Patient Participation in Psychodynamic Psychotherapy: Contributions of Alliance and Therapist Technique

Introduction

The contemporary understanding of alliance is largely derived from the “here and now” interactions during therapy (Horvath, 2006) and Bordin’s (1979) conceptualization of agreement on goals, tasks, and the bond between patient and therapist. A strong alliance can help the patient delve with greater ease into understanding and exploring more difficult dynamics (Byrd, Patterson & Turchik, 2010). Extensive research demonstrates that this collaborative relationship is significantly related to outcome (Horvath, Del Re, Fluckiger, & Symonds, 2011). Horvath et al. (2011) stressed that key components in fostering a strong alliance include the therapist implementing a collaborative approach, possessing a nondefensive style, and demonstrating flexibility in his or her technique. Intervention and alliance building are not separate; often, they are related constructs, as effective treatment builds on the alliance (Horvath et al., 2011).

It is important to note that a patient concealing difficult emotional material is especially understandable if the therapeutic alliance is a poor one; thus, self-concealment can be viewed as self-protective in the presence of a poor therapeutic alliance (Pattee & Farber, 2008). Substantial research supports a related assertion by Hill, Gelso, and Mohr (2000) that honest involvement during a session is important for the development of a positive alliance. In addition, Kahn et al. (2008) demonstrated that patient disclosure is most profound when the therapist addresses the patient’s emotional experience. Emotional content, or a high level of affect during sessions, was correlated with a high-impact session (Kahn et al., 2000).

The aim of this study was to investigate the relationship between patient participation and the working alliance during early treatment session process as well as to examine the association of such a relationship with therapist technique. Patient engagement during session is an important facet of effective treatment, since disclosure in therapy is beneficial to the therapeutic process (Farber et al., 2004). Therapist interventions affect alliance building, especially early in treatment, from the psychological assessment, to the initial interview, and the first session (Hilsenroth, Cromer & Ackerman, 2012). Both therapeutic alliance and the techniques implemented in psychotherapy seem important to promote positive change. However, effective technique can go only so far in treatment without the presence of a strong alliance (Goldfried & Davila, 2005; Owen & Hilsenroth, 2011). They are likely linked constructs, as effective treatment in and of itself builds on alliance, and adequate alliance early in therapy is important for continued exploration and working through (Owen, Quirk, Hilsenroth, & Rodolfa, 2012).

The first hypothesis of this study is that in early sessions of psychodynamic psychotherapy, greater use of psychodynamic technique will be related to greater patient participation. Second, we hypothesized that higher levels of the overall working alliance will be associated with greater patient participation in early session process during psychodynamic psychotherapy. The third hypothesis is that specific psychodynamic-interpersonal techniques will be related to greater patient participation in early session process during psychodynamic psychotherapy. Fourth, we hypothesized that higher levels of various facets of working alliance (e.g., subscales such as Bond) will be related to greater patient participation in early session process during psychodynamic psychotherapy. Finally, we decided that if results demonstrated that both alliance and technique were related to patient participation, then mediation analyses would be conducted to ascertain the unique contribution of alliance and technique with patient participation. We anticipated that the working alliance would mediate any relationship between technique and patient participation.

Method

Participants

Eighty-eight participants were drawn from a patient sample at a university-based community outpatient psychological clinic. The majority of the participants were female (female, 71%; male, 29%). In terms of their relationship status, most identified as single (61%), followed by married (24%), divorced (12%), and widowed (1%). A large majority of the participants identified their ethnic background as Caucasian (90%). The most common DSM-IV diagnosis was Mood Disorder (51%), followed by V-Code (17%), Adjustment (14%), Anxiety (13%), Eating (3%), Substance-Related (1%), and Impulse Disorders (1%). Moreover, slightly more than a majority of participants were diagnosed with a comorbid personality disorder diagnosis (PD diagnosis, 55%; No PD diagnosis, 45%; see Table 1).

Clinicians

Clinicians in the study were 26 advanced doctoral students (13 male and 13 female) enrolled in an American Psychological Association approved Clinical Psychology Ph.D. program. Each clinician received a minimum of three-and-one-half hours of supervision per week (one-and-one-half hours of individual and two hours of group) on the Therapeutic Model of Assessment ([TMA], Finn & Tonsanger 1997; Hilsenroth 2007), clinical interventions, the organization of collaborative feedback, psychodynamic theory and review of videotaped case material. Individual and group supervision focused heavily on the review of videotaped case material and technical interventions. All clinicians were trained in psychodynamic psychotherapy using guidelines delineated by Book (1998), Luborsky (1984), McCullough et al. (2003), and Wachtel (1993), as well as selected readings on psychological assessment, psychodynamic theory, and psychodynamic psychotherapy (for a more detailed description of this training process, see Hilsenroth, DeFife, Blagys, & Ackerman, 2006).

Treatment

Patients first received a psychological evaluation using a Therapeutic Model of Assessment ([TMA]; Finn and Tonsanger, 1997; Hilsenroth, 2007). The TMA uses a multi-method assessment (i.e., interview, self-report, performance tasks and free response measures). Moreover, the TMA focuses sharply on the development and maintenance of empathic connections with patients, factors contributing to maintenance-of-life problems (often relational), collaboration in defining individualized treatment goals and tasks, and the sharing and exploration of assessment results with patients. The TMA used in this study consisted of four components, including three meetings between the patient and clinician that totaled approximately four-and-one-half hours, and an additional patient appointment to complete a battery of self-report measures. The three meetings included (1) a semi-structured diagnostic interview (Westen & Muderrisoglu, 2003, 2006), (2) a follow-up interview, and (3) a collaborative feedback session. The collaborative feedback session emphasized prominent inter- and intra-personal themes derived from the test results, the patient-therapist interaction, and factors contributing to the maintenance-of-life problems. This session also provided an opportunity to explore these new understandings and apply them to the patient’s current problems in living. The patient and clinician also reviewed a socialization interview (SI) (developed by Luborsky, 1984) on what to expect in psychodynamic psychotherapy. The SI delineates the patient’s and clinician’s roles during treatment and highlights the relational focus of the therapeutic process. An aim is for the patient to become more cognizant of dynamics that were not known before the start of psychotherapy. Finally, the clinician and patient work together to develop treatment goals and to negotiate an explicit treatment frame (i.e., scheduling session times, frequency of treatment session[s], missed sessions and payment plan). For this study, the clinician who carried out the psychological assessment was also the clinician who conducted the formal psychotherapy sessions.

Individual psychotherapy consisted of once- or twice-weekly sessions of psychodynamic treatment that were organized, aided, and informed (but not prescribed) by the technical guidelines delineated in the treatment manuals detailed above. Key features of the psychodynamic treatment model utilized in these sessions included (Blagys & Hilsenroth, 2000):

In addition to these areas of treatment focus, relational patterns, case presentations, and symptoms were conceptualized in the context of cyclical patterns (Book, 1998; Luborsky, 1984; McCullough et al., 2003; Wachtel, 1993). Also, the Safran and Muran (2000) model of intervention was used in responding to treatment ruptures and repair as they occurred in the therapeutic relationship. Treatment was open-ended in length, rather than of a fixed duration. Setting a termination date became a frequent area of intervention, since issues related to the termination were often linked to key interpersonal, affective, and thought patterns prominent in that patient’s treatment.

Treatment goals are first explored during the assessment feedback session, and a formal treatment plan is reviewed with each patient early in the course of psychotherapy. This treatment plan is subsequently reviewed at regular intervals for changes, additions, or deletions. Prior to these review points, we reassessed patient functioning on a standard battery of outcome measures as well as process ratings completed by patients and therapists immediately after selected sessions. Patients were informed—both verbally and in writing—that that their therapists would not have access to their responses on any psychotherapy process measure (i.e., alliance, session process, etc.). Also, all treatment sessions (not just the sessions during in which reassessment ratings were completed) were videotaped. Patient process ratings as well as the independent technique ratings for this study were collected at the same point in time, early in therapy (post-TMA assessment), predominantly at the third or fourth sessions. Mean number of sessions attended by these 88 patients was 25 sessions during (an average of) eight months. However, the median number of sessions and length of treatment were somewhat shorter, 21 sessions and six months.

Procedure

All participants were representative of individuals seeking outpatient treatment at a university-based community clinic. Individuals were asked to participate in a research study about patients’ interactions with their therapists. Cases were assigned to treatment practicums and clinicians in an ecologically valid manner based on real world issues including aspects of clinician availability, case load, etc. Moreover, patients were accepted into treatment regardless of disorder or comorbidity. All patients provided informed consent, and they received no monetary compensation for their participation in the study.

All sessions of treatments were videotaped, not just the sessions during which the process measures were completed. Videotapes of an early session (third/fourth) of each patient’s treatment were coded by external raters using the Comparative Psychotherapy Process Scale (CPPS). The Vanderbilt Psychotherapy Process Scale-Short Form (VPPS-S) and The Combined Alliance Short Form-Patient Version (CASF-P) were administered directly after these sessions, and the CPPS was rated by independent coders at a later date, by viewing a videotape from the therapy session.

In order to highlight the contribution and relationship among alliance, technique, and participation, we employed a mediational analysis. Baron and Kenny (1986) explained that the mediator is the catalyst that enables the independent/predictive variable to influence the dependent variable. To establish mediation the following conditions need to be met:

Thus, the following three steps were performed to rule out possible mediation.

Measures

The constructs that were assessed included treatment interventions, which were coded using the Comparative Psychotherapy Process Scale (CPPS; Hilsenroth, Blagys, Ackerman, Bonge & Blais, 2005), and patient participation in the therapeutic process, which was assessed by utilizing the Patient Participation subscale of the Vanderbilt Psychotherapy Process Scale-Short Form (VPPS-S; Smith, Hilsenroth, Baity & Knowles, 2003). The Combined Alliance Short Form-Patient Version (CASF-P; Hatcher & Barends, 1996) was used to assess the therapeutic alliance. Therefore, the data for this study were the patient responses on the VPPS-S, independent ratings of therapeutic techniques using the CPPS from videotaped sessions early in treatment, and patient responses to the CASF-P—all of which were taken from the same early treatment session.

Comparative Psychotherapy Process Scale-External Rater Form (CPPSER; Hilsenroth, Blagys, Ackerman, Bonge, & Blais, 2005). The CPPS is a brief descriptive measure designed to assess therapist activity and technique used during the therapeutic hour. The CPPS-ER is based on the findings of two empirical reviews of the comparative psychotherapy process literature (Blagys & Hilsenroth, 2000; 2002). Based on these reviews, a list of interventions were developed that represent characteristic features of psychodynamic-interpersonal therapy ([PI]; defined broadly to include psychodynamic, psychodynamic-interpersonal, and interpersonal therapies) and cognitive-behavioral ([CB]; defined broadly to include items that are significantly more characteristic of cognitive-behaviorally oriented therapy [Blagys & Hilsenroth, 2002], cognitive, and behavioral therapies). The PI subscale measures the seven domains of therapist activity previously described as key features of the psychodynamic treatment model (Blagys & Hilsenroth, 2000). The CB subscale consists of items which include:

The CPPS measure consists of 20 randomly ordered techniques rated on a 7-point Likert scale ranging from 0—not at all characteristic, 2—somewhat characteristic, 4 — characteristic, to 6—extremely characteristic. The CPPS may be completed by a patient (P), therapist (T), or an external rater (ER). Ten statements are characteristic of PI interventions, and ten statements are characteristic of CB interventions. These interventions can then be organized into two scales: one measuring PI features (CPPS-PI, 10 items), and one measuring CB features (CPPS-CB, 10 items).

The reliability and clinical validity of the CPPS has been well established (see Hilsenroth, 2007 for a review). We have recently reported (Hilsenroth et al., 2005; Slavin-Mulford, Hilsenroth, Weinberger, & Gold, 2011; Stein, Pesale, Slavin, & Hilsenroth, 2010) on the excellent inter-rater reliability and internal consistency of the CPPS, as well as the results of validity analyses conducted across several different contexts and samples. The CPPS data we utilized in the current study are derived from these reports, follow procedures detailed there, and were provided by trained external raters who have demonstrated the ability to rate these individual techniques in the good (ICC [intra-class correlation coefficient] = .60–.74; Fleiss, 1981) to excellent range (ICC ≥ .75; Fleiss, 1981). Several sets of external raters demonstrated good to excellent reliability on the CPPS for the sessions used in the current study (Stein et al., 2010). All Spearman-Brown corrected mean ICCs for the individual CPPS-PI and CPPS-CB techniques were in the good-to-excellent range (and thus were examined individually), as were the ICCs for the CPPS-PI and CPPS-CB scale scores. Corrected average ICCs were reported for both CPPS technique items and subscales; two external raters evaluated all of the sessions, allowing us to use their more reliable average ratings across their pairs. In the current study, the mean CPPS-PI scale score for the rated sessions was 3.30 (SD = 0.72), and the mean CPPS-CB scale score was 1.28 (SD = 0.57), representing a significant level of therapist technique that was consistent with a psychodynamic treatment model (degrees of freedom [df] = 87, t = −20.46, p < .0001, d = 3.1), in the same session that patient alliance was rated. Coefficient alphas for the CPPS-PI and CPPS-CB scales from the 88 sessions rated in this study were .80 and .75, respectively.

Videotapes of an early treatment session (third/fourth) for each patient were arranged in random order, and entire sessions were watched/rated by two raters independently. Raters were graduate students in clinical psychology. Immediately after viewing a videotaped session, judges independently completed the CPPS; each subscale (PI and CB) was coded in random order. Regular reliability meetings were held during the coding process to prevent rater drift (for a more detailed description of this rater training process, see Stein et al., 2010).

The Vanderbilt Psychotherapy Process Scale-Short Form: Patient Participation Subscale ([VPPS-S-Patient]; Smith, Hilsenroth, Baity & Knowles, 2003). The VPPS-S-Patient is a Likert-scale (1 = very poor, 2 = poor, 3 = fair, 4 = good, 5 = excellent), in which patients assess their behavior, demeanor, and feelings during session. There are eight items on the Patient Participation subscale. In the section of the VPPS titled “characterize your behavior during this hour,” these subscale items include: “I actively participated in the interaction;” “I took the initiative in bringing up the subjects that were talked about;” “I was motivated for the session;” “I was logical and organized in expressing thoughts and feelings;” and “I talked about my feelings.” Items on the patient participation subscale in the section titled “describe your demeanor and feelings during this hour” include rating the level of “inhibition (reverse scored);” “spontaneity;” and “passivity (reverse scored).”

The VPPS-S has been instrumental in better understanding patient participation during sessions. There have been many studies that have built upon this instrument, such as the studies conducted by Windholz and Silberschatz (1988) as well as Smith, Hilsenroth, Baity and Knowles (2003). Suh, O’Malley, Strupp and Johnson (1989) discussed the evolution of the Vanderbilt Psychotherapy Process Scale (VPPS). These researchers explained that the VPPS is a clinical rating system that examines positive and negative dimensions of patient and therapist “behavior and attitude” and how these dimensions impacts overall therapeutic outcome (Suh et al., 1989). The subscales investigate both patient and therapist characteristics (i.e., patient participation and therapist emotional involvement) during session. The subscales of the VPPS demonstrate both high internal consistency and inter-rater reliability. The VPPS is unique in that it is effective irrespective of the theoretical orientation to which a clinician subscribes (Suh et al., 1989). Smith and colleagues (2003) reported a coefficient alpha of .76 for the patient participation subscale using a subset of the current participants, as well as the factor structure, psychometric characteristics and convergent validity data. For the current sample, the mean VPPS Patient Participation scale was 4.00 (SD = 0.46; range = 2.75 to 4.88) from the early treatment sessions (i.e., third or fourth) used in this study.

The Combined Alliance Short Form-Patient Version (CASF-P; Hatcher & Barends, 1996). The CASF-P is a patient-rated alliance measure created from a factor analysis of the responses from 231 outpatients at a university-based community clinic and was based on three widely used measures of alliance: (a) the Penn Helping Alliance Questionnaire (HAQ; Alexander & Luborsky, 1986), (b) the Working Alliance Inventory (WAI; Horvath & Greenberg, 1986), and (c) the California Psychotherapy Alliance Scales (CALPAS; Gaston, 1991). The CASF-P consists of 20 items rated on a 7-point scale consisting of 1—never, 2—rarely, 3—occasionally, 4—sometimes, 5—often, 6—very often, and 7—always. The psychometric properties, reliability and validity of this measure are outlined by Hatcher and Barends (1996). Hatcher and colleagues (Hatcher & Barends, 1996; Hatcher, Barends, Hansell, & Gutfreund, 1995) reported initial evidence on both the construct and incremental validity of this scale with regard to outcome. An examination of the internal consistency of this measure has demonstrated a coefficient alpha of .91 for the Total Scale using a subset of the current participants (Ackerman, Hilsenroth, Baity & Blagys, 2000). Both Ackerman et al. (2000) and Clemence et al. (2005) reported convergent validity data with related measures of psychotherapy process, as well as criterion validity with regard to the prediction of treatment outcome using a sample of clients at the same university-based clinic as the patients in the current study.

The CASF-P consists of four subscales: Confident Collaboration (e.g., “What I am doing in therapy gives me new ways of looking at my problems”); Goals and Tasks (e.g., “My therapist and I are working on mutually agreed upon goals”); Bond (e.g., “My therapist and I trust each other”), and Idealized Relationship (e.g., “How much do you disagree with your therapist about what issues are most important to work on during these sessions? [this particular item is reverse scored]).”

Results

Patient Participation and Technique

To investigate the relationship between psychodynamic technique (CPPS-PI) and patient-rated participation (VPPS-S), pairwise correlation coefficients were calculated between each subscale of the CPPS (i.e., psychodynamic-interpersonal and cognitive-behavioral technique) and the patient participation subscale of the VPPS-S (Patient Participation; N = 88). Results demonstrated that greater use of psychodynamic technique is related to more patient participation in early session process during psychodynamic psychotherapy (r = .23, p = .03). The degree of use of cognitive–behavioral intervention, by contrast, was not related to patient participation in early session process during psychodynamic psychotherapy (r = .04, p = .71)

Given the relationship between overall utilization of psychodynamic technique and patient participation, we sought to evaluate the relationship between specific psychodynamic interventions and patient participation. Pairwise correlation coefficients were calculated, and these data are presented in Table 2. Psychodynamic-interpersonal items that were found to be significantly correlated with patient participation were:

A Bonferroni adjustment for these exploratory analyses on individual technique would lead to the more conservative level of significance of p < .005.

Patient Participation and Alliance

In order to evaluate the relationship between patient-rated overall working alliance (CASF-P) and patient participation (VPPS-S), pairwise correlation coefficients were calculated between the CASF-P and the Patient Participation subscale of the VPPS-S. Results demonstrated that higher levels of overall working alliance were related to greater patient participation (r = .50, p < .0001; see Table 3). Likewise, to examine the specific components of the alliance (CASF-P) and patient participation (VPPS-S), pairwise coefficients were calculated between each subscale of the CASF-P and the Patient Participation subscale of the VPPS-S. The strongest correlations between VPSS-S and the CASF-P were as follows: Goals and Tasks (r = .43, p < .0001), Confident Collaboration (r = .39, p = .0002), Bond (r = .37, p = .0003) and Idealized Relationship (r = .34, p = .001). Therefore, results demonstrated that higher levels of various facets of the working alliance were related to greater patient participation during psychodynamic psychotherapy.

Thus, a higher level of overall working alliance is related to greater patient participation. The findings indicated that there was a large effect¹ between overall patient working alliance and patient participation, more so than the individual subscales of patient-rated alliance with patient-rated participation. A Bonferroni adjustment for these exploratory analyses on alliance subscales would lead to the more conservative level of significance of p < .01.

Mediation Analysis

Results highlighted that alliance and participation were significantly correlated, and that the use of psychodynamic technique and participation were significantly correlated. However, alliance and technique were not significantly related, and therefore, due to this finding, the results did not meet criteria for a mediation model. As presented in Figure 1, results demonstrated that in path a, the independent variable (technique) was not significantly correlated with the mediator variable (patient alliance; r = .11 p = .32).

In light of this lack of a demonstrable relationship between technique and alliance, we instead examined partial correlations between CPPS-PI and CASF-Total with VPPS Patient Participation. In these analyses, the relationship between the employment of psychodynamic technique and patient participation was examined while partialling (i.e., covarying) out the effects of alliance, and vice versa. Results of these partial correlations demonstrated that the relationship between early session psychodynamic technique and patient participation remains significant (pr = .21, p =.05) when partialling out the effects of alliance on this relationship. Conversely, the relationship between early session alliance and patient participation remains unchanged (pr = .50, p <.0001) when partialling out the effects of psychodynamic technique on this relationship. In summary, both implementation of psychodynamic technique and alliance appear to represent significant and separate pathways associated with higher levels of patient participation in early sessions of psychodynamic psychotherapy.

Discussion

Results demonstrated that greater use of psychodynamic techniques is related to greater patient participation in early session process during psychodynamic psychotherapy. Moreover, a higher level of overall working alliance is also related to greater patient participation. The findings indicated that there was a large effect between overall patient working alliance and patient participation, more so than the individual subscales of alliance with patient participation. Third, individual therapist techniques from a psychodynamic model of treatment were found to significantly correlate with patient participation. In particular, “The therapist encourages the exploration of feelings regarded by the patient as uncomfortable,” and “The therapist focuses attention on similarities among the patient’s relationships repeated over time, settings or people” indicated a specific relationship between both an affective treatment focus and exploration of cyclical relational patterns with patient participation.

Therefore, an important applied clinical implication of these findings is that interventions focusing on a patient’s uncomfortable feelings and emphasizing a discussion of the patient’s Cyclical Maladaptive Pattern (CMP) are strongly associated with the degree of patient participation. The results of the current study are consistent with a preponderance of research, indicating therapist technique in facilitating and providing sustained follow-up of the patient’s level of affect/difficult emotional experiences has a strong relationship with patient engagement (Diener, Hilsenroth & Weinberger, 2007; Kahn et al., 2008; McCullough et al. 2003). Moreover, deriving a patient’s CMP can be a collaborative process because it focuses on exploration of the patient’s presenting issues and difficulties (for examples, see Diener & Pierson, 2013; Lingiardi, 2013; Slavin-Mulford, 2013). Patients may find great meaning from the focus on the CMP since interpersonal difficulties often derive from dysfunctional patterns of understanding self and relationships with others (Levenson & Strupp, 2007).

We are not, however, suggesting that this set of techniques used in isolation would constitute an effective treatment. To do so would be a very concrete interpretation of the data presented here and clinically unsophisticated. What these data do suggest is that within a psychodynamic model of treatment, delivered in an optimally responsive manner and in which patient alliance was found to be high (see Owen & Hilsenroth, 2011, 2014), the use of these specific interventions at moderate levels (i.e., CPPS mean scores between 2 and 4, or between “somewhat characteristic” and “characteristic”) was related to greater degrees of patient participation during the course of therapy. Thus, the current findings provide support for accentuating these therapeutic techniques within a comprehensive approach that is consistent with psychodynamic theory, research, and practice.

The results also demonstrated a significant relationship between overall patient-rated alliance and patient-rated participation, and this relationship represented a large effect. Also, of note, the strongest correlation was between patient participation and the CASF-P subscales of Goals and Tasks agreement. It seems intuitive that increased patient participation may be due in large part to a patients’ sense of being included in collaboratively developing the focus of their treatment and key tasks, or in areas of focus designed to facilitate the achievement of these goals (also see Goldman, Hilsenroth, Owen & Gold, 2013; Olivera, Braun, Penedo, & Roussos, 2013). The use of the Therapeutic Model of Assessment (described earlier) explicitly engaged these patients in exactly this type of process.

Although results demonstrated that both technique and alliance were significantly related to patient participation, alliance and technique were not significantly related with each other early in treatment. Therefore, technique and alliance were independently related to patient participation, but not to one another. This is a clinically relevant example of the statistical term heteroscedasticity, which refers to cases in which independent variables (alliance and technique) are not related to each other, but rather independently predict the same dependent variable, which in this case was patient participation. This lack of relationship between early session alliance and technique means that the results did not meet criteria for mediation (Barron & Kenny, 1986). That is, the independent variable (technique) was not significantly correlated with the mediator variable (alliance). However, it is important to note that alliance demonstrated a large effect in relation to patient participation.

The current study strengthens our knowledge base of the relationship among the use of psychodynamic technique, alliance, and patient participation. Although this study has several strengths, it is important to discuss its limitations. First, patients reported mild to moderate levels of distress. Although the participants were representative of individuals seeking outpatient treatment at a university-based community clinic, the participants may have had less severe pathology than would be found among patients in an inpatient setting. Second, this sample provided a narrow representation in terms of the participants’ ethnicity/race (90% of the participants identified their ethnic/racial background as Caucasian). The lack of minority patients in this study therefore hinders the generalizability of its results. Third, clinicians were advanced doctoral students enrolled in an APA-approved Clinical Psychology Ph.D. program. The fact that these students had a relatively limited degree of training may have yielded different results than might be found in a study of therapists who have a greater level of expertise. Future research should utilize a more experienced subset of clinicians to investigate if the degree of therapist experience impacts overall treatment outcome and the relationship between alliance and patient participation.

It is also very important to understand the nonsignificant relationship between cognitive-behavioral interventions and patient participation in this study. These findings indicate that the limited amount of cognitive-behavioral interventions used within this larger psychodynamic treatment did not significantly contribute to early patient participation. This finding should not be generalized beyond this limited context, and future work should examine the relationship between these psychotherapy process variables in samples from other treatment orientations. However, the findings clarify that cognitive-behavioral interventions were not related to the development of greater levels of patient participation early in psychodynamic treatment.

It would be useful to examine how therapist variability affects the formation and maintenance of the therapeutic alliance. Research has demonstrated that in relation to the therapeutic alliance, therapist variability more readily predicts treatment outcome than patient variability (Baldwin & Imel, 2013). Using the statistical procedure of multilevel modeling would be beneficial in addressing these issues. The clinical implications are of importance, since therapists may benefit from supervision or further training in alliance as well as monitoring and managing ruptures.

In summary, the current study demonstrated that patients feel more involved and engaged when affective and relational issues are addressed early in treatment, and when these issues are explicitly discussed in relation to potential changes that patients can apply in their important relationships. The findings also support the importance of facilitating alliance building during the early treatment stages—from psychological assessment, during the initial interview, and in the first session (Hilsenroth, Cromer, & Ackerman, 2012). Results support the premise that a more explorative, engaged, and explicitly collaborative process within psychodynamic psychotherapy elicits a greater degree of patient participation. This type of therapist activity engages the patient in the therapeutic process and may, in turn, foster a stronger alliance and deeper participation.