The Impact of Interpersonal Style and Interpersonal Complementarity on the Therapeutic Alliance Between Therapists and Offenders in Sex Offender Treatment

Abstract

Therapist and treatment process variables affect the effectiveness of offender rehabilitation programs. This study examined the influence of therapists’ and offenders’ interpersonal styles (IPSs) and interpersonal complementarity on therapeutic alliance (TA). Seventy-five sex offenders and their therapists evaluated each other’s IPSs and the TA after 3 weeks of treatment. Offenders evaluated the TA more positively than therapists. Regarding the impact of IPS, therapist affiliation was positively correlated and therapist control was negatively correlated with offenders’ ratings of the TA; in other words, offenders evaluated the TA more strongly when therapists were perceived as affiliative, and weaker when therapists were viewed as controlling. Offender affiliation was positively correlated with therapists’ ratings of TA; in other words, therapists evaluated the TA more strongly when offenders were viewed as more affiliative; perceptions of offender control were unrelated to offenders’ ratings of TA. Complementarity in IPS between offenders and therapists did not affect TA.

Keywords

therapeutic alliance interpersonal style complementarity sexual offender treatment process

Treatment process variables affect therapeutic change, with the therapeutic relationship accounting for approximately 25% variance in treatment outcomes in non-forensic populations (Morgan, Luborsky, Crits-Christoph, Curtis, & Solomon, 1982). The therapeutic relationship is also important for offenders’ engagement in treatment (Holdsworth, Bowen, Brown, & Howat, 2014), including sexual offender treatment (Frost & Connolly, 2004; Levenson, Macgowan, Morin, & Cotter, 2009). The treatment process includes factors such as therapist characteristics and clients’ perception of therapists, the therapeutic climate (group climate), and the therapeutic alliance (TA; Marshall & Burton, 2010). Compared with interrogation of treatment content, research into process-related issues in sex offender treatment programs is limited (Serran, Fernandez, Marshall, & Mann, 2003). However, there is evidence that a positive therapeutic relationship facilitates a safe environment in which therapists can engage the sexual offender about crime minimization/denial and potentially improve outcomes (Serran et al., 2003). Marshall and Burton (2010) suggested that treatment process may be more important for offender populations than non-offender populations because offenders typically have a higher base rate of interpersonal difficulties. Furthermore, the influence of hostile, uncaring prison environments has been shown to negatively affect the development of trust and offender engagement (Ross, Polascheck, & Ward, 2008); such difficulties in treatment processes can lead to treatment non-completion or reductions in treatment gain (Holdsworth et al., 2014; Martin, Garske, & Davis, 2000). This is significant for offender populations as treatment non-completion is associated with an increased likelihood of recidivism (Olver, Stockdale, & Wormith, 2011).

The TA

One important treatment process factor that has attracted substantial research focus is the TA (also called the working alliance, therapeutic bond, or helping alliance). The TA involves the collaborative and affective bond between therapists and their clients (Martin et al., 2000), with a positive TA having been found to be associated with better treatment outcomes (Martin et al., 2000). TA is most commonly understood through Bordin’s (1979) Pantheoretical Model, which articulates three core components: collaboration on tasks, goals, and the affective bond (Bordin, 1979).

Meta-analysis has shown an overall positive relationship between TA and therapy outcomes (Horvath & Symonds, 1991). Although the TA typically fluctuates over time, TA measured early in treatment is significantly associated with treatment outcomes (r = .31), consistent with TA measured later in treatment (r = .30; Horvath & Symonds, 1991). The available evidence suggests there is typically a lack of agreement between how clients, therapists, or third-party observers view the working alliance (Tichenor & Hill, 1989). Therapists often rate the TA more poorly than clients (Auerbach, May, Stevens, & Kiesler, 2008; Belding, Iguchi, Morral, & McLellan, 1997), with one study finding a similarity in ratings (Johansen, Melle, Iversen, & Hestad, 2013). It has been suggested that therapists and clients may have different ideas about the nature of the TA, with therapists placing greater importance on tasks and goals (Hatcher, Barends, Hansell, & Gutfreund, 1995). Few studies have examined whether this discrepancy is related to poorer outcomes; extant research includes one study that found no relationship between client–therapist discrepancy and treatment retention (Meier & Donmall, 2006), whereas Auerbach and colleagues (2008) found that a larger discrepancy between raters was associated with poorer behavioral functioning for adolescents during substance abuse treatment.

Beyond Bordin’s conceptualization of the TA, research has increasingly expanded to incorporate an understanding of other features of therapy, such as therapist and client characteristics, and their impact on therapeutic engagement. Ross and colleagues (2008) proposed a Revised Theory of the TA specifically for the forensic context, extending Bordin’s Pantheoretical Model, incorporating greater focus on both therapist and client characteristics (including their interpersonal style [IPS], attachment style, professional and personal skills, treatment readiness, and how these variables affect the core TA factors: goals, task, and bond). They also proposed that the interaction between clients and therapists is crucial to understanding the TA, including their behavior, and emotional and cognitive responses toward each other. Of note, they suggest this interaction is a dynamic one that can change over time.

IPS

IPS describes the consistent pattern of interpersonal behaviors that a person brings to different interpersonal situations (Kiesler, 1987). There are two core dimensions of interpersonal behavior (Kiesler, 1987; Leary, 1957): (a) power/control (ranging from dominance to submission) and (b) affiliation (ranging from hostility to friendliness/warmth; Blackburn & Renwick, 1996). People typically have a preferred pattern of interpersonal behaviors; however, adaptive interpersonal functioning is marked by flexibility in response to situational demands (Kiesler & Auerbach, 2003). A revised model of the circumplex, suggested by Horowitz and colleagues (2006), proposed that interpersonal responses are not automatically evoked but provide an invitation for someone to respond. They note that interpreting the motive behind the interpersonal behavior is important and that ambiguous behaviors can be interpreted differently.

Within offender populations, interpersonal difficulties are prominent, with higher levels of hostility, self-centeredness, irritability, callousness, self-defeating behaviors, and anxiety (Ross et al., 2008). More extreme, pervasive, and inflexible interpersonal behaviors are associated with personality disorder (American Psychiatric Association, 2013). For example, offenders with personality disorder have been found to be noncompliant, dominant, and coercive in their IPS (Blackburn, 1998). Furthermore, offenders are higher in interpersonal dominance than community samples (Edens, 2009). Edens (2009) noted that dominance is related to sex offender treatment non-compliance (Edens, 2009). High levels of dominance (high control/power) and low warmth (low affiliation) are also related to antisocial traits, as are aggressive behaviors within prison (Edens, 2009). Daffern and colleagues have shown that interpersonal dominance and hostile-dominance in particular (elevations on both dominance and hostility) are associated with aggression and violence in mental health units (Daffern, Duggan, Huband, & Thomas, 2010; Daffern, Thomas, et al., 2010; Daffern, Tonkin, et al., 2010). It is unknown what the optimal levels of dominance/submission is for offenders to build a strong TA, or whether complementarity with therapists (a matching between therapists’ and clients’ IPS) is important. Interpersonal complementarity may be particularly important in treatment. For example, submissive clients may prefer dominant therapists and vice versa (reflecting complementarity between the IPSs of the two interactants).

Offender characteristics also affect treatment engagement and the TA (Holdsworth et al., 2014). Holdsworth and colleagues reported that higher levels of hostility were related to an increased likelihood of treatment dropout, reduced participation in treatment, reduced rapport with therapist, reduced treatment satisfaction, and reduced peer support. Offenders who complete treatment have been found to have lower levels of hostility (Derks, 1996), psychopathy (McCarthy & Duggan, 2010), and impulsivity (McCarthy & Duggan, 2010; McMurran, Huband, & Duggan, 2008). A low-level of friendliness (low affiliation) is also related to reduced benefit from the therapeutic process in non-offenders (Hardy et al., 2001). However, if a TA was established despite this low affiliation, the negative impacts were ameliorated. It should be noted that there is limited published research that has explored the impact of offenders’ IPS on the TA task, goals, and bond, although Cookson, Daffern, and Foley (2012) found that hostile-dominance was negatively correlated with TA in a sample of mental health patients.

Therapist Characteristics

Therapist characteristics affect treatment outcomes with sex offenders (Marshall et al., 2002). Therapist empathy and warmth are significantly related to treatment benefit, as are asking open-ended questions, and being directive and rewarding (verbal encouragements; Marshall et al., 2003). Challenging by therapists that is firm but supportive has also been shown to be related to a reduction in victim blaming by sex offenders (Marshall et al., 2002); however, Karver, Handelsman, Fields, and Bickman (2005) warned that too much directiveness (in a non-offender population) can lead clients to view therapists as less warm and empathic, negatively influencing the TA. With a sex offender population, it has been found that a confrontational therapeutic style (aggressive and/or derogatory challenging, critical, hostile, sarcastic) negatively affects treatment-induced change and participation (Drapeau, 2005; Marshall et al., 2003). A controlling and aggressive therapeutic style can also lead to lowered group cohesiveness, which may affect therapeutic quality for sex offender group programs (Beech & Fordham, 1997). However, with sex offenders, high levels of denial and minimization may encourage a confrontational approach by therapists (Serran et al., 2003).

Therapists are often required to change their interpersonal approach in different settings as well as in response to the different IPSs of clients. Flexibility is an important feature for effective therapy, which includes being able to adapt to the clients’ interpersonal behavior and emotional needs (Marshall, 2005). Marshall and colleagues (2003) noted that therapists’ ability to move between a directive and non-directive style was an important feature in adjusting to a clients’ interpersonal approach, and it has been suggested that a reflective style is more effective with aggressive or defensive clients (Ashby, Ford, Guerney, & Guerney, 1957). However, if therapists deem that a supportive approach with offenders is ineffective, they could quickly change to a more confrontational approach (Marshall et al., 2002), which is likely to reduce the TA.

Therapy-Related Interactions

Although much of the previous research looks at either therapist or offender characteristics in isolation, Ross and colleagues (2008) proposed that therapist and client characteristics are not static, isolated factors, but rather lead to cognitive processes and emotional responses affecting the clients’ and therapists’ behaviors toward each other. They note that these processes affect the task, bond, and goal components of the TA. However, research in this field remains limited.

Interpersonal theory explores the interaction between the IPS of therapists and clients by way of measurement of interpersonal complementarity. Complementarity is a core principle of interpersonal theory and relates to covert transactional interactions designed to evoke overt behaviors from others (Kiesler & Schmidt, 2006). When an interaction is complementary, interactants have either a corresponding (friendliness evoking friendliness) or reciprocal (dominance evoking submission) response (Lillie, 2007). Research shows that interpersonal complementarity affects satisfaction with the therapeutic process and engagement (Dryer & Horowitz, 1997). Within a medical setting, complementary affiliation between doctors and patients has been shown to lead to greater treatment compliance (Kiesler & Auerbach, 2003). Greater complementarity is also related to a higher bond rating of the working alliance (Kiesler & Watkins, 1989). However, findings are inconsistent, with more recent research finding no relationship between complementarity and the working alliance in an adolescent substance treatment population (Auerbach et al., 2008). Research has shown that therapists are more likely to show complementarity in their IPS when patients are friendly, with complementarity potentially increasing patients’ self-esteem and self-evaluation (Caspar, Grossmann, Unmüssig, & Schramm, 2005). There is no research on the impact of interpersonal complementarity in offender population treatment studies.

Current Study

It is known that therapeutic process variables account for a significant proportion of treatment outcome; however, more research needs to be conducted to determine the impact of therapists’ and clients’ characteristics on the therapeutic relationship in offender populations. Understanding the TA in this population is important due to the increased level of interpersonal difficulties in those offenders and the increased risk of recidivism associated with disengagement from or non-completion of treatment.

This study has four aims: (a) to describe and compare the nature of the TA from the perspective of therapists and offenders, (b) to determine the relationship between therapists’ and offenders’ IPS and the TA, (c) to explore whether offender hostility and dominance is related to assessments of the TA, and (d) to explore the level of complementarity in IPS between therapists and offenders and determine whether complementarity is related to TA. It was hypothesized that both hostile and dominant therapist and offender IPSs would be negatively associated with the TA. It was further hypothesized that complementarity between therapists and offenders would be positively associated with TA.

Method

Participants

All participants were male, above the age of 18 years, and recruited from the Sex Offender Program run by the Department of Justice (Corrections Victoria is part of the Department of Justice). All participants were assessed as moderate-low to high risk of sexual reoffending (Australian Institute of Criminology, 2011) according to STATIC-99 assessment (with some participants’ risk rating increased from low risk due to clinical over-ride despite STATIC-99 assessment). The treatment program targets dynamic risk factors related to sexual offending, with the community-based Program comprising a 3-hr session once a week and the prison-based program consisting of 2 weekly 3-hr sessions. Total hours in the program range from 72 to 150 hr, between 3 and 8 months, depending on level of risk posed by the offender and their treatment needs.

The sample comprised 75 participants, with 60 recruited from Marngoneet Correctional Centre and 15 from Community Corrections; in the community, the offenders were either on a community order or parole. Those offenders recruited from Marngoneet Correctional Centre started the program toward the end of their sentences, whereas community-based offenders started the program near the beginning of their community or parole orders. Offenders ranged in age from 21 to 73 years (M = 44.97, SD = 15.15), with 61 (53%) offenders born in Australia, 13 (11.3%) overseas, and 1 (0.9%) unspecified. Twenty-seven (23.5%) offenders completed lower than Year 10 level of schooling, 30 (26.1%) completed lower than Year 12, 8 (7%) completed schooling to Year 12 level, with 10 (8.7%) offenders having engaged in tertiary education to a complete or incomplete level. Sixty-seven offenders completed the treatment program, 5 were non-completers, and 3 were ongoing at the time of the study. Reasons for non-completion were (a) re-offense while undertaking the program, (b) offenders withdrawing after hostile and aggressive behavior toward therapists, and (c) mental health issues (active symptoms of psychosis). Most offenders were classified as moderate-high or high risk of reoffending (based on the Static-99), with 28 (24.3%) offenders rated as moderate-high and 26 (22.6%) rated as high. Eighteen (15.7%) offenders fell within the low, moderate-low, or moderate risk of reoffending (with 3 [2.6%] not specified although regarded as moderate to high risk given their inclusion in the treatment program). A one-way between-subjects ANOVA was conducted to compare the effect of risk level on TA, with no significant difference found, F(2, 61) = 0.29, p = .75.

Measures

The Impact Message Inventory (IMI)

The Impact Message Inventory–Circumplex (IMI-C; Schmidt, Wagner, & Kiesler, 1999) is a 56-item inventory that assesses the covert emotional experience of interpersonal behaviors or interactions (Kiesler & Auerbach, 2003). The IMI-C was designed to measure the nature of the dyadic interaction (Kiesler & Auerbach, 2003) and has been used to measure the impact of clients’ interpersonal behaviors on their therapists (Schmidt et al., 1999). The IMI-C consists of an octant scale that follows the traditional interpersonal circle format, consisting of 7 items per subscale: Dominant (e.g., “when I am with this person he makes me feel bossed around”), Hostile-Dominant (e.g., “when I am with this person he makes me feel that I want to get away from him”), Hostile (e.g., “when I am with this person he makes me feel like an intruder”), Hostile-Submissive (e.g., “when I am with this person it appears to me that he thinks he’s inadequate), Submissive (e.g., “when I am with this person it appears to me that he sees me as superior), Friendly-Submissive (e.g., “when I am with this person he makes me feel that I could ask him to do anything”), Friendly (e.g., “when I am with this person he makes me feel that I could lean on him for support”), and Friendly-Dominant (e.g., “when I am with this person it appears to me that he wants to be the charming one”). Each octant scale is rated on a 4-point rating scale (ranging from not at all to very much so). The IMI-C has shown acceptable internal consistency, with Cronbach’s alpha ranging from .69 to .89 across the eight subscales (Schmidt et al., 1999).

The Working Alliance Inventory–Short Form (WAI-SF)

The WAI-SF (Tracey & Kokotovic, 1989) is a self-report measure of the TA between therapist and client. The short-form version consists of 12 items and can be completed by the client or the therapist. The WAI has three subscale scores, consisting of 4 items per subscale; Bond (“[therapist] and I trust one another”), Goal (“[therapist] and I have different ideas on what my problems are”), and Task (e.g., “I believe the way we are working with my problem is correct”), all measured on a 7-point Likert-type scale (ranging from never to always). The WAI-SF has good internal consistency, with a Cronbach’s alpha of .98 (client’s form) and .95 (therapist’s form; Tracey & Kokotovic, 1989).

Procedure

The study was approved by the Monash University Human Research Ethics Committee (MUHREC) and the Department of Justice Human Research Ethics Committee (JHREC). Sexual offenders participating in sex offender treatment through Corrections Victoria (the state of Victoria, Australia corrections department) and their therapists were invited to participate in the study. Participation was voluntary and the study was not associated with the requirements of their program. In June 2012, permission was received to begin recruitment and data collection at Marngoneet Correctional Centre (a medium security prison in Victoria, Australia) and through Victorian Community Corrections services. Questionnaires were administered 3 to 4 weeks after the commencement of treatment. Participants at Marngoneet Correctional Centre completed the IMI-C and WAI-SF while the researcher was present, outside of treatment program time. Community participants were given the IMI-C and WAI-SF, with return paid and addressed envelopes, to complete outside of the treatment program time. At this time, treatment therapists were also given the IMI-C and WAI-SF to be completed about the participants within their treatment group. There were two therapists per group, with each offender being asked to complete questionnaires based on their perception of IPS and TA on both therapists. Therapists were randomly allocated as Therapist 1 or Therapist 2 for each participant.

Data Preparation

Missing data were determined to be random, and four cases were discarded due to having more than 10% missing data. Of the remaining cases, 18 had 1.8% missing, 5 had 3.6% missing, and 8 had 8.3% missing; in these cases, the participant’s mean response for the specific subscale was used. To keep data independent, data were consolidated to use only one offender–therapist response. Data were eliminated first if there was no offender or therapist response; if both full sets were available, “Therapist 1” was chosen. All Therapist 1 and Therapist 2 data sets were correlated except the therapists’ view of the Bond aspect of the TA (see Table 1). This suggests that there was no difference in the way the two therapists viewed the TA with the offender except for therapists’ assessment of the bond; the two therapists’ perceptions of the bond differed. The reason for this difference is unclear. It may be because forensic therapists have different ideas about the importance of the bond, whereas they may agree more readily on the tasks and goals; the therapeutic bond may be seen by some therapists as means through which they can be exploited or manipulated by offenders (Kozar & Day, 2012). It may be that some therapists are reluctant to acknowledge a strong bond because they want to appear distant and professional and focused on goals and tasks. It may also be that within a therapeutic relationship, the bond that is shared with the two therapists is different, that participants share a stronger bond with one therapist compared with another. As therapists’ perceptions were correlated, with the exception of assessment of bond, to aid data analysis, the data were consolidated, although some loss of variance in terms of the WAI-Bond was lost in this process.

Table 1.

Correlations of Therapist 1 and Therapist 2 IMI and WAI-SF Ratings.

	Therapist 1
Therapist 2
WAI-Goal	.33*
WAI-Task	.35*
WAI-Bond	.24
WAI-Total	.33*
IMI Dominant	.76**
IMI Hostile-Dominant	.71**
IMI Hostile	.71**
IMI Hostile-Submissive	.53**
IMI Submissive	.55**
IMI Friendly-Submissive	.87**
IMI Friendly	.72**
IMI Friendly-Dominant	.42**
IMI Control	.62**
IMI Affiliation	.74**

Note. IMI = Impact Message Inventory; WAI-SF = Working Alliance Inventory–Short Form.

p < .05 (two-tailed). **p < .01 (two-tailed).

Approach to Analysis

A one-way between-subjects ANOVA was conducted to analyze the difference between therapists’ and offenders’ ratings of the TA. Independent-samples t tests were run to compare WAI-SF normative samples with the current WAI-SF findings, and to compare IMI Complementarity normative samples with the current IMI Complementarity findings. Correlations were computed to explore the relationship between therapists’ and offenders’ ratings of the TA and IPS. The following IPS calculations were conducted based on instructions from the IMI-C manual (Kiesler & Schmidt, 2006). The Affiliation axis was calculated by combining the dimensions of hostility and friendliness. The Control axis was calculated by combining the dimensions of dominance and submission. Complementarity along the Control axis was computed using the formula: ABS (offender control axis + therapist control axis). Complementarity along the Affiliation axis was computed using the formula: ABS (offender affiliation axis − therapist affiliation axis). A score of 0 equals perfect complementarity, with a score of 6 being perfect non-complementarity. Overall complementarity was computed by summing the two axes’ complementarity scores. A score of 0 equals perfect complementarity, with a score of 12 equating perfect non-complementarity on the total complementarity score. Regression analysis was used to explore whether a hostile or dominant IPS would predict ratings of the working alliance for sex offenders undergoing treatment.

Results

Therapists’ and Offenders’ Ratings of the (WAI-SF)

Offenders’ ratings of the TA Goal, Task, and Total alliance score were all significantly higher than therapists’ ratings. There was no significant difference between offenders’ and therapists’ ratings on the TA Bond (Table 2). Table 3 shows normative samples of clients’ and therapists’ ratings on the WAI-SF. A significant difference was found between therapists’ ratings of the TA in the current sample and the schizophrenia spectrum sample. A significant difference was found between therapists’ ratings of the TA Goal, Task, and Total alliance in the current sample and the adolescent substance treatment sample; however, no difference was found in therapists’ TA Bond ratings. There was no difference found between offenders’ ratings of the TA and clients’ ratings in the two samples.

Table 2.

WAI-SF: Offender and Therapist Ratings.

		Offender			Therapist
	n	M	SD	n	M	SD	t(122)
WAI-Goal	64	21.06	4.12	60	17.40	3.89	25.83**
WAI-Task	64	22.02	4.80	60	16.10	4.59	49.18**
WAI-Bond	64	19.44	5.21	60	17.92	3.73	3.46
WAI-Total	64	62.52	12.74	60	51.42	10.96	26.91**

Note. WAI-SF score range: Subscales = 4-28, total = 12-84. WAI-SF = Working Alliance Inventory–Short Form.

p < .05 (two-tailed). **p < .01 (two-tailed).

Table 3.

WAI-SF: Mean Difference Between Current Study WAI-SF Ratings (Table 2) and Normative Samples.

		Adolescent sample (Auerbach, May, Stevens, & Kiesler, 2008)			Schizophrenia sample (Johansen, Melle, Iversen, & Hestad, 2013)
	n	M	SD	t(101)	n	M	SD	t(104)
Client ratings
WAI-Goal	39	21.23	6.24	0.17	42	20.9	3.8	0.20
WAI-Task	39	21.92	5.93	0.09	42	20.4	4.1	1.80
WAI-Bond	39	20.61	6.01	1.04	42	20.3	4.9	0.85
WAI-Total	39	63.76	16.4	0.43	42	61.6	11.1	0.38
Therapist ratings				t(97)				t(100)
WAI-Goal	39	20.38	4.42	3.53**	42	20.6	3.0	4.48**
WAI-Task	39	19.43	4.34	3.60**	42	20.4	3.4	5.16**
WAI-Bond	39	19.48	4.24	1.93	42	21.5	2.6	5.37**
WAI-Total	39	59.28	12.1	3.35**	42	62.4	7.7	5.59**

Note. WAI-SF score range: Subscales = 4-28, total = 12-84. WAI-SF = Working Alliance Inventory–Short Form.

p < .05 (two-tailed). **p < .01 (two-tailed).

Therapists’ and Offenders’ Ratings of the IMI-C

The covert impacts of the IPS of offenders and group therapists were evaluated using four scales (Dominance, Submission, Friendliness, Hostility); Table 4 shows mean complementarity scores, with normative data (Kiesler & Schmidt, 2006) shown in Table 5. The therapist/offender population showed a lower level of Control Complementarity compared with Affiliation Complementarity, and showed a lower level of Control Complementarity than the normative samples.

Table 4.

Mean and Standard Deviation of IMI Complementarity Scores.

	n	M	SD
Complementarity Control	56	2.48	1.44
Complementarity Affiliation	56	1.82	1.26
Complementarity Total	56	4.29	2.06

Note. IMI complementarity score range: Axes = 0 (perfect complementarity) to 6 (perfect non-complementarity), total = 0 (perfect complementarity) to 12 (perfect non-complementarity). IMI = Impact Message Inventory.

Table 5.

Preliminary Normative Data on IMI Complementarity Scores (Kiesler & Schmidt, 2006) and Mean Difference With Current Sample (Table 4).

	n	M	SD	t(102, 103)
Diabetic patients and physicians
Complementarity Control	48	1.23	0.76	5.40**
Complementarity Affiliation	49	1.38	0.80	2.10*
Complementarity Total	48	2.61	0.76	5.34**
TBI-Patient/Intervention–Admission				t(82)
Complementarity Control	28	1.34	1.12	3.67**
Complementarity Affiliation	28	2.53	1.80	2.10*
Complementarity Total	28	3.86	2.33	0.86
TBI-Patient/Intervention–Discharge				t(82)
Complementarity Control	28	1.62	1.00	2.83**
Complementarity Affiliation	28	2.61	1.46	2.57*
Complementarity Total	28	4.24	1.82	0.11

p < .05 (two-tailed). **p < .01 (two-tailed).

Intercorrelations Between Therapists’ and Offenders’ Ratings of the TA and IPS

Offenders’ perception of therapist Affiliation was positively correlated (r = .57) with the offenders’ rating of the TA (including Total TA and Goal, Task, and Bond subscales). There was also a significant negative correlation between offenders’ perception of therapist Control (r = −.39) and the offenders’ rating of the TA, (including Total TA and Goal, Task, and Bond subscales). Therapists’ perceptions of offender Affiliation was positively correlated (r = .61) with the therapists’ rating of the TA (including Total TA and Goal, Task, and Bond subscales). There was also a significant negative correlation between therapists’ perception of offender Control (r = −.45) and the therapists’ rating of the TA (including Total TA and Goal, Task, and Bond subscales; Table 6).

Table 6.

Correlations of WAI and IMI Scores.

	Therapist affiliation (rated by client)	Therapist control (rated by client)	Client affiliation (rated by therapist)	Client control (rated by therapist)	Affiliation complementarity	Control complementarity	Total complementarity
WAI-Client
Goal	.43**	−.37**	.32*	−.05	−.09	−.21	−.20
Task	.47**	−.25*	.22	−.02	.14	−.10	.02
Bond	.62**	−.41**	.39**	−.14	.19	−.04	.09
Total	.57**	−.39**	.35**	−.08	.10	−.12	−.02
WAI-Therapist
Goal	.09	−.26	.55**	−.39**	−.17	.04	−.07
Task	.02	−.26	.50**	−.29*	−.23	.00	−.14
Bond	.07	−.22	.63**	−.55**	−.17	.03	−.08
Total	.06	−.28*	.61**	−.45**	−.21	.03	−.11
WAI-Client/Therapist Difference
Goal	.17	.20	−.24	.11	.09	.10	.12
Task	.25	.17	−.16	.09	.10	−.06	.02
Bond	−.09	−.01	−.11	−.05	−.17	.29*	.10
Total	.17	.12	−.21	.03	−.01	.13	.08

Note. WAI = Working Alliance Inventory; IMI = Impact Message Inventory.

p < .05 (two-tailed). **p < .01 (two-tailed).

There was no statistically significant association between Complementarity of IPS and the ratings of TA for therapists or offenders. There was, however, a significant association between Complementarity on the Control axis (dominance and submission) and differences in therapist–offender ratings on the TA Bond (r = .29). An increase in non-complementarity on the Control axis was associated with an increase in difference in bond ratings between therapists and offenders.

Hostility, Dominance, and the TA

Therapists’ ratings of offender hostility were associated with lower offenders’ rating of the TA (see Table 7). However, therapists’ ratings of offender dominance were not significantly associated with the offenders’ rating of the TA (Table 7).

Table 7.

Regression Analysis of Therapist Rating of Offender Hostility and Dominance and the Related Offender Rating of Working Alliance.

	Standardized B	T
Constant		15.41**
Therapist rating of offender hostility	−.40	−3.16**
Adjusted R²		.14
n		54
Constant		14.85
Therapist rating of offender dominance	−.17	−1.28
Adjusted R²		.01
n		54

Note. Dependent variable: Client-rated working alliance total.

p < .05 (two-tailed). **p < .01 (two-tailed).

Discussion

The TA

The first aim of this study was to explore and compare the nature of the TA from therapists’ and offenders’ perspectives. Results revealed a significant difference between therapists’ and offenders’ assessments of the Task and Goals elements of the TA, as well as overall TA. In general, offenders rated the TA higher than therapists, except for the Bond component, where there was no statistically significant difference between offenders’ and therapists’ assessments. This difference may be due to offenders feeling less comfortable about the process or connection with therapists when compared with the more practical goal/task components of the therapeutic relationship; this may be due to the fact that there were two therapists and the treatment was provided in a group format so the strength of the bond may have been dispersed as offenders connected emotionally with two therapists rather than one. The discrepancy between ratings of TA by offenders and therapists is consistent with a study of therapists and adolescents in substance abuse treatment, and a study of short-term psychotherapy with non-offenders (Auerbach et al., 2008; Tichenor & Hill, 1989). It has been suggested that therapists and offenders see the TA differently due to the differing filter through which each may see the relationship (Horvath, 2000). Horvath (2000) suggested that therapists would potentially be influenced by their training to see the TA through a theoretical lens, whereas clients would assess the TA based on their past relationship experiences.

The TA alliance ratings were compared with two normative samples (Auerbach et al., 2008; Johansen et al., 2013), with offender ratings of the TA being consistent with both adolescents in substance abuse treatment (Auerbach et al., 2008) and clients on the schizophrenia spectrum undergoing treatment (Johansen et al., 2013). However, therapists’ ratings of TA with sex offenders in this study were significantly lower than therapists’ ratings of the TA in the two normative samples. This may be due to the challenges of working in a forensic setting, with therapists potentially dealing with more difficult IPSs, or due to the challenges associated with exploring sexual offenses. The bond ratings of therapists treating adolescent substance abusers was similar to the bond ratings of therapists treating sex offenders, suggesting there may be similar challenges with developing, or a reluctance to develop, a therapeutic bond in these populations.

IPS and the TA

The second aim was to determine the relationship between therapists’ and offenders’ IPSs and the TA. There was a positive association between therapists’ perception of offender Affiliation and their rating of the TA. In addition, therapists’ perception of offender Control/Dominance was associated with poor TA; that is, therapists who assessed their offender clients as friendly rather than hostile reported higher TA, and therapists who assessed their offender clients as more dominant reported a lower TA. This is consistent with previous research suggesting that offender hostility and dominance negatively affect rapport and engagement (Edens, 2009; Holdsworth et al., 2014). There was also a positive association between offenders’ perception of therapist Affiliation and all aspects of the TA. However, if offenders assessed therapists as being more Controlling, then this negatively affected their perception of the TA. This suggests that therapist hostility and dominance both negatively affect the TA. This is consistent with previous research showing that therapist characteristics can affect offender engagement and the quality of the working relationship (Beech & Fordham, 1997; Drapeau, 2005; Marshall et al., 2002, 2003).

Therapists may be perceived as hostile if they are overly confrontational, use aggressive and/or derogatory challenging, and/or are overly critical or sarcastic (Drapeau, 2005, Marshall et al., 2003). An overly dominant therapist is controlling and/or overly directive; he or she has an authoritarian rather than an authoritative IPS. When dealing with offenders with a hostile IPS, therapists may benefit from seeking an understanding of the function of the offenders’ hostility and its developmental origins. It is the offenders’ past relational experiences that likely lead to current interpersonal difficulties, with offenders’ early life experiences affecting how they view current relationships. Hostility will likely stem from relational schemas (expectations regarding relational interactions, including cognitions, emotions, and strategies/procedures for negotiating relationships; Safran et al., 2000). Therefore, if therapists can understand the nature of offenders’ maladaptive relational patterns, they may be able to understand the function the hostility serves for offenders and its developmental origins, and seek to address the problematic (albeit seemingly acceptable and familiar) interpersonal behavior collaboratively, rather than simply reacting to hostility with hostility or control. Naturally, when therapists and other offender clients are unsafe, controlling tactics are required, but if there is no imminent risk of harm, therapists should work collaboratively and directly with hostility and offender control, and attempt to exercise other, more functional interpersonal behaviors.

Hostility, Dominance, and the TA

The third aim of this study was to determine whether offender hostility or dominance affected their own assessment of the TA. It is important to note that client-rated TA has been shown to have a greater relationship with therapy outcome (Horvath, 2000). It was hypothesized that offenders who were higher in hostility and dominance would have worse ratings of the TA; this hypothesis received partial support. Offender Hostility was associated with the TA as perceived by offenders, whereas Dominance was not. The first finding is consistent with research that has shown higher levels of offender hostility is related to reduced participation in treatment and rapport with therapists (Holdsworth et al., 2014). Conversely, the finding that dominance was not associated with the TA is somewhat dissimilar to the findings reported by Edens (2009). In Edens (2009), sex offenders who had higher levels of dominance showed greater non-compliance of treatment, as rated by staff. That study used an offender reported measure of dominance and looked at aspects of attendance, assignment completion, and within group behavior as their outcome. This study may also have had an overlap of participants who were dominant, but also hostile, whereas the current study included only those who were perceived as dominant or hostile, not an overlap of both hostile-dominance. Based on the current findings, therapists perceiving hostility in sex offenders within their group program should be aware that this may be associated with offenders having a more negative view of the TA. However, this may not be the case with those who are displaying dominant, yet friendly, behaviors. These findings suggest that within the sex offender population, hostile IPS rather than dominance is more important to TA, with dominant presentations in this population less likely to affect the development of the TA. With hostile sex offenders it is particularly important for therapists to be flexible in their interpersonal approaches and to maintain awareness of their own interpersonal behaviors (Marshall et al., 2003). Ideally a warm, empathic approach is recommended, with flexibility to move from directive (but non-confrontational) to non-directive (using open-ended questions) interactions (Marshall et al., 2003).

Complementarity and the TA

The fourth aim was to determine whether complementarity in IPS was related to TA. Therapists’ and offenders’ complementarity for Control was lower than complementarity for Control in the normative data, reported by Kiesler and Schmidt (2006). This suggested that the degree of match or complementarity between offenders and therapists was less than that observed in a non-forensic setting; this may be because forensic therapists are reluctant to relinquish control for fear of collusion or manipulation, whereas non-forensic practitioners may be more comfortable negotiating issues of control. It was hypothesized that greater complementarity between therapists and offenders would be positively related to TA. This hypothesis was not supported; Complementarity in IPS was unrelated to the Goal, Task, Bond, or overall TA. These results suggest that interpersonal complementarity between sex offenders and therapists within a group program does not affect TA. This finding is inconsistent with previous findings by Dryer and Horowitz (1997; regarding satisfaction of interactions) and Kiesler and Watkins (1989), however both these studies were with non-forensic populations. The current finding is consistent with findings from an adolescent substance abuse program (Auerbach et al., 2008), where complementarity was also unrelated to TA. It may therefore be that offender clients who have high levels of dominance are more familiar with non-complementary interactions (having their dominant actions met by dominant reactions, particularly from authority figures). In addition, given the mandated forensic population, offenders may have also been more inclined to be overly compliant or to endorse responses consistent with positive impression management. Finally, it is possible that therapists’ and client offenders’ IPS characteristics have greater impact on TA than complementarity.

An increase in Control non-complementarity was related to a greater difference in the ratings of the TA Bond by offenders and therapists, but not with the ratings of Goal and Task. Although it is not uncommon to have a discrepancy in perceptions of the TA (Tichenor & Hill, 1989), the current sample had no statistically significant difference in perceptions of the TA Bond. Potentially, an increase in difference in bond rating could be due to a lack of reciprocation on the Control axis. If one interactant responded to a dominant approach with dominance, or a submissive approach with submission, this may influence one of the parties to rate the bond lower. Therefore, responding to a dominant action with an attempt to control or challenge may, although necessary, be problematic with regard to TA, and effort will be required to acknowledge the underlying reasons for the rupture in the alliance and attempts will need to be made to repair the relationship. Alternatively, these results imply that therapists may be able to use a non-directive approach with those with a dominant IPS, with use of reflection and open-ended questions, to preserve the TA bond.

Limitations/Future Research

There are several limitations which should be considered when interpreting the results presented here. The first limitation was the measurement of the TA and IPS at Week 3 of the treatment program. Although this is when TA is commonly measured, in a group setting, the TA may not have been fully developed at this point in time. Further to this, the context of a group program, with the presence of other group members and two therapists, may have influenced perceptions of the TA. This may make the findings difficult to apply to one-therapist relationships due to the coexisting, and potentially overlapping, therapeutic relationships. Furthermore, the use of data from one of the two therapists may have affected the findings due to reducing the variability within the data; however, our use of random allocation as Therapist 1 or 2 will help ameliorate such differences. Another limitation was that social desirability was not assessed; it may also be that offenders in this population are inclined to be overly compliant or use impression management, thus potentially affecting the accuracy of the findings. However, it is noted that the WAI is a commonly used tool in offender research. Another limitation was the small sample size; although smaller sample sizes are not uncommon in forensic settings, this may have affected statistical power and therefore, our ability to detect significant findings; generalizability is also affected. Furthermore, some of the data were incomplete due to non-responses by therapists and by offenders within the Community Corrections setting. Although specific reasons for non-completion were not gathered, this may potentially be associated with participants’ IPS and TA.

Future research could expand the current findings by exploring the link between sex offenders’ IPS and treatment outcomes (completion and treatment gain), as well as their impact on recidivism. The current study used a short-form version of the WAI; future research might incorporate the full version to increase validity and further explore TA. The TA has been found to change over the course of treatment (Horvath & Symonds, 1991), therefore looking at whether IPS impacts on the TA changes over time may also be informative for treatment providers. People with different interpersonal presentations may also respond differently to any ruptures in the TA, or reduce their ability to repair the relationship. It may be at this stage that complementarity can be beneficial; however, this proposition would require further empirical investigation.

Conclusion

The current study explored the impact of IPS on the TA between therapists and offenders in a sex offender treatment program. The results expand our current understanding of the TA to incorporate interpersonal characteristics of those involved in the working relationship. Offenders viewed the TA more positively than therapists on all aspects except Bond. In addition, when therapists or offenders assessed the other as more Affiliative, then they also viewed the TA more positively, whereas viewing the other as being more Controlling had a negative impact on how that interactant (therapists or offenders) perceived the TA. This suggests that the perception of the interpersonal behaviors of others is important, and that friendliness/affiliation is conducive to a more positive TA; interpersonal dominance worsens perceptions of the therapeutic relationship. Complementarity in IPS was not found to affect the TA, which suggests that choosing therapists to match offenders’ IPS in an attempt to increase complementarity is unnecessary. It was further found that offender hostility was associated with a more negative perception of the TA.

Therapists should keep in mind that a dominant/overly directive approach may create compliance but potentially impair the TA. Therapists should aim to lessen engagement in offenders’ dysfunctional interpersonal patterns, to be aware of their own hostile and dominant behaviors, and to be flexible in their responses to offenders; making sarcastic, critical or hostile comments; or having a hostile tone or body language in response to offenders’ hostility will negatively affect the TA. Rather, warmth, empathy, using a reflective style with verbal encouragements and having the capacity and preparedness to shift from a non-directive (using open-ended questions) to a directive approach (but non-confrontational) is important for building and maintaining beneficial therapeutic relationships. This approach may be easier to maintain if therapists can understand the underlying function of offenders’ behaviors and to recognize that dysfunctional interpersonal patterns stem from earlier adverse life experiences.

Due to the higher prevalence of difficult interpersonal functioning in forensic populations, support and supervision for therapists who are working with hostile and domineering sexual offenders is critical. Supervision and support may help minimize frustration and hostile responses, while maximizing opportunities for working effectively with this challenging offender population. It is recommended that in addition to gaining knowledge and skills concerning program content and facilitation, a higher focus needs to be placed on training therapists to deal with responsivity factors, such as offender dominance and hostility, and managing ruptures which occur in the therapeutic relationship as a consequence of problems within the relationship.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Arlington, VA: American Psychiatric Publishing.

Ashby

J. D.

Ford

D. H.

Guerney

B. G.

Guerney

L. F.

(1957). Effects on clients of a reflective and leading type of psychotherapy. Psychological Monographs, 71, 1-32. doi:10.1037/h0093778

Auerbach

S. M.

May

J. C.

Stevens

Kiesler

D. J.

(2008). The interactive role of working alliance and counselor-client interpersonal behaviors in adolescent substance abuse treatment. International Journal of Clinical and Health Psychology, 8, 617-629. Retrieved from http://www.redalyc.org/pdf/337/33712016001.pdf

Australian Institute of Criminology (2011). Prison-based correctional rehabilitation: An overview of intensive interventions for moderate to high-risk offenders. Trends & Issues in Crime and Criminal Justice, 412, 1-6. Retrieved from http://www.aic.gov.au/media_library/publications/tandi_pdf/tandi412.pdf

Beech

Fordham

A. S.

(1997). Therapeutic climate of sexual offender treatment programs. Sexual Abuse: A Journal of Research and Treatment, 9, 219-237. doi:10.1177/107906329700900306

Belding

M. A.

Iguchi

M. Y.

Morral

A. R.

McLellan

A. T.

(1997). Assessing the helping alliance and its impact in the treatment of opiate dependence. Drug and Alcohol Dependence, 48, 51-59. Retrieved from http://dx.doi.org/10.1016/S0376-8716(97)00103-8

Blackburn

(1998). Relationship of personality disorders to observer ratings of interpersonal style in forensic psychiatric patients. Journal of Personality Disorders, 12, 77-85. doi:10.1521/pedi.1998.12.1.77

Blackburn

Renwick

S. J.

(1996). Rating scales for measuring the interpersonal circle in forensic psychiatric patients. Psychological Assessment, 8, 76-84. doi:10.1037/1040-3590.8.1.76

Bordin

E. S.

(1979). The generalizability of the psychoanalytic concept of the working alliance. Psychotherapy: Theory, Research and Practice, 16, 252-260. doi:10.1037/h0085885

10.

Caspar

Grossmann

Unmüssig

Schramm

(2005). Complementary therapeutic relationship: Therapist behavior, interpersonal patterns and therapeutic effects. Psychotherapy Research, 15, 91-102. doi:10.1080/10503300512331327074

11.

Cookson

Daffern

Foley

(2012). The relationship between aggression, interpersonal style and therapeutic alliance during short-term psychiatric hospitalization. International Journal of Mental Health Nursing, 21, 20-29. doi:10.1111/j.1447-0349.2011.00764.x

12.

Daffern

Duggan

Huband

Thomas

(2010). Staff and patient’s perceptions of each other’s interpersonal style: Relationship with severity of personality disorder. International Journal of Offender Therapy and Comparative Criminology, 54, 611-624. doi:10.1177/0306624X09335111

13.

Daffern

Thomas

Ferguson

Podubinski

Hollander

Kulkarni

. . . Foley

(2010). The impact of psychiatric symptoms, interpersonal style and coercion on aggression during psychiatric hospitalization. Psychiatry and Biological Processes, 73, 365-382. Retrieved from http://dx.doi.org/10.1521/psyc.2010.73.4.365

14.

Daffern

Tonkin

Howells

Krishnan

Ijomah

Milton

(2010). The impact of interpersonal style and perceived coercion on aggression and self-harm in personality-disordered patients admitted to a secure psychiatric hospital. Journal of Forensic Psychiatry & Psychology, 21, 426-445. doi:10.1080/14789940903505951

15.

Derks

F. C. H.

(1996). A forensic day treatment program for personality disordered criminal offenders. International Journal of Offender Therapy and Comparative Criminology, 40, 123-134. doi:10.1177/0306624X96402005

16.

Drapeau

(2005). Research on the processes involved in treating sexual offenders. Sexual Abuse: Journal of Research and Treatment, 17, 117-125. doi:10.1177/107906320501700203

17.

Dryer

Horowitz

L. M.

(1997). When do opposites attract? Interpersonal complementarity versus similarity. Journal of Personality and Social Psychology, 72, 592-603. doi:10.1037/0022-3514.72.3.592

18.

Edens

J. F.

(2009). Interpersonal characteristics of male criminal offenders: Personality, psychopathological, and behavioral correlates. Psychological Assessment, 21, 89-98. doi:10.1037/a0014856

19.

Frost

Connolly

(2004). Reflexivity, reflection, and the change process in offender work. Sexual Abuse: Journal of Research and Treatment, 16, 365-380. doi:10.1177/107906320401600408

20.

Hardy

G. E.

Cahill

Shapiro

D. A.

Barkham

Rees

Macaskill

(2001). Client interpersonal and cognitive styles as predictors of response to time-limited cognitive therapy for depression. Journal of Consulting and Clinical Psychology, 69, 841-845. doi:10.1037/0022-006X.69.5.841

21.

Hatcher

R. L.

Barends

A. W.

Hansell

Gutfreund

M. J.

(1995). Patients’ and therapists’ shared and unique views of the therapeutic alliance: An investigation using confirmatory factor analysis in a nested design. Journal of Consulting and Clinical Psychology, 63, 636-643. doi:10.1037/0022-006X.63.4.636

22.

Holdsworth

Bowen

Brown

Howat

(2014). Offender engagement in group programs and associations with offender characteristics and treatment factors: A review. Aggression and Violent Behavior, 19, 102-121. doi:10.1016/j.avb.2014.01.004

23.

Horowitz

L. M.

Wilson

K. R.

Turan

Zolotsev

Constantino

M. J.

Henderson

(2006). How interpersonal motives clarify the meaning of interpersonal behavior: A revised circumplex model. Personality and Social Psychology Review, 10, 67-86. doi:10.1207/s15327957pspr1001_4

24.

Horvath

A. O.

(2000). The therapeutic relationship: From transference to alliance. Journal of Clinical Psychology, 56, 163-173. doi:10.1002/(SICI)1097-4679(200002)56:2<163::AID-JCLP3>3.0.CO;2-D

25.

Horvath

A. O.

Symonds

B. D.

(1991). Relation between working alliance and outcome in psychotherapy: A meta-analysis. Journal of Counseling Psychology, 38, 139-149. doi:10.3389/fpsyg.2011.00270

26.

Johansen

Melle

Iversen

V. C.

Hestad

(2013). Personality traits, interpersonal problems and therapeutic alliance in early schizophrenia spectrum disorders. Comprehensive Psychiatry, 54, 1169-1176. doi:10.1016/j.comppsych.2013.05.016

27.

Karver

M. S.

Handelsman

J. B.

Fields

Bickman

(2005). A theoretical model of common process factors in youth and family therapy. Mental Health Services Research, 7, 35-51. doi:10.1007/s11020-005-1964-4

28.

Kiesler

D. J.

(1987). Manual for the impact message inventory. Palo Alto, CA: Consulting Psychologists Press, Inc.

29.

Kiesler

D. J.

Auerbach

S. M.

(2003). Integrating measurement of control and affiliation in studies of physician-patient interaction: The interpersonal circumplex. Social Science and Medicine, 57, 1707-1722. doi:10.1016/S0277-9536(02)00558-0

30.

Kiesler

D. J.

Schmidt

J. A.

(2006). The Impact Message Inventory–Circumplex (IMI–C) manual: Sampler set, manual, test booklet, scoring key, work sheets. Menlo Park, CA: Mind Garden.

31.

Kiesler

D. J.

Watkins

L. M.

(1989). Interpersonal complementarity and the therapeutic alliance: A study of the relationship in psychotherapy. Psychotherapy, 26, 183-194. doi:10.1037/h0085418

32.

Kozar

C. J.

Day

(2012). The therapeutic alliance in offending behavior programs: A necessary and sufficient condition for change? Aggression and Violent Behavior, 17, 482-487. doi:10.1016/j.avb.2012.07.004

33.

Leary

(1957). Interpersonal diagnosis of personality. New York, NY: Ronald Press.

34.

Levenson

J. S.

Macgowan

M. J.

Morin

J. W.

Cotter

L. P.

(2009). Perceptions of sex offenders about treatment: Satisfaction and engagement in group therapy. Sexual Abuse: Journal of Research and Treatment, 21, 35-56. doi:10.1177/1079063208326072

35.

Lillie

(2007). Getting clients to hear: Applying principles and techniques of Kiesler’s Interpersonal Communication Therapy to assessment feedback. Psychology and Psychotherapy: Theory, Research and Practice, 80, 151-163. doi:10.1348/147608306X115198

36.

Marshall

W. L.

(2005). Therapist style in sexual offender treatment: Influence on indices of change. Sexual Abuse: A Journal of Research and Treatment, 17, 109-116. doi:10.1177/107906320501700202

37.

Marshall

W. L.

Burton

D. L.

(2010). The importance of group processes in offender treatment. Aggression and Violent Behavior, 15, 141-149. doi:10.1016/j.avb.2009.08.008

38.

Marshall

W. L.

Serran

Fernandez

Y. M.

Mulloy

Mann

Thornton

(2003). Therapist characteristics in the treatment of sexual offenders: Tentative data on their relationship with indices of behaviour change. Journal of Sexual Aggression, 9, 25-30. doi:10.1080/355260031000137940

39.

Marshall

W. L.

Serran

Moulden

Mulloy

Fernandez

Y. M.

Mann

Thornton

(2002). Therapist features in sexual offender treatment: Their reliable identification and influence on behaviour change. Clinical Psychology and Psychotherapy, 9, 395-405. doi:10.1002/cpp.335

40.

Martin

D. J.

Garske

J. P.

Davis

M. K.

(2000). Relation of the therapeutic alliance with outcome and other variables: A meta-analytic review. Journal of Consulting and Clinical Psychology, 68, 438-450. doi:IO.I037//0022-006X.68.3.438

41.

McCarthy

Duggan

(2010). Engagement in a medium secure personality disorder service: A comparative study of psychological functioning and offending outcomes. Criminal Behaviour and Mental Health, 20, 112-128. doi:10.1002/cbm.758

42.

McMurran

Huband

Duggan

(2008). A comparison of treatment completers and non-completers of an in-patient treatment programme for male personality-disordered offenders. Psychology and Psychotherapy: Theory, Research and Practice, 81, 193-198. doi:10.1348/147608308X288762

43.

Meier

P. S.

Donmall

M. C.

(2006). Differences in client and therapist views of the working alliance in drug treatment. Journal of Substance Use, 11, 73-80. doi:10.1080/14659890500137004

44.

Morgan

Luborsky

Crits-Christoph

Curtis

Solomon

(1982). Predicting outcomes of psychotherapy by the Penn Helping Alliance Rating Method. Archives of General Psychiatry, 39, 397-402. doi:10.1001/archpsyc.1982.04290040013002

45.

Olver

M. E.

Stockdale

K. C.

Wormith

(2011). A meta-analysis of predictors of offender treatment attrition and its relationship to recidivism. Journal of Consulting and Clinical Psychology, 79, 6-21. doi:10.1037/a0022200

46.

Ross

E. C.

Polascheck

D. L.

Ward

(2008). The therapeutic alliance: A theoretical revision for offender rehabilitation. Aggression and Violent Behavior, 13, 462-480. doi:10.1016/j.avb.2008.07.003

47.

Schmidt

J. A.

Wagner

C. C.

Kiesler

D. J.

(1999). Psychometric and circumplex properties of the octant scale Impact Message Inventory (IMI-C): A structural evaluation. Journal of Counseling Psychology, 46, 325-334. doi:10.1037/0022-0167.46.3.325

48.

Serran

Fernandez

Marshall

Mann

(2003). Process issues in treatment: Application to sexual offender programs. Professional Psychology: Research and Practice, 34, 368-374. doi:10.1037/0735-7028.34.4.368

49.

Tichenor

Hill

(1989). A comparison of six measures of working alliance. Psychotherapy, 26, 195-199. doi:10.1037/h0085419

50.

Tracey

T. J.

Kokotovic

A. M.

(1989). Factor structure of the Working Alliance Inventory. Psychological Assessment, 1, 207-210. doi:10.1037/1040-3590.1.3.207