Nzps Elizabeth Ross

Examining the relationship between client factors,

therapeutic alliance, and treatment outcome in high-risk

violent offender treatment

Dr Elizabeth Ross, Dr Devon Polaschek, Dr Marc Wilson

Victoria University of Wellington

What we know

• Therapeutic alliance is a collaborative relationship between therapist and client that can facilitate positive change for the client (Ross, Polaschek & Ward, 2008)

• Therapeutic alliance accounts for 25% variance in outcome in therapy (Horvath & Symonds, 1991)

• A client’s motivation to change linked to both a positive alliance and positive treatment gains

What we know

• Therapeutic alliance and other process factors overlooked in offender rehabilitation

• Despite compelling reasons for salience of these factors

• No studies with incarcerated high-risk violent offenders but partner-violent community treatment studies promising (Taft et al., 2003, 2004, 2007)

Research Question

• What is the relationship between motivation to change, therapeutic alliance and treatment outcome for high-risk incarcerated violent offenders in treatment?

Method: Setting

• Rimutaka Violence Prevention Unit• 30-bed low-medium security Special

Treatment Unit at Rimutaka Prison, Wellington, New Zealand

• 36-week CBT based programme• Aim is to reduce violent re-offending• Voluntary but heavily influenced by parole

(external motivation)

Method: Participants

• Cohort of 70 men across seven treatment groups

• 31 yrs, 54% indigenous Maori, 52 % psychopathic, estimated risk over 5 years = .65, 8.7 years average sentence

• 11 therapists (2 per group)• 2 clinical supervisor observers

Method: Measures

• Therapeutic Alliance = Working Alliance Inventory (WAI) - Short version, 12 items, Goals, Tasks and Bond

• Motivation to Change (MTC) = item 13 of Client Attributes Scale therapist-rated

• Outcome = Violence Risk Scale Change (VRS), Treatment Completion/Time in Programme

Method: Procedure

• Therapeutic alliance and motivation measured at four time points Week 2, 10, 18, 30

• WAI by observer and MTC by therapist• VRS by trained consultant pre and post-

programme• Treatment completion and time in the

programme recorded

Results: Can therapists and offenders form an alliance?

• Yes

• The average score for the WAI was 65/80

• All raters scored the alliance highly and it only increased over time

Results: Is motivation to change related to alliance?

• Yes• At time 1 motivation and alliance correlated

r =.53, p < .01• In multiple regression motivation to change

was the only significant client variable accounts for 29% variance in WAI total scores at Week 2 of the programme

Results: Is therapeutic alliance related to outcome?

• Mixed Results• Chose to use Hierarchical Linear

Modeling - good for “nested” data and mediation

• In Hierarchical Linear Models, no significant relationship between alliance at any time point and change in the VRS

Results: Is therapeutic alliance related to outcome?

• In order to look at predicting the outcome of treatment completion need to assess the “odds” using Logistic Regression

• The alliance at specific time points did not predict in HLM so looked at the change in alliance scores across time

• In Logistic Regression Change in WAI scores predicted odds of client completing (Odds ratio = .08, 83% versus 70% chance)

Results: Is motivation to change related to outcome?

• Yes

• In a logistic regression, MTC predicted the odds of client completion better than the therapeutic alliance (Odds ratio = 1.19, 90% versus 70% chance)

Results: What is the relationship between motivation, therapeutic

alliance and treatment outcome? • Structural Equation Modeling allows us to

build models of mediation relationships• Conceptual pathways were tested using: • Motivation (MTC) at time 1, therapeutic

alliance (WAI) at time 2, change in risk of violence (VRS), treatment completion and time in programme

Mediation Model 1:Motivation-Alliance-Change

.28* .44*

MOTIVATIONTime 1

ALLIANCETime 2

CHANGE

Mediation Model 2:Motivation-Alliance-Completion

.30* .39*

MOTIVATIONTime 1

ALLIANCETime 2

COMPLETION

Mediation Model 3:Motivation-Alliance-Prog Time

.34* .60*

MOTIVATIONTime 1

ALLIANCETime 2

PROG TIME

Results: Comparing models

FIT INDICES

2

df

p

CFI

RMSEA

Results: Comparing models

FIT INDICES CHANGE

2 .04

df 1

p .83

CFI 1

RMSEA .00

Results: Comparing models

FIT INDICES CHANGE COMPLETE

2 .04 4.17

df 1 1

p .83 .04

CFI 1 .68

RMSEA .00 .18

Results: Comparing models

FIT INDICES CHANGE COMPLETE TIME

2 .04 4.17 2.47

df 1 1 1

p .83 .04 .12

CFI 1 .68 .87

RMSEA .00 .18 .13

What if motivation is the mediator?

.47* .28*

FIT INDICES MOTIVATION MED ALLIANCE MED

2 .98 .04

df 1 1

p .32 .83

CFI 1 1

RMSEA .00 .00

ALLIANCETime 1

MOTIVATIONTime 2

CHANGE

Results Summary

• Strong alliances formed • Motivation to change at time 1 most

strongly related to time 1 alliance • Alliance not explaining variance in

outcome in HLM’s• Change in alliance across time predicts

odds of treatment completion but not as well as average motivation across time

Results Summary

• Motivation to change at time 1 strongly related to VRS change and time in programme mediated by alliance at time 2

• However support for bi-directional relationship with motivation also acting as a mediator between alliance and outcome

Implications: Forensic work

• Alliance is high despite these offenders being “challenging” psychopaths

• Motivation to change is vital

• Clients may form motivation in treatment

• Therapeutic alliance is important but may be more facilitative of change rather than directly related to change

• May work through motivation, protection

Implications: Future Research

• Next step of current research is to look at recidivism: Are alliance and motivation related to recidivism?

• Important to continue research in this challenging area

Acknowledgements

• Dr Devon Polaschek and Dr Marc Wilson

• Victoria University of Wellington

• Department of Corrections

• Staff and inmates of RVPU