.McKeoun, B. and Thomas, D. (1988). 0 Methodology. NewburyPark, CA: Sage Publications.Meador, B. D. & Rogers, C. R. (1984). Person-centeredtherapy. In R. J. Corsini (Ed.), Current Psychotherapies (3rd ed.) (pp. 142-195). Itasca, ILL: F. E. PeacockPublishers.Milgram, Stanley (1974). The experience of living in cities.In R. H. Moos & P. M. Insel (Eds.), Issues in socialecology (pp. 189-207). Palo Alto, CA: National PressBooks.Mischel, W. (1968). Personality and assessment. New York:John Wiley and Sons.Murray, J. & Abramson, P. R. (1983). An investigation of theeffects of client gender and attractiveness onpsychotherapists' judgements. In J. Murray & P. R.Abramson (Eds.), Bias in psychotherapy (pp. 129-167). NewYork: Praeger.Murray, J. & Abramson, P. R. (Eds.). (1983). Bias inPsychotherapy. New York: Praeger.O'Donohue, W. & Fisher, J. E. (1990). Treatment decisions:Their nature and their justification. Psychotherapy, 27,421-427.Park, B. & Rothbart, M. (1982). Perception of out-grouphomogeneity and levels of social categorization: Memory forthe subordinate attributes of in-group and out-groupmembers. Journal of Personality and Social Psychology, 42,1051-1068.93Pedersen, P. B. & Marsella, A. J. (1982). The ethical crisisfor cross-cultural counseling and therapy. Professional Psychology, 13, 492-500.Pedersen, P. B. (1991). Counseling international students.Counseling Psychologist, 19, 10-58.Rinn, J. L. (1961). Q-methodology: An application to groupphenomena. Educational and Psychological Measurement, 21,2, 315-329.Ross, L. (1977). The intuitive psychologist and hisshortcomings. In L. Berkowitz (Ed.), Advances inexperimental social psychology (vol. 10, pp. 173-220). NewYork: Academic Press.Rotter, J. B. (1966). Generalized expectancies for internalversus external control of reinforcement. Psychological Monographs, 80, (1, Whole No. 609).Sattler, J. M. (1977). The effects of therapist-client racialsimilarity. In A. S. Gurman & A. M. Razin (Eds.),Effective psychotherapy: A handbook of research. New York:Pergamon Press.Saxon, J. P. & Spitznagel, R. J. (1992).^Age bias: Anintervening variable in vocational rehabilitationfeasibility decision-making? Journal of AppliedRehabilitation Counseling, 23(1), 22-26.Schlossberg, N.' K. (1977). Hide and seek with bias.Personnel and Guidance Journal, April, 481-484.Schneider, D. J. (1973). Implicit personality theory: Areview. Psychological Bulletin, 79, 294-309.Seligman, M. E. P. (1975). Helplessness: On depression, development and death. San Francisco, CA: Freeman.Sidman, M. (1960). Tactics of scientific research: Evaluatingexperimental data in psychology. New York: Basis Books.Skinner, B. F. (1974). About behaviorism. New York: VintageBooks.Sloane, R, Staples, F., Cristol, A., Yorkston, N. & Whipple,K. Psychotherapy vs. behavior therapy. Cambridge, MA:Harvard.Soder, M. (1990). Prejudice or ambivalence? Attitudes towardpersons with disabilities. Disability, Handicap, & Society, 5, 227-241. 94Stephenson, W. (1953). The study of behavior. Chicago, Ii:University of Chicago Press.Strohmer, D. C. & Shivy, V. A. (1990). Information processingstrategies in counselor hypothesis testing: The role ofselective memory and expectancy. Journal of CounselingPsychology, 37, 465-472.Sue, D. W. & Sue, D. (1990). Counseling the culturallydifferent: Theory and practice. New York: John Wiley &Sons.Taylor, S. E., & Fiske, S. T. (1975). Point of view andperceptions of causality. Journal of Personality andSocial Psychology, 32, 439-445.Weary, G., Stanley, M. A., & Harvey, J. H. (1989).Attribution. New York: Springer-Verlag.Weiner, B., Frieze, I., Kukla, A., Reed, L., Rest, S., &Rosenbaum, R. M. (1972). Perceiving the causes of successand failure. In E. E. Jones, D. E. Kanouse, H. H. Kelly,R. E. Nisbett, S. Valins, & B. Weiner (Eds.), Attribution: Perceiving the causes of behavior. Morristown NJ: GeneralLearning Press.Yin, R. K. (1989). Case study research: Design and methods.London: Sage.Zimbardo, P. G., & Leippe, M. R. (1991). The psychology ofattitude change and social influence. Philadelphia: TempleUniversity Press.Zuckerman, M. & Lubin, B. (1965). Manual for the MultipleAffect Adjective Check List. San Diego, CA: Educationaland Industrial Testing Service.Zuriff, G. E. (1985). Behaviorism: A conceptual reconstruction. New York: Columbia University Press.Appendix ARecruitment of Subjects9596The following is the presentation given to classes inCounselling Psychology at the University of British Columbiarecruiting subjects for the study.Recruitment Presentation to be Used in SolicitingVolunteers from Counselling Psychology ClassesHello. My name is Glen Grigg. I am a student incounselling psychology and I am here today to ask you toprovide data for my thesis project. My advisor on thisstudy is Mary Westwood from this department. Theresults will be kept confidential. The data will bepresented in the form of aggregate statistical summariesor anonymous case profiles.Why should you participate in this study? Well, I hopeyou will be willing to help a fellow student with a taskwe all must complete, specifically, a thesis! But, muchmore importantly, I think that participating as subjectsin this study may be useful to you. You see, I aminterested in the kinds of decisions that counsellorsmake, and their process of judgment, especially early onin the counselling process. Participating in this taskmay be very useful to you in thinking about, andlearning about, assessment issues in our work withclients.Specifically, I would like you to read a series of fourintake interview summaries, each of which describes afictitious client. After reading each one, the task isto do a card sort. Intervention possibilities arewritten on a series of cards. A standard format isprovided to help you indicate which counsellor actionsyou see as most appropriate, and those you view as leastappropriate. While I don't see any particularpreferences as right or wrong, or even better or worse,I am betting that the theoretical orientation of thetherapist, as well as the client and the client'sdifficulties, makes a difference in how clinicaldecisions are made. And I also want to control fortherapist gender and level of training. That's why, aspart of the task, I will also ask you to fill in a shortdemographic sheet that includes facts such as your agerange and sex, your theoretical orientation--Adlerian,family systems, and so on--as well as a checklist ofcourses and labs you have completed in the program. I'dalso like to follow up your work with a short debrief,so I would like to get a phone number where you can bereached.97Some people have completed the card sorts in as littleas thirty minutes, but most people have taken somewherebetween a half hour and an hour to complete this task.I would like to remind you that you are free to withdrawfrom this study at any time. Participation in thisstudy is voluntary and does not have any bearing on anycourse work or program requirements.Any questions? Okay! Thanks your time, your attention,and thanks in advance for your contribution to my study.I would like to be called to confirm my participation asa subject in "Clinical Judgment in Counselling andPsychotherapy: Extending Theory and Research Design". Iunderstand that I am free to withdraw from this study atany time and that participation in this study isvoluntary and does not have any bearing on any coursework or program requirements. I can also call thefollowing people for more information:Glen Grigg, Master's Candidate, Dept. of CounsellingPsychology, 986-0534.Dr. Marvin Westwood, Dept. of Counselling Psychology,University of British Columbia, 822-0870.The following letter was sent to prompt the return ofdata.Dear [Subject],About two months ago I visited you during acounselling class at UBC to recruit volunteer subjectsfor my thesis project. You were kind enough to commitsome of your time and your expertise to this project.We were in touch by phone, following which I sent you apackage of materials: four "Q-sorts" and a demographicssheet.So far I haven't heard back from you. So, I'm writingto touch base again and offer any help I can incompleting the card sorts. I understand that this hasbeen a very busy time of year with lots to do! If I canbe of any assistance, please give me a call at 986-0534.I work some odd hours, so you may end up talking to ananswering machine, but I will get back to you aspromptly as possible.Appendix BInstruction Sets and Experimental Materials9899TIIE UNIVERSITY OF BRITISH COLUAIIIIAF2EE A4 13 THIS^I Ft ST"In this package you should find the following items:1. Four descriptions of counselling clients.2. Seven envelopes, numbered 1 to 7 attached to eachdescription.3. Four sets of 32 cards bearing words or phrases describingcounselling interventions.4. A large brown envelope, stamped and addressed, by which toreturn your completed work.5. A blank envelope on which to write your address so that Ican send you a summary of the results of the study.6. Some tea, hot chocolate, and a granola bar in case you gethungry or thirsty while sorting the cards.7.^A demographic, by which you can share a litte informationabout yourself --if you wish.B.^This instruction set.THE BASIC TASKIn a nutshell, you are asked to consider the needs of eachclient, and then sort the counselling options from most preferredto least preferred and return the sorted cards by mail.IF YOU NEED HELPIf you have questions or problems, call me--Glen Grigg—at 986-0534. You may get my answering machine, but in that case leave amessage and I will get back to you right away.RETURNING THE RESULTSThe postage is already paid and the envelope addressed, so allyou need to do is put everything back in the return envelope anddrop it in the mail.100THE UNIVERSITY OF BRITISH COLUMBIAINSTRUCTIONSThank you for participating in my study! My name is Glen Griggand I can be reached to answer any questions by calling 986-0534.My advisor is Dr. Marvin Westwood, Associate Professor in theDepartment of Counselling Psychology; Dr. Westwood's number is822-5259. This questionnaire is part of my thesis projectundertaken towards the completion of my Master of Arts degree incounselling psychology. The purpose of this study (titledClinical Judgeeent in Counselling and Psychotherapy: ExtendingTheory and Research) is to enhance the understanding of howcounsellors and psychotherapists make clinical judgements. It ismy hope that by participating and later reviewing the results youwill gain information and insight about client assessment andprofessional choice.I would like to remind you that participation is voluntary andthat you may withdraw at any time. The task takes somewherebetween thirty minutes and an hour to complete. I have providedyou with a large brown envelope already stamped and addressed tome. When you have finished, please put the materials back in theenvelope and return it to me by mail. The return of thisquestionnaire will be taken to indicate your consent toparticipate.To have a summary and results of my study mailed to you, pleaseput your name and address on the small envelope provided.Page 1101THE UNIVERSITY OF BRITISH COLUMBIAIn this package are four client descriptions that you are toimagine to be actual summaries of screening interviews taken at aclinic for treatment of depression. Attached to each clientdescription is a set of thirty-two cards and seven envelopes.Your task is to read these case descriptions as if you wereplanning to conduct an initial counselling session based on this"intake" information. Then you use the cards and the envelopesto indicate what actions you would choose and which you wouldavoid in helping this client. The paragraphs below will explainexactly how to do this.Each of the cards has been printed with a few words or a phrasedescribing a counselling technique or procedure. Your task is tosort these techniques and procedures from most preferred to leastpreferred. You may wish to begin by sorting them into two piles--preferred and not preferred--but to complete the task you mustsort them into seven groups from most preferred to leastpreferred. Each group must contain a specific number of cards,as listed below. Since I need to know what your preferenceswere, one envelope has been provided for each group of cards.When you're finished sorting, you can seal that group of cards inthe envelope that is labelled for that pile. Envelope 401 islabelled "Most favoured" and enveloped #7 is labelled "leastfavoured". The other five envelopes represent points in betweenthese extremes. When you have finished sorting, you must havethe following numbers of cards for each envelopesEnvelopeEnvelopeEnvelopeEnvelopeEnvelopeEnvelopeEnvelopeNumber 1 "Most Preferred" 2 cardsNumber 2 4 cardsNumber 3 6 cardsNumber 4 El cardsNumber 5 6 cardsNumber 6 4 cardsNumber 7 "Least Preferred" 2 cardsYou might picture the stacks of cards as looking like thisxxx^x^xx^x^xx^x^x^x^xx^x^x^x^x^x^x^x^x^x^x^x^x ^x^x^x^x^x^x ^Most preferred 1^2^3^4^5^6^7^Least preferredPage 2102THE UNIVERSITY OF BRITISH COLUMBIAamisedrhsitorskr' 17When you are finished, put the cards in the envelopes, seal theenvelopes, clip them together, and go on with the next casedescription.You can organize this sorting task any way you want. Many peoplefind that arranging the envelopes in front of them like thishelps them along:•You may feel that you have insufficient information to make acategorical judgement. You may also have ideas for practices andprocedures that are not indicated on the cards. I would be veryinterested if you would make a note of these ideas and pass themalong, but please give the best answer you can given the optionssupplied.It is very important that you put the cards in the envelopes andseal them up. Otherwise, there will be no record of what yourpreferences wereAlong with each case description is a five-point scale by whichyou can indicate your assessment of how severe the client'sproblem is. Here's an example:11 Problem Severity^ 1I^ II 1^2^3^4^5^II^Mild Moderate Severe IIn the example above, the therapist judgement was moderateseverity.When you have finished each sort, please put the envelopescontaining the cards and the client descriptions in the returnenvelope. Once you have completed all +our sorts and have allthe cards, envelopes, and client descriptions in the returnenvelope, just seal it up and drop it in the mail. If you wouldlike to hear back with a results summary, put your address on theblank envelope provided. I'll send you a summary of the resultswhen they are availablePage 3103T II E UNIVE It SITV^II It I 'I' I S II C 0 I, ti MI! I AYour time is important and valuable--to you and to me. So pleasebe aware that you don't have to finish this task in one sitting.You can go through the clients one at a time, or do several atone sitting. I much more concerned with thoughtful responsesthan speed.Once again, thank you for your participation!Page 4104'III E IJ N I V 11,', It S I 'I' Y 0^IIRITIS II C^I,^NI II I ADemographics SheetPlease answer the questions below. The intent of the questionsis to allow me to report more precisely on the nature of thoseparticipating in my study, and also to see if age or gender orlevel of training makes a systematic difference in the waysubjects respond.However, if for any reason you don't wish to answer any of thequestions, just leave it blank._FemalePlease check opposite your age range20-24^25-29^_._.30-3435-39 40-44 45-49•50-54^55-59^60+Please check both those courses you have completed as well asthose in which you are currently enrolled.CNPS 504 Elementary School CounsellingCNPS 508 Review of Research in Educational MethodsCNPS 514 Counselling Adolescents.CNPS 531 Interview and Nonstandard Measures in Counselling_.CNPS 532 Tests in Pupil Personnel Services.CNPS 534 Gender and Sex Role Issues in CounsellingCNPS 544 Family Counselling I.CNPS 545 Family Counselling II_GNPS 561 Laboratory PracticumGNPS 564 Group Counselling.CNPS 574 Career Planning and Decision-Making Counselling_CAPS 578 Counselling Theories and Interventions_CNPS 579 Research in Guidance Sbrvices_CNPS 580 Problems in EducationCNPS 584 Program Development in Counselling_CNPS 588 Supervised Training in CounsellingCNPS 594 Cross-Cultural CounsellingCNPS 598 Field ExperiencePlease check the description that most closely corresponds toyour orientation to counselling..Psychodynamic^ BehavioralAdlerian_^ _ Gestalt.RealityPerson-Centered_ _._Analytical Existential_Cognitive^ FamilyMay I call you for a brief follow-up interview if needed?Yes^NoThank you for all your help!Home Address:Postal Code:Referred by: se/fziG L._ I hi IThEEU13 -^-105Intake Screening Form7Date:_me:  ,/ Name:^/1,%7‘rA 111AMalinglaiparAge:   Sex: 0 FPhone #: fOMMONOO(home) ^ (work)Occupation: ^/1/d" ^Currently Employed? YReligious affil. (optional): ^Ethnicity (optional):Referral to ^Medical Assessment^Testing (Specify tests: ^^Counselling^Other (Specify:^Counsellor Assigned: AmmoRIMPWSummary of Intake Interview:(Please type or write CLEARLY)In the screening interview Joe slouched in his chair- as ifextremely tired or unmotivated to invest energy in the encounter.When questioned, Joe revealed that he was very pessimistic aboutthe future. Joe told the interviewer that he wakes up earlier inthe morning than he used to. Asked about his social life, Joeresponded that he felt bored most Crf the time. Joes says that helives on a combination of welfare payments and the occasional oddjob. Joe indicated his agreement when the interviewer suggestedcalling his feelings "broken-hearted", "downcast", and"afflicted"., 1i Problem Severity^ 11^ 11 1^2^3^4^5^11^Mild Moderate^Severe 1Home Address:Postal Code:Referred by:106.0011.11111=101=11.r.cl__sT-FtEE -rlerill"12111111"^-allighICCILIVEFR ,^El.Intake Screening Form^ Time: .^07 • 006-1111111111111,Age: ^ Sex :^FPhone #: IMMOMIWome)Occupation: C274.5 A177 ^Currently Employed?Religious affil. (optional):^Ethnicity (optional):Referral to^Medical AssessmentTesting (Specify tests:^Other (Specify:Counsellor Assigned: __11111,1119110461QPSummary of Intake Interview:(Please type or write CLEARLY)-1During the screening interview Christopher's eyes appearedsomewhat reddened, and he disclosed that he had been experiencingsome episodes of crying. He told the interviewer that he feelsbad or unworthy a good part of the time. Chris told theinterviewer that he was concerned with a general feeling of"unwellness" and unpleasant feelings in his body. He alsorevealed an employment concern saying, "I don't work as well as Iused to." At points in the interview Christopher's feelings werereflected to him as "broken", "grieved", and "hopeless". Chrisreported that these words accurately described his feelings.Problem Severity1Mild^2^ 4Moderate SevereDate:Name:^CA(Y5.g 5 Sex:Time:1- I hlOMMSOMCOMMEMOMMV ET-FEEEElrV "A^u v FRIntake Screening Form107Occupation:Home Address:Date:Name:Age:Phone #:^(home) (work)Currently Employed? NjPostal Code:Referred by:  Religious affil. (optional):  AF7,e7LEthnicity (optional): Referral to ^Medical Assessment^Testing (Specify tests:^X  Counselling^Other (Specify: ^Counsellor Assigned:Summary of Intake Interview:(Please type or write CLEARLY)Although able to talk to the interviewer, Benjamen's speech wasnoted to be high pitched and his verbal responses sparse.Reporting a sense of self-hate, Ben said, "I just don't likemyself." Benjamen reported a gradual loss of appetite, but couldnot be specific about the period of time. He also told of beingless interested in people than he used to be. Ben talked ofproblems getting his welfare check to "go the distance" eachmonth. Feeling wards such as "weary", "oppressed" and"downhearted" were endorsed by Benjamen during the interview.^I Problem Severity^ -11^2^3^4^5M i1 ,1108101111111111MMINENSIri—I"ICinilinglESEMENSIV -1-fREE -rVANICOL-IVEIR•13.Date: ialIllaEIMEIIIWime:Name: ‘,1 Age: ^ Sex: (M) FPhone #: ^ home)e) 101101.(work).Occupation:^47/1 42/X74- ^Currently Employed? 6 NHome Address:Postal code:A101111111,i=11111.116111,Referred by:^--_e_Z--'^n^_.....Religious affil. (optional): 1'V - Ethnicity (optional): Referral to Medical AssessmentTesting^(Specify testes_ CounsellingOther (Specify: ^11111•11111111Counsellor Assigned: 4^F—Summary of Intake Interview:(Please type or write CLEARLY)The interviewer noted during the screening that James was veryslow to respond to questions or st.atemants, often taking fiveseconds or more be.fore replying either verbally or nonverbally.James stated, "I have a feeling something bad may happen to me."James also told of waking up in the morning without feelingreally rested. James revealed that he felt that at some point inthe past he had begin to get annoyed or irritated more easilythan he used to. James confirmed for the interviewer that hefelt "heartsick", "glum", and "desolate".Problem Severity1^2^3^4^5Mild Moderate^SevereIntake Screening Form;e4L5-Appendix CQ-Sort Items and Their Rater Reliabilities109110SUMMARY OF RELIABILITY DATANumbers to the left of the items indicate how many times,out of ten, this item is categorized into the quadrant thatheads the list. The exceptions are the two items labelled"R". This designates the two items that were generated on arational basis to make up two more items for a category thathad only six nominations in total.In each other category, the highest rated eight itemswere used as the content of the Q-sort.Internal StableReliability^Item9^Analysis of characterological type8 Analyze cognitive style and cognitivedistortions8^Directing client to change questions tostatements8^Discover and correct irrational ideation8 Examine premises and suppositions underlyingclient's expressed beliefs8^Examine premises of client's world view8 Values clarification8^Free association7^Cultural identity consciousness-raising7 Gender identity consciousness-raising7^Positive asset search7 Analytic dream analysis7^Focussed free association6 Analysis of client's developmental tasks6^Relaxation training6 Analysis of transference and counter-transference111Internal UnstableReliability^Item8^Charting and self-monitoring7 Analyzing elements of a dream as metaphors forparts of the self7^Analysis of client's developmental stage7 Client self-monitoring of thoughts andassociated emotions7^Thought stopping6 Analysis of resistance6^Directing client to exaggerate the symptom6 Direction client to exaggerate symptoms andproblems6^Redirect client's use of language6^Direct client's attention to "here and now"6 Gestalt "two-chair" technique6^Questioning6 Emotional intensification through advancedaccurate empathy6^Having the client talk to the "empty chair"6 Ref ramingExternal StableReliability^Item8^Constructing a "life line"8 Tracking interactional patterns7^Assertiveness training7 Collecting a family history7^Generating options and desirable scenarios foroutcomes6^Multigenerational pattern analysis throughgenogram constructionExploring client's social and cultural context,including support network.Exploring and analyzing client's current lifetasks.112External UnstableReliability^Item10^Assigning therapeutic tasks to disruptproblematic interactions10^Contingency and reinforcement management10 Contracting for new behaviors10^Generating a specific behavioral problemdefinition10^Mediating communication between family members10 Pinpointing behavioral targets for change10^Role-playing of alternative behaviors10 Setting specific, observable, measurable goalsfor counselling outcome9^Analyzing the prevailing contingencies ofreinforcement9^Circular questioning9 Social skills training8^Directing client's attention to aspects of lifeother than the problem8^Teaching communication skills8 Teaching decisional strategies8^Role playing and/or behavioral rehearsal7 Systematic desensitization7^Primary empathy and attending skills7 Re-enactment of critical incidents6^Joining in the rule structure and languagepractices of the family6^Paradoxical intention6 Creating a metaphor for description of aproblem6^Re-labelling "problems" behaviors as strengthsor justifiable protests against unjusttreatment or circumstancesAppendix DAnalysis and Content of Patterns113114Summary of Analysis of Variance PerformedOn Q-Scores for Each Counsellor TypeFamily^Person-^Person-Counsellor 1 Centered 1 Centered 2^Condition Source of^Signif^Signif^SignifVariation F of F^F of F^F of FFACTOR 1^FACTOR 3^FACTOR2POOR Main Effects 7.383 .003* 14.376 .000* 2.308 .118NATIVE IE 11.016 .003* 7.611 .010* .462 .502SU 3.749 .063 21.141 .000* 4.154 .051*2-way Inter. 6.197 .019* 3.383 .077 .205 .654FACTOR 1 FACTOR 2 FACTOR 3MIDDLE Main Effects 8.235 .002* 12.156 .000* 2.014 .152CLASS IE 8.235 .008* 10.144 .004* .806 .377NATIVE SU 8.235 .008* 14.168 .001* 3.223 .0832-way Inter. 8.235 .008* 1.341 .257 .201 .657FACTOR 2 FACTOR 1 FACTOR 4POOR Main Effects 6.072 .006* 12.634 .000* 3.151 .058CAUCASIAN IE 9.359 .005* 8.537 .008* 3.570 .069SU 2.785 .106 16.732 .000* 2.733 .1092-way Inter. 9.359 .005* 1.366 .253 1.394 .248FACTOR 1 FACTOR 3 FACTOR 2MIDDLE Main Effects 4.439 .021* 10.884 .000* 3.323 .051*CLASS IE 3.346 .078 8.092 .008* .054 .817CAUCASIAN SU 5.532 .027* 13.676 .001* .591 .0162-way Inter 6.829 .014* 2.023 .166 .218 .644Effects significant at p<.05 are shown by *115FACTOR 1 - POOR NATIVE: FAMILY COUNSELLOR PATTERN* * * ANALYSIS OF VARIANCE * * *PATTERN Q-SCORE (PA)BY INTERNAL/EXTERNAL (IE)STABLE/UNSTABLE (SU)^Sum of^Mean^SignifSource of Variation Squares DF Square F^of FMain Effects 24.125 2 12.062 7.383 .003IE 18.000 1 18.000 11.016 .003SU 6.125 1 6.125 3.749 .0632-way Interactions 10.125 1 10.125 6.197 .019IE^SU 10.125 1 10.125 6.197 .019Explained 34.250 3 11.417 6.987 .001Residual 45.750 28 1.634Total 80.000 31 2.581*** CELL MEANS ***Stable^Unstable^TOTALInternal 3.13 3.38 3.25External 5.75 3.75 4.75TOTAL 4.44 3.56116FACTOR 1 - POOR NATIVE: FAMILY COUNSELLOR PATTERN*** CONTENT OF Q - SORT ***Item Content^ Type Z-ScoreExploring client's social and culturalcontext, including support network^ES^2.294Multigenerational pattern analysis andgenogram construction^ ES^1.537Constructing a "life line" ES^1.426Exploring and analyzing client'slife tasks^ ES^1.327Tracking interactional patterns^ES^1.229Generating options and desirablescenarios for outcomes ES^1.163Thought stopping^ IU^-1.050Discover and correct irrationalideation IS^-1.079Directing client to change questionsto statements^ IS^-1.130Free association IS^-1.152Contingency and reinforcementmanagement EU -1.310Analysis of characterological type^IS^-1.547Analysis of resistance^ IU^-1.607Classification of Items:US - Internal-Stable^IU - Internal-UnstableES - External-Stable EU - External-UnstableItems with standard (Z) scores of greater or less than one arepresented. Scores with positive scores are most preferred;those with negative scores are least preferred.117FACTOR 1 - MIDDLE CLASS NATIVE: FAMILY COUNSELLOR PATTERN* * * ANALYSIS OF VARIANCE * * *PATTERN Q-SCORE (PA)BY INTERNAL/EXTERNAL (IE)STABLE/UNSTABLE (SU)^Sum of^Mean^SignifSource of Variation Squares DF Square F^of FMain Effects 25.000 2 12.500 8.235 .002IE 12.500 1 12.500 8.235 .008SU 12.500 1 12.500 8.235 .0082-way Interactions 12.500 1 12.500 8.235 .008IE^SU 12.500 1 12.500 8.235 .008Explained 37.500 3 12.500 8.235 .000Residual 42.500 28 1.518Total 80.000 31 2.581* * * CELL MEANS * * *Stable^Unstable^TOTALInternal^3.38 3.38^3.38External^5.88^ 3.38 4.63TOTAL 4.63 3.38118FACTOR 1 - MIDDLE CLASS NATIVE: FAMILY COUNSELLOR PATTERN* * *^CONTENT^OF^Q-SORT * *^*Item Content Type Z-ScoreExploring client's social and culturalcontext, including support network^ES 1.806Collecting a family history^ ES 1.774Multigenerational pattern analysis andgenogram construction ES 1.619Exploring and analyzing client'scurrent life tasks^ ES 1.462Constructing a "life line" ES 1.359Tracking interactional patterns^ES 1.186Contracting for new behaviors EU -1.017Directing client to change questionsto statements^ IS -1.098Thought stopping IU -1.252Analysis of characterological type^IS -1.321Contingency and reinforcementmanagement^ EU -1.642Analysis of resistance^ IU -1.721Classification of Items:US - Internal-Stable^IU - Internal-UnstableES - External-Stable EU - External-UnstableItems with standard (Z) scores of greater or less than one arepresented. Scores with positive scores are most preferred;those with negative scores are least preferred.119FACTOR 1 - POOR CAUCASIAN: PERSON-CENTERED PATTERN 1* * * ANALYSIS OF VARIANCE * * *PATTERN Q-SCORE (PA)BY INTERNAL/EXTERNAL (IE)STABLE/UNSTABLE (SU)^Sum of^Mean^SignifSource of Variation Squares DF Square F^of FMain Effects 37.000 2 18.500 12.634 .000IE 12.500 1 12.500 8.537 .007SU 24.500 1 24.500 16.732 .0002-way Interactions 2.000 1 2.000 1.366 .252IE^SU 2.000 1 2.000 1.366 .252Explained 39.000 3 13.000 8.878 .000Residual 41.000 28 1.464Total 80.000 31 2.581* * * CELL MEANS * * *Stable^Unstable^TOTALInternal^5.25 4.00^4.63External^4.50^ 2.25 3.38TOTAL 4.88 3.13120FACTOR 1 - POOR CAUCASIAN: PERSON-CENTERED PATTERN 1*** CONTENT OF Q - SORT ***Item Content^ Type Z-ScoreExamine premises of client's worldview^ IS^1.889Examine premises and suppositions under-lying client's expressed beliefs^IS^1.855Exploring and analyzing client'scurrent life tasks^ ES^1.538Exploring client's social and culturalcontext, including support network^ES^1.400Values clarification IS^1.280Mediating communication between familymembers^ EU^-1.144Directing client to exaggerate thesymptom IU^-1.306Setting specific, observable, measure-able goals for counselling outcome^EU^-1.364Contingency and reinforcementmanagement^ EU^-1.371Contracting for new behaviors^ EU -1.540Generating a specific behavioralproblem definition^ EU^-2.072Classification of Items:US - Internal-Stable^IU - Internal-UnstableES - External-Stable EU - External-UnstableItems with standard (Z) scores of greater or less than one arepresented. Scores with positive scores are most preferred;those with negative scores are least preferred.121FACTOR 1 - MIDDLE CLASS CAUCASIAN: FAMILY COUNSELLOR PATTERN* * * ANALYSIS OF VARIANCE * * *PATTERN Q-SCORE (PA)BY INTERNAL/EXTERNAL (IE)STABLE/UNSTABLE (SU)^Sum of^Mean^SignifSource of Variation Squares DF Square F^of FMain Effects 16.250 2 8.125 4.439 .021IE 6.125 1 6.125 3.346 .078SU 10.125 1 10.125 5.532 .0262-way Interactions 12.500 1 12.500 6.829 .014IE^SU 12.500 1 12.500 6.829 .014Explained 28.750 3 9.583 5.236 .005Residual 51.250 28 1.830Total 80.000 31 2.581* * * CELL MEANS * * *Stable^Unstable^TOTALInternal^3.50 3.63^3.56External^5.63^ 3.25 4.44TOTAL 4.56 3.44122FACTOR 1 - MIDDLE CLASS CAUCASIAN: FAMILY COUNSELLOR PATTERN* * * CONTENT OF Q-SORT * * *Item ContentExploring client's social and culturalcontext, including support networkCollecting a family historyConstructing a "life line"Examine premises and suppositions under-lying client's expressed beliefsMultigenerational pattern analysis andgenogram constructionTracking interactional patternsFree associationThought stoppingAnalysis of characterological typeDiscover and correct irrationalideationAnalysis of resistancePinpointing behavioral targets forchangeContingency and reinforcementmanagementType Z-ScoreES 1.861ES 1.669ES 1.575IS 1.319ES 1.281ES 1.276IS -1.121IU -1.164IS -1.197IS -1.226IU -1.406EU -1.689EU -1.702Classification of Items:US - Internal-Stable^IU - Internal-UnstableES - External-Stable EU - External-UnstableItems with standard (Z) scores of greater or less than one arepresented. Scores with positive scores are most preferred;those with negative scores are least preferred.123FACTOR 2 - POOR NATIVE: PERSON-CENTERED PATTERN 2* * * ANALYSIS OF VARIANCE * * *PATTERN Q-SCORE (PA)BY INTERNAL/EXTERNAL (IE)STABLE/UNSTABLE (SU)^Sum of^Mean^SignifSource of Variation Squares DF Square F^of FMain Effects 11.250 2 5.625 2.308 .118IE 1.125 1 1.125 .462 .502SU 10.125 1 10.125 4.154 .0512-way Interactions .500 1 .500 .205 .654IE^SU .500 1 .500 .205 .654Explained 11.750 3 3.917 1.607 .210Residual 68.250 28 2.437Total 80.000 31 2.581*** CELL MEANS ***Stable^Unstable^TOTALInternal 4.25 3.38 3.81External 4.88 3.50 4.19TOTAL 4.56 3.44124POOR NATIVE - FACTOR 2: PERSON-CENTERED PATTERN 2* * * CONTENT OF Q-SORT * * *Item Content^ Type Z -ScoreExploring client's social and culturalcontext, including support network^ES^2.327Examine premises of client's worldview^ IS^1.681Examine premises and suppositions under-lying client's expressed beliefs^IS^1.473Collecting a family history^ ES^1.263Analyze cognitive style and cognitivedistortions^ IS^1.164Mediating communication between familymembersDirecting client to change questionsto statementsAnalysis of characterological typeAnalysis of resistanceEU^-1.164IS^-1.681IS^-1.774IU^-1.932Classification of Items:US - Internal-Stable^IU - Internal-UnstableES - External-Stable EU - External-UnstableItems with standard (Z) scores of greater or less than one arepresented. Scores with positive scores are most preferred;those with negative scores are least preferred.125FACTOR 2 - MIDDLE CLASS NATIVE: PERSON-CENTERED PATTERN 1* * * ANALYSIS OF VARIANCE * * *PATTERN Q-SCORE (PA)BY INTERNAL/EXTERNAL (IE)STABLE/UNSTABLE (SU)Source of VariationSum ofSquares DFMeanSquare FSignifof FMain Effects 36.250 2 18.125 12.156 .000IE 15.125 1 15.125 10.144 .004SU 21.125 1 21.125 14.168 .0012-way Interactions 2.000 1 2.000 1.341 .257IE^SU 2.000 1 2.000 1.341 .257Explained 38.250 3 12.750 8.551 .000Residual 41.750 28 1.491Total 80.000 31 2.581* * * CELL MEANS * * *Stable^Unstable^TOTALInternal^5.25 4.13^4.69External^4.38^ 2.25 3.31TOTAL 4.89 3.19126FACTOR 2 - MIDDLE CLASS NATIVE: PERSON-CENTERED PATTERN 1* * *^CONTENT^OF^Q-SORT * *^*Item Content Type Z -ScoreExamine premises of client's worldview^ IS 1.928Examine premises and suppositions under-lying client's expressed beliefs^IS 1.729Exploring client's social and culturalcontext, including support network^ES 1.566Exploring and analyzing client'scurrent life tasks^ ES 1.358Values clarification IS 1.358Mediating communication between familymembers^ EU -1.258Directing client to exaggerate thesymptom IU -1.267Contracting for new behaviors^ EU -1.348Contingency and reinforcementmanagement^ EU -1.358Setting specific, observable, measure-able goals for counselling outcome^EU -1.829Generating a specific behavioralproblem definition^ EU -2.037Classification of Items:US - Internal-Stable^IU - Internal-UnstableES - External-Stable EU - External-UnstableItems with standard (Z) scores of greater or less than one arepresented. Scores with positive scores are most preferred;those with negative scores are least preferred.127FACTOR 2 - POOR CAUCASIAN: FAMILY COUNSELLOR PATTERN* * * ANALYSIS OF VARIANCE * * *PATTERN Q-SCORE (PA)BY INTERNAL/EXTERNAL (IE)STABLE/UNSTABLE (SU)^Sum of^Mean^SignifSource of Variation Squares DF Square F^of FMain Effects 19.625 2 9.812 6.072 .006IE 15.125 1 15.125 9.359 .005SU 4.500 1 4.500 2.785 .1062-way Interactions 15.125 1 15.125 9.359 .005IE^SU 15.125 1 15.125 9.359 .005Explained 34.750 3 11.583 7.168 .001Residual 45.250 28 1.616Total 80.000 31 2.581*** CELL MEANS ***Stable^Unstable^TOTALInternal 3.00 3.63 3.31External 5.75 3.63 4.69TOTAL 4.38 3.63128FACTOR 2 - POOR CAUCASIAN: FAMILY COUNSELLOR PATTERN*** CONTENT OF Q-SORT ***Item Content^ Type Z -ScoreExploring client's social and culturalcontext, including support network^ES^2.025Collecting a family history^ ES^1.662Multigenerational pattern analysis andgenogram construction ES^1.615Constructing a "life line" ES^1.477Generating options and desirablescenarios for outcomes^ ES^1.323Tracking interactional patterns ES^1.049Discover and correct irrationalideation^ IS^-1.011Free association IS^-1.206Thought stopping IU^-1.292Directing client to change questionsto statements^ IS^-1.365Analysis of resistance IU^-1.761Analysis of characterological type^IS^-1.762Classification of Items:US - Internal-Stable^IU - Internal-UnstableES - External-Stable EU - External-UnstableItems with standard (Z) scores of greater or less than one arepresented. Scores with positive scores are most preferred;those with negative scores are least preferred.129FACTOR 2 - MIDDLE CLASS CAUCASIAN: PERSON-CENTERED PATTERN 2* * * ANALYSIS OF VARIANCE * * *PATTERN Q-SCORE (PA)BY INTERNAL/EXTERNAL (IE)STABLE/UNSTABLE (SU)Source of VariationSum ofSquares DFMeanSquare FSignifof FMain Effects 15.250 2 7.625 3.323 .051IE .125 1 .125 .054 .817SU 15.125 1 15.125 6.591 .0162-way Interactions .500 1 .500 .218 .644IE^SU .500 1 .500 .218 .644Explained 15.750 3 5.250 2.288 .100Residual 64.250 28 2.295Total 80.000 31 2.581*** CELL MEANS ***Stable^Unstable^TOTALInternal 4.63 3.50 4.06External 4.75 3.13 3.94TOTAL 4.69 3.31130FACTOR 2 - MIDDLE CLASS CAUCASIAN: PERSON-CENTERED PATTERN 2*** CONTENT OF Q- SORT ***Item Content^ Type Z-ScoreExploring client's social and culturalcontext, including support network^ES^2.373Examine premises and suppositions under-lying client's expressed beliefs^IS^1.474Exploring and analyzing client'scurrent life tasks^ ES^1.228Analyze cognitive style and cognitivedistortions^ IS^1.109Generating options and desirablescenarios for outcomes^ ES^1.072Setting specific, observable, measure-able goals for counselling outcome^EU^1.054Directing client to exaggerate thesymptom^ IU^-1.558Mediating communication between familymembers EU^-1.786Analysis of resistance^ IU^-2.199Classification of Items:US - Internal-Stable^IU - Internal-UnstableES - External-Stable EU - External-UnstableItems with standard (Z) scores of greater or less than one arepresented. Scores with positive scores are most preferred;those with negative scores are least preferred.131FACTOR 3 - POOR NATIVE: PERSON CENTERED PATTERN 1* * * ANALYSIS OF VARIANCE * * *PATTERN Q-SCORE (PA)BY INTERNAL/EXTERNAL (IE)STABLE/UNSTABLE (SU)^Sum of^Mean^SignifSource of Variation Squares DF Square F^of FMain Effects 38.250 2 19.125 14.376 .000IE 10.125 1 10.125 7.611 .010SU 28.125 1 28.125 21.141 .0002-way Interactions 4.500 1 4.500 3.383 .077IE^SU 4.500 1 4.500 3.383 .077Explained 42.750 3 14.250 10.711 .000Residual 37.250 28 1.330Total 80.000 31 2.581* * * CELL MEANS * * *Stable^Unstable^TOTALInternal^5.13 4.00^4.56External^4.75^ 2.13 3.44TOTAL 4.94 3.06132FACTOR 3 - POOR NATIVE: PERSON CENTERED PATTERN 1* * * CONTENT OF Q-SORT * * *Item Content^ Type Z -ScoreExamine premises of client's worldview^ IS^2.032Exploring client's social and culturalcontext, including support network^ES^1.732Examine premises and suppositions under-lying client's expressed beliefs^IS^1.590Values clarification^ IS^1.417Exploring and analyzing client'scurrent life tasks ES^1.229Mediating communication between familymembers^ EU^-1.290Contingency and reinforcementmanagement EU -1.417Contracting for new behaviors^ EU -1.544Setting specific, observable, measure-able goals for counselling outcome^EU^-1.811Generating a specific behavioralproblem definition^ EU -1.844Classification of Items:US - Internal-Stable^IU - Internal-UnstableES - External-Stable EU - External-UnstableItems with standard (Z) scores of greater or less than one arepresented. Scores with positive scores are most preferred;those with negative scores are least preferred.133FACTOR 3 - MIDDLE CLASS NATIVE: PERSON-CENTERED PATTERN 2* * * ANALYSIS OF VARIANCE * * *PATTERN Q-SCORE (PA)BY INTERNAL/EXTERNAL (IE)STABLE/UNSTABLE (SU)^Sum of^Mean^SignifSource of Variation Squares DF Square F^of FMain Effects 10.000 2 5.000 2.014 .152IE 2.000 1 2.000 .806 .377SU 8.000 1 8.000 3.223 .0832-way Interactions .500 1 .500 .201 .657IE^SU .500 1 .500 .201 .657Explained 10.500 3 3.500 1.410 .261Residual 69.500 28 2.482Total 80.000 31 2.581* * * CELL MEANS * * *Stable^Unstable^TOTALInternal^4.13 3.38^3.75External^4.88^ 3.63 4.25TOTAL 4.50 3.50134FACTOR 3 - MIDDLE CLASS NATIVE: PERSON-CENTERED PATTERN 2*** CONTENT OF Q-SORT ***Item Content^ Type Z -ScoreExamine premises of client's worldview^ IS^1.916Exploring client's social and culturalcontext, including support network^ES^1.495Examine premises and suppositions under-lying client's expressed beliefs^IS^1.380Generating options and desirablescenarios for outcomes^ ES^1.351Client self-monitoring of thoughts and^IU^1.105Setting specific, observable, measure-able goals for counselling outcome^EU^1.097Collecting a family history^ ES^1.095Directing client to exaggerate thesymptom^ IU^-1.069Assertiveness training^ ES^-1.098Analysis of characterological type^IS^-1.321Mediating communication between familymembers^ EU^-1.381Directing client to change questionsto statements^ IS^-1.647Analysis of resistance IU^-2.057Classification of Items:US - Internal-Stable^IU - Internal-UnstableES - External-Stable EU - External-UnstableItems with standard (Z) scores of greater or less than one arepresented. Scores with positive scores are most preferred;those with negative scores are least preferred.135FACTOR 3 - POOR CAUCASIAN: FAMILY COUNSELLOR PATTERN 2* * *^ANALYSIS^OF^VARIANCE * *^*PATTERN Q-SCORE (PA)BY^INTERNAL/EXTERNAL (IE)STABLE/UNSTABLE (SU)Source^Sum of^Meanof Variation^Squares DF^Square^FSignifof FMain Effects 22.625 2 11.312 6.180 .006IE 12.500 1 12.500 6.829 .014SU 10.125 1 10.125 5.532 .0262-way Interactions 6.125 1 6.125 3.346 .078IE^SU 6.125 1 6.125 3.346 .078Explained 28.750 3 9.583 5.236 .005Residual 51.250 28 1.830Total 80.000 31 2.581*** CELL MEANS ***Stable^Unstable^TOTALInternal 3.50 3.25 3.38External 5.63 3.63 4.63TOTAL 4.56 3.44136FACTOR 3 - POOR CAUCASIAN: FAMILY COUNSELLOR PATTERN 2* * * CONTENT OF Q-SORT * * *Item Content^ Type Z -ScoreExploring and analyzing client'scurrent life tasks^ ES^2.282Collecting a family history ES^1.513Values clarification IS^1.448Setting specific, observable, measure-able goals for counselling outcome^EU^1.376Tracking interactional patterns^ES^1.239Generating options and desirablescenarios for outcomes ES^1.074Free association^ IS^-1.226Directing client to exaggerate thesymptom IU^-1.241Analysis of characterological type^IS^-1.622Analysis of resistance^ IU^-2.005Classification of Items:US - Internal-Stable^IU - Internal-UnstableES - External-Stable EU - External-UnstableItems with standard (Z) scores of greater or less than one arepresented. Scores with positive scores are most preferred;those with negative scores are least preferred.137FACTOR 3 - MIDDLE CLASS CAUCASIAN: PERSON-CENTERED PATTERN 1* * * ANALYSIS OF VARIANCE * * *PATTERN Q-SCORE (PA)BY INTERNAL/EXTERNAL (IE)STABLE/UNSTABLE (SU)^Sum of^Mean^SignifSource of Variation Squares DF Square F^of FMain Effects 33.625 2 16.812 10.884 .000IE 12.500 1 12.500 8.092 .008SU 21.125 1 21.125 13.676 .0012-way Interactions 3.125 1 3.125 2.023 .166IE^SU 3.125 1 3.125 2.023 .166Explained 36.750 3 12.250 7.931 .001Residual 43.250 28 1.545Total 80.000 31 2.581* * * CELL MEANS * * *Stable^Unstable^TOTALInternal^5.13 4.13^4.63External^4.50^ 2.25 3.38TOTAL 4.81 3.19138FACTOR 3 - MIDDLE CLASS CAUCASIAN: PERSON-CENTERED PATTERN 1* * *^CONTENT^OF^Q-SORT * *^*Item Content Type Z-ScoreExamine premises and suppositions under-lying client's expressed beliefs^IS 1.947Examine premises of client's worldview^ IS 1.763Exploring client's social and culturalcontext, including support network^ES 1.433Client self-monitoring of thoughts and IU 1.103Exploring and analyzing client'scurrent life tasks^ ES 1.046Values clarification IS 1.046Pinpointing behavioral targets forchange^ EU -1.204Setting specific, observable, measure-able goals for counselling outcome^EU -1.275Directing client to exaggerate thesymptom^ IU -1.424Contingency and reinforcementmanagement EU -1.433Contracting for new behaviors^ EU -1.534Generating a specific behavioralproblem definition^ EU -1.956Classification of Items:US - Internal-Stable^IU - Internal-UnstableES - External-Stable EU - External-UnstableItems with standard (Z) scores of greater or less than one arepresented. Scores with positive scores are most preferred;those with negative scores are least preferred.139FACTOR 4 - POOR CAUCASIAN: PERSON CENTERED PATTERN 2* * * ANALYSIS OF VARIANCE * * *PATTERN Q-SCORE (PA)BY INTERNAL/EXTERNAL (IE)STABLE/UNSTABLE (SU)^Sum of^Mean^SignifSource of Variation Squares DF Square F^of FMain Effects 14.125 2 7.062 3.151 .058IE 8.000 1 8.000 3.570 .069SU 6.125 1 6.125 2.733 .1092-way Interactions 3.125 1 3.125 1.394 .248IE^SU 3.125 1 3.125 1.394 .248Explained 17.250 3 5.750 2.566 .075Residual 62.750 28 2.241Total 80.000 31 2.581*** CELL MEANS ***Stable^Unstable^TOTALInternal 4.63 4.38 4.50External 4.25 2.75 3.50TOTAL 4.44 3.56140FACTOR 4 - POOR CAUCASIAN: PERSON CENTERED PATTERN 2* * * CONTENT OF Q-SORT * * *Item Content^ Type Z -ScoreExploring client's social and culturalcontext, including support network^ES^2.032Analyze cognitive style and cognitivedistortions^ IS^1.669Examine premises and suppositions under-lying client's expressed beliefs^IS^1.233Client self-monitoring of thoughts and IU^1.233Thought stopping^ IU^1.113Discover and correct irrationalideation IS^1.113Assigning therapeutic tasks to disruptproblematic interactions^ EU^-1.113Role-playing of alternative behaviors^EU -1.233Tracking interactional patterns ES^-1.354Directing client to change questionsto statements^ IS^-1.475Mediating communication between familymembers^ EU -1.911Analysis of resistance IU^-2.032Appendix EAnalysis of Variance for Q-Scores ofIndividual Subjects141142Analyses of Variance For Individual SubjectsSubject's Q-ScoresBY Ethnicity (ET)Socioeconomic Status (SS)Internal/External (IE)Stable/Unstable (SU)Subject # 1Source of VariationSum ofSquares DFMeanSquare FSignifof FMain Effects 43.281 4 10.820 5.281 .001ET .000 1 .000 .000 1.000SS .000 1 .000 .000 1.000IE 32.000 1 32.000 15.617 .000SU 11.281 1 11.281 5.505 .0212-way Interactions 43.594 6 7.266 3.546 .003ET^SS .000 1 .000 .000 1.000ET IE 2.000 1 2.000 .976 .325ET^SU 2.531 1 2.531 1.235 .269SS IE .500 1 .500 .244 .622SS^SU .281 1 .281 .137 .712IE SU 38.281 1 38.281 18.682 .000Explained 90.500 15 6.033 2.944 .001Residual 229.500 112 2.049Total 320.000 127 2.520143Subject # 2Source of VariationSum ofSquares DFMeanSquare FSignifof FMain Effects 80.094 4 20.023 9.918 .000ET .008 1 .008 .004 .951SS .008 1 .008 .004 .951IE 6.570 1 6.570 3.254 .074SU 73.508 1 73.508 36.409 .0002-way Interactions 12.172 6 2.029 1.005 .426ET^SS .008 1 .008 .004 .951ET IE .945 1 .945 .468 .495ET^SU 2.258 1 2.258 1.118 .293SS IE .383 1 .383 .190 .664SS^SU .070 1 .070 .035 .852IE SU 8.508 1 8.508 4.214 .042Explained 94.867 15 6.324 3.133 .000Residual 226.125 112 2.019Total 320.992 127 2.527Subject # 3Sum of Mean SignifSource of Variation Squares DF Square F of FMain Effects 56.562 4 14.141 8.848 .000ET .031 1 .031 .020 .889SS .031 1 .031 .020 .889IE 24.500 1 24.500 15.330 .000SU 32.000 1 32.000 20.022 .0002-way Interactions 84.562 6 14.094 8.818 .000ET^SS .031 1 .031 .020 .889ET IE 2.000 1 2.000 1.251 .266ET^SU .000 1 .000 .000 1.000SS IE .125 1 .125 .078 .780SS^SU 1.125 1 1.125 .704 .403IE SU 81.281 1 81.281 50.858 .000Explained 142.969 15 9.531 5.964 .000Residual 179.000 112 1.598Total 321.969 127 2.535144Subject # 4Source of VariationSum ofSquares DFMeanSquare FSignifof FMain Effects 79.906 4 19.977 13.239 .000ET .000 1 .000 .000 1.000SS .000 1 .000 .000 1.000IE 66.125 1 66.125 43.822 .000SU 13.781 1 13.781 9.133 .0032-way Interactions 42.844 6 7.141 4.732 .000ET^SS .000 1 .000 .000 1.000ET IE 18.000 1 18.000 11.929 .001ET^SU .281 1 .281 .186 .667SS IE 12.500 1 12.500 8.284 .005SS^SU 11.281 1 11.281 7.476 .007IE SU .781 1 .781 .518 .473Explained 151.000 15 10.067 6.671 .000Residual 169.000 112 1.509Total 320.000 127 2.520Subject # 5Sum of Mean SignifSource of Variation Squares DF Square F of FMain Effects 57.250 4 14.312 7.023 .000ET .000 1 .000 .000 1.000SS .000 1 .000 .000 1.000IE 21.125 1 21.125 10.366 .002SU 36.125 1 36.125 17.726 .0002-way Interactions 24.250 6 4.042 1.983 .074ET^SS .000 1 .000 .000 1.000ET IE 2.000 1 2.000 .981 .324ET^SU .500 1 .500 .245 .621SS IE .125 1 .125 .061 .805SS^SU .500 1 .500 .245 .621IE SU 21.125 1 21.125 10.366 .002Explained 91.750 15 6.117 3.001 .000Residual 228.250 112 2.038Total 320.000 127 2.520145Subject # 6Source of VariationSum ofSquares DFMeanSquare FSignifof FMain Effects 62.344 4 15.586 7.623 .000ET .031 1 .031 .015 .902SS .000 1 .000 .000 1.000IE 42.781 1 42.781 20.924 .000SU 19.531 1 19.531 9.552 .0032-way Interactions 26.469 6 4.411 2.158 .052ET^SS .031 1 .031 .015 .902ET IE .125 1 .125 .061 .805ET^SU .000 1 .000 .000 1.000SS IE 1.531 1 1.531 .749 .389SS^SU .281 1 .281 .138 .711IE SU 24.500 1 24.500 11.983 .001Explained 91.000 15 6.067 2.967 .001Residual 229.000 112 2.045Total 320.000 127 2.520Subject # 7Sum of Mean SignifSource of Variation Squares DF Square F of FMain Effects 50.781 4 12.695 5.479 .000ET .000 1 .000 .000 1.000SS .000 1 .000 .000 1.000IE 26.281 1 26.281 11.343 .001SU 24.500 1 24.500 10.574 .0022-way Interactions 9.094 6 1.516 .654 .687ET^SS .000 1 .000 .000 1.000ET IE .031 1 .031 .013 .908ET^SU .000 1 .000 .000 1.000SS IE .031 1 .031 .013 .908SS^SU .000 1 .000 .000 1.000IE SU 9.031 1 9.031 3.898 .051Explained 60.500 15 4.033 1.741 .053Residual 259.500 112 2.317Total 320.000 127 2.520146Subject # 8Source of VariationSum ofSquares DFMeanSquare FSignifof FMain Effects 144.000 4 36.000 24.000 .000ET .000 1 .000 .000 1.000SS .000 1 .000 .000 1.000IE 72.000 1 72.000 48.000 .000SU 72.000 1 72.000 48.000 .0002-way Interactions 8.000 6 1.333 .889 .506ET^SS .000 1 .000 .000 1.000ET IE .000 1 .000 .000 1.000ET^SU .000 1 .000 .000 1.000SS IE .000 1 .000 .000 1.000SS^SU .000 1 .000 .000 1.000IE SU 8.000 1 8.000 5.333 .023Explained 152.000 15 10.133 6.756 .000Residual 168.000 112 1.500Total 320.000 127 2.520Subject # 9Sum of Mean SignifSource of Variation Squares DF Square F of FMain Effects 62.156 4 15.539 7.236 .000ET .000 1 .000 .000 1.000SS .000 1 .000 .000 1.000IE 7.031 1 7.031 3.274 .073SU 55.125 1 55.125 25.672 .0002-way Interactions 11.469 6 1.911 .890 .505ET^SS .000 1 .000 .000 1.000ET IE .281 1 .281 .131 .718ET^SU 2.000 1 2.000 .931 .337SS IE .031 1 .031 .015 .904SS^SU .125 1 .125 .058 .810IE SU 9.031 1 9.031 4.206 .043Explained 79.500 15 5.300 2.468 .004Residual 240.500 112 2.147Total 320.000 127 2.520147Subject # 10Source of VariationSum ofSquares DFMeanSquare FSignifof FMain Effects 49.906 4 12.477 5.339 .001ET .000 1 .000 .000 1.000SS .000 1 .000 .000 1.000IE 36.125 1 36.125 15.457 .000SU 13.781 1 13.781 5.897 .0172-way Interactions 2.594 6 .432 .185 .980ET^SS .000 1 .000 .000 1.000ET IE .125 1 .125 .053 .818ET^SU .781 1 .781 .334 .564SS IE 1.125 1 1.125 .481 .489SS^SU .281 1 .281 .120 .729IE SU .281 1 .281 .120 .729Explained 58.250 15 3.883 1.662 .069Residual 261.750 112 2.337Total 320.000 127 2.520Subject # 11Sum of Mean SignifSource of Variation Squares DF Square F of FMain Effects 63.125 4 15.781 7.207 .000ET .000 1 .000 .000 1.000SS .000 1 .000 .000 1.000IE 8.000 1 8.000 3.653 .059SU 55.125 1 55.125 25.174 .0002-way Interactions 11.375 6 1.896 .866 .522ET^SS .000 1 .000 .000 1.000ET IE .125 1 .125 .057 .812ET^SU 2.000 1 2.000 .913 .341SS IE .125 1 .125 .057 .812SS^SU 1.125 1 1.125 .514 .475IE SU 8.000 1 8.000 3.653 .059Explained 74.750 15 4.983 2.276 .008Residual 245.250 112 2.190Total 320.000 127 2.520148Subject # 12Source of VariationSum ofSquares DFMeanSquare FSignifof FMain Effects 152.500 4 38.125 29.860 .000ET .000 1 .000 .000 1.000SS .000 1 .000 .000 1.000IE 8.000 1 8.000 6.266 .014SU 144.500 1 144.500 113.175 .0002-way Interactions 24.500 6 4.083 3.198 .006ET^SS .000 1 .000 .000 1.000ET IE .000 1 .000 .000 1.000ET^SU .000 1 .000 .000 1.000SS IE .000 1 .000 .000 1.000SS^SU .000 1 .000 .000 1.000IE SU 24.500 1 24.500 19.189 .000Explained 177.000 15 11.800 9.242 .000Residual 143.000 112 1.277Total 320.000 127 2.520


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items