Psychotherapy Research
                 ISSN: 1050-3307 (Print) 1468-4381 (Online) Journal homepage: http://www.tandfonline.com/loi/tpsr20
             Goldilocks on the couch: Moderate levels
             of psychodynamic and process-experiential
             technique predict outcome in psychodynamic
             therapy
             Kevin S. McCarthy, John R. Keefe & Jacques P. Barber
             To cite this article: Kevin S. McCarthy, John R. Keefe & Jacques P. Barber (2016) Goldilocks
             on the couch: Moderate levels of psychodynamic and process-experiential technique
             predict outcome in psychodynamic therapy, Psychotherapy Research, 26:3, 307-317, DOI:
             10.1080/10503307.2014.973921
             To link to this article:  http://dx.doi.org/10.1080/10503307.2014.973921
                 Published online: 03 Nov 2014.
                 Submit your article to this journal 
                 Article views: 276
                 View related articles 
                 View Crossmark data
                 Citing articles: 1 View citing articles 
                         Full Terms & Conditions of access and use can be found at
                    http://www.tandfonline.com/action/journalInformation?journalCode=tpsr20
    Download by: [University of Pennsylvania]              Date: 23 May 2017, At: 09:34
                Psychotherapy Research, 2016
                Vol. 26, No. 3, 307–317, http://dx.doi.org/10.1080/10503307.2014.973921
                EMPIRICALPAPER
                Goldilocks on the couch: Moderate levels of psychodynamic and
                process-experiential technique predict outcome in psychodynamic
                therapy
                                            1,2                     2                               3
                KEVINS. MCCARTHY ,JOHNR.KEEFE ,&JACQUESP.BARBER
                1Department of Psychology, Chestnut Hill College, Philadelphia, PA, USA; 2Department of Psychology, University of
                Pennsylvania, Philadelphia, PA, USA & 3Derner Institute, Adelphi University, Garden City, NY, USA
                (Received 8 November 2013; revised 5 August 2014; accepted 29 September 2014)
                Abstract
                Objectives: Greater symptom change is often assumed to follow greater technique use, a “more is better” approach.
                We tested whether psychodynamic techniques, as well as common factors and techniques from other orientations, had a
                curvilinear relation to outcome (i.e., whether moderate or “just right” intervention levels predict better outcome than lower
                or higher levels). Methods: For 33 patients receiving supportive-expressive psychodynamic psychotherapy for depression,
                interventions were assessed at Week 4 using the multitheoretical list of therapeutic interventions and symptoms were rated
                with the Hamilton Rating Scale for Depression. Results: Moderate psychodynamic and experiential techniques predicted
                greater symptom change compared to lower or higher levels. Conclusion: This “Goldilocks effect” suggests a more complex
                relation of intervention use to outcome might exist.
                Keywords: psychoanalytic/psychodynamic therapy; experiential/existential/humanistic psychotherapy; outcome research
                Modern psychodynamic therapy incorporates mul-           outcome (cf., Stiles, 1996; Stiles, Honos-Webb, &
                tiple theories from its 120-year history to help         Surko, 1998; Stiles & Shapiro, 1989, 1994)ora
                patients change (Mitchell & Black, 1996; Summers         “more is better” approach (Barber, 2009; Stiles &
                &Barber, 2009). Even with this diversity of thought,     Shapiro, 1989). For instance, quantitative investiga-
                most psychodynamic thinkers and practitioners            tions of the relation of techniques to outcome have
                agree on several core techniques that constitute the     almost exclusively relied on the use of linear correla-
                work of psychodynamic therapy (Blagys & Hilsen-          tion models (for reviews, see Stiles & Shapiro, 1989;
                roth, 2000; Summers & Barber, 2009). Among               Webb, DeRubeis, & Barber, 2010).
                them, supportive interventions reinforce adaptive           However, the empirical association of dynamic
                responses (ego defenses) in the patient. These           interventions to symptom change remains unclear
                interventions include relationship building, sugges-     (for reviews, see Barber, Muran, McCarthy, &
                tion or advice giving, and boundary setting. Express-    Keefe, 2013; Høglend, 2004). By and large, investi-
                ive techniques explore and uncover (express) the         gations of global measures of psychodynamic inter-
                unconscious conflict behind the patient’s symptoms       ventions and symptom change have been equivocal
                (Luborsky, 1984). These interventions include            (for no relation, see Barber, Crits-Christoph, &
                exploration of affect and interpersonal themes and       Luborsky, 1996; DeFife, Hilsenroth, & Gold, 2008;
                making connections between different relationships       Ogrodniczuk & Piper, 1999; Ogrodniczuk, Piper,
                in the patient’s life (transference interpretations).    Joyce, & McCallum, 2000; for a favorable relation,
                Most often it is assumed that greater use of these       see Ablon & Jones, 1998; Gaston, Thompson,
                techniques is likely to lead to improved patient         Gallagher, Cournoyer, & Gagnon, 1998; Hendriksen
                Correspondence concerning this article should be addressed to Kevin S. McCarthy, Department of Psychology, Chestnut Hill College,
                Philadelphia, PA, USA. Email: kevin.mccarthy@chc.edu
                ©2014 Society for Psychotherapy Research
                    308 K. S. McCarthy et al.
                    et al., 2011; Hilsenroth, Ackerman, Blagys, Baity,                          meta-analysis that varied from strongly negative to
                    & Mooney, 2003; Luborsky, McLellan, Woody,                                  strongly positive, which lends partial support to a
                    O’Brien, & Auerbach, 1985; for an unfavorable                               curvilinear hypothesis.
                    relation, see Barber et al., 2008). In the only meta-                          It is interesting and perhaps surprising that only a
                    analysis to date on the subject, Webb et al. (2010)                         few studies have tested a curvilinear relation of
                    found no significant relation between adherence                             dynamic interventions and symptom change. In the
                    (i.e., the degree to which therapists deliver theory-                       first  empirical study of hypothesis, only partial
                    specific interventions which are consistent with a                          support for a curvilinear relation between interpreta-
                    therapy manual) and symptomatic improvement                                 tion and outcome was found (Piper et al., 1991).
                    across 32 studies of psychotherapy outcome, regard-                         Two later correlational studies did not produce
                    less of the therapeutic modality under study (e.g.,                         evidence that moderate dynamic technique use was
                    dynamic, interpersonal, cognitive-behavioral, and                           associated with better outcome than lower or higher
                    process-experiential).                                                      levels of technique use (Barber et al., 2008; Ogrod-
                       While the disappointing link between technique                           niczuk & Piper, 1999). Indeed, the opposite effect
                    use and outcome is not unique to psychodynamic                              was observed for a sample of patients in dynamic
                    therapy alone (Stiles, 1996; Wampold, 2001; Webb                            therapy for cocaine dependence (very high and very
                    et al., 2010), it raises the question about how                             low level of dynamic interventions were related to
                    techniques might have their effect in therapy. Some                         better outcome than were moderate levels; Barber
                                                                                                                  1
                    researchers have productively explored the interac-                         et al., 2008).      A unique experimental study tested
                    tion of dynamic techniques and therapeutic alliance                         whether patients with mixed diagnoses improved
                    in predicting outcome (e.g., Barber et al., 2008;                           more when randomly assigned to psychodynamic
                    Gaston et al., 1998; Høglend et al., 2011; Owen &                           treatment with a moderate level of transference
                    Hilsenroth, 2011). Others have examined how com-                            interpretations (1–3 per session) or a low level of
                    petent delivery of dynamic techniques relates to                            interpretations (Høglend et al., 2006). Patients
                    symptomchange(e.g., Barber et al., 1996). Notably,                          receiving both low and moderate levels of interpreta-
                    Stiles and colleagues (1989, 1994, 1996, 1998) have                         tion improved significantly over time, but there was
                    cautioned against the expectation of a simple cor-                          little difference in the amount that patients in each
                    relation between process and outcome factors and                            condition changed (although moderating effects of
                    have suggested the responsiveness of the therapist to                       patient characteristics were later found, Høglend
                    the patient’s current needs is what might facilitate                        et al., 2011). The researchers chose not to include
                    change. This hypothesis agrees with the experience                          a high interpretation level condition in their design
                    of many dynamic practitioners. The psychodynamic                            because their review of naturalistic studies of psycho-
                    clinical literature has long suggested that too many                        dynamic process and outcome suggested that such a
                    interpretative interventions may be disruptive to the                       condition was unlikely to be effective in terms of
                    patient’s functioning (Strachey, 1934; Gill, 1982).                         outcome and cost (Per Høglend, personal commun-
                    Too many supportive interventions may overly grat-                          ication 02/15/10). A curvilinear relation may there-
                    ify the patient and mask his or her symptoms without                        fore still exist between psychodynamic techniques
                    bringing longer term relief or might trigger an                             and outcome in patients with internalizing disorders.
                    enactment of his or her conflict in the therapeutic                            Delivery of any psychotherapy is also likely to
                    relationship (Freud, 1919; Gill, 1951; Kohut &                              involve interventions from numerous schools of
                    Wolf, 1978).                                                                treatment. For example, it has been demonstrated
                       Accordingly, some researchers have hypothesized                          that psychodynamic therapy contains modest levels
                    a curvilinear relation between psychodynamic inter-                         of techniques from other therapy systems, like drug
                    ventions and outcome (Piper, Azim, Joyce, &                                 counseling (Barber et al., 2008), cognitive therapy
                    McCallum, 1991; see also Barber et al., 2008;                               (DeFife et al., 2008), and a number of different
                    Høglend et al., 2006; Ogrodniczuk & Piper, 1999).                           other theoretical orientations (McCarthy & Barber,
                    Namely, moderate levels of dynamic interventions                            2009; Trijsburg et al., 2002). The effect of dynamic
                    might be related to greater symptom improvement                             interventions on symptom improvement in other
                    than might very high and very low levels (a “just                           types of therapies has been reviewed elsewhere (see
                    right” hypothesis). The equivocal results of previous                       Shedler, 2010), but considerations of how techni-
                    studies might have either represented a single leg of                       ques from other therapies might influence outcome
                    the curvilinear relation (either the positive or negat-                     in dynamic therapy have been less systematic. In a
                    ive association of techniques to symptom improve-                           handful of studies, unintended techniques have had
                    ment) or might have represented a linear model                              a modest to substantial contribution to outcome in
                    being forced on curvilinear data. Webb and collea-                          dynamic psychotherapy (Ablon & Jones, 1998;
                    gues (2010) observed individual effect sizes in their                       Barber et al., 2008; DeFife et al., 2008; Luborsky
                                                                                                          Psychotherapy Research     309
                 et al., 1985; but see also Hilsenroth et al., 2003).              Fifty-one patients were initially randomized to
                 However, in each of these studies interventions from           psychodynamic therapy. However, nine dropped
                 only a few psychotherapy systems have been exam-               out of treatment before the first process measure-
                 ined (e.g., Luborsky et al., 1985), which under-               ment point, five of whom never attended their first
                 represents the variety of interventions from different         therapy appointment. Of these nine patients, one
                 systems that dynamic therapists might possibly use.            described practical reasons for dropping out of
                 A curvilinear relation of unintended techniques to             treatment (e.g., childcare and transportation), two
                 outcome might also exist as we hypothesize for                 stated that the time commitment was too great, three
                 dynamic therapy techniques. Very high levels of                wanted a treatment other than dynamic therapy, and
                 interventions from any system might represent tech-            three did not say why they exited the study or could
                 nical rigidity or inflexibility to the patient’s needs on      not be reached. Eight of the nine patients leaving
                 the part of the therapist (Miller & Binder, 2002),             treatment before Week 8 were of minority status
                 whereas very low levels of interventions from any              (i.e., African-American or Latino). An additional
                 system may not be sufficient to motivate change in             nine patients discontinued treatment after Week 4 or
                 the patient.                                                   declined to participate in a termination interview.
                   The present study tests the curvilinear relation of          Onemovedawayfromthearea, one dropped out for
                 a wide range of interventions from a number of                 practical reasons, two felt improved and did not
                 theoretical   orientations   to  subsequent outcome            want further treatment, three felt they were not
                 in psychodynamic therapy for depression. More                  improving fast enough, one was dissatisfied with
                 specifically, we hypothesize that moderate levels of           her treatment, and one did not say. Five of these
                 psychodynamic interventions (e.g., exploration and             nine patients leaving treatment before termination
                 interpretation) will predict more symptom improve-             were of minority status. The subsample of 33
                 ment than will higher or lower levels of dynamic               completers did not differ significantly from the larger
                 interventions. We also hypothesize that supportive             sample on demographic characteristics, process
                 interventions (e.g., common factors like providing             measures, or outcome measures.
                 hope and fostering therapeutic alliance) will be                  Sixty-one percent of patients were female (n = 20).
                 associated with outcome, although we have no                   Mean age was 35.5 years (SD = 12.0, range = 19–
                 strong predictions for whether this relation will be a         58). Six percent identified their primary race or
                                                                                ethnicity as Asian (n = 2), 49% as African-American
                 linear or curvilinear function. We will also explore           (n = 16), 3% as Latino/a (n = 1), and 42% as
                 the contribution of interventions not intended to be           Caucasian (n = 14). The majority of clients were
                 found in dynamic therapy (i.e., behavioral, cognitive,         single (n = 21, 64%), with fewer clients separated/
                 dialectical-behavioral,   interpersonal,    person-cen-        divorced (n = 2, 6%) or married or cohabiting (n =
                 tered, and process-experiential) to outcome. We do             9, 27%), and 1 (3%) widowed. Average number of
                 not have specific directional hypotheses for the               years of education completed was 14.3 (SD = 2.2,
                 potential relations of interventions from these sys-           range = 10–19). Fifty-five percent worked either full-
                 tems to outcome, but will explore for their linear and         (n = 14) or part-time (n = 4), 36% (n = 12) were
                 curvilinear relation to outcome for interventions              unemployed/disabled, and 9% (n = 3) were students.
                 from these different systems separately.
                                                                                   Therapists. Therapy was provided by four Ph.D.-
                                                                                level psychologists (three were female) with an
                                        Methods                                 average of more than 15 years of psychotherapy
                 Participants                                                   experience at the beginning of the trial. All therapists
                                                                                were between the ages of 40 and 50. All had received
                   Patients. Participants were 33 patients complet-             training in psychodynamic therapy prior to participa-
                 ing treatment in the psychotherapy arm of a rando-             tion in this study and had achieved acceptable levels
                 mized     controlled      trial   (RCT)      comparing         of adherence and competence using the Penn Adher-
                 psychodynamic therapy versus pharmacotherapy ver-              ence-Competence Scale (Barber & Critis-Christoph,
                 sus pill placebo. To be included in the study                  1996). The median number of clients that each
                 participants were required to have a primary dia-              therapist treated in this sample was eight.
                 gnosis of major depressive disorder based on their
                 responses to the structured clinical interview for             Treatment
                 DSM-IV(Axis I) administered by a trained diagnos-
                 tician. They could not have lifetime history of                The psychodynamic therapy conducted in this study
                 bipolar or psychotic disorder nor a substance abuse            followed a supportive-expressive (SE) treatment
                 or dependence disorder in the previous 6 months.               model (Luborsky, 1984) with specific adaptations