The Relationship Between Therapist Epistemology, Therapy Style, Working Alliance, and Interventions Use

Introduction

A growing area of interest in counseling psychology research explores the philosophical underpinnings of different approaches to counseling and psychotherapy (DisGiuseppe & Linscott, 1993; Lyddon, 1990; Mahoney, 1991). More specifically, recent literature has investigated the translation of “epistemology into practice”, exploring the relationship between philosophical commitments and theories and concepts related to change processes (Arthur, 2000; Botella & Gallifa, 1995). One expression of this work looks at the epistemic assumptions that underpin the theory and practice of cognitive therapy (Lyddon, 1991a). In particular, a developing literature addresses the distinct differences between contemporary cognitive therapies according to their epistemic assumptions: rationalist, empiricist, and constructivist (Hollon & Beck, 1986; Lyddon, 1991a).

The notion that counselors maintain different perspectives regarding the processes and methods of human change because of differing philosophical commitments has been considered in the epistemic style literature (Lyddon, 1989). Personal epistemological commitments have been linked to a variety of different features in psychotherapy, such as preferences for particular types of therapy (Lyddon, 1989; G. Neimeyer & Morton, 1997) and specific therapeutic interventions (Winter & Watson, 1999).

The link between personal epistemologies and particular therapist behaviors is just beginning to receive attention. In fact, there has been relatively little empirical research addressing the translation of therapist epistemological leanings into corresponding therapeutic practices. The purpose of such exploration would be to gain a better understanding of therapist epistemic style and how it relates to the methods and processes of therapy. There is reason to believe that specific epistemic commitments would be related to and may potentially direct a) particular psychotherapeutic styles, b) the structuring of particular types of therapeutic relationships, and c) the selection of particular forms of psychotherapy interventions.

Literature Review

Epistemic Style

Royce has developed a long-standing line of research investigating people’s “ways of knowing” (1964; Diamond & Royce, 1980; Royce & Powell, 1983). Throughout this extensive program of research, Royce and his colleagues developed a conceptual model that specifies three fundamental classes of knowing. These three primary approaches to knowing are referred to as the three epistemic styles: rationalism, empiricism, and metaphorism.

Rationalism maintains the dominant assertion that thought has superiority over the senses with regards to obtaining knowledge. Those with a rational epistemic style are devoted to testing their views of reality in terms of logical consistency. Rationalism is the epistemological worldview that underlies cognitive-rational therapy (Lyddon, 1989; Mahoney, 1991).

Empiricism is primarily concerned with sensory experience as the main way of knowing, in this style, people know to the extent that they are able to perceive accurately. The empirical view of knowledge is primarily inductive and determined mostly by the reliability and validity of observations (Diamond & Royce, 1980; Vincent & LeBow, 1995). Empiricism is the epistemological worldview that underlies behavioral therapy (Mahoney, 1991; Schacht & Black, 1985).

The metaphorist perspective sees knowledge neither as firm nor rigid, but as more flexible, and as embedded within individually and socially constructed symbolic processes. Metaphorism thus takes the stance that reality is personal and mutable, rather than fixed, and that individuals construct their bases of knowledge from their personal learning histories, external experience, and their own personally constructive processes (Vincent & LeBow, 1995). Metaphorism is the epistemological worldview that underlies constructivist therapy (Neimeyer, Prichard, Lyddon, & Sherrard, 1993).

Royce’s conceptual framework holds that the processes of conceptualizing, perceiving, and symbolizing are interdependent processes, where the meaningful convergence of these three processes makes up a person’s view of reality or worldview. Although the processes are interdependent, people tend to show a leaning towards a dominant epistemic style (Royce & Mos, 1980; Royce & Powell, 1983).

Epistemology and Rationalist-Constructivist Therapies

Mahoney (1991) distinguishes and extends epistemic-style research by suggesting that current cognitive therapies are distinguished by their differing epistemological commitments (rationalism and constructivism). Rationalism argues there is a single, stable, external reality, and thoughts are held superior to senses when determining the accuracy of knowledge (Mahoney, 1991; Mahoney & Gabriel, 1987, Mahoney & Lyddon, 1988). Winter and Watson (1999) further depict rationalists as believing the therapist’s role is to instruct the client to think more rationally, thus increasing the correspondence between an individual’s perceptions and the reality of the events confronted. Thus, rationalist therapies are more persuasive, analytical, and technically instructive than the constructivist therapies (Neimeyer, 1993b). Successful rationalist therapy occurs when clients are able to control their negative emotions through rational thinking (Mahoney & Lyddon, 1988). Lyddon (1989) goes on to note that rationalist cognitive theories, due to their epistemological commitment to reason and logical-analytic processes, depict a rational epistemic style.

Constructivism, however, argues that individuals are proactive in their personal constructions of their realities. From this point of view, knowledge is comprised of meaning-making processes in which the individual is in charge of organizing his or her experiences. Constructivists believe that reality is not single, stable, or external, and instead assert that individuals’ feelings and actions cannot be meaningfully separated from human thought (Lyddon, 1988; Mahoney, 1991; Mahoney & Gabriel, 1987; Mahoney & Lyddon, 1988). Thus, constructivist therapies are more personal, reflective, and elaborative than the rationalist therapies (Neimeyer, 1993b).

Lyddon (1990) notes the different role that emotions play in psychotherapy for rational and constructivist therapists. Rationalists view negative emotions as representing problems that need to be controlled or eliminated, whereas constructivist therapists see emotion as playing a functional role in the change process and “encourage emotional experience, expression, and exploration” (p. 124). Lyddon (1989) further notes that constructivist cognitive theories, due to the primacy placed on the construction and alteration of personal meaning, is most representative of a constructivist epistemic style.

Epistemic Style and Preferences

The influence of epistemic style on preference for rational and constructivist therapies have been noted in recent research (Arthur, 2000). The primary implications of this research reveal an existing match between the rational epistemic style and rational therapies, as well between the constructivist epistemic style and constructivist therapies. Lyddon (1989) noted that, for example, people with a dominant rational epistemic style tend to prefer rationalist therapy because rational therapy facilitates clients approaching emotional and personal troubles in a rational and logical way that is congruent with their ways of dealing with difficulties in other aspects of their lives. Thus, when considering the findings of Royce and Mos (1980)— people tend to have a leaning towards a dominant epistemic style—it naturally follows that a match would exist between therapists’ epistemology and their theoretical orientation, reflected in the underlying epistemology of that therapy orientation (Lyddon, 1989).

In the broader literature, the impact of additional therapist variables on therapist’s selection of different therapeutic approaches has also been considered (Scaturo, 2005). Selection of theoretical orientations by therapists has been conceptualized by looking at primitive (grounded in one’s personal experience) versus higher order belief systems (received from formal education) or worldviews. Scaturo (2005) supports the influence of therapist worldview on treatment approach and further discusses how the therapist variable of repression versus sensitization (component of world-view looking at how one defends against threat to one’s domain) impacts therapy orientation. This author proposes that therapists who identify more with repression from their worldview may align with behavioral approaches whereas those that identify more with sensitization may align with insight-oriented approaches. This highlights how certain therapist variables may underlie therapist development of different epistemological leanings.

In considering the epistemology literature, a much broader range of theoretical, strategic, and technical distinctions have been conceptualized in relation to differing epistemological positions than have actually been documented in research literatures (Mahoney & Lyddon, 1988; R. Neimeyer, 1993b). These conceptual differences include expected differences in the characteristic style of therapy, differences in the nature and enactment of the therapeutic relationship (R. Neimeyer, 1995), and differences in the actual interventions associated with different therapy orientations (Lyddon, 1990). Despite the many different conceptual differences that have been noted, relatively few of these have received careful empirical documentation (Neimeyer, Saferstein, & Arnold, 2005).

Working on the basis of current conceptual distinctions in the literature, it is possible to identify and test expected differences between rational and constructivist therapists in relation to

Therapy Style

There has been some literature investigating conceptual differences in the relationship between therapist epistemic assumptions and therapy style. Granvold (1996), for example, suggests that traditional cognitive behavioral therapists tend to target irrational beliefs for modification, educate and guide the client, and take an active and directive position with the client. On the other hand, a constructivist therapy style is characterized by the therapist who is less directive and who engages in more exploratory interaction in their behavior with clients.

In addition, empirical work has begun exploring differences in therapy style and epistemic assumptions. For example, Winter & Watson (1999) found that rationalist therapists showed a more negative attitude towards their clients, while the personal construct therapists showed greater regard for them. Additionally, clients involved in personal construct therapy showed greater overall involvement in therapy. These differences are in line with the collaborative nature of the personal construct therapist as originally depicted by Kelly (1955).

In another study, Neimeyer and Morton (1997) compared the commitments to epistemological assumptions; they found personal construct therapists demonstrated a significantly higher commitment to a constructivist epistemology (and a lower commitment to a rationalist perspective) compared to rational-emotive therapists, who supported the translation of epistemic commitments into therapeutic practice.

Further efforts to build upon these findings can be developed in relation to the conceptualization by Fernandez-Alvaraez, Garcia, Bianco, & Santoma (2003) of therapists’ personal style. These authors describe therapists’ personal style as the, “… imprint left by each professional in his work” and note that it “has a relevant impact on the outcomes of the treatment.” (p. 117). This can be considered in relation to how therapy style manifests differently in various theoretical approaches. For example, Granvold (1996) notes the marked differences between cognitive behavioral (e.g. more directive) and constructivist therapy styles (e.g. more exploratory and experiential) regarding how these different orientations view treatment goals.

Fernandez-Alvaraez et al. (2003) created the Personal Style of the Therapist Questionnaire to measure therapists’ personal style. It assesses five specific dimensions of therapist style:

Instructional (flexibility-rigidity),

Expressive (distance-closeness),

Engagement (lesser degree-greater degree

Attentional (broad focused-narrow focused

Operative (spontaneous-planned).

Thus, the first hypothesis makes predictions regarding the influence of therapist epistemology on therapists’ particular therapy style. According to these authors’ definitions, our first hypothesis is that therapist epistemology will be a significant predictor of their therapy style. More specifically, when compared to therapists with a constructivist epistemology, therapists with rational epistemologies would have a therapy style depicting greater rigidity on the Instructional subscale, greater distance on the Expressive subscale, a lesser degree of Engagement, a narrower focus on the Attentional subscale, and more planned on the Operative subscale.

Working Alliance

These differences in therapy style reflect broader differences regarding the nature and role of the therapeutic relationship. In addition to therapy style, cognitive behavioral and constructivist therapies maintain notable differences in the nature of the working alliances they form with their clients. The working alliance is defined by Bordin (1979) as the combination of (a) client and therapist agreement on goals (Task), (b) client and therapist agreement on how to achieve the goals (Goal), and (c) the development of a personal bond between the client and therapist (Bond).

While rationalist and constructivist therapies both value the working alliance, the empirical literature suggests that rationalist and constructivist therapies value different qualities within the working alliance. A conceptual depiction of the differences between cognitive behavioral and constructivist therapists in the therapeutic relationship comes from Beck, Rush, Shaw, and Emery (1979), who state that the therapist is a “guide who helps the client understand how beliefs and attitudes influence affect and behaviour” (p. 301). This assertion highlights the differences between cognition, affect and behavior in Beck’s approach, compared to the holistic perspective maintained in the constructivist approach.

Further empirical studies have addressed key distinctions between cognitive behavioral and constructivist therapies with regard to emphasis on working alliance. For example, a study by Winter and Watson (1999) found that constructivist therapists were “less negatively confrontative, intimidating, authoritarian, lecturing, defensive, and judgmental” (p. 17). In addition, constructivist therapists had greater use of exploration, silence, open questions and paraphrasing, along with decreased use of approval, information and direct guidance, compared to cognitive behavioral therapists.

Additionally, the working alliance has been noted to have an important role in cognitive behavioral therapy (Raue, Goldfried, & Barkham, 1997). The CBT therapists value working toward a common goal more important than the bond of the relationship itself. Consensus on the tasks and goals of therapy is inherent in Beck’s (1975) basic notion of collaborative empiricism, which highlights the collaboration between client and therapist in achieving therapeutic gains. Consensus, which is highly valued within the rationalist therapies, falls in line with Bordin’s (1979) definition of the Task and Goal components in the working alliance.

Constructivists tend to have less narrowly defined tasks or goals compared to cognitive behavioral therapists (Granvold, 1996). Mahoney & Lyddon (1988) depict constructivist therapists as viewing the human connection within the therapeutic relationship as a crucial component of therapeutic change, a connection that “functions as a safe and supportive home base from which the client can explore and develop relationship with self and world” (p. 222). This is directly in line with Bordin’s (1979) depiction of the Bond component of the working alliance, as comprising the key elements of rapport: trust, acceptance, and confidence.

Thus, the second hypothesis is that therapist epistemology will be a significant predictor of working alliance (Task, Bond, and Goal) and that rationalist therapists will have higher scores on the Task and Goal subscales and lower on the Bond subscale than therapists with constructivist epistemologies.

Therapeutic Interventions

Both rationalist and constructivist therapies view psychotherapy as occurring within a therapeutic relationship, however the nature of this relationship is somewhat different (e.g. instruction vs. exploration, correction vs. creation, etc.). Thus, the specific techniques used by rationalist and constructivist therapists might be expected to fit within these broad relationship differences.

For example, Mahoney and Lyddon (1988) point out that rationalist interventions tend to focus on the “control of the current problems and their symptomatology” (p. 217). In contrast, constructivist interventions tend to focus on “developmental history and current developmental challenges” (p. 217). They highlight the key differences between these two therapy interventions as reflecting a “problem-versus-process” distinction that itself is reflected in the implicit and explicit goals of these two types of therapy.

Additionally, Granvold (1996) notes that traditional cognitive behavioral interventions are geared at controlling, altering or terminating negative emotions (e.g. anxiety, depression, anger, worry, etc.). In contrast, constructivist interventions use more creative than corrective interventions (e.g. exploration, examination, and experience).

Winter and Watson (1999) noted empirical evidence for these conceptual distinctions between cognitive behavioral and constructivist therapy techniques. Findings indicated that cognitive behavioral therapists used interventions that seemed to be “more challenging, directive and to be offering interpretations that do not always lead directly from what the client has said,” (p. 17), whereas, constructivist therapists asked more questions than made statements and used interpretation more as a means of facilitating the client’s elaboration.

Consequently, psychotherapy research investigations have found a theoretical allegiance according to which techniques therapists use in their practice. Personal construct therapists were found to use techniques with greater relying on exploration, open questions and paraphrase (Winter & Watson, 1999). The current study plans to extend this line of research according to therapist epistemology.

Thus, for the third hypothesis, epistemology will be a significant predictor of therapy techniques used by the therapists in the sample. More specifically, therapists with rationalist epistemologies are expected to report using techniques associated with cognitive behavioral therapy (e.g. advice giving) more than constructivist epistemologies, and therapists with constructivist epistemologies will report using techniques associated with constructivist therapy (e.g. emotional processing) more than therapists’ with rationalist epistemologies.

In the present study, we investigate the potential influence of epistemic style (rational vs. constructivist) on therapist therapy style, nature of the working alliance, and use of specific interventions. These therapist variables were included according to noted importance in translating epistemology into practice (Neimeyer et al. 2005).

Method

Participants

Most participants were professional psychologists recruited online through membership in different professional organizations, the American Psychological Association (APA)—Practice Organization online practitioner directory (www.apapractice.org of approximately 15,057 members).

In addition to a number of APA-approved counseling centers, participant solicitation e-mails were also sent to APA Division 17 (Counseling Psychology, 355 members); APA Division 29 (Psychotherapy, approximately 224 members); APA Division 32 (Humanistic Psychology, approximately 130 members); The North American Personal Construct Network (NAPCN) list serve (approximately 95 members); the Albert Ellis Institute e-mail list (approximately 57 members). The solicitation e-mail also encouraged participants to forward the e-mail survey on to other eligible practitioners; therefore the response rate of approximately 13.5% has to be considered with reservations (approximately 15,918 surveys were solicited and 2,149 returned). Because we encouraged recipients to forward the email on to others, it is an approximation.

Therapist participation was voluntary, it took therapists approximately 30 minutes to complete the instruments, and the study was conducted in accordance with APA ethical guidelines.

Demographics

The sample consisted of 1151 therapists (733 women, 418 men) with a M_age of 45.09 (SD = 12.54). The sample was primarily European American, 88.8% (N =1030), followed by Multiracial, 2.9% (N = 34), Hispanic, 2.7% (N = 31), African American, 2.4% (N = 28), Asian American, 2.1% (N = 24), and Other, 1.1% (N = 13).

Participants were asked to indicate the level of their highest degree held, which consisted of Ph.D., 60.1% (N = 700); followed by M.A./M.S., 18.6% (N = 216); Psy.D., 11.0% (N = 128); B.A./B.S., 4.3% (N = 50); Ed.D., 1.7% (N = 20); M.S.W., 1.4% (N = 16); and Other, 2.9% (N = 34). Additionally, the average year participants obtained their highest degree was 1992.55 (SD =11.1), along with the average total number of years spent in clinical practice being 14.01 (SD = 11.03). The majority of participants were no longer in school, 93.5% (N = 1105); 6.5% (N = 77) were graduate students.

In addition, participants were asked their dominant theoretical orientation and most participants indicated that their dominant theoretical orientation was cognitive behavioral, 35.9% (N = 414); followed by integrative, 18.1% (N = 209); psychodynamic, 15.2% (N = 175); interpersonal, 7.6% (N = 88); humanistic, 7.2% (N = 83); constructivist, 3.2% (N = 37); existential, 2.2% (N = 25); rational emotive, 1.7% (N = 20); gestalt, 0.7% (N = 8); and other, 8.2% (N = 95).

Measures

Members from the aforementioned organizations were sent an online survey containing an informed consent form, a brief demographics information sheet, and the five measures: Therapist attitudes questionnaire-short form (TAQ-SF); Constructivist assumptions scale (CAS); Personal style of the therapist questionnaire (PST-Q); Working alliance inventory-short form (WAI-S); and the Techniques list (TL). Participants were debriefed at the end of the study and were provided with the contact information for further inquiries.

Therapist Attitudes Questionnaire-Short Form (TAQ-SF)

The TAQ-SF, developed by Neimeyer and Morton (1997), is a revision of the Therapist Attitudes Questionnaire (TAQ) developed by DisGiuseppe and Linscott (1993). The TAQ-SF measures philosophical, theoretical, and technical dimensions of rationalist and constructivist therapies. The instrument contains 16 items rated on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree), and requires approximately five minutes to complete. The TAQ-SF replicates the basic factor structure of the original TAQ and has shown its predictive validity by predicting the therapeutic identifications and descriptions of a group of practicing professionals (Neimeyer & Morton, 1997). TAQ-SF scores in the present study yielded a Chronbach’s alpha of .72 for rationalist scale and a Chronbach’s alpha of .63 for the constructivist scale (see Table 1).

Constructivist Assumptions Scale (CAS)

The Constructivist Assumptions Scale (CAS) was developed by Berzonsky (1994), and was designed to assess constructivist epistemological assumptions. This is a 12-item self-report measure with each item being rated on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). The CAS has internal reliability estimated to be .61, and a 2-month test-retest reliability (N = 78) of .68. CAS scores in the present study yielded a Chronbach’s alpha of .72 (see Table 1).

Personal Style of the Therapist Questionnaire (PST-Q)

The Personal Style of the Therapist Questionnaire (PST-Q) was developed by Fernandez-Alvarez et al. (2003), and was created to assess “the set of characteristics that each therapist applies in every psychotherapeutic situation, thus shaping the main attributes of the therapeutic act” (p. 117). The questionnaire assesses five different dimensions: flexibility-rigidity (Instructional subscale), distance-closeness (Expressive subscale), lesser engagement-greater engagement (Engagement subscale), broad focused-narrow focused (Attentional subscale) and spontaneous-planned (Operative subscale). This is a 36-item self-report measure filled out by therapists with answers rated on a scale ranging from 1 (total disagreement)to 7 (total agreement). The measure has shown a test-retest reliability of .79, with Chronbach’s reliability coefficients for each subscale, as follows; instructional, .69; expressive, .75; engagement, .78; attentional, .80; operative, .78. Factor Analysis revealed a KMO = .756 (See Table 1 for alphas in the current study).

Working Alliance Inventory-Short Form (WAI-S)

The Working Alliance Inventory (WAI), developed by Horvath & Greenberg (1986), is a 36 item questionnaire that can be administered to both clients and therapists and is rated on a 7-point Likert type scale from1 (never)to 7 (always). Tracey and Kokotovic (1989) proposed a client and therapist Working Alliance Inventory-Short Form (WAI-S), which contains four items per subscale (Task, Goal, and Bond) and overall WAI-S scores, demonstrating high reliability with alpha levels similar to, and even better than the WAI for the therapist subscales and overall average scores (Task, alpha = .83; Bond, alpha = .91; Goal, alpha = .88; and General Alliance, alpha = .95). See Table 1 for alphas in the current study.

Techniques List (TL)

The Techniques List measure was adapted from Hollis (1995), who catalogued an extensive list of counseling and psychotherapy techniques representing a broad spectrum of philosophical bases. In order to refine this extensive list according to techniques used specifically by cognitive behavioral and constructivist therapy orientations, we recruited counseling psychology graduate students to read through the total list of 108 therapy techniques and rate the extent to which each technique is used by each therapeutic approach (cognitive behavioral and constructivist), using a 5-point Likert type scale from 1 (Never or Almost Never) to 5 (Always or Almost Always).

Procedures

Sixteen counseling psychology graduate students participated in these ratings (six men, 10 women), with the M_age = 28.44 (SD = 2.67). Results of a paired differences analysis for all 108 items indicated that there were 77 techniques rated as being used with significantly differential frequency by cognitive behavioral and constructivist therapies. We then divided this distribution of 77 techniques into quartiles and retained the top and bottom quartiles. This resulted in 20 cognitive behavioral techniques (e.g.advice giving, rational restructuring) and 20 constructivist techniques (e.g. emotional processing, reflection) that were rated most significantly different (cognitive behavioral vs. constructivist). This final list of 40 items of therapy techniques (20 cognitive behavioral techniques; 20 constructivist techniques) was used in the current study.

These 40 items were listed alphabetically and participants were asked to rate the extent to which they use each technique in their practice of therapy along a 5-point scale from 1 (Never or Almost Never)to 5 (Always or Almost Always). The ratings of the 20 rationalist items were summed and a mean was calculated to reflect the average frequency of using rationalist interventions (possible range = 1-5), and the same procedure was applied in relation to the 20 constructivist interventions.

The raw data was used to conduct a confirmatory factor analysis on the Techniques List measure. The current analysis was examined for multi-variate normalcy and the assumptions were met. All kurtosis estimates for the variables fell between 1 and –1 variables except constructivist items 12, 13, 16, and 20 and cognitive behavioral item 3, which had a kurtosis values between 2 and –2. Consequently, these five items were removed from the measure prior to running the confirmatory factor analysis.

A confirmatory factor analysis was utilized to fit a model of two types of therapy technique factors (constructivist and cognitive behavioral therapy techniques). Thirty-five indicators were included in the model (16 constructivist techniques and 19 cognitive behavioral techniques). After running the analysis with the 35 items, and two factors (constructivist techniques and cognitive behavioral techniques), factor loadings revealed eight items (six constructivist items and two cognitive behavioral items) loading at less than .40. These eight items were removed and the confirmatory factor analysis was then re-run with the remaining 27 items (10 constructivist and 17 cognitive behavioral items). No further model modifications were made because there was no other compelling theoretical rationale for additional changes and these 27 items were used in all subsequent analyses using this measure. Items were constrained to load only on to their respective factors (constructivist techniques and cognitive behavioral techniques), and the two factors were allowed to correlate.

The measurement model was examined utilizing LISREL (8.7) and was evaluated based on multiple goodness of fit indices, with the maximum likelihood as the estimation method. Examination of the results revealed that the fit of the model was a fairly good fit although not necessarily a superior fit for the data, χ² (323, N = 914), = 2249.37, p < .001, SRMR = .066, RMSEA = .08, NFI = .91, and CFI = .93, suggesting overall a good fit. The final standardized solution factor loadings were all significant (p < .05) and ranged from .40 to .81 for the constructivist techniques and from .47 to .71 for the cognitive behavioral techniques. The correlation between the two factors was –.30. p < .03. See Table 1 for alphas in the current study

Results and Discussion

Correlational Analyses

Person Product Moment correlations, using a criterion level of .05 (1-tailed), were computed between the two epistemology subscales (Rationalist and Constructivist) and each of the criterion variables in an attempt to confirm that the relationships were in the predicted directions. Results were in the predicted directions, revealing a significant positive correlation between the TAQ-SF constructivist subscale and the CAS, r = 0.30, P ≤ 0.001 and a significant negative correlation between the TAQ-SF rationalist subscale and the CAS, r = – 0.36, P ≤ .001. Additionally, a Pearson Product Moment correlation was conducted on the TAQ-SF rationalist and constructivist subscales to justify their use as two separate continuous subscale scores, r = –.09, p < .001.

For therapist style, the rationalist and constructivist subscales were significantly correlated with the subscales of the PST-Q, all in the predicted directions. For the WAI-S, rationalist epistemologies were not significantly correlated with any of the WAI-S subscales (e.g. Task, Bond, and Goal); however, the constructivist epistemology was significantly positively correlated with all three subscales of the WAI-S and in the predicted direction. When looking at types of techniques therapists use in treatment, the rationalist epistemology was significantly negatively correlated with the use of constructivist techniques (r = – .32, P ≤ .001) and significantly positively correlated with the use of cognitive behavioral techniques (r = 0.43, P ≤ .001), which was in the predicted directions. While, constructivist epistemologies were significantly positively correlated with the use of constructivist techniques (r = 0.22, P ≤ .001), which was in the predicted direction; however, constructivist epistemology was not significantly correlated with cognitive behavioral techniques. (See Table 2 for a complete listing of correlations).

Regression Analyses

In order to assess the capacity of the data to be in line with the normality assumptions of multiple regressions, the data was subjected to tests of skewness and kurtosis. Results of these analyses indicate that the assumptions for multivariate normalcy were met. In addition, Bonferroni corrections were utilized.

Hypothesis 1—Therapist Epistemology as a Predictor of Therapy Style

The first hypothesis concerned therapist epistemology as a predictor of therapy style. We hypothesized that therapists with rational epistemologies would have a therapy style depicting more rigidity on the Instructional subscale, greater distance on the Expressive subscale, a lesser degree of Engagement, a narrower focus on the Attentional subscale, and be more planned on the Operative subscale compared to therapists with a constructivist epistemology. Separate regression analyses were conducted for each of the five PST-Q scores measuring therapy style.

The Instructional Subscale The epistemology scores accounted for significant variation in Instructional scores, F(2, 1061) = 7.06, p < .001 (R²= .013). The standardized beta coefficient for the rationalist epistemology (β = .053) was in the positive direction, but was not significant— t(1061) = 1.73, p < .084. The standardized beta coefficient for the constructivist epistemology (β = – 0.097) was significant and in the negative direction for the Instructional subscale— t(1061) = –3.15, p < .002. The direction of the effect indicated that the more a therapist endorsed constructivist epistemology, the less likely that therapist was to use an instructional approach to therapy. This supported the hypothesis that a constructivist epistemology tends toward the direction of flexibility on the Instructional subscale; however, the small effect size of approximately 1% of the variance needs to be considered.

The Expressive Subscale Epistemology was also a significant predictor of the therapy style along the Expressive subscale F(2, 1080) = 94.27, p < .001 (R² = .15). The standardized beta coefficient (β = –0.177) was significant for the rationalist epistemology t(1080) = – 6.28, p <.0001 and in the negative direction, whereas the significant standardized beta coefficient for the constructivist epistemology (β = 0.326), was significant t(1080) = 11.56, p < .0001 and in the positive direction along the Expressive subscale. This supported the hypothesis that the rationalist epistemology tends towards distance on the Expressive subscale, whereas, the constructivist epistemology tends towards greater closeness on the Expressive subscale.

The Engagement Subscale Epistemology was also significant predictor of the therapy style along the Engagement subscale, F(2, 1096) = 47.26, p < .001 (R² = .08). The significant standardized beta coefficient (β = –0.245) for the rationalist epistemology, t(1096) = – 8.42, p < .001, was in the opposite direction compared to the significant standardized beta coefficient (β = 0.119) for the constructivist epistemology, t(1096) = 4.08, p < .001, along the Engagement subscale. This supported the hypothesis that the rationalist epistemology tends towards a lesser degree of engagement on the Engagement subscale and the constructivist epistemology tends towards a greater degree of engagement on the Engagement subscale.

The Attentional Subscale Epistemology was also significant predictor of the therapy style along the Attentional subscale, F(2, 1096) = 118.33, p < .001 (R² = .18). The significant standardized beta coefficient (β = 0.396) for the rationalist epistemology t(1096) = 14.41, p < .001, was in the positive direction; whereas the significant standardized beta coefficient (β = –0.129) for the constructivist epistemology t(1096) = –4.12, p < .001, which was in the negative direction along the Attentional subscale. This supported the hypothesis that the rationalist epistemology has more of a leaning towards a narrow focus on the Attentional subscale, and the constructivist epistemology leans more towards a broad focus on the Attentional subscale.

The Operative Subscale Lastly, epistemology was a significant predictor of the therapy style along the Operative subscale, F(2, 1093) = 187.86, p < .001 (R² = .256). The standardized beta coefficient (β = 0.461) for the rationalist epistemology was significant, t(1093) = 17.61, p < .0001 and in the positive direction, compared to the significant standardized beta coefficient (β = –0.170), for the constructivist epistemology, t(1093) = –6.50, p < .0001, which was in the negative direction along the Operative subscale. This supported the hypothesis that the rationalist epistemology tends towards more planning on the Operative subscale and the constructivist epistemology tends towards more spontaneity on the Operative subscale.

Thus, epistemology (rationalist vs. constructivist) was found to be a significant predictor of therapy style. In particular, the most robust findings provide provisional support for the notion that there are specific differences in the personal style of the therapist according to the therapists’ epistemic assumptions. More specifically, the current study found that therapists with rationalist epistemologies tended towards more distance, a lesser degree of engagement, a narrower focus, and a greater degree of planning in their sessions with clients, whereas, the constructivist epistemology tended towards having a greater degree of closeness, a greater degree of engagement, a broader focus, and more spontaneity in their therapy sessions. Additionally, there was some support for the notion that therapists with constructivist epistemologies tend toward the direction of flexibility rather than rigidity in their therapy style; however this was not a particularly strong finding in the current study.

These findings are helpful when considering the potentially inherent differences maintained by rationalist versus constructivist epistemologies according to therapy style. More specifically, current findings support the notion that cognitive-behavioral therapies, which represent the best depiction of the rationalist epistemology, maintain an “active-directive” and systematic approach to therapy (Granvold, 1988) with specific goals used to plan the course of the session (Mahoney & Lyddon, 1988).

Hypothesis 2—Epistemology Influences the Therapeutic Relationship

According to the second hypothesis—therapists with rationalist epistemologies will score higher on the Task and Goal subscales and lower on the Bond subscale than the constructivist epistemologies—another multiple linear regression model was conducted to determine if the same predictor variable (therapist epistemology) would influence therapists’ ratings of the criterion variables (working alliance) based on therapists’ scores of three subscales—Task, Goal, and Bond.

The Task Subscale Epistemology was a significant predictor of therapist emphasis on the working alliance along the Task subscale (e.g. client and therapist agreement on goals), F(2, 1080) = 8.34, p < .001 (R² = .015). The standardized beta coefficient for the rationalist epistemology (β = 0.042) was in the positive direction, but was not significant t(1080) = 1.39, p < .164. The significant standardized beta coefficient (β = 0.120) for the constructivist epistemology, t(1080) = 3.96, p < .0001, was also in the positive direction along the Task subscale. This was inconsistent with the hypothesis that the rationalist epistemology would place a greater emphasis on the Task subscale in the working alliance than therapists with a constructivist epistemology. However, the small effect size of approximately 2% of the variance needs to be considered when interpreting these findings.

The Goal Subscale Epistemology was also a significant predictor of therapist emphasis on the working alliance along the Goal subscale (e.g. client and therapist agreement on how to achieve the goals), F(2, 1093) = 4.92, p < .007 (R² = .009). The significant standardized beta coefficient (β = 0.065) for the rationalist epistemology t(1093) = 2.16, p < .031, was in the positive direction. The significant standardized beta coefficient (β = 0.075) for the constructivist epistemology t(1093) = 2.47, p < .014, was also in the positive direction along the Goal subscale. This was again inconsistent with the proposed hypothesis that the rationalist epistemology would have stronger leanings towards the Goal subscale in the therapist emphasis on working alliance compared to therapists with a constructivist epistemology.

The Bond Subscale Lastly, epistemology was also a significant predictor of the therapist emphasis on the working alliance along the Bond subscale (the development of a personal bond between the client and therapist), F(2, 1089) = 19.49, p < .001 (R² = .035). The standardized beta coefficient for the rationalist epistemology (β = – 0.034) was in the negative direction, but was not significant, t(1089) = –1.15, p < .249. For the constructivist epistemology, the standardized beta coefficient (β = 0.179) was significant t(1089) = 5.99, p < .0001, and in the positive direction along the Bond subscale. This supported the hypothesis that the rationalist epistemology is less inclined towards therapist emphasis on working alliance on the Bond subscale than the constructivist epistemology.

The current study indicated that therapist epistemology was a significant predictor of at least some aspects of the working alliance. The strongest finding was in relation to the development of a personal bond between the client and therapist (Bond subscale). Therapists with a constructivist epistemology tended to place more emphasis on the personal bond in the therapeutic relationship compared to therapists with a rationalist epistemology. This supports the notion in the literature that constructivist therapists place a greater emphasis on building a quality therapeutic relationship characterized by, “acceptance, understanding, trust, and caring.

Hypothesis 3—the Selection of Specific Therapeutic Interventions

The third and final analysis is designed to address the prediction that epistemology will be a predictor of therapist use of specific therapy techniques. More specifically, that the rationalist epistemology will report using techniques associated with cognitive behavioral therapy (e.g. advice giving) more than constructivist epistemologies, and therapists with constructivist epistemologies will report using techniques associated with constructivist therapy (e.g. emotional processing) more than therapists with rationalist epistemologies). A multiple linear regression analysis was conducted to determine if the predictor variable (therapist epistemology) will influence therapist ratings of the criterion variables (therapy techniques).

Epistemology was a significant predictor of cognitive behavioral therapy techniques F(2, 993) = 112.34, p < .001 (R² = .185). The standardized beta coefficient for the rationalist epistemology (β = 0.430) was significant, t(993) = 14.96, p < .001 and in the positive direction. The standardized beta coefficient for the constructivist epistemology (β = 0.057) was significant and in the positive direction t(993) = 1.98, p < .05. This supported the hypothesis that the rationalist epistemology would have stronger leanings of therapist use of cognitive behavioral techniques when conducting therapy than constructivist epistemologies.

Finally, epistemology was a significant predictor of constructivist therapy techniques F(2, 1012) = 80.82, p < .001 (R² = .138). The standardized beta coefficient for the rationalist epistemology (β = – 0.297) was significant t(1012) = –10.09, p < .0001 and in the negative direction. The standardized beta coefficient for the constructivist epistemology (β = 0.195) was significant t(1012) = 6.63, p < .0001, and in the positive direction. This supported the hypothesis that the constructivist epistemology would place a stronger emphasis on therapist use of constructivist techniques when conducting therapy than rationalist epistemologies.

Findings in the current study regarding therapists’ epistemology and their use of specific techniques revealed that therapists’ with rationalist epistemologies tended to favor the use of cognitive behavioral techniques and also tended to reject the use of constructivist techniques. Similarly, therapists’ with constructivist epistemologies tended to favor the use of constructivist techniques in their practice of therapy; however they did not as strongly reject the use of cognitive behavioral techniques. This notion is supported by literature that suggests that constructivist therapists value having “a rich set of possibilities that can be engaged at any moment depending on the client’s need.” (R. Neimeyer, 2005, p. 83). Thus, findings from the current study may suggest that while the constructivist therapist is more likely to use constructivist therapy techniques, they are also more open to using other techniques depending on the individual client compared to rationalist therapists.

Limitations and Future Research

This study is not without limitations. For example, this study was conducted on a voluntary basis and those who volunteered to participate may have been a biased sample and compromised the external validity. Rosenthal and Rosnow (1975) suggest that volunteers tend to differ from non-volunteers in behavioral research regarding their level of education, intelligence, and desire of social approval. Additionally, the external validity may have been compromised by the data collection, which was conducted via the Internet, and may further distinguish the characteristics of the participants who volunteered to participate in the study from non-volunteers. Another limitation regarding the generalizability of the findings in the current study is the self-report nature of the study. Rosenthal and Rosnow (1991) indicate that self-reports are subject to distortion and social desirability effects. In addition, self-reports may not correlate well with participants’ actual behavior.

However, in light of these limitations, having an overall sample size of more than one thousand practicing psychologists representing all of the 50 United States may have improved the representativeness of the sample and subsequently, the generalizability of the findings.

In addition, greater confidence in the representativeness of the sample in the current study is found by using the closest approximation to what would be a comparison with the bulk of our sample (e.g. psychologists) to members of the American Psychological Association along demographic dimensions (e.g. gender, ethnicity, and age). For example, in the current study, women were 64 % of the sample and men 36% of the sample, which is roughly comparable to APA members (approximately 53% women and 47% men). The mean age of participants in the current study was 45.09 (SD = 12.54), which again, is roughly comparable to APA members (mean age = 53.30, SD = 13.6). The ethnicities in the current study were Caucasian, 88.8%, Multiracial, 2.9%, Hispanic, 2.7%, African American, 2.4%, and Asian American, 2.1%. Again, this is roughly comparable the APA members reported ethnicities as Caucasian, 67.6%, Multiracial, 0.3%, Hispanic, 2.1%, African American, 1.7%, and Asian American, 1.9% (http://research.apa.org/profile2005t1.pdf, 2005).

It is also important to highlight the fact that the findings in the current study are associations between the variables of interest and do not imply causal relationships. Therefore, current results can only suggest potential relationships and cannot imply causality.

Further research could aim to investigate client’s perceptions of cognitive-behavioral and constructivist therapists’ therapy style, emphasis on the working alliance, and use of particular therapeutic interventions to see if clients corroborate therapists’ self-reported styles with their experience of the therapists’ style.

Finally, while the fit of the two factors (constructivist techniques and cognitive behavioral techniques) to the Techniques List was relatively good, future work on the Techniques List measure might also benefit from some revision of the current instrument and the addition of more psychometrics.

Nonetheless, the findings of this study contribute to the literature addressing the translation of epistemology into practice. The current study supports the notion that therapists with rationalist epistemologies are consistently different in their approach to therapy, including the emphasis on the therapeutic relationship and use of particular interventions, from therapists with a constructivist epistemology in ways consistent with their epistemological underpinnings.

The current findings are important because they (1) demonstrate the translation of epistemology into practice; (2) provide information that could be useful to clients in selecting a therapist whose orientation may enable them to anticipate stylistic features; and (3) provide the opportunity to further study the translation of these perceptions into actual behaviors and behaviors into different impacts or outcomes.

Conclusion

The current study extended the developing literature on therapists’ epistemology as a factor relating to psychotherapists’ practice of therapy. Further, more outcome-related research is required to understand how therapists’ epistemological beliefs impact the successfulness of work with clients. The current study was the first empirical investigation of therapists’ epistemological values and the specific translation of epistemology into the practice of therapy in relation to therapists’ style, working alliance, and use of specific techniques. While some of the results failed to support the expected directions for the specified subscales, most results were in the expected directions supporting the overall coherence of the epistemological commitment with therapeutic enactments. Further work may benefit from focusing on how therapist epistemological viewpoints might affect the effectiveness of practicing psychotherapy in accordance with therapists’ epistemic commitments.