CBT and Ex Offenders

7/31/2019 CBT and Ex Offenders

1/82

Home Office Research Study 171

Changing offendersattitudes and behaviour:what works?

Part I:

The use of cognitive-behavioural approaches with

offen ders: m essages from th e research .

Julie Vennard, Darren Sugg and Carol Hedderman

Part II:

The influence of cognitive approaches: a survey of

pr obation pr ogramm es.

Carol Hedderman and Darren Sugg

A Research and Statistics Directorate Report

London: Home Office

Home Office

Research and

StatisticsDirectorate


2/82


3/82

Foreword

O ffend er p ro grammes wh ich draw on the p sych o l o gical the ories and

t e chn iques of cognitive - b e h avi o urism seek to modify patterns of thinking

and reduce offending behaviour. This research was undertaken with a view

to increasing understanding of the effe c t i veness of cognitive - b e h av i o u ra l

interventions with offenders and providing information about their use byprobation services in England and Wales.

The first part of the report reviews the findings of relevant research studies.

It cons ide r s the r a n ge of t echniques subsumed w i th in the

c o g n i t i ve b e h av i o u ral fra m ew o rk, th e me thods use d to gauge th eir

e ffe c t i veness and the conclusion wh ich can be safe ly drawn about their

impact on offending behaviour.

The second part of the report describes the results of a national survey of

p rob ation service p ro grammes wh ich draw on cogn itive - b e h av i o u ra l

methods. What types of offender are selected for these programmes, how

a re the pro grammes man aged and re s o u rced and to w hat exte nt do

probation services measure their impact on offenders attitudes and ways of

behaving?

Together the literature review and the survey provide a valuable source of

information for policy makers, researchers and those involved in managing

or d elive ring offender pro grammes. They also sugggest ways in w hich

p robation services can strengthen pro grammes wh ich draw on cognitive -behavioural approaches and improve their understanding of the components

and techniques which work best with different types of offender.

CHRIS LEWIS

Head of Offenders and Corrections Unit

Research and Statistics Directorate

July 1997

Foreword

iii


4/82

Acknowledgements

The authors would like to thank all of the Chief Officers of Probation who

agreed to their services taking part in the survey and the individual

respondents tasked with providing the info rmation we req uired. We are

also grateful to James McGuire for valuable comments on an early draft of the

literature review.

Carol Hedderman

Darren Sugg

Julie Vennard

iv

Changing offenders attitudes and behaviour: what works?


5/82

Contents

Page

For eword iii

Ackn owledgem en ts iv

Sum m ary vii

The use of cognitivebehavioural app roaches with offenders:

messages from the research

The influence of cognitive approaches: a survey of probation

programmes

Th e use of cogn itivebeh aviour al app r oaches w ith offen ders:

m essages from the r esearch

Introduction 1

The use of cognitivebehavioural app roaches with offenders:

messages from the research 1

Cognitive-behaviourism: theory and techniques 5

How cognitive-behavioural approaches have been evaluated 8

The use and effectiveness of cognitive-behavioural approaches

with offenders 13

Designing and delivery effective programmes 27

Conclusion 33

Th e in fluence of cognitive app r oaches: a survey of

pr obation p rogramm es

Introduction 39

The number and nature of cognitive skills programmes 39

Selection criteria and take-up rates 43

Management, training and staffing 46

Partnerships 48

Monitoring, evaluation and costs 50

Conclusion 51

Referen ces 53

Publication s 61

Contents

v


6/82

vi

Changing offenders attitudes and behaviour : what works?


7/82

Summary

The use of cognitivebehavioural approaches withoffenders: messages from the research

O ver th e last de cade , re s e a rch has suggeste d that some fo rm s of

re h ab i l i t a t i ve wo rk can be effe c t i ve in reducing re o ffending. Amon g therange of methods of working with offenders, those encompassed within the

t e rm cognitive - b e h av i o u ral are incre a s i n g ly favo u red. The y are widely

v iewed as offering considerable adva nt ages over more traditional fo rms of

intervention. Because this term is so broad it is difficult to define precisely,

but it invo l ves helping offe n de rs to face up to th e consequences of their

actions, to understand their motives, and to de velop new ways of controlling

their behaviour.

This review examines the evidence on the application and effectiveness of

cognitive-behavioural approaches with offenders. It also draws lessons from

th e litera t u re abou t th e succ essful d elive ry and managemen t of such

programmes.

The main points to emerge from the review are:

The effectiveness of interventions with offenders varies significantly

according to the type of approach adopted. Programmes which seek

to modify offenders patterns of thinking and behaving are generally

m o re successful t han tech niques such as group or individualcounselling and non-directive therapy.

E ven these appro a ches r a re ly p ro duce major r educ t ions in

reoffending among offenders in general. This supports the view that

p ro gramme components and styles of delive ry need to be cl o s e ly

matched to the needs and learning styles of particular offenders.

The use of cognitive-behavioural methods in programmes which also

include training in social skills (for example, teaching sex offenders to

fo rm appro p riate relationships w ith othe r adults) show the most

positive results with both juvenile and adult offenders.

Summary

vii


8/82

L a ck of agreement about how to classify the ra n ge of tech n i q u e s

encompassed by the cognitive-behavioural approach means that it is

no t p ossible to ident ify p re c i s e ly wh ich par ticular fo rm s of

intervention are most promising.

The research shows that the impact of cognitive-behavioural methods

of working with offenders can be enhanced by:

t a rgetting on fa c t o rs which have contributed to the offe n d i n g

behaviour

using active, participatory, problem solving methods of working

matching the intensity and duration of intervention to likely future

risk of offending

running programmes as intended, without allowing the goals and

methods to shift over time.

Im plicat ions for policy, practice and future research

The positive, but inconcl us ive, findings of the revi ews summarised in the

re p o rt raise a nu mber of issues for th ose invo l ved in th e de live ry and

management of offender p rogrammes.

R e s e a rch e rs and pra c t i t i o n e rs wo rking with offe n d e rs need to

d evelop and agree definition s for t he va rious co mpo ne nts o f

cognitivebehavioural methods. More also needs to be known about

what type of approach works best, under what conditions and types

of setting, with whom.

Offenders should be assessed and allocated to programmes according

to their particular needs and learning styles. Although c ognitive -

b e h av io u ral meth ods can yield the most successful outcomes, the y

a re unlike ly to wo rk unless othe r p rinciples concern ing effe c t i ve

programmes are followed.

E ffe c t i ve use of cognitive b e h av i o u ral app ro a ches re q u i res a fuller

knowledge of the underlying principles, theories and techniques than

most p robation officers curren tly possess.

Those undertaking and managing wo rk with offe n d e rs need to

be come more re s p o n s i ve to the ne ed to evaluate the ir wo rk .

P ro grammes included in an evaluation nee d to make explicit the

viii



9/82

u n d e r lying the ory as to how the y exp ect to effect ch a n ge, the

con ten t and met hod of in ter vention , an d th e fra m ewo rk fo r

implementation.

The influence of cognitive approaches: a survey ofprobation programmes

In Fe b ru a ry 1996 the Offenders and Corrections Unit of the Home Offi c e

sent out a questionnaire to all probation areas asking for information about

the extent to which the programmes they operated, or had access to, made

use of cognitive skills training. Cognitive skills was defined for the purposes

of this survey as an app ro a ch which attempted to reduce re o ffend ing byteaching offenders to analyse and modify their thinking.

The results of the survey are:

Fo rt y - t h ree areas replieda response rate of 78 pe r cent. Of these

four did not run programmes with a cognitive skills component. The

remaining 39 areas ran 191 such programmes. Only half of these had

be en run ning for more than fi ve ye a rs; and a quarte r h ad b een

running for less than a year.

The decision to set up cognitive behavioural programmes was often

influenced by a combination of factors: 53 per cent of respondents

cited successful results in other probation areas; 44 per cent said the

approach had been championed by a member of staff; and a third of

programmesmainly those dealing with sex offenders or self control

programmeshad been influenced b y research from abroad.

Only 35 programmes in 17 areas were bought-in as opposed to being

developed locally. The fact that the intensity and duration of these(h igh ly struct u red) programmes we re so often altered suggests that

ins u fficient attention was being given to ensuring that program m es

were delivered as intended.

Gene ral programmes we re more popular than those which focused

on a particular form of offending. Among the latter, those for treating

sex offenders and for tackling poor driving behaviour were the most

common.

Programmes seemed to function without reference to risk and needs

principles. While most were prim a ri ly run for those atten ding as a

con dition of a cou rt ord e r, few ex cluded part icip ation by othe r

o ffe n d e rs attending on a vo l u n t a ry basis. Mixing offen d e rs in this

Summary

ix


10/82

way may be an efficient use of resources, but is unlikely to achieve

programmes which are well matched to levels of risk, and offending-

related needs.

Sex offender programmes tend to be (comparatively) well-organised,

run by well-trained staff, and able to draw on outside expertise, but

they too are rarely restricted to serious offenders.

P ro grammes sh ould eithe r be staffed by appro p ri a t e ly tra i n e d

p rob ation offi c e rs; or invo l ve spe cialists th rou gh p art n e rs h i p

a rra n gements. Yet 24 per cent of probation staff h ad re c e i ved no

extra training at all and 105 programmes were run without any other

agencies being involved.

The limited information provided on costs shows that specialist such

p ro grammes do not come cheap, yet areas have spent surpri s i n g ly

little time or effort on examining whether such programmes work. In

most areas it seems that they do not even know how many people

attend, who drops out and why, and who succeeds and why.

Perhaps the most encouraging aspect of our findings was that respondents

were far from complacent about the way programmes were operating. Many

of those responsible for running the programmes on a day-to-day basis would

cl e a r ly welcome more training, more advice and better evaluations. One

obvious improvement would be for a more strategic approach to setting up

and running programmes at senior levels, with pro grammes being p iloted

and evaluated before being delivered on a larger scale. Another might be to

create a standard set of evaluation measures which would ensure that similar

programmes were being compared fairly.

x



11/82

Par t I:

Th e use of cogn itive-beh avio ur al ap p ro a c h e s

with offen ders: m essages fr om th e research .

Julie Vennard, Darren Sugg and Carol Hedderman


12/82


13/82

The use of cognitive-behavioural approaches withoffenders: messages from the research

Introduction

Community and p rison based programmes designed to rehabilitate offenders

have been criticised over recent decades on two main grounds. First, that

t h ey over simplify the roo ts o f offend ing behaviou r and t ake t oo little

account of adverse social and economic circumstances. Second, that there

is a lack of re s e a rch evidence demonstrating the effe c t i veness of such

i n t e rventions. This so called nothing wo rks position has gi ven way in

recen t ye a rs to a cau tious opt imism, suppo rted by a growing body o f

re s e a rch evidenc e, th at some type s of inte rven tion can be effe c t i ve inreducing offending.

Among the range of offender programmes designed to reintegrate offenders

into society, those which are increasingly favoured, both in institutional and

c o m munity settings, seek to address deficiencies in offe n d e rs ways of

thinking, reasoning and associated beh aviour thro ugh what are t erm e d

c o g n i t i ve - b e h av i o u ral tec hnique s. The main aim of th is rev i ew is to

s u m m a rise th e ava i l able re s e a rch findings on th e effe c t i vene ss of such

i n t e rventions in reducing offending and ach i eving ben eficial attitude or

behavioural changes. This introd u cto ry section describes the context and

scope of the review. The next considers what is commonly understood by

the term cognitive behaviourism in the context of rehabilitative programmes

undertaken with offenders. The third section summarises the ways in which

re s e a rch ers have appro a ched the broad question what wo rks, since it is

within this body of research that any evidence of the impact of cognitive-

behavioural techniques is to be found. Consideration is then given to the

findings of the re search, fo l l owed by a brief discussion of some possibl e

implicatio ns of th e find ings for th e de sign and de live r y o f effe c t i ve

programmes with offenders. The concluding section summarises the broadm e s s ages and th e gap s in th e re s e a rch litera t u re w ith re g a rd to th e

e ffec t iveness of cognitive - b e h av i o u ral interventions and ident ifies areas in

which further research is needed.

D u r ing th e 1970 s and 1980s scep ticism re g a rding th e efficacy of

individualised treatment of offenders, which attributes criminal conduct to

the particular personality and circumstances of the individual, was heavily

influenced by a study undertaken in the United States by Martinson (1974).

Unde r pre s s u re to demon strate th at time spent in pr ison can ach i eve

reh ab i l itati ve as well as punitive aims, in 1966 the p rison service in New

Yo rk State commissioned Robert Martinson to undert a ke a rev i ew of the

available research . The central message to emerge from the review, which

was based upon 231 studies of re h ab il i t a t ive programmes carried out in a

The use of cognitive-behavioural approaches with offenders: messages from the research

1


14/82

variety of institutional and non-institutional settings published in the English

language over a 22 - year period, was that, with few and isolated exceptions,

the treatment studied had no significant impact on further offending.

In seeking to explain his findings Martinson postulated that there may be a

radical fl aw in our present stra t e giesthat education at its b est or that

psychotherapy at its best cannot overcome, or even appreciably reduce, the

p owerful tendency for offenders to continue in their criminal behavi o u r .

M a rt inson ente red a number of caveats about his findings, including th e

poor quality of many of the ori ginal stud ies, lack of comparability in the

measurement of reoffending and the heterogeneity of the groups studied. It

is also noteworthy that in his original arti cle Martinson ackn owle dged the

possibility that some treatment programmes a re working to some extent, butthat methodological weaknesses in the original studies prevent any pattern

being detected. These caveats tended, ho wever, to be overlooked by those

wh o saw in th e art i cle stron g emp irical supp ort for the rejection of a

treatment approach and for return to a just deserts perspective, in which

the principle of appropriate and proportionate punishment is paramount in

the sentencing decision.

Perhaps unsurpri s i n g ly, many pra c t i t i o n e rs and specialists wo rking with

o ffe n d e rs du ring th is p eriod con tinue d to believe that app ro p ri a t e ly

targetted interventions could bring about a reduction in offending. McGuire

and Priestley (1995) observe that, from the outset a number of researchers

( n o t ably Gend reau and Ross, 1979) ex p ressed re s e rvations abou t the

nothing works doctrine and were able to find some evidence of positive

outc omes from bet ter cont rolled studies, often the same typ es of

i n t e rven tion as t ho se e valuate d b y Martinson . Gendr eau (1995) h as

commented more recently that, despite the disaffection with rehabilitation

in the US, in Canada the rehabilitative agenda continued to flourish through

the 1970s and 1980s, academics and gove rnm ent officials continuing to

promote and research rehabilitative programmes.

The re s u rgence of confidence in Britain in the re h ab i l i t a t i ve po tent ial of

probation can be attributed in large part to the findings from Canadian and

North American research. Recent years have also witnessed an increasing

nu mbe r of c on fe ren ces in th is cou ntr y, atten ded by po licy-make rs ,

re s e a rch e rs and p ra c t i t i o n e rs, on the sub ject of re s e a rch and pra c t i c e

concerning effective interventions with offenders. At one such conference,

spons o red by the Home Office, Association of Chief Officers of Prob at ion

and the Central Probation Council, delegates met to consider how the results

of this emerging research might best be put into practice through improved

management of community supervision programmes.

Fo l l owing this Managing What Wo rks confe ren ce , Probation Circ u l a r

2



15/82

77/1995 re q u i red Chief Probation Offi c e rs to rev i ew their cur rent and

planned programmes against a list of critical success factors adapted from

research and from the conference proceedings. These principles have been

widely endorsed by a number of researchers who have reviewed some of the

bet ter designed studies of re h ab i l i t a t i ve pro grammes (e.g. Andrews et al.,

1990; Hollin, 1994; Lipsey, 1992; McGuire, 1995) and are now familiar to

probation of ficers in this country. They are described in the literature in

the following terms:

Risk classificationmore intensive programmes should be targetted at

high risk offenders, while those of lower risk should receive lower or

minimal intervention;

Ta rgetin g cri m i n o geni c n eedsit is imp ort an t to d iffe re n t i a t e

b e t ween cri m i n o genic nee ds (th ose fa c t o rs wh ich con tri b u t e

d i re c t ly to cr iminal be hav i o u r, such as ant i-soc ial attitude s, dru g

dependency, limited cognitive skills) and needs which have no direct

relationship w ith p ropen sity to re o ffend. If the pu rpose of the

p ro gramme is to reduc e offen ding th e focus should be up on

criminogenic needs

R e s p o n s i v i t y m a t ch the learning styles of the offe n d e r s to the

t e a ching styles of those providing the pro grammes. The learn i n g

styles of most offe n d e rs re q u i re active, part i c i p a t o ry app ro a ch e s

rather than traditional client-centred counselling

Treatment moda litylinked with responsiveness, this principle states

that app ro p riate types of intervention are typically skills ori e n t e d

(designed to improve pro blem-solving, social interaction and other

types of coping skills) and draw upon cognitive - b e h av i o u ral and

social learning methods

C o m mu ni ty ba se ge n e ra l ly com munity-based pro grammes have

s h ow n more p ositive results, bu t p ro gramme s w hich take into

account risk, criminogenic need and responsiveness can work in any

treatment setting

P ro gra mm e integri t y p ro gramme s should be pro p e r ly manage d

andcarried out by trained staff in accordance with clearly ident ified

aims and objectives that do not change over time.

In promoting these principles throughout the probation service in the form

of a Home Office circ u l a r, it is h op ed th at the y will be translated into

e ffe c t i ve local p ractice w he n assessing risk and nee d and allocat ing

offenders to suitable programmes. The Prison Service is similarly seeking to


3


16/82


17/82

h o w h a s t h e im p ac t o f t h is a p p r o a c h , a n d o t h e r s u se d in

interventions with offenders been evaluated

wh at evidence exists of the ben efits of cognitive - b e h av i o u ra l

approaches in reducing further offending

what is the available evidence on the types of approa ch which are

most effective with different groups of offenders

wh at lessons can be drawn from the li tera t u re re g a rding the

re q u i rements for successful delive ry and manageme nt of cognitive -

behavioural programmes.

It should be said at the outset that it is not an intention of this paper to

p rovide a detailed rev i ew of the full ra n ge of individual stud ies (most of

which are Canadian or North American) which have examined the impact of

c o g n i t i ve - b e h av i o u ral pro grammes. Most of th e published stud ies have

a l re a dy bee n examined and ge n e ra l ly, the more rob ust one s have bee n

synthesised in traditional or meta-analytic literature reviews reported over

the past decade. This paper draws pri m a ri ly on the most ex t e n s i ve and

more recent of these reviews in seeking to answer the above questions.

Cognitive-behaviourism: theory and techniques

C o g n i t i ve behav i o u rism is not a un ified, distinct psych o l o gical th eor y or

meth od bu t a term gi ven to a ra n ge of interventions deri ved from the

following three psychological theories:

b e h av i o u ri s m , which stresses the role of ex t e rnal, env i ro n m e n t a l

fac t o rs in shaping an individuals actions so that , for offe n d e rs, fo r

example, encouragement from peers and/or the lack of immediatepunishmen t from auth ority fi g u res re i n fo rces criminal beh av i o u r

(Blackburn, 1995)

cognitive theory, in which the importance of an individuals thought

p roc esses, such as reasonin g, me mor y and p ro blem solving are

s tressed. As Hollin notes, cognition is implicit in the theories of

w ri t e r s wh o suggest th a t va rio u s styles of th in kin g, su ch a s

i m p u l s i ve a n d con cret e, a re cha ra c t e r ist ic o f offe n d e r

populations(Ausubel, 1958; Glueck and Glueck, 1950; Hollin, 1990,

p 10)

social learning theory, w hich is de ri ved from beh av i o u rism but

states that, while the environment remains a key factor, learning may


5


18/82

also occ ur indire c t ly, th rou gh obse rvation of o ut com es of the

behaviour of others. The impact of the environment on behaviour is

thus mediated by cognition (Hollin, 1990).

In a cour se manual on cognitive -behavioural theory and research McGuire

(1996) describes the way in which the ideas from behavioural and cognitive

t raditions within psych o l o gy and psych o t h e ra py we re integrated, in the

1970s, into a new app ro a ch to un der stand ing the complex dy n a m i c

relationships between thoughts, feelings and behaviour. As an approach to

wo rking with offe n d e rs, cognitive - b e h av i o u ral modification assumes that

offenders are shaped by their environment and have failed to acquire certain

cognitive skills or have learnt inapp ropriate ways of behaving. Thinking may

be impulsive and egocentric, concrete rather than abstract and rigid ratherth en fl ex i ble. McGuire stre sses, howeve r, tha t a cognitive - b e h av i o u ra l

a p p ro a ch does no t attr ibute the causes of criminal be haviour solely to

individual or psych o l o gical fa c t o rs. It also takes into account the social

conditions which affect individual development and is not in conflict with

sociological explanations of criminal activity, such as those which view such

behaviour as acquired from influential delinquent peer groups.

Since it is considered that such cognitive deficits are learned rather than

i n h e rent, offender programmes based on this conceptualisation of human

b e h aviour are intended to teach offe n d e rs to face up to what they have

done, to understand their motives and to develop new coping strategies and

ways of con trolling their behaviour. This can be accomplished using a range

of be hav i o u ral and cognitive tec hnique s. B e h av i o u r a l m odi fi c a t i o n

typically seek to reinforce socially acceptable behaviours th rough techniques

s u ch as con tingency con tracts, according to w hich offe n d e rs agree to

b e h ave in cer tain ways an d are rewa rded for doin g so. Rew a rds fo r

achieving specific targets may take the form of approval, desired activities or

token reinforcers which provide access to such activities. Self-reinforcement

techniques may also be used, whereby offenders monitor and evaluate theirown behaviour (Hollin, 1993). Behaviour therapy is also based upon the

theory of conditioned learning but focuses on reducing unwanted patterns

of t hought and beh aviour through such tech niques as relaxation tra i n i n g

(reduction in levels of arousal), training and assertiveness training designed

to rep lace anxiety and similar emotions with positive cop ing re s p o n s e s

(McGuire, 1996). The distinctive characteristic ofcognitive therapy, which

M c G u i re descr ibe s as the most cogn itive ly or iente d fo rm of co gnitive -

b e h av i o u ral the ra py, is that it seeks to ident ify and modify dy s f u n c t i o n a l ,

distorted thoughts, replacing them with more adaptive thinking patterns.

Cognitive-behavioural interventions wh ich seek to impr ove self control draw

on behavi ou ral techniques such as self-monitoring, self-re i nforcement and

self-instructional training (SET), in which covert inner speech is used to

6



19/82

control overt behaviour. McGuire (1996) describes self-instructional training

as a standard feature of the cognitive-behavioural approach, though he notes

that it is widely used in comb ination with o ther types of method. As an

a p p ro a ch to self regulation, SET may, in fact , be viewed as more cl o s e ly

linked to cognitive learning than to behavioural conditioning and illustrates

the difficulty in categorising the ran ge of techniques e ncomp assed b y the

cognitive-behavioural label.

Also subsumed w ithin the cognitive-behavioural framework are certain other

related and overlapp ing app ro a che s. Socia l skills tra i n i n g , a fo rm of

b e h av i o u ral thera p y, is sometimes used in co njunct ion w ith cognitive

t e chn ique s as a met ho d of inc reasing skills for de aling w ith social

e n count ers. This technique is commonly employed w ith young offe n dersand sex offenders whose offending is thought to derive, at least in part, from

their lack of self-esteem and inability to deal with other adults appropriately

or effe c t i ve ly. The objective may be to develop micro skills such as eye

contact and body posture, or macro skills such as communication with

parents and resisting pressure from delinquent peers. The methods adopted

in social skills p ro grammes include instruction con cern ing standards o f

behaviour and expectations, modelling (presentation of a coping response in

a gi ven so cial situat ion ), ro le p lay, re h e a rsal, fe e d b a ck and so cial

reinforcement. Observing that some category boundaries are still somewhat

fluid, Palmer (1994) notes that social skills training is classified variously as

c o g n i t i ve skills, cognitive - b e h av i o u ral, social cogn itive or skills

oriented, the latter sometimes being used as a generic term to encompass

life skills ( a wide range of practical skills such a literacy and numeracy) as

well as soc ial skills. Skills training pro grammes may be blende d with

p ro blem-solving tra i n i n g , w h i ch seeks to improve pro blem awa re n e s s ,

perspective taking and the ability to foresee the likely consequences of ones

actions. Here again, the skills training techniques of instruction, modelling,

discussion and feedback are employed.

As Hollin (1996) and others have observed, p rogrammes for young offenders

frequently incorporate a variety of methods, including cognitive-behavioural

a p p ro a che s, into a mu l t i m o d a l p ro gramme. Key element s are, t ypically

p ro blem solving training, social skills training, anger co ntrol and mora l

education. Although multimodal programmes are increa singly rega rded as

offering the best chances of successful reduction in reoffending, since they

a re able to address a wide ra n ge of pro blems and needs, difficulties ari s e

when seeking to identify which components are and are not effective in any

gi ven co mb ination. As will b e seen , this u nce rt aint y has cre a t e d

fundamental difficulties for those seeking to discriminate betw een va ri o u s

a p p ro a ches and to dete rmine wh at wo rks be st w ith diffe ren t types of

offenders.


7


20/82


21/82

Oth er difficulties include the fa i l u re of re c o nviction studie s to measure

changes in the severity of offending and inter vals between offences; and the

imp act th at p olice and prosecution practices have on wh eth er furt h e r

offences result in criminal proceedings.

H aving themselves carr ied ou t one of the largest and most soph isticated

British reconviction studies, Lloyd, Mair and Hough caution against uncritical

reliance on them as a measure of effectiveness and suggest that in studying

the impact of diffe rent sentence s or pro grammes the scope for dev i s i n g

additional measures should always be considered. For example, in the case

of sex offender treatment programmes offenders can be scored on a range of

psychometric tests before and after therapy designed to measure changes in

attitudes believed to contr ibute towa rds re o ffending (see, for ex a m p l eB e cket t et al., 1994). It is ge n e ra l ly acce pted , howeve r, that eva l u a t i o n

studies must seek to demonstrate that any such changes hold up over time

and do, in fact, translate into reduced offending rates. Hence, the evaluation

of sex offender treatment programmes undertaken by Beckett et al., allowed

for the calculation of rec onviction rates two yea rs after comp letion of the

programme (Hedderman and Sugg, 1996).

There is some evidence to suggest that the methodological rigour of studies

of wo rk with offende r group s has improved in recen t ye a rs. Some large

scale p ro grammes, suc h as the Sex Offender Treatm en t Eva l u a t i o n

Programme (SOTEP) in California, have used randomised controls, extensive

fo l l ow up p eriods and comp lex statistical pro c e d u res (Marques, 1994).

F u rt h e rm o re, evaluations are be ginning to study how pro grammes are

d e l i ve red (as opp osed to how the y we re designed to op erate), and what

consequences this has for the outcome as, for example, in a study by Mair et

al., (1994) of int en sive p rob ation in England and Wales. I t is also

i n c re a s i n g ly th e case th at evalu ation stud ies includ e ot he r out co me

measures, such as educational achievement, attitude or behavioural change.

Few ind ividu al stu dies h ave, h ow eve r, p rovide d a suffi c i e n t ly cl e a runderstanding of the causal links between intermediate outcome measures

and reconviction to enable replication of the conditions in which successful

outcomes have been o bserved. It was w ith a view to synthe sising the

findings from better designed studies, there by increasing understanding of

the ingredients of successful interventions with offenders, that a number of

researchers turned to the technique of meta-analysis.

Put simply, this technique involves reducing the characteristics of individual

studies into a number of summary statistics, such as number of offenders,

the type of treatment given, and recidivism rates (however measured) after

treatment. The summary statistics are then analysed to p roduce an overall

e ffect size statistic w h ich re p resen ts th e a m o u n t of diffe re nc e in

recidivism that ex ists bet ween the inter vention p ro gramme and the


9


22/82

respective control programme. One of the main advantages claimed for this

t e chn ique is that the results from small scale studies, w hich may be of

l imited use w he n co nsidered in isolation, b ecome inf luent ial wh en

combined with the results of similar programmes.

There are, however, a number of problems associated with meta-analysis and

w ith th e w ay in wh ich it has been used t o ag gregate th e find ings of

p ro grammes in th e offender corrections field. Without examining these

p ro blems in detail it is impor tant to be awa re of th e limitations of meta-

analysis as a method of determining what types of inter vention are effective

in reducing reoffending.

In their critique of meta-analysis Mair and Copas (1996) distinguish betweenthe use of this technique in order to review research literature systematically

for example, to see if certain programme characteristics are associated with

h igher effect sizesand using it to draw broade r infe renc es from t he

c om bin e d res u lts ab ou t t he re l a t i ve effe c t i ven e ss of re h ab i l i t a t i ve

programmes. They conclude that the former, a purely descriptive technique,

is relatively uncontroversial and useful. The latter, which they term Level 11

analysis, is more problematic for several reasons. The main difficulty arises

from the use of formal statistical procedures to calculate mean effect sizes

for subsets of the studies, grouping, for example, according to design

ch a ra c t e ristics or typ e of p ro gramme . Sinc e effect sizes are art i fi c i a l

c o n s t ructs they may fail to take accou nt of impor tant diffe rences in the

design of the studies and the offenders sampled. Sample subjects are not, as

in con trolled tr ials, selected ra n d o m ly from a cl e a r ly defined po pu lation .

R a t h e r, the collect ion of p ri m a ry studies comprises the sample and it is

assumed that the totality of offen d e rs across the studies is re p rese n tat i ve .

S i m i l a r ly, Level 11 analysis assumes th at the int er vent ions are b ro a d ly

comparable and are delivered in equivalent conditions. Since both types of

assump tion are unsafe, according to Mair and Copas it fo ll ows that meta-

analysis cannot predict what would happen if a particular intervention wereto be tra n s fe rred elsew h e re or delive red more w idely. Others (such as

Fa rrington, 1996) have observed that the way in w hich meta-analysis has

been used to ag gregate the results of pro grammes with offe n d e rs glosses

over the disparate nature of much of this work, over the different research

designs e mployed in the ori ginal studies, and over t he va rious succe ss

criteria used.

In fact, the more recent large-scale meta-analyses, such as that car ried out b y

Lipsey (1992), have included an examination of the relationship betw een the

study design, sample size and characteristics, and the effect size distribution.

This enabled Lipsey to conclude that effect size does depend substantially

up on th e met ho dological ch a ra c t e ristics o f the study, bu t it is also

i m p o rt a n t ly influenced by the nature and circumstances of the tre a tm e n t

10



23/82

under study. Lipsey and others concede, however, that treatment modality

and the organisational stru c t u res and settings in which pro grammes are

delivered, are often described rather poorly in the original studies and are

c o rre s p o n d i n g ly difficult to code in a meta-analysis. Consequent ly, the

results are unlikely to be capable of replication elsewhere.

Other difficulties associated with meta-analysis arise from shortcomings in

the ori ginal pro gramme evaluations and although some re s e a rch e rs have

been highly selective (for example, on ly 46 of the 317 p rogrammes Izzo and

Ross (1990) iden tified met their evaluation cri te ria) others may have been

less rigorous in their selection p rocess. These difficulties exist even w hen

meta-analysis is used simply to describe differences and similarities between

studies in order to see what kinds of treatment have been tested with whatresults.

A particular problem in evaluating cognitive-behavioural programmes arises

from the absence of consensus over what this approach encompasses, and

whether the types of intervention unde rt a ken in the ori ginal studies can

l egit imat ely be entitled cognitive-b e haviou ral (Palme r, 1994). Progra m me s

which adopt this approach are often combined with other components and

have tended to be defined somewhat differen tly by meta-analysts. As Palmer

o b s e rves, th is gi ves rise to p ro blems of inter p retation . For ex a m p l e ,

Panizzon et al., (1991) found little evidence of success with what they

t e rmed social cognitive inte rvent ions but impre s s i ve results for ro l e

playing/modelling, wh ich others (e.g. Izzo and Ross, 1990) have classified as

a form of social cognitive intervention.

When evaluating the results of individual studies meta-analysts and reviewers

ge n e ra l ly con sider that diffe rence s betwee n ex p e rimental and con tro l

groups which achi eve the pro bability level p


24/82

uncertain because of the paucity of studies of interventions which are both

intensive and suitably tar getted on offence-related behaviour. Furthermore,

inadequate matching between the ex pe rimental and the control group on

factors pertaining to risk of reoffending will obviously reduce the chances of

a c c u ra t e ly assessing the impact of a pro gramme. This is thought to have

o c c urred in some of the studies included in meta-analyses, although once

again the primary studies have, in fact, tended to say little about methods of

selection or suitable participants.

Meta-analyses may also overlook the possibility that work undertaken with

comparison groups has been reasonably effective, thus reducing the ability

of the re s e a rch to gauge the impact of the ex p e rimental interve n t i o n .

Indee d, as Palmer (1994) p oints o ut, s tandard pro grammes ru n bycompetent staff, may make a considerable contribution to the goal of

reducing reoffending and will thus reduce the difference in the recidivism

rates betw een the exp erimental and the control group. This is confirmed by

Lipsey (1992) who found that control groups that received some attention

(which is common practice in juvenile justice settings) showed less contrast

with treatment groups than where control groups received no treatment of

any kind.

Finally, critics of meta-analysis have warned that it is unsafe to assume that

the reports of programmes included in the analysis are representative of all

similar pro grammes. This is because stud ies yielding positive resu lts are

more likely to be published and hence available to meta-analysts. Some meta-

analysts (Lipse y, 1992; Redondo et al., 1996) have sought to overcome this

p ro blem by bro adening the c ove rage of the litera t u re to include bo th

published and unpublished repo rts. Redondo found, moreover, that there

were no significant differences between the unpublished and the published

studies in terms of the effectiveness of the programmes included. It must be

acknowledged, moreover, that publication bias is also a feature of orthodox

literature reviews.

Given the limitations in the primary studies on which meta-analytic reviews

are based and the above qualifications regarding the use of this technique as

a means of quant ifying differences between offender programmes, what can

be gleaned from the research literature about the effectiveness of cognitive-

b e h av i o u ral technique s? If there is evidence to suggest th is app ro a ch is

significantly more effective than others, what, if anything, does the literature

reveal about the main forms of successful cognitive-behavioural intervention

with different types of offe n der, such as those convicted of sex or violent

o ffences or invo l ved in drug abuse? This sect ion considers the ava i l abl e

evidence with a view to clarifying what conclusions it is safe to draw and

which claims should be treated with a degree of caution.

12



25/82

The use and effectiveness of cognitive-behaviouralapproaches with offenders

Since the most extensive and recent meta-analyses have focused wholly or

p ri m a ri ly on juveniles and youn g o ffe nd er p ro grammes, th e fi n d i n g s

reported in this ove rv i ew regarding interventions with adult offenders are

limited. As will also be apparent, the majority of the original studies were

u n d e rt a ken in North Amer ica in the 1970s and 1980s. Unsurpri s i n g ly,

t h e re fo re, some commen tator s have ex p ressed conside rable scep ticism

about the app licability of the results and conclusions to adult offe n d e rs

sentenced in this countr y (Mair, 1995; Losel, 1993). In fact, Losel et al.

(198 7), Maye r et al. (1 986) and Lipse y (199 2) fou nd no signifi c a n t

relationship between a ge and treatment effect in their meta-analyses. Since,how ever, the age range for most studies included ended at 21, this will have

effectively excluded many offenders with serious drug and alcohol problems

w h i ch have develop ed over time or w ith long-standing r elationship or

employment difficulties. These kinds of personal factors can be exp ected to

h ave a bearing on p ro gramme succ ess irre s p e c t i ve of th e p ro gra m m e

content and approach. Despite this, in one of the largest meta-analyses to

have included juvenile and adult offender programmes Andrews et al. (1990)

found no significant difference in their effectiveness according to age. And,

as will be seen, certain of the factors identified in meta-analytic reviews as

being associated with successful outc omes h ave been confi rmed by the

results of individual studies with adults (e.g. Ross et al., 1988) and by the

work with sex offenders.

The first part of this section summarises the findings of individual studies or

meta-an a lyses which are not focused up on any particular type of offe nc e .

The age catego ry or catego ries encompassed are mentioned wh ere this is

k n own, togeth er w ith any re c o rded diffe ren ces in th e effe c t i veness of

cognitive-behavioural approaches according to the age of the offenders. The

following sub-sections consider the use and impact of cognitive-behaviouralinterventions w ith alcohol and drug abusing, violent and sexual offenders.

Juvenile and adult offender s

Individual studies have produced conflicting results about which forms of

behavioural and cognitive-behavioural work are most effective with juveniles

and young offe n d e rs. Hollin (1990, 1993, 1996) and McGuire (1996a)

rev i ewed a wide ran ge of studies and concluded that interventions based

s o l e ly on social skills t raining (e.g. Klein et al., 1977) have had limited

success while those based solely on individual beh aviour th era p y have

sometimes bee n succe ssful b ut are unp op ular w ith young offe n d e rs .

Approaches which combine behavioural techniques with social learning and

cognition, including teaching young offenders self-control, appear to be the


13


26/82

most promising in reducing reoffending. For example, Feindler and Ecton

(1986) developed a pro gramme for young offe n d e rs which successfully

taught them how to control anger, while McDougall et al. (1987) reported

significant anger reductions among inmates in young offenders institutions

who had under gone an anger management p rogramme combining cognitive

p ro c e d u res, specifi c a l ly self-inst ruc tion and self-statem en ts to use in

situations of anger, with role playing to improve behavioural responses and

an edu cational module to he lp offe n d e rs under stand the ir ange r. The

sample size in the McDougall study was small (n=18) and the measurement

of effectiveness was limited to a comparison between treatment and controls

in the use of Gove rn o r s re p o rts th ree mon th s after comp letion of the

programme. Lochman (1992), moreover, assessed a similar programme in a

school and found that, although self-esteem and social problem solving skillswere increased, there w as no long-term impact on recidivism.

A study by Bordu in et al. (1995) examined the long-term effe c t i veness of

multi systemic th era p y (MST), an appro a ch w hich dire c t ly add re s s e s

p e rson al (including cognitive) and social (fa m i ly, peer group and sch o o l )

fa c t o rs th at are be lieved to be associated with adolescen t antisocial

behaviour. A sample of 176 families with a juvenile offender (aged 1217)

were ra n dom ly assigned to MST and convent ional individual t hera py (IT).

All of the offenders had records of serious criminal involvement. Following

therapy the two groups were compared on a number of measures related to

th e goals of improving individual adjustme nt and fa m i ly relations and

reducing involvement in criminal activity. On all outcome measures the MST

group showed significant impr ovements which were not reported in respect

of the control group . At the end of the four year follow-up the overall arrest

rate for MST completers (22%) was less than one - third the rate for the

individual the ra py group (71%). Those w ho droppe d out of MST were at

higher risk of arrest (47%) than those who completed but were still at lower

r isk th an th ose wh o un derw en t IT. MST was also associated w ith

significantly fewer serious cr imes among those w ho were arrested.

B o u rdin et al. c on sider th at th e succ ess of MST w ith h igh risk yo u n g

o ffe n d e rs may be due to its recognition of the multiple deter minants of

ant isoc ial b eh aviour and its exp licit focu s on ameliorating be hav i o u ra l

problems within the context of problematic family relations.

Turning to the findings from meta-analyses of programmes which have been

u n d e rt a ken with juveniles and young offe n d e rs, Losel (1993 , 1995) has

estim ate d an ove rall effec t size o f 0.20 for all th e me ta-analyses he

considered, which means that recidivism for treatment groups was 10 per

cen t less than for the con trol groups. Pro grammes that yielded th e best

results we re c o g n i t i ve - b e h av i o u ral, skills or ienta ted a n d mu l t i m o d a l .

Similarly, Palmer (1994) found that approaches classified as behavioural were

14



27/82

among the most successful with juveniles (i.e., had the largest average effect

sizes or reductions in recidivism when comparing experimental and control

groups). Programmes that used cognitive- b e h av i o u ral, or combinations of

cognitive and life skills approaches also sho wed successthough often under

other labels such as social skills trainingas did those simply classified as

multimodal. Palmer re p o rts that the least effe c t i ve app ro a ches are those

w h i ch employ con frontation, group counselling/thera p y (unless care f u l ly

focused), and individual counselling/therapy.

L i p s ey s rev i ew of published and unpublishe d re s e a rch concerning 443

programmes for juvenile offenders has been influential because it is one of

the most systematic and rigorous to date (Lipsey 1992, 1995; Palmer, 1994).

Lipsey s sample included offenders in the 1221 age range, although mostwere aged 18. For 285 (64%) of the programmes recidivism was lower for

the treatment groups. Even more import ant was the finding that in 131

(30%) studies treatment programmes actually increased delinquency. These

we re pun ishment or de terrence based interventions, such as shock

i n c arce ration and boot camps, which provided little by way of education,

t rain ing or thera py. Having contro lled for fa c t o rs such as the diffe re n t

measures used by each study to judge rec idivism (e.g. arrest, conviction etc .)

and the type of research design (randomised or non-randomised treatment

and control groups), Lipsey concluded that programmes which were multi

modal and had a more concrete, behavi o u ral or skills-orien ted characte r

had the most impact both within and outside the juvenile justice system

with effect sizes of 0.200.32 (equivalent to a 1016% reduction in

recidivism against untreated controls). Lipsey did not attemp t to disentangle

the ingredients of successful multi modal packages and cautioned that the

inherent fuzziness of these coded categories makes futile any discussion of

whether particular forms of intervention are universally superior to others,

( L i p s ey, in McGuire, 1995 ed). Like Pa l m e r, howeve r, Lipsey found th at

concrete, structured approaches focused directly on overt behaviour, were

more successful than were traditional counselling and casework techniques.

An earlier meta-analysis by Whitehead and Lab (1989) is notable for the fact

that the authors reached more pessimistic conclusions about the success of

rehabilitative work with juveniles than in subsequent meta-analyses. On the

basis of an analysis of 50 programmes for juvenile offenders Whitehead and

Lab found that only 2432 per cent of programmes were successful in terms

of achieving significant reductions in recidivism. They concluded that no

single type of inter vention displayed ove r w h e l m i n g ly positive results on

recidivism. It should be noted, however, that of the 50 studies in the meta-

a n alysis, 30 involved juvenile dive rsion, which may not have included any

fo rm of intervention intended to address offending beh av i o u r. Prev i o u s

studies have indicated that diversionary approaches, which tend not to be

used with re l a t i ve ly serious offe n d e rs, produce smaller effect sizes than


15


28/82

other forms of intervention, perhaps because the offenders are judged to be

fairly good risks in the first p lace.

This finding is generally considered to be a consequence of the researchers

use of very strict criteria for success that would require large reductions in

rec idivism, no t simp ly a stat istically significant diffe ren ce b etw e e n

ex perimental and control groups. The equivalent effect sizes we re larger

th an th ose no te d in oth er met a-analyses (Pa l m e r, 1994; Losel, 1995 ;

Farrington, 1996). When Andrews and colleagues (1990) re-analysed 45 of

the 50 studies considered by Whitehead and Lab and included a further 35

studies (a third of which cove red pro grammes for adult offe n d e rs) their

c o n clusions we re more in line with those re a che d by Losel, Lipsey and

others. The authors speculate that effect sizes may have risen th roughout the1980s precisely because of the increasing use of cognitive methods within

behavioural programmes.

D rawing on the findings of previous studies, Andrews et al. diffe rentiat ed

b e t ween app ro p riate and inappro p riate services. Ap p ro priate serv i c e s

were defined as those which delivered more intensive treatment to higher

risk offenders; targetted the criminogenic needs associated with offending;

and used cognitive or behavioural approaches. Inappropriate services were

those which delivered treatment to low risk offenders, or failed to match the

t e a ch ing styles of staff to offe n d e r s learn ing styles; or used gro u p

a p p ro a ch es wit h no spe cific aims (Pa l m e r, 199 4). For ap p ro p ri a t e

p ro grammes recidivism rates were on average 53 per cent lower than fo r

inapp ropriate services. Andrews et al., like Lipsey, also reported increased

recidivism rates where the intervention was intended to punish or deter

rather than to rehabilitate through appropriate service.

Although Whitehead and Lab we re cr itical of Andrews et al. analysis and

c o n clusions re g a rding app ro p riate cor rec tional app ro a che s (Lab and

Whitehead, 1990), in a rejoinder Andrews et al. (1990a) re-asserted that theirempirical findings were robust and reliable. They conceded, however, that

Whitehead and Lab were correct in noting the need for further research to

support the theory that the connection between treatment and recidivism is

mediated by change in criminogenic factors. In other words, there is a need

for more research ...on the links among treatment, intermediate change

and recidivism.

Izzo and Rosss (1990) analysis of 46 juvenile offender studies conducted

d u ring the pe riod 19701985 also indicated th at pro grammes wh ich

i n c o r p o rate a cognitive compone nt we re more than tw ice as effe c t i ve as

those which did not, again, using effect size as a measure of the difference

b e t ween ex p e rimental and control groups. Izzo and Ross described the

individual cognitive pro grammes as using one or more of the fo l l ow i n g :

16



29/82


30/82

of offender rehabilitation programmes. The research indicated that effective

p ro grammes included an intervention technique w hich could infl u e n c e

cognitive deficits whereas ineffect ive programmes did not. This suggested

to the re s e a rch e rs th at cognitive skills training is a ke y e lement in any

successful programme. The programme is multimodal in that it is designed

to mod ify many aspec ts of offe n d e rs th inking, includ ing ego c e n t ri c i t y,

impulsiveness, and failure to understand the views and feelings of others.2

M o re details can be found in the Reasoning and Rehabilitation handbook

(Ross et al. 1989).

In 1988 the R & R pro gramme, wh ich consists of 80 hour s of inten sive

t raining delivered to groups of foursix offenders by five specially tra i ned

p robation offi c e rs, was fi rst piloted in Ontario, Canada. Offe n d e rs we rera n d o m ly assigned to a con trol group of regular pro b a t i o n e rs (n=23); a

group of regular probationers given additional life skills training (n=17); and

to the Reasoning and Rehabilitation group (n=22). The average age of the

o ffe n d e rs was 24. All membe rs of the sample had similar levels of ri s k

(calcu lated th rough use of the Leve l of Sup ervision Inve n t o ry or LSI),

although the R & R group had slightly more previous convictions. The nine-

month reconviction results were very impressive as only 18 per cent of the R

& R group we re re c o nvicted com pared with 69 pe r ce nt of re g u l a r

probationers and 47 per cent of the life skills group.

Although the results are favourable, the sample sizes for each group are small

and the follow up period (9 months) is short. Nonetheless, the programme

has been widely adopted across North America and Spain and is on the core

curriculum for the Correctional Services of Canada (CSC) which includes 47

prisons and community residencies across Canada (Ross & Ross ed., 1996).

By late 1993 programmes based on Reasoning and Rehabilitation had been

d eveloped by 13 prob ation areas in England and Wales. In most are a s

evaluation has only been carried out via client feedback (McGuire, 1995a),the excep tion being the pro gramme imp lemented in 1991 by th e Mid-

G l a m o rgan p robat ion service en titled STOP - S t ra ight Thin kin g On

P ro b a t i o n . STOP comprises 35 intensive sessions based on the R & R

programme. To date, over 150 offenders have received STOP orders from

courts (Raynor & Vanstone 1996, Knott 1995) and their performance on the

project has been subjected to independent evaluation.

The research design consisted of comparing the patterns of reconviction for

a group of offenders attending STOP with that of comparison groups subject

18


2 The full list, from Ross et al. (1988), is: structured learning theory (to teach social skills); lateral thinking (to teach

creative problem solving); critical thinking (to teach logical, rational thinking); values education (to teach values,

concern for others); assertiveness training (to teach non-aggressive ways to meet ends); negotiation skills training

(to teach alternatives to violent behavio urs in interpersonal conflict situations); interp ersonal cognitive pro bl e m

solving (to teach thinking skills requ i red to deal with inter personal pro blems and conf licts); social pers p e ct ive

t raining (to reco gnise and und erstand ot her peo ples views an d feelings); and ro l e - p l aying and mod elling (to

demonstrate socially acceptable behaviours).


31/82

to : s tandard p robat ion ord e r s ; pro bat ion ord e r s wi th day cent re

re q u i rements; community service ord e rs; immediate custody of up to 12

mon ths; and susp ende d sent enc es. For each group p redicted risk of

reconviction scores were also calculated. These scores enabled Raynor and

Vanstone (1994, 1996) to assess how each group fared compared with their

expected reconviction scores.

After 12 months the re conviction rate for those who completed the STOP

programme was better than predicted (35% compared with 42% predicted).

This pattern was not evident for the comparison groups. For example, 49

per cent of the custodial group were reconvicted compared with a predicted

rate of 42 per cent. However, the results were not sustained in the second

year wh en 63 per cen t of STOP c omp leters we re re c o nvicted (against apredicted rate of 61%) (Raynor and Vanstone, 1996).

Raynor and Vanstone also noted that after 12 months only five (8%) of the 59

STOP completers has been reconvicted of a serious offence (violent/sexual

offences or burglary) compared with 34 (21%) of those given custody and 19

(18%) of tho se sent ence d to STOP but wh o did not co mplet e the

programme. Again, these differences were not sustained in the second year:

after 24 month s the perc e n t ages re c o nvicted for serious offences for the

STOP full sample, custody and STOP completers were 27 per cent, 25 per

cent and 22 per cen t respectively. However, the STOP completers were still

more likely to avoid prison when reconvicted, which Raynor and Vanstone

believe may reflect a lesser degree of seriousness than simple offence codes

indicate. Raynor and Vanstone consider that better results, both in terms of

l o n ger term outc ome and comp letions, could be ach i eved by more

a p p ro p riate offen der selectio n, by offe ring sup p ort and fo l l ow -up to

offenders who comp lete the programme and by app lying the reasoning and

rehabilitation model outside the actual group sessions.

A recent British study of fi ve prob ation run group pro grammes for adultoffenders investigated their short-term effectiveness in achieving their stated

aims and attempted to identify the links betw een process characteristics, e.g.

integrity, coherent aims etc., and e ventual outcome, as measured by pre and

post treatment assessment (McGuire et al., 1995). The fi ve pro gra m m e s

considered were: a seven - week programme for drink driving offenders; a

structured activity programme for motoring offenders; a day centre activity

programme of the type used by many probation services; a cognitive skills

t raining pack age based on the Reasoning and Rehabilitation progra m m e ;

and a group for women offe n d e rs. The drink driving p ro gramme, the

cognitive skills programme and the programme for women offenders all used

cognitive-behavioural methods to some degree.

The short timescale of the study did not permit examination of reconviction


19


32/82

rates fo l l owin g com pletion of the p ro gramme bu t a num ber of othe r

measures were used including a range of pre/post treatment psychological

tests, the cr ime-pics risk of re o ffend ing questionnaire and the Rosenbury

self-esteem scale. There were other tests specific to each p rogramme (e.g.

alcohol know l e d ge tests, driving b eh aviour etc. ). The cogn itive skills

programme was assessed via six additional tests which measured alternative

thinking, impulsiveness and locus of control (the degree to which offenders

p e rc e i ve events to be out of their con trol). Comp arison groups we re

selected among offe n d e rs who had been recommended for placement on

th e p ro gramme s bu t w ho re c e i ved oth er disposals (m ostly c ustod ial

sentences or community service orders).

The resu lts of th is p ro gramme we re mixe d. Only th e drin k dri v i n gprogramme achieved statistically significant pre and post test differences on

the va rious te sts, while no significant diffe renc es we re det ected in t he

compa rison group . Other programmes, including the course on cognitive

training, showed promising, but non-significant results. The authors believe

that the failure to demonstrate statistically significant improvement could be

a function of the small number of offenders (a total of 65 offenders across

the fi ve pro grammes) and the dearth of standardised tests for eva l u a t i n g

these type s of intervention. In addition, some of the pro grammes lacke d

on e or more of th e fe a t u res desc ribed ab ove wh ich meta-analysts have

i d e n t i fied as distinguishing th e more successful offender pro gra m m e s .

Whereas the drink driving programme incorporated all the success factors,

the cognitive training programme lacked specific targetting and programme

i n t e gr i ty (or at least the se re q u i re ment s w e re no t eviden t in the

q u e s t i o n n a i re responses). The women offe n d e rs group used cognitive -

b e h av i o u ral met hod s and had a d ire c t i ve style o f wo rking bu t lacke d

p ro gramme integri t y, had no clear aims or spec ific treatment targets, and

was not highly stru c t u red . The day cen tre pro gramme had a dire c t i ve

approach but lacked any of the other success criter ia.

Despite the inconclusive findings from this study, the researchers consider

that the y lend supp ort to the con clusions of large scale meta-analy s e s

re g a rding the com bination of ingredients which yield the most pro m i si ng

results with high, or relatively high risk offenders.

To summarise this subsection, the avai lable re se arch on juvenile and adult

o ffend er p ro grammes p oints t o a b road c onsensu s as to th e types of

a p p ro a ch w hich ach i eve the greate st imp act on offend ing beh av i o u r

(expressed in terms of experimental studies achieving lower recidivism rates

th an contro ls). Those which comb ine cognitive - b e h av i o u ral tech n i q u e s

with the other success fa c t o rs ident ified in the meta-analysis (targe t t i n g ,

structured approaches, programme integrity) appear to offer the best chance

of redu cing rates of rec idivism. Although the findings are limited with

20



33/82

regard to adult offenders, the message with regard to cognitive-behavioural

approaches is consistent with that reported in the more numerous studies of

young offenders. The reviews by Lipsey (1992) and Andrews et al. (1990)

also indicate that, while some forms of intervention are associated with fairly

large reductions in recidivism, those based on the use of punitive measures

actually appear to increase the chances of reoffending.

As Lipsey (1995) concludes, however, only general and broad brush advice

on effe ct i ve intervent ions is possible on the basis of meta-analyses w hich

aggregate over a wide range of studies. All interventions must be de veloped

and delivered in accordance with the particular needs and circumstances of

the offe n d e r. This is con sisten t with Pa l m e r s (1994) obse rvation that

w h i ch part icular combinations of ex p e riment al p ro gram fe a t u res havecommonly yielded positive results with large p ortions of their pop ulations

and which have nothas seldom been systematically explored and is largely

u n k n own. More ove r, in order to make pro gress in determining wh ich

p a rticular fo rms of cognitive - b e h av i o u ral training and combinations seem

promising, Palmer stresses that researchers and others need first to develop

and agree definitions for the various components of any given combination.

In short, although the simple, brief labels typically allocated to programmes

do not permit fine grain re p o rting of the comp onen ts associated with

reduced recidivism, Lipsey and others find sufficient evidence in the pattern

of results to permit some broad inferences. The consistently strong effect of

multimodal combinations reported in several meta-analyses does suggest that

there may be advantage, having assessed an offenders criminogenic needs,

in seeking to address them within a single, integrated programme.

Substance abusers

Once again evaluation of programmes delivered to drug abusing offendershas, in the main, been conducted in North America. In a rev i ew of the

l i t e ra t u re of drug misuse and the criminal justice system Hough (1995)

points out that these reviews have largely been of programmes rather than

programme componentsa general difficulty with the literature concerning

re h ab i l i t a t i ve wo rk with offe n d e rs. More ove r, many studies of diffe re n t

fo rms of treatme nt have no t evaluated pro grammes located within the

c riminal justice system and have no t examined effe c t i veness in re d u c i n g

reoffending. For example, a review carried out by Husband and Platt (1993)

of app ro a ch es u sed to address dr ug and alcoh ol abuse ind icates that

c o g n i t i ve skills app ro a ches wh ich include pro blem solving training are

succe ssful in redu cing alcoho l intake . The rev i ew do es no t ex a m i n e

wh eth e r th is type of appro a ch was also e ffe c t i ve in redu cing furt h e r

offending, but an evaluation by Platt, Perry and Metzger (1980) on a heroin


21


34/82

treatment programme which included behav iou ral the ra py and training in

i n t e r p e rsonal pro blem-solving show ed promising results with re g a rd to

re o ffending. Youn g offe n d e rs who attended the pro gramme, wh ich wa s

located within a youth correctional centre, had significantly lower rates of

p a role revocation for further offence s (both drug and non-drug offe n ce s )

than a matched con trol group. One othe r p ublished study, re p o rted by

Johnson and Hun ter (1995) ex amined p attern s of parole revocation fo r

groups of drug abusing offe n d e rs, some of whom attended a pro gra m m e

adapt ed from the Ross et al. R & R p ro gramme. Drug offe n d e rs we re

ra n d o m ly assigned to eith er regular p rob ation, a n on-cognit ive dru g

programme or the cognitive skills programme. After one year 6/32 (19%) of

the cognitive sample had been revoked, compared with 8/23 (35%) for the

regular p robation and 10/33 (30%) of the non-cognitive drug pro gram m e .With such small nu m b e rs, howe ve r, the se results are no t statistically

significant.

An appro a ch advocated by McMurran and Hollin (1993) for addre s s i n g

subst ance abuse is that of a mod ular p ro gramme involving: thoro u g h

assessment; behavioural social skills training (BSCT); skills training; relapse

prevention; and lifestyle modification. According to this approach the client

acts as a persona l scientist, monitoring alcohol consumption, setting goals,

and modifying his or her expectations from alcohol consumption. BSCT has

be en succ essful in redu cing alcoho l co nsump tion in non -offe n d e r

populations, but again, there is no evidence of its effect on recidivism in

offender populations. Furthermore, the approach has apparently not been

attempted with illicit drugs users (McMurran, 1996). McMurrans (1996)

rev i ew of the litera t u re on alcohol, drugs and crime also con fi rmed that

there is insufficient research evidence to permit conclusions about precisely

w hat com binations of cognitive, be hav i o u ral and skills elemen ts are

necessary for successful intervention, measured in terms of both substance

intake reduction and in reduced recidivism. This is unsurprising, given the

complexity of the research task and the relatively few studies in this field ofenquiry.

While programmes such as methadone maintenance and detoxification focus

on the immediate need to treat the addiction, relapse prevention techniques

are, as the label suggests, concerned with longer term coping strategies and

with enabling drug misusing offenders to recognise situations where there is

a high risk of a recurrence of the problem behaviour. Cognitive-behavioural

approaches are used in relapse prevention, techniques including role playing

of high risk situations, positive self-statements etc. There is some evidence

of their success in alcohol treatment groups: Allsop and Saunders (1989)

found that offenders who underwent relapse prevention were less likely to

relapse to heavy drinking after 6 months. However, as far as we are aware,

t h e re is no p ublishe d evidence of the success of relapse p reve n t i o n

22



35/82

t e chn iques w ith dru g users nor of th eir impac t o n recidivism among

substance abusing offenders.

Violent offenders

T h e re is a growing body of re s e a rch evidence wh ich links ag gres s ive and

violent behaviour with early fa m i ly ex p e riences, in particular poor ch i l d

re aring and parenting, harsh and authorita rian discipline. Cohort studies,

s u ch as that undert a ken by Fa rrington (1996) indicate th at the roots of

ag gression and vio lence, like oth er fo rms of anti-soc ial be hav i o u r, are

learned, and that the family plays a key role in determining whether early

patterns of aggressive behaviour become established. While not denying thei m p o rtance of early env i ronmental and socialisation fa c t o rs, in order to

i m p rove the effe c t i veness of inter ventions with tho se whose ag gre s s i ve

beh aviour has become established, p sychologists have also sought a fuller

u n d e rstanding of the cognitive processes which unde rlie and tri g ger this

form of behaviour.

One of the most prolific researchers in the area of violent offending has been

N ovaco (1975, 1978). Novaco d escribes anger as be ing cau sed by a

p hy s i o l o gical arousal which is then labelled by an offe n d e r, the lab e l l i n g

being dep end ant on h ow an offend er pe rc e i ves h is or he r social o r

e nv i ro nme nt al s i tu ation . Hence , i t i s sugges ted tha t cogni t ive

re stru c t uring of a violent persons perception of social events, and their

relationships with others, can help in reducing aggressive behaviour and

hostility (Brown and Howells, 1996). Where cognitive techniques are used

with violent offe n d e rs the se would no rm a l ly fo l l ow Nova c o s app ro a ch

( B l a ck b u rn, 1995). This combines the use of cognitive self-con trol with

relaxation tech niques and an ed ucational element to help offe n d e rs

u n d e rstand th eir beh av i o u r. Novaco found that the self-con trol element

combined with relaxation had a greater impact on anger than re l a x a t i o nalone. Later research by Schlicter (1978) confirmed that a combination of

relaxation training with self-instructional methods of anger control and

coping skills had a greater impac t on ag gre s s i ve adolescen ts than basic

group counselling. Feindler and Ecton (1986) also obtained pro m i s i n g

resu lts w ith ag gre s s i ve ado lesce n ts w hen using cogn itive - b e h av i o u ra l

techniques in a programme of anger-control and social skills training.

G l i ck and Goldste in (198 7) rev i ew ed and evaluated man y of th ese

techniques in developing their Aggression Replacement Training (ART) for

young violent offenders, which has been adapted for use with violent young

adults. The combination of anger control training techniques (based in part

on Nova c o s wo rk) with social skills training, pro blem solving and mora l

education has been found to improve self-control and there is some evidence


23


36/82

of a reduction in recidivism (e.g., Leeman, Gibbs and Fuller, 1993).

However, whether programmes such as ART can significantly reduce the rate

of violent offending is uncertain. One study found that the beneficial effect

failed to re a ch statistical significanc e on ce allow ance w as made fo r

d iffe rences in levels of supervision recei ved by the treatment and contro l

group following the programme (Glick et al. 1989).

The influe nc e of the fa m i ly up on adolesc en t ant isocial and ag gre s s i ve

b e h aviou r has led some psych o l o gists to invo l ve th e fa m i ly in th era py

through the use of cognitive skills training and problem solving methods. In

the study re p orted by Borduin et al. (1995) on the effe c t iveness of mu lt i

systemic treatment (which addressed behaviour problems in the context offa m i ly, p ee r group and scho ol), offe n d e rs aged 1217 we re ra n d o m ly

assigned to multisystemic treatment (MST) or individual therapy. Outcome

m e a s u res include d arrests for violent cr imes (rape , atte mp ted ra p e ,

ag gravated assault etc .) dur ing a fou r year fo l l ow-up . The youth s wh o

participated in MST, many of whom had previously committed offences of

violence , we re signifi c a n t ly less like ly to be arrested for further violen t

crimes than were youths who received individual therapy. The researchers

concluded from this finding that interventions with violent young offenders

a re like ly to be mo re effe c t i ve if th e y are co mp re h e n s i ve, add re s s

behavioural problems within the context of problematic family relations, and

are delivered in the familys home or at a community location.

Some evidence that cognitive-behavioural work with violent adult offenders

can reduce reoffending has emerged from a recent evaluation funded by the

Sc ottish Office and Home Office of two re-ed ucation p ro grammes fo r

perpetrators of domestic violence (Dobash et al. 1996). Both p rogrammes

adop ted a cognitive - b e h av i o u ral app ro a ch and consisted of we e k ly gro u p

sessions over a six to seven month p eriod. The aim was to increase the

o ffe n de rs insight into their violent behaviour and to model new ways ofthinking and acting through the use of a range of behavioural and cognitive

techniques. These included: teaching cognitive techniques for recognising

the sequence of events and the emotions associated with the on-set of

violence; use of continuous fo rms of self-assessment and mon itoring as a

means of reinforcing group wo rk; p ractising new behaviour through ro l e

p l ay; and didactic meth ods to enh ance offe n d e rs un der stand ing of the

nature of violent behaviour towards women.

In order to assess the effe c t i vene ss of these two pro grammes the study

followed up two groups of men who participated in the programmes (a total

of 41 men) and comp ared them with a group of 71 men who received other

c o urt disposals. Reoffending by the p ro gramme and other court disposal

group was assessed by interviewing women partners three and 12 months

24



37/82

after the completion of the p rogramme, thus significantly reducing the effect

of undetected offending inherent in most reconviction studies. The research

cou ld no t c on trol for the effect s of selec tion eithe r b y She ri ffs or b y

pro gramme staff and although there was no statistical difference betwee n

the programme and control groups in terms of the mens use of violence,

there were a few differences between them in their ages, employment status

and background of parental violence. Bearing in mind that the sample was

quite small and may have been subject to some selection bias, the results are

viewed by the researchers as encouraging. Only 33 per cent of those who

p a rticipated in the pro gramme had committed a violent act against their

partners after 12 months, compared with 75 per cent of those who received

oth er criminal justice disposals. Wome n w hose p art n e rs did und ergo

treatment also noted reductions in the coerci ve and controlling behaviourknown to be associated with domestic violence, such as threats, shouting

and restrictions on the womens social life.

On th e basis of an examinat ion of the use of beh av i o u ral and cognitive

techniques with violent offenders, Blackburn (1995) concluded that training

in anger management has indeed been shown to reduce aggression, at least

in the short term, and its usefulness in maintaining order in institutions has

been established. However, evidence of long term success is inconclusive,

particularly for the more persistent violent offenders who may require more

intensive interventions. Hence Blackburns observation in relation to violent

o ffe n d ers that issues such a s offender hetero geneit y, effec t ive tre a t m en t

com pon ents a n d t r ea t m ent in tens i ty h a ve ye t to be a ddre s s e d

(Blackburn, 1995, p382).

Sex offenders

Cognitive-behavioural methods with sex offenders attempt to modify three

sets o f p ro blem s: de viant sex u a l i t y, social co mp et en ce andattitudi n a l /cogni t i ve distort ions (Epps, 1996). Deviant sexuality is tackled

t h rough be hav i o u ral me th ods such as ave rsion thera py, w her e dev i a n t

d e s ires are associated with unpleasant consequen ces. Social competence

training includes improving empathy and assertiveness, anger management,

communication and conflict resolution. Techniques used include problem

solving and social skills training. Attitudinal and cognitive distortions can be

a d d ressed using group discussions, for example, wh ere each offender is

videoed talking about his or her offending behaviour which is then

discussed by members of the group (Epps, 1996; Barker and Morgan, 1993).

Evidence of the impact of these techniques has mainly come from the US

Furby et al. (1989) reviewed studies which included recidivism as a measure

but found little evidence of success. This may, ho wever, have been due to


25


38/82

poor research designs, which prevented any firm conclusions being reached

(Marques, 1994), and to the low rates of reconvictions by sex offenders in

the non -treatmen t control groups. For example, Marshall and Barbe re e s

(1 990 ) rev i ew o f th e ou tc om e of co gnit ive - b e h av i o u ral tr e a t m e n t

programmes with sex offenders includes only one study in which outcome

results are show n for both treated and untreated groups. The results of

some studies are also qualified by the high drop out rates from the

p ro grammes w hich may have ex ag ge rated th eir suc cess in re d u c i n g

recidivism. These methodological weaknesses may account in part for the

number of unexplained inconsistencies noted by Marshall and Barberee and

their conclusion that better indices of treatment effectiveness are required,

i n cluding det ailed info rmation on ch a n ges in sexual pre fe renc es, social

competence and cognitive distortions.

An American study which does include a control group and is based on a

re a s o n able sample size and inco rporates a long ter m fo l l ow up is the

C a l i fo rn ian Sex Offender Treatment Evaluat ion Pro gram (SOTEP). The

results of th is longitud inal study of re o ffending rates for offe n d e rs w ho

attended the pro gramme have re c e n t ly been published (Marques, 1994).

SOTEP uses a multi modal approach which includes social skills training and

d ru g/alcoh ol tre atment acco rding to offen der nee ds and a c ognitive -

b e h av i o u ral dimension wh ich foc uses on relapse pre vention, p rov i d i n g

o ffe n d e rs w ith sufficient con trols to anticipate and avoid future ri s k

situations. The programme is delivered in hospital and is followed by one

ye a r s community fo l l ow -up with mu ch emp hasis of relapse preve n t i o n

when offe n d e rs are released. The re s e a rch design included a tre a t m e n t

group of 98 offen d e rs, a volunteer con trol group of 97 offe n d e rs (who

vo l u n t e e red fo

CBT and Ex Offenders

Documents

Transcript of CBT and Ex Offenders