• Aucun résultat trouvé

lf il

N/A
N/A
Protected

Academic year: 2022

Partager "lf il"

Copied!
111
0
0

Texte intégral

(1)
(2)
(3)
(4)
(5)

ANASS ESSMENTOF THERELI ABILITYAND VALIDI TYOF THE SECURITY RATINGANDINMATE RIS KAS SE S SMENTS USEDIN CORR ECT ION AL INSTITUT IO NSIN NEWFOUNDLANDANDLABRADOR

BY

KELLYL. BURDEN

Athes issubmit t ed totheScho ol ofCr a d uate seud t ee in parti al fulf ilment of the

req uirements for thedegree of Ma s t er of Science

Department of Ps y.::ho l og y Memorial Unive rsity ofNewfoundland

199 6

St John's Newfo undland

(6)

1+1

NalionalLibrary

01Canada Bibliolhequenaliona1e

cccereoe

Acquisilion~and Djrechondes~qui sitionseI Bibliographic sevcee Blanch desseMC(~Sblbhograp hlques 39SW~lirlglooStrPe1 39S.ruewel,ngton

~1c:'

•.

pr"(arlo ~..r~Onlano)

The author has granted an irrevocable non-exclusive licence allowing the NationalLibraryof Canada to reproduce. loan, distribute or sell copies of his/herthesisby any means and

In any form orformat, making

this thesisavailabletointerested persons.

The author retain sownershipof the copyright in his /herthesis.

Neither the thesisnor substantial extracts fromit may be printed or otherwise reproduced without his/herpermission.

L'auteur a accorde une licence irrevocable at non exclusive permettant it la Bibliotheque nation ale du Canada de reproduir e, prater,dlstribuerou vendre des copies de sa these de quelque manlere et sous quelque forme que ce soltpour mettredes exemplalresde cette these it la disposition des personnes Interessees.

L'auteurconserve la proprlete du dro it d'auteur qui protege sa these.Nila these nldes extraits substantiels de celle-cl ne dol vent etre lmprl rnes ou autrement reproduits sans son autorisation .

ISBN 0-612-13884-4

Canada

(7)

Abs tract

The Sec ur i ty Ratingand Inmate RiskAssessments, two ins t r ume nt s used in correctionalinstitutions in Newf oundla ndand Labrador,were examinedfor theirability topr1!dict recidi v i s m.The secur ityRating Assessment is primarilycomposed of itemsrelated to an offender' s past and present criminalbehaviou r. In contrast, theInma t e Risk As sessment is compos e dof "n e e d" items: itemsthatde a l wi t h the type and seve rityof social, emotional, andeconomic pr oblems expe rie nc e d by an offender. Rec i d iv ism wa sdefined as re Incai-ce r etLon and both a dichotomous crit e rio n , recid iv istvs . non-r ecidivist ,and aco nt i nuo us cr i te rio n, number ofvio lation-fr eeday s, were used.Threeinstitutions withdiffe r e nt male inmatepopulat ion s we r e sepa ratel y exami ned to determinethe relia bilityand validity of the two inst ruments.Th e correlations obtained at the three institutionswere covertedto z-s c o r e s and tests of dif fe re nc e swe re conduc ted . In ca s e s whe r e si g ni f ican t dif fe renceswere not found, a single co r r e l a t i o n was calcu lat e d us ingthe comb inedsamples.The recid i v ism rate forall the InetLt ut.I cna combinedwas18.57 %~litha lo w of 11.74 %and a highof 29.21%. Bothassessmentswere found to be reliableas shownby the intercorrelationsand Cronbach's alpha. In general,the assessmentswere also found to be valid asshown by the criterion-to-totalscore correlations .

i i

(8)

Al so, the re c idi v i s m ra t escorrelatedpositive lywithth e diffe rent se c ur ity andrisk le vels asme a s ure d by the securi ty Rati ngand Inma t eRis kAsses s me nts . Specificall y, mor e medium-security inmates thanminimum-securi tyinmates wer e rec idiv ists.As well, low-r i s k inma teswerele s s like ly to ::.-ec idiva t etha nme di umor hig h-r i sk inmat es.The Inmate Risk Assessmentcouldnot di s cr i mi na t e between the medium and high-r i s k categories.Inconc lusion,mo r e accurate pred i ct i o ns of reci d ivism canbemade byusing the Security Rat i ng andInma t eRiskAs sess men t s .

iii

(9)

Acknowle dgements I thank Hr. Marvin McNutt, Director of Adult Cor rectio ns, for providinq thef\~.·'Hn9for thetra vel expensreinc ur r ed during the research. I alsoth a nkthe classificationofficers at the Labrador CorrectionalCentre andHer Majesty1s penite ntiaryfor their assistance, and especiallyMrs.Susan Dun ne . classificationofficer at the West Coast CorrectionalCentr e ,who suggested the topic .

Mysi ncerestappreciationis extended toMr .Malcolm Gr a nt,mysupe r visor, whohasspent man ymonths reading and re-re a dingdr a ft s of my thesis andofftlri ng guid anceev e r y ste p of the 'Way.I alsothank Dr. Cathy But t o n and Dr. Andr e e Liddell for be i ng on)!I¥thesiscommittee. Last but notlea s t, I thankmy parents and husband for their unfailing encourager-ant and support duringmy Masters.

iv

(10)

Contents

page

ABSTRACT ii

ACKNOWLEDGEMENTS iv

LIST OFTABLES vii

Chapter1 INTRODUCTI ON 1

1.1Risk/NeedsInstrumen ts 4

.:..2 Areas of Inmate Research 5

1.3The ClassificationProcess 5

1.4validi tyof Instruments 7

1.5 charact.erIetIceof the Institut ions 9

1.5.1Her Majesty'speniten tiary 10

1.!:>.2West Coast CorrectionalCe n t r e 12 1.5.3 La bra d or Correc tionalCentre 13 1.6ne v el op me n t of th e SecurityRatingandInmate

HiskAssessments 13

1. 6 .1 securicv RatingAssessment 16

1.6.2Inmate RiskAssessment 20

1.7Present Study 21

1.7. 1Reliability 22

1.7.2Valid ! ty 22

Chapter2 MEASURES 24

2.1Definitio nof Recidivism 25

2.2 Classificationof Recidivism 27

2.3 Lengthof the Fo llow-Up Period 29

Chapter3METHOD 31

3.1Participants 32

3.2 DataColle c t i o n 33

3.JAn a l y s e s a a

Chapter 4 RESULTS )6

4.1 Recid i vi sm Rate s 37

4.1.1Recidivism Rates AcrossSecurityRating

Levels 41

4.1.2Recidivism RatesAcross Inma t e Ri s kLevels 4) 4.2 Relationshi pBetweenthesecuri tyRatingand

Inma t eRi s k Assessments 46

4.3 securityRatingLevels 46

(11)

page

4.4 Inmate Risk Levels 48

4. 5 Reliability andvalidity Analyses 50 4.5.1Reliabil ityof theSe c uri t y Rat ing

Assessment 51

4.5.1.1All ,t,amples Combined 51 4.5 .1.2Differences Amongthe Insti tutio ns 51 4.5.2validity ofth e secur ityRating Ass essment 51 4.5.2.1Allsampl esCombined 51 4.5.2.2Differen cesAmongthe Institutions 56 4.5.3Reliability of the InmateRis k Assessment 56

4.5.3.1All samples Combined 56

4.5.4Validityof the Inmate Ri s k Assessment 61

4. 5 .4.1All Samples Combined 61

Chapter 5DISCUSSION 62

5.1Reliability of th e Assessments 65

5.2Va lid i t y of the Assessments 65

5.2.1 Recidivis mRatesAcrossSecurity Rating

Levels 66

5.2.2RecidivismRatesAc ros s Inmate RiskLe ve l s 68 5.2.3criteria-to-Total Sc o r e Correlation s of

the security Rating Asses s ment 69 5.2. 4 Criter i a-to-TotalCorrelat i o ns of the

Inmate Ri skAssessment 71

5.3Pr o b l e msin andLi mitation s of theRese arch 72

5.4FutureResearch 76

REFERENCES 78

APPENDICES

AppendixA 89

Appendix B 91

vi

(12)

TABLES

Table1.1.Characteristics of Her M<.,jesty's Penitentiary, th e West CoastcorrectionalCentre, and the La bra do rCo r r ec tiona l Centre.

Table 4.1. Number ofinma t es on whom the security Rating and Inmate Ris kAssessments were completedat each institution.

Ta ble 4.2.Percentageof recidivistsateach institution . Ta ble 4.3.Recidivismrates for each securityle ve l foreach

inst i tution.

Ta bl e 4. 4 .Recidivism rates for each risk level for each institution.

Table 4.5. Number of inmatesat each institutionat the minimum, medium, and maximum security level as determi ned by theSe c urity Rating Assessment.

Table 4.6 . Number of inmates at each institutionat the low, medium, and high-risklevel as determinedby the InmateR.l.sk Asse ssment.

Table 4.7. Intercorrela tio ns of theitemsonthe Secur ity Rating Assessmentfor the three samples.

Table 4.8. Differencesamong the institut ions on the secur ityRating intercorrelat! ons.

Table 4.9. Intercorrelationsof theit emson theInmateRisk Assessmentfo r thethree samples.

vii

(13)

CHAPTER 1 Introduction

(14)

An As seasnen t;ofthe Reliabilit yand Validi t yof the securityRat i ngand Inmate RiskAssessments Used in

Co rrect i ona l institutionsin Nc·:.rfoundiandand Labrador

Over the pastseveral decadesthere have beenmany attempts to deveLcp statistica l or actuarialmethDds to pre d i ct the le velofris k that,"ri ml nal offenderspresent to the community. Itha s be e n evide nt inthe literatur esince the 19 409and 195 0s tha t a few wel l -chosenrisk factors can predict criminalrecidivismwi th animpr ess ivelevelof ac curacy. However , it wasnot until the 1970s and1980s that statistical pred ictioninstruments weresystematically int r od uc e d into correctionalpract iceinCanada (Andrews, 1989; Ha:-ris, Rice, &Qu i ns e y , 1993).

Ideally,classification ins t rument s incorrec tions should assessno t onlyri s k rector-sbutneedfactors as wel l (Bonta &Motiuk, 1985).Risk is defined in terms of an offende r'spote ntia l for committ inga subsequent offence or technical violationupon release froma correctiona l institution. The assessment of risk is us ua ll y based primarilyonthe of f e nder' s criminal hi s t o r y. Need , a subset of ri s k, is usuallydefinedas the type and severity of social,emotional, and economicproblemsexperienced by the offender. Need is a subset of risk in the sensethat the r e existsa positive relationshipbetween the ne e ds of an

(15)

offender and his risk ofreo ffe ndi nq (AdultCo rre c t i on s Division , 1991 ; Andre ws, Bonta, , Hoga, 1990; Bo nta ' Motiuk , 1985).

Thereare two re aso ns for assessing bot hrisk and need fac tors. Firs t,a combinedassessment of risk and nee d le vels si g nificant ly improv es theabilitytopredict whois likely toreoffendbeca us e a gre ater nue e e- andva rietyof fa c t o r s are examined (Andre ws, 19 89 ; Bont a&Motiuk, 1987;

Mo t i uk , 1993 ). Res earch has found that cr imi nal history fact or s are pos itivelyrel ate dto recidivi s m andthat a consi ste nt relatio nsh ipexistsbe'tveenthe typeandnumber ofne ed s off e nde rs exh ib it and thelikelihoodof re of f e ndi ng (Ad ult Corre c t i onsDivision , 1991 ; Andrews etaL, , 1990 ; Bonta ' Moti uk 198 5 ) .

,.second re a s o n forassessingboth ri s k andnee ds is tha t it he l ps to achieveabal ancebetweenthe goalsof ensuringcommuni tysaf ety and re h abil itatingthe of fe nde r (Clear ' Gallagher ,1983). To ens ur ecommun ity safety , the offender must be pla c e dunde r appropriate sur ve i l lanc e . To rehabilita te the offende r , problemsor nee ds havetobe ide nt if ied and resolvedin order to reintegrate the pers on intothe communi t y. As pointed outbyAndre ws (1 98 9 ),the researchonne e ds issma llinvo l umeincompari so n to the abundan c e of re s earchon ri s k factors.

(16)

1.1 Risk/Need sInstruments

According to Gendreau, Cullen, and Bonta (1994), there are onlythr e e risk/needs instrumentsin widespread use.

Theseinclude the Levelof SupervisionInventory ';LSI) which is used in probation and parolein ontario , the Wisconsin c1assificat:l.onsystemwhichis used in probation in several statesand Canadianprovinces, and theCo mmu n i t y Risk/Needs Managem~ntSc ale whichis used to class ifyfederal offenders on co nd i t i ona l releasein Canada.

The risk instrumentsusually consist of stat i c items that pertainto an offender'spriorcrim inalhistory such as the number and severity of pr i or offences, age at first conviction, and record of escapesor attempts (Bonta, 199 3 ). In contrast, ne e d s instrumentsusuallyinclude dynamic items whichassess "criminogenicll needs (Andrews, 1989 ; Gendreau et aI., 1994).criminogenicor dynami cneedsare cha r a cte r i s t i c s of the offender that can chanqe or be modifiedover time suchas attitUdes , associates and companions, subs t a nc e abuse, and educationa l and vocational skills (Andrews et al., 1990).The importanceof the se criminogenicne e ds is that they serveastr e a t me nt goals.

When programssuccessfullytarget anddiminishor resolve offenders ' needs, one can expect; a decreasein recidivism (And r ews at aL;, 1990;Gendreau et aL., 199 4 ) . Conversely , offenderswhos e nee ds arele f t unresolvedare more likelyto

(17)

cont i nue their invo l veme nt incri.ina l beha viour .

1.2 Areas of Inmate Assessmen t Resear c h Many studieshave focusedonclassi fica t ionasse ssments used inthe areas of probation, parole , or hal f way- ho u s e programs (Andrews, Kiessling, Robins on, &Hickus , 1986;

"'.shford &t.ecrcv , 1988; Baird, 1981 ; Bont a, 1993; Bonta &

Motl u k , 1985, 1987, 1990; Clear&Gallagher, 198 3, 1985;

Eaglin &Lombard, 198 1 , 1982;Mo t iuk,Bonta , &Andrews, 1986; Wright , Clear , &Dick s on , 1984). Incontrast , relatively little inmateassessment rese archhas been condu c t ed wi th i n co r rec tional insti tu t i o ns and wha t hasbeen donehasde alt wit h insti t ution a ladjustmen t (Cla r k, Fis her,

&McDougall, 1993; Han s on, Hos s, Hosford, &Jo h nson, 1983;

Wright, 1988).Althoug h thisis thecase,cla s si f i c atio n instrume ntsservethe same purpose in cor rec t i o na l insti t u tio ns as they do in probation and paro le. That is, the instru mentspurpo rt toide ntifythose ris kand nee d fa ctorsthatarepos it ive lyrela te dtore c id ivi sm.

1.3The Class ificationProce ss Animportant ta s k in corre ct ions isto deal wit h offend ersinsuc hillway thatlow-r i s k ca sesre ma inlowand higher-riskca s esmovetowar d lower ri sk.Identifying offende r sas lo W', medi um, orhigh inris korne ed is

(18)

ach ieved thr ou gh apr'o c e ua known ascl a s s i fIc atIon, Af t e r iden tify ingoffenders' risk/need levels, spec Lr Ic pLe ne can be devised topl a ceoffendersinappropriateprograms. As Andre ws (1989) sta ted :

.. • (cla s s i f i c at i on ] involvesthe managementand tr e atment of of f e nde r s according to their ri s k leve l (the risk pr inci p le), choosing appropriat etargets of rehabilitative pr og r a mm i ng (thene ed principle), and employingstylesand modesof trea tmentthat are appropriate for offend ers (t h eresponsi vityprinci ple ) (p. 14).

Ther e fore, the pr i ma r y goa l of inmate classification systems is to placeinmatesin correctionalsettings which maximize the probabilityof rehabilitationandensure the safetyand security of correctio na l insti t ut i ons (OntarioMinistry of Correctiona l Services, 1990 ).

Perha ps the most importantcharacteristicof inst ruments tha t measureri s k and needsis their abi li tyto pred ict re cidi vism.Hence, tobeus e f ul , the instrume nts' predic t ive val id ity mustbe demonstrated. Interms of offe nderclassificat ion, validity ref e r s to the deg r e eto whh::hoffe nde rscl assifie das high risk and/orneedaremore likel y tore of fe nd than thos e classifiedas lo w risk and / o r

(19)

need. It is important to notetha t inmates' scores on the s e instr ument s ma ke itpossible only to rankthe inmates in te rms of their potentialfor reoHending.The scorescannot be used to estimate, in absolute te rms, the probability that any parti c u lar inmatewillreoffend (Clear&Gallagher , 198 3 ; Nuttallet a1., 1977; Wrightet al. , 1984).

1.4 Validi t yofIns t r ume nts

There is a wealthof evidenc e inthe literaturethat thevalidi tyof ins t rume n tsca nnot be assumeil to tr an s f e r fr om one po pulatio n to anothe r:valid itymust be es tablis he d separately for eachpopulation (Anastasi , 1982; Ashfor d&

LeCroy, 1988;Clear'Gallagher ,1983,198 5 ; Ea g li n &

Lombard, 1981, 1982: Got tf redson, 1977;Gottf redso n , 1987:

Gottfredson&Gottfredson, 1986; Kane, 1986; Rhodes, 198 5 : Simo n , 1971: Talmage&Rasher, 1981; wright, 1988; wrightet a1., 1984).For example, the Wisconsin classificationsystem was fou nd to bepr ed i c t i ve ofrevoc a t i onrate s for prob at ioners inLos Ange les county (Gl aser, 1987), but not valid fo r a populationof new York City pro bat i oners (wr i gh t et aI., 1984 ).

There ar eseveral pos s ibleexplanat h~nswhy a cl ass ifica tionsystem that pred i cts well for one popUlation pred i c ts poorly for anothe r. Firs t, the remay be littl e or no vari abi lityin one or mo re pr edicto r s inapa r ticular

(20)

pop ulat i on . For example, i fa la r ge majori tyof th e offendersin a populationareunemp l oyed, employmentstatus is not likelytobeus e f u l inpr edic ti ngt-ecf dfv Lea, However , inanot her populationwhere seve r aloff ende rs have ful l - timejobs, others are empl oyed seasona lly, and othe rs areunemplo y e d and re lyonsocialassistanc e,employmen t st a t us may bea more usefulpredictor .

A second re a sonfor differenc e s acro sspcpuLatIone is thatone or more pred i ctorsmay not beculturally relevant toa particulllrpopUlation . For example, alcoho l abus e roilY be amajor dete r mi nantof cr i mi nal be havi ou r a~ongoffe n ders insome cul t ur e s butnot in others. The r e fo r e, thespeci fi c soc i al and envi r o nmental infl u enc e s on cri mi na lit yllIa y va r y acr os s po pUlati ons .AsGo t.tfredson andGot tfredson(1980 ) exp la i n:

.••th e grea testlimitationof predictio n me thods is that the de vices ar edev e l ope dand va li dat e d wi th re s pect to specific cri t e ria , using availab ledata, in a spec if icjuris diction, duringaspecifi c t imeperi od.

Thus, anygen eralization toot h er outcomesof interest , or after modit'icationsof the i temdefinitionsused,or to otherjurisd i c tionsor popUlation s, or to other time period s , are tobe questioned (p. 328 ).

(21)

Lastly, thepowe r of diffe r e ntpredictorsmay vary acrosspopul atio ns be c au s e of simplechance fac t or s . Instrume nts developedusing techn iquessuch as multiple regression , logit analysis, di s cr i mi nan t analysis, or predictiveat tributeanalysis are more susceptibl e toType I errors thanare simpler techniquessuch as the Burgess or simple summatio nmet hod (Benda, 19 81 ; Copas&Tarling, 198 6 ; (':.,)t tfre dson& Gott f redson , 1980; Hoff ma n, 198 3 ; Loeber &

Diahion, 1983;NUffi e ld , 198 2 ; Simon, 1971; wr ightet al., 19 8 4).Be c au s e measurementand recording errors will inevitaoly be prese nt in thedata (Benda,1987;Copas &

Ta r li ng , 1986), repeated valida tionsare essential.

1.5Chara cte ri s t i c s of theInstitutio ns In Newfou nd l a nd andLabrado r , the Inst i t u t iona l Services 8ra nc hof theAdul tCorrectiona lDivisionis comprisedof seven cor r e c t iona l centres andtwoloc k-up s (Adul t Co r r e ct i onsDi v i s i o n, 1991).The data in the presen t st udy we r e gatheredat the thre e largest correc tional centres, Her Majesty'speniten tiaryin st. John 's , th e West Co astCo r r ec tional Ce nt r e in ste p henvi l l e, and the Labrador Correctiona l Ce ntre inHa pp y va l ley/ Goose nay.These ins t itu t i ons were sele c t e d part l ybecauseof their sizebu t mainly beca us e itwa s known thatthe classif ication officer (s) ateachinstitut ionwer e usi ng the Sec uri ty

(22)

10 Rating an d Inma teRisk As s es s ment s.

Somestat i sticsconcern ing the thr e einst i t ut i onsare pre sented in Table1.1. Inthi s table, the standa r dcapacity refers to thenumber ofinmates theins t itut i o n iscapabl e ofhou sing.Rat e of admiss ionsrefer stothepe rc en ta g e of inma t e s that were incarc eratedat theinsti t u tion duri ngth e year. As ca n besee n, He rMaj e s ty's pen i t ent i ar yisthe la rge stofthe Insti t l. tio ns and in carc erated the most inma tesdu rin g 1990- 19 91. or all the in mates inca rc e rated during19 9 0-19 !;1, 57% were inc arcerat ed at Her Ma je sty· s Peni tenti~.:cy.

1.5.1 Her Mjlj e st y ' s peni ten ti ary Thi s institution is cla ssified as a min i mum/medi um secu rity pris on . Pris o ns with thi s secu r ityclassification can hous e inmat e s for the ma ximum periO d of two yea r-n le s s a da y. Inma te sincarceratedat thi s fac ilityha vebeen co nvict e d ofoffe nce s rang ing fro m tow eeverLt. y (e.g., impaire ddriving, cau singa distu r ba nce , etc .) to hi g h sev e r i ty (e. g.•armed robbe r y , rape,eco.}. Although most inmateslive on theea s tcoa st of Newfoundland , the r e are se v era l excep t i o ns. Forexa mple, perso n s fro macross the provi nce whohave bee n conv icted ofsexu al and/o r ph ys ica l abuse against child re nor ad u l tsarein carce r a t edat lIer Maj e s t y's penitentiarybecaus ethe ins t i t ut ion hasa

(23)

11 Table1.1.

~aJ:a.cterj s t1c sofHf.u:....Ha,.ksty's Pe n i t e n tia ry the West C-Qa s t Co rrectj o nalc.entu...-.a ndthe Iab ra dQTCgrrect jonal

Ins titut ion capaci tyStandar d Comple meStaf f nt No. of Admis sions (~99 0 -19 91 )

Rateof Ad miss ions

(')

HerMajesty1s

Penitentiary 147 106 1296 57.0

We st Coast Correctional

Cen tre 50 27 441 19.4

Labrador Correcti onal

Centre 38 29 U6 5.1

.t!.Q..t&.. Fro m tnma te class ifi c a ti on sys t em (n.p .),by Adu l t Correctio ns Div i s i on, 1991, stJohn ' s , Gover n ment of New f oundl a nd andLabrador, Department of Just ic e. Copyright 199 1by Depart me nt of Justice.Ad a pt ed with permis s ion of theaut h or.

(24)

12 protectivecustodyunit . Second, inma te s whohavepreviously been incarcerated at anotherinstitutionin the provi ncebut whohave cau s ed problems whileserving theirsentences are bro ught toHe r Majesty's penitentiary. Thi r d ,inma t es who needpsychiatric care, e.g ., arehigh lysuic idal, arc brough t toHe r Majesty's penite ntiary.Psychiatric as s es s me nt s arocompleted at the Waterford Hospital.Lastl y, inmateswho ha ve le ngt hy cri mi n a lrecords and ha v e committed vio l e n t of f e nc es (i.e ., "hard-co re" criminal s )arehous edat Her Majesty·s penitentiary.

J..5.2 West Coast cprrectional Ce nt r e

Thisins titu tion is also class i fi ed as aminimum /med iu m securityprison. Gen erally,inmates at thi s inst i t uti o n , compared to thos e at Her Majesty'sPeniten tiary , have commit ted le ssviolent of fences and haveshortercri mina l records. In contrastto bothHer Majesty's Penitentiaryand the LabradorCnrrectionalCe ntre,inmates whoare remanded into custody whi l e awaitingtrialor sentenc ing ca n only be housedat theWestCoas t cor r ect io na l Centrefor amaximum of seven days.At Her Majes ty's penitentiary and the Labrado>:Co r r e ct i o nal Centre, inmateswhoare remanded into custod y can behoused atthe s e institutionsfor an indefiniteperiod of time.Although inmates at theWest Coast Cor rectional Cent r eus u a llyli v e westofCe n t r al

(25)

13 Newf oundl a nd, there ar e exception s, For example,when the Labrado r Correctio n alCentreis fil led , offendersfrom Labradorare moved to theWe s t Coast Correctiona l Centre.

1.5.3 Ia h r ad oT Corre c ti ona l Cent re Th isins t i tu tion ,the smallest ofthethr e e, is also classi fiedas a minimum/ mediumsecu rityprison.The la rg e st percen t a g e of inmatesarenat i ve (e.g•• Innu and In nui t).

Liketh e West coast Correctiona l Centrebut incont ras t to Her Majest y's Pe ni t e nt i a r y, inmat eshavecommit tedless violentoffencesand do not havelen g t h y criminalrecords. Although the institutiondoes nutha ve a pro tectivecustody un it, it does hous e some sexual offe nde rsinthe general inmatepopulation . Wi th in thenative popUlation,this type ofinmate isnot at as hdqha risk of be in gha r me d as are Caucas iansexualoffenders in a ca uc asi a n popu lation.

1.6 Develo pment of the Security Rat i ngand InmateRisk Assessments

Thecl a s s if i c ati on ins t rumen tswere developed in Ne wfoundl a nd in1991by a committeeconsis tingof the supervisorof Classificatio n , the supe r visorof Commun i ty Co rre ctions , anda Cl as s if i c a t ion Of f i c e r atHerMa j e s t y. s Pen i t e n tiary. One oftheprimaryrea s o n s forthe development of the ne w ins t rume ntswas to move away from the subjective

(26)

14 system, which hadbeen used for severa l years in Newfoundlandand Labrador, toward a more objectivesystem.

Relianceon sUbjectiveclassification instruments with informal criteria (e.g.,escape risk) often leadsto inconsistent assessments(AdUltCorrectionsDivision, 1991).

On the othe r hand, objectiveclassificationinstrumentssuch as the Le vel of Supervisory Inve ntory (LSI) and the Statistica l In f orma tion onRe c i d i v i s m (SIR) Scale consist of wel l -definedfactors which include the following: Ca) le g a l items dealing with such things asth e severity of the presentotfence(s)I le ng t hof thesentence, prior criminal record,and incide ntsofviolence; (b) dynamic it e ms dealing with attitUdes, associatesand companions,and substance abuse; and (c) fixed i temsthat assessthi ngs such as age.

The response alter nativesof the items areassigned different numericalvalues which reflectthe assessment valueofth e information. These values are thenused to determine aninmate's le vel of ris kand/orneed. Objective classificationinstruments the r e byhelp to ens uremore fa i r ne s s and consistencyinthe decision-makingprocess because all decisions are based on the same factors and criteria.

Duri ng theear lystages of developinga classi! ication instrument it became evident to the commit tee that the di vi si o nalman dateandthe institutiona l requirementscould

(27)

15 no t be thoroughlyfulfilledbyth e deveLo paerrt; ofoneinmato cl assifi c a tio n assessment (AdultCorrecti onsDi vision , 199 1).The divi siona l llanda t eis "to pr o vide theneces s ary custody, superv ision, and controlofof f e nders whi l e af f o r di ng the mprogra m opp ort un iti esto r reintegration int o thecommuni tyto becom elawabid ing cit i ze ns" (Adult Cor r e ctio nsDivision , 1991 , p. 1).noe ce , to fulfil the div i siona l req uirementsthe rewasaneed tode vi se a risk instru ment toassess the likelihood that releasedoffenders wou ldreoffend . Regard i ngtheinst i t uti o n al requirements . the r ewas aneed to de vi se asecurityinstrume ntto prov ide a sys t e ma t ic,compr ehensive , andcon s i stentmethod of comp l e t ing asecuri tyclass if icatio non sent encedof fe nders. The r e fo r e . wha t resu l t edweretwo classi fica t ion instrume nts that are indepe nden t of ea c hothe r;eac h us e s di s t inct objective scori ng.e t hods, and each has clea r ly establi sh ed purpos e s andgoals . Howeve r.bothare us e d inmak i n g decisio ns regardi ngtra ns fers betwee n instit ut ions andabou t re l e aseoptions such asparole andgra nt ingtempor a r y absences.

The two assessmentscurrentlyused in correctio nal inst itutions in Newfou ndlandand Labradorarethe Securi ty Rati ngAssessmentwhich assessesth eris k of off end ersand theInmateRisk As sessment which,despiteits name. as se s ses th eneedsofot' t'ende r s.Theseinstr ume ntshave bee n usedat

(28)

16 the West coastCorrectional Centre in Stephenville,the Labrador Correctiona lCentrein Goose Bay, andHer Majesty's Penitentiary in st. John'ssince November ~992 . Wittl!n JO days of an offender'sin c a r cer a t i on , the assessmentsare completedthrough interviews conducte dbythe Classi fication Officer. The Security Rating Assessment is completed onall inmat e s except those serving intermittent sentences,those incarcerated due to parole suspensions, or those remanded into custody while awaiting trial. The Inmate Risk Assessment is completed on all inmateswho received a sentenceof93daysor greaterwiththe exception of inmat e s who are servingin t e r mi t t e nt sentencesor inmateswho are incarcerateddue to parolesus pe ns i o ns. As well, recidivist inmates who have beenassessed within the past12months are not assessedagain unless important changes in the offender's circumstances are knowntohave occurred. The instruments will now be discussedin detail.

1.6.1SecurityRatingAs.s..e..s.s.m.en The Security Rating Assessmentwasderived from the S~curityRating Scale used in Alberta. However, information

(e.g., items and item weights) contained in the followin g sources was also reviewedand assessed specificallyforit s relevanceto Newfoundland and t.abredorts Provincial Correctional System: The Correctional servicesof Canada

(29)

17 cus todyRatingSca l e , Mani t oba' sAdmi s s ion Assessment Scale , the Secur i ty Rat i ngSc a lefr omsask atchewa n, andthe classific a tionsys tem us edbyontar io 's Ministry of corr ections (Adu l t Correctio nsDivi sion, 19 91).

Thei te msincluded inthe assessment are ba s ed on evide nce inthel it er atu r e that the fr eq u enc y andseverity ofpas tcrimi nal behavio uris the bestindicatorof similar behaviou r in the futur e ("ndrews, 1.983;Ashford&LeCroy, 1988,1990;Cl a r k etaL,, 1993; Cor n ish& Cl arke , 19751 Far r ington &West, 199 0;Gabo r,198 6;Got t f re dso n&

Got t f re dson ,1986 ; Hansonat al., 19 83 ; Hi ll, 19 85; Hoffman, 1983;}{lein s ca g gia no, 198 6: Loebe r &Dishlon, 1983: NUf field, 19 82; Owe ns&Schoenfeldt , 1979; Si mo n , 1971) . spec i fic a l ly , the foll o wingnine fac tors are assessed:

(a) na t ure of out stan d ingchar ges and crown appea ls, (b) severi tyof presentoffence, (C) lengthof sentence, (d) nat ure ofpr i or off e nces , (e) re c ord of esc apesor attempts, ef )histor y ofvi o l e nce, (g) ag e, (h ) pre-t rial stat us ,an d (i) psychi at ricstabili ty . Th e ful l assessment instr ument is re p rinted in Appendix A.

Four of the factors, (a ) , (d), (hI , and (i ) , requi r ea brief expla nat i o n. Fact or (a), nature of out standing charg es andcro wn appeal s , needsto be definedse parate l y.The natureof out s tandingchar ge s ref ers to theseveri ty of off e n ces aninmate has beencharge d wi t h, but ha s not been

(30)

1B convictedof, at the time of his inc arce rat i on . Forexamp l e , an inmat emay beincarceratedbe ca u se he was conv i c t edof breakingand en t e ri ng. Hemayalsobe ch arged wi t h possessionof a na r cot i cbut thi s charg e may notye tha ve been heardby a jUdg e. Thisof f e nce represents an outstandingcharge.Thenature ofCrownappe al srefersto theseve rityof offencesan whichan inmate wasor igina lly co nvict e d or acquittedbut onwhich the Pros e c utionha s lodged an appeal.Factor (d), natur e of pr ior off e nc es, refers tothese verityof offenc e san in mat e commi ttedin the past.Theseve ri t y of offenc es forfa c t ors (a ) and (d) is dete r mined usi ng theseverityof Offenc eScale (Adult Co rrec t io nsDi v is i on , 1991).Factor (h). pre-trial sta t u s, refers to the offender's status prece dingand dur ingthe trial pe r iod for the curre ntof f e nc es (e.g ., whe t h e ror not a ba ilor recognizanceorder was in ef fect). Final l y, factor

(i) ,psychiatricstabil ity , isscored 'y e s' or tno'whe reas the otherfactorsare assignedweig hted numeric al valuesto re fl ectthesi g n if i c a nc e of thefactorto rec idiv ism .

Two StflPS arerequire dto ca l c ul a t e the tota 1ris k score. Firs t, thescore s on the firstsevenfact o rs, (a) to (g), are added together . Second , the score for factor (h), pre-tria lst.atus, is subt r a c ted from the total obtainedin thefirstste p. I fthe resu l t isnega t ive , zero is as signed as the tot al risk sc o re. The following sca l e showshow the

(31)

19 riskscoreis usedto assIgn an inmate to anap p ro priate securi ty level:

Spentjty[.rye]

minimum med ium max imum

AssessmentSCQr e

oto15points 16to24 points 25+points

Exceptions to the as signment rules someti me s occur. If the r e areconcer nsabout inmates'me ntal,emot i o na l, or ps ychia tricstabllit y, those whowouldothe rw i s ehav ebeen cat e gorizedas minimumsecuri t yareaut oma t i ca lly placedin themedium-riskcategory. Forexample,an inmatewho ha san assessmen t scorebetween0and 15, but who has displayedor isdisplayingsuicidal te nd e nc I e s wouldbe placedinthe medium- riskcategory. It Is importa nt to note, however ,that the psychiatricstabili ty facto r isonlyused asa te mpo r a r y me asu r e until a psychologica l or ps ych i at r i c rep ort is comple t ed. Pl acing a minimum- riskinmatein the med i um-risk cat e g ory is al s okno....nas overriding thesc ore. over rid i nga score refersto plac inganinmate ina higherorlowe r se c u r ityra ting categorythan was identif iedby the as sessmentbas edoninf orn at io n that isusu allynot conta i nedon the assessment.Otherexamplesof factors which mayjustify over ridi ngan initia l class if icatio ninc lu de the

(32)

20 needto separate co-accused off e nde rsand theneed to provi d e spe c i a l facilities for eld e rl y persons orthose \li t h di s abilit i e s . The override opt i onsho uld onlybe used in lot to154of al l cases (Adu l t Cor rectio ns Div i si on, 19 91) .

1.6 . 2In mate8isk As s e s s me nt The InmateRisk As sessment was de ri ve dfromthe Wis consin class i fi c a t ion system.Howeve r, information (e.g ., itemsand item weights ) co n t ai n e d inth e followingsource s wasalso reviewed andasse sse dspe c i f ica lly for its relevanceto Newfo u nd l a nd and Labrador's Provinc ial Correctional Sys tem: The Corr e c tional service sof Cana da Case Manage me nt St ra teg ies (ris k / need assessment ) . the stati s t i cal Infarn at i on onRecidivism (SIR) ,and other provinc i al classif icationsys t e ms (AdultCo r rect i on s Di vi s i on, 1991).The eeeeeesenc, cons isting of bot h dyn a lD i c and st a tic fa c t ors, includes the fol l o wing13 ca t eg ories:

(a) at t i tude , (b) alcoholusage , ec) ot he r dr uginvolvement, Cd)assoc iate s/compan i on s, (e )livi ng arrange ments, (f) self-management skills, (9) interpersonal relationships, (h) earlyfamilyof origin, (i) age of firstconviction, (j) numb e r of prior periods of probationand/or parole su pe r v i s i on, (k )numberof prior bre a ch e sof probation, parole, and / or temporaryleave of abs en c e , (1) number of priorconv ict ions for indictable off e nces , and (m) prior or

(33)

21 current convictionsfor spousal or sexu alassaul t.Th e comp leteinstru ment can be seeninAppendixB.

The fac to r sthatha v ebe en found to be most clo s e l y relate d to recidiv i sm areweight ed mos t heavily. As wel l , thedynami c fa c t o r sar e weightedsl ight lymore hea vily than thestatIc fa c t or s.Totalne e d scoresare calculateduy summing thescoresforea ch factor.The followingrule s are us ed to assignthe inmateto oneof thre e need le v el s :

AssessmentScore

lo w me d i um hi gh

o to8 points 9 to18 points 19 +points

There isanexcept ion tothe r-mes,howe ve r. Offe nde rs whohaveprior or cur re ntconvictionsfor spousa l or sexual assau ltareautoma ticallyplacedinthehigh-ne e d category. specific prog r amsca nthe n be imple me nt ed with thi styp e of offen de r.

1.7 Pre s e nt Study Taking intoconsideration that thepur po s e of Class ificat ioninstrume n ts is to predictre cidiv i s m and tha t thei r valid i tymust bees t ab lishe d separatelyfor each popUl a t i on , theprese ntstudywa sund erta k e nto assess the

(34)

22 reliabilityand validityof the securit yRatingandInmate RiskAssessment swhich areus ed. in three cor rectiona l ins ti tut ion s in Newfoundlandand Labrador. The pr esent rese a rchis thefirstat tem p t to assess thereli a b il i tyand validity of the s e in s trument~.

1.7.1 ~

Re lia bil i ty is a ne cessary al t ho ug h not su f fi c i e nt prerequisiteof validity .In the pre s ent st ud y

intercorrelati on s andCr o nba ch 's alphawere empha s i ze d.The items on theassessmentsare potenti al predic to rsof re c i d i vi s m and do not presume to measurea singl e dime ns i o n or const ruct.

1.7.2~

Validity. likere lia b i l it y canbe assessed inmany ways. In the presentstudy , the typ e of validityrer er eed to as crit e r ion- re latedorpre dict i v e validi ty is eepnaeteed, Nunna lly (1978) state d tha t crite r i on-relatedvalidi ty "is at is sue....henthe purpos e is touse aninst r ument to estimate some importantformof beh a vi ourthatis externa l to the measuring instrument itself, the latterbeing rsferred to asthecrite r i o n" (p , 87 ) . The refore, predi ct i vo val i di t y is di rec ted towardans wer i ng the question "Ar e tho ind ivid u a l items inclassif i c ationins trume ntspredic tiveof

(35)

23 inma t e or offender behaviour? " (Aus t i n. 1986,p, 303 ) .The fac tors includedon the ins t r umentsare th epredicto rsand the beh a v iour thatonewishestopred i ct , inthi s case, re cid i vi s m, isthe crite rio n. If theins t r ume ntsar evalid, the r e wil l be aposit i verel a tion s hi p bet wee n an offe nde r 's scoreson the in s trumentsandtherate of rec idivism (Clea r

&Gallagher, 19 8 3 : wright, 1988). Validpredictive

instrumen tsshould make it pos s i b l e to dist i nguishinmates in terms of theirpo t e nt io!ll risk of recidivism (Aus t in . 19 8 3)•

(36)

CHAPTER2 Measures

(37)

25 Beforeprocee di ngwithth e detailsofthestudy, several issues related toth e measurement of recidivism ne e d to beex e nd n ed,

2. 1 De fi nit i o n of Recid ivism Althoughresearche rsag reeon the impor tanceof re ci d i v i s m, ther e islit tle agreementonit s operat iona l definition (Allen , Eskridge, Latessa ,&vito, 19 85; Benda , 19 8 7 ; cavior&Cohen,19 7 5 ) . Recidivismcan meanr-eer-r-eeb , reconviction, or reincarceration .Acontroversy existsin thelitera t ur e concerning the consequencesofus i ng different def i nitionsof recidivism.Hoffman,

stone-Meierhoefer, and Beck (1978) and Klein andCaggia no (198 6 ) compared differentmeasuresof recidivismand found that they produced similarresult s.In contrast, Ge erken and Hayes (199 3) ,Ha....kins, ca ssidy, Light, andMiller (1977), and Hoffmanand Stone -Meierhoefer (1980 ) found different resul tswit hdifferent definitions.

Reppucci andcUngempeel (1978) statethatre c i d ivi s m is often defined as rei nc arc e r at i on fol lowingrelease froma correctional setting. Reincar cerationma y resultfr o m vi olati ons of releaseconditions (e .g ., paroleor problltion) and/orconvictio ns for ne .... offences.Severalstudies have oper atio nalized recid i vi s m asre incarc era t i on (Benda, 1987;

Bonta&Motiuk, 1985, 1987, 199 0; Carlson, 1981;Got tfredson

(38)

26

&Gottfredson, 1980; Harriset aL; , 19 9 3 ; Hoffma n, 1983;

Motiuk et al. . 1956; Motiuk,Motiuk, &Bonta, 19()2). Th e majordrawback of using the reinc>:.rceration definitionisthe difficultyof determiningwho are tr u l y recidivists.Thereare numeroussitua tionsin whichan offender would be classifiedas a non- rec idivistwhen in fact he did commit anot heroffenceuponrele a s e . First, some crimes may go undetected bythe police. Second, an offender may be arrested but not reincarcerated. Instead, he may re c e ive probation or he may be admit tedto an alternative prog ramsuch as one offered in a menta l hospital. Third, he may die whi lecommittingan offence or duringthe follow-up period. Fourth, anoffender may move out of the provinc e and continue his crimina l behaviour,und etectedat Ne wf o undla nd and Labrador 'sprovinciallevel. Finally, an offender may receivea federal sentence(Le., over 2 years) which again may not be detected at the provincial level. In all these instances, the offender would be inc o r r e c t l y identified as a non-recidivist. such errors areknown as Type II errorsor false ne g at i v e s (Repp ucci &Clingempeel, 1978).

Despite thepro b l e ms just men tioned, reincarce ration was chosenas the ope ra tionaldef in i tionof re c i d i v i s m in the pre s e nt study. Re i nc arc e r a t i o n data for Newfoundland and Labrador are re a d ily availablevia the cor rectiona l insti tutions ' compu te r system, the Canadia nCriminalJustice

(39)

27 system (CCJS ) . whe r e a sre a rre st and re c o nvict i o n data mu st be obtai ned through the Royal Ca nadi a nMounte d Pol ice(RCMP) and/or the Ro yalNewfo undla nd Constabu lary (RNe) andare not easi ly ac c e ss i bl e.

2.2 Cla s sific a t io D of Recidiyism

Relatedto theopera t iona ldefin i t ion of re c idivi smare the differen t classificationsof reci di vismthatma y be us e d. Re cid i vi s m can be measured as an all -or-none di ch otomy or as a cont i nuo usvariablewithtim e to recidivism measured inday s, weeks,months, etc.Time to recidivism is also ref e rred to as"vio l a t i o n- f r e eti me" (Ea g l i n&Lombard, 1981, p , 26). Themajori tyof studieshaveus e d the bina ry Classificatio n , re cidi vi st versus no n- r e cidi v i st(And rews at a1. 198 6;Andrews , Wor mith, &Kiessling , 19 8 5 ; Ashford&

LeCroy, 1990 ; Benda, 198 7; Bonta &Mot iuk, 1985 , 198 7 , 199 0 ; Eag li n&Lomba rd, 19 81; Gendrea u , Ma dde n, & Leip ciger, 198 0 ; Gottfre d s o n&Gottf redson, 198 0; Harriset aL, , 199 3;

Hoffman, 198 3; Mot i uket al ., 19 8 6 ; Mo t iuket aL,, 1992;

Nuttall et aL,, 1977 ; Simo n , 1971; Wright et aL, 1984). However, it ha s beenarg ue d that a cont inuou s sca lemay providea moreaccurateassessment or recIdivism(Gend reau &

Leipciger, 1978; Gottfredson&Go ttfredson , 1980; Har ri s &

Mol t r a , 1978; Holosko&Car lson , 198 6 ; f'l'altz , 1984) . Dicho tomousmeasuresare co nsideredto beove r-s i mpl i fi e d;

(40)

2.

their use impliesthat there areno degreesof successor failure. Conversely. ifa continuo us ae aeu reis used, the assumption that low-rIskoffenders will have longer periods of vio l a tion-free timetha n high-r i s koffe nders needs tobe empirica lly tested (taglin&Lomba r d, 1981). Relatedtothi s pointis the fInding tha t numerousoffendersdo not convert to non-crim inalbehaviourin a sing lest e p. Instea d, they prog r essina step....ise seriesfrom seriousoffences tole ss seriou s offen c e sto no cont a c t wit hthe la w. AsMoberg and Ericso n (1972) st ate d :

•. . the typical rehabilitationproce s s for criminal offenders se e ms to involvea serIes of gradual steps away froJIItheir past levelsand types of crim ina lis t ic behaviourand tovard law-abidingbehaviour (p. 51).

In spiteof the arguments in favour of a continuo us me a s u r e , se ve r alst udie s have foundthat a dichotomous measureyieldspredictions that areas good or bette r (Burden, 199 4; Eag11 n, Lo mbard, 1981, 1982; wormi t h , Golds t o ne , 198 4 ).Oneex planatio n fo r this find i ng ha s be e n sugge s t e d by Maltz (198 4). As he pointed out, time to re i nc arcera t i on isthe sum of severa l ti me inte rva ls wh ich inc ludethe fallowi ng: (a ) releaseto ar rest ,

(b) arrestto ahea r i ng, (c) a hearing to a trial (if there

(41)

2.

is a trial), (d) a trial tosentencing ,and (e) sentencing to reinca rcerat i on. Release toarr e s t is the only interval thatreflects an offenderIs behav i o ur;the others reflect the behaviour of the criminal jus t i c esys t e m. Hence, muc hof the variab ility in theco nt i nuousmeasur emaysimp l y be nois e.

Bothdich otom ou sandcon tinuo u s criteria ar euse d in the pres e nt stud y.us ingbothcr i teria makesit pos sibleto comparethe twomea s ures to determinewhich providesa more accura t eass e s s me nt ofre c i d iv ism. As well. the assumpt i on that low-risk offenders have longer periodsof

violation -freetime than high-ri skoff enderscan be tested.

Thus,an of f e nde r is classifiedasa reci d ivist or non-recidivist, and i fhe isclas s i f i edas a recidivist, the lengthofvi ol a tion-f r ee time, measur edinda ys , is also recorded.

2.3 lengthQftheFQl l g w-UpPeriQd The final issuethatne eds to becl a r i f ied is the le ng t hof time in the- -ericv-ceper-Lea.Studieshaveus e d one-ye arfollow-ups (Bonta" Mot!uk, 1985, 1987, 19 90; Motiuket aI., 1986; Motiuk etal . , 19 92) , two- y e a r fo11ow- ups (Gendreauet aI., 1980; Got tfredson, Gottfredson , 198 0 ; Gottfredson , Wilkins,&Hof f ma n, 1978; Hoffman, 19B3), or follow-upperiodsof var iablelength (And r e ws, Kies sling ,

(42)

30 Ro b ins o n , &Mickus,1986; Bend a, 1987; Van Voorhis, 1988;

Wrightetal. , 1984).Maltz (1984) stated that the one-year fol l o w- up periodhas been most freque nt ly us e d and a one-yearfollow-up was originallypl a nn e d butbecausean adequa tesamplecould no t beobt a ined, a six-mont hfollow-up was chosen in orde r to ha ve scoresfor as many participan ts as possible. Thestar ti ngdateisthe day of release. Fo r instance, ifaninma t e wasre l e a s e d on March I, 1992, the follow -upwouldproceed unti lAugust 30, 199 2 . Offende rs who ha v e remainedout of prison for at leastsix months are classified as non- r ecidivists.

(43)

CHAPTER3 Method

(44)

32 3.1 partici pa n t s

Three correct i on al ins titutio n populati onswer e ex a mined : the Wes t Coast Correctio na lCent re , Her Maj e sty' s Penitentia ry, andthe La br a d or Correctiona l Centr e.The original inte nt i on wastoobta i nfr o m ins titu tio na l re c o rd s ara ndo m sampleof saomale offende rswho had beenassessed us i ng the Secur i tyRatingand InmateRisk Assessments.

Howe v e r. this numbe rwas no t attained for severalrea s ons . Alt hough th e assessmentswere avail able for use in No v em b e r 1992, severalclassificationofficers used the assessments onl ysporadicallyunt illolay 1994.star ting inMa y1994 classif icatio n of f i c e r swere required todocument,on mo nthlylogs , offenders andtheirassessmentscores. This resulted intheas s e s s ments be ing completed on all eligible inmates. A secondreasonwhy the int endedsamplecouldno t be obtaine dapp lies totheInmateRisk As se s smen t only. In Nov e mber199 4 , use of the Inmate Risk Assessment was stopped. Wi th the ons etof theElec tronic Mon ito ring system in St.Jo h n 's, a validated ins trume nt wasneeded to determineWhichinmateswere eligibleto participateinthe program. Because the Inmate Risk Assessment ha d notbe e n validated , the classificationofficerswe r e toldto usetho Wi s c o ns in Proba tionassessment,wh i c h hadbeen valida te d [or Newf oundla nd.

(45)

33 3.2Data Collect io D

Inma t e s' scores on the security Rating andthe Inmate Risk Assessments wer e draw nfromth e inst i t utiona lflIes . The items included ontheassessmentsrepresen t the predictor var i ab l e s. Two dependentvar ia b l e s weremeasured which repre s ent the recid ivismcrite rion . The first mea s u r e was whe t heror not the pa r tic i pan t s were reincarcerated duringthe six-mont hfollow-upperiod; the secondmeasure was the numberof violation- free day s.Ascore of18 3 was used fo r the non- recidi vists.

An offendermay be reincarcerated ina different provincialcorrectional institutionfrom theprevious one.

Hence,all inma t e s' criminalre co r ds wereaccessedvia the Canadian crimi na l Justicesyst em(CCJS).ThUS, ve rif i c at i on of provincia l incarcera tion du r i ng the follow-up per i o d was possible. Nonnumer i c variab l essuch as the identityofthe correctional inst itut ionwere coded us i ng dummycoding (Keppel & Zedeck , 1989).

3. 3 ~

All anal yses we r e conduc tedforeach as s e s s ment separately. The first st epin the evaluatio nwasto determine the rmmbet-of offendersWhofel l into the thr e e risk and ne edcate go r i e s and to calculate , within each category , the percentage of offenderswhore c i d ivated .From

(46)

J4 the s efigure s, th e ove rallviolati o nrate wa s calc ula t e d. In the pres e ntstudy, the vio lationrate isthepercentageof off ende r s whorecidivatedwith inshemonthsofthe ir re le a s e . Forre search purpose s,the ideal bas e ratewou ldbe SOl, but thi s rarel y occursin practice.The morethe bas e ra t e de v iate s fr om50l, the les sus e f u lany pr e dic t i on instrumentwill be (c o p es&Tarling, 198 6 ; Gabor, 1986;

Got t f r e ds on,1987;Han l e y, 1979).Ho we ver , Simon(1911 ) sugg e s t s thatpredicti oninstrume ntsarestill useful when the baserateIsmoder a t e,sp ecif ically, in the30 \ to70\

ra nge. Inanat tempt toobta i naba s eratewi t hi nth is ran g e, thedefini tion of re cidivi s m should be broad(e.g., retu r nto prison)asre c ommend edbyBonta and Mot iuk (1985). Usingmorespecificde fi n i t ions canre sult indiffic u lti es associat edwith pred i c t ingrareevents. Chi-squaretests wereco n du cted todetermineif thefa ilu r e rates

significa nt l y di f f ere dacros s im'la tesat diff ere nt lev els of riskand ne ed.

Inte rc orr elat io ns amo ng the ite ms on theas sessments and Cro n bach's alphawereca lcula tedtodeterminethe rel iabilityofthe inst r umen t s.

To determine thevalid ityof the inst r u ments, total ins t r ume n t scores....ere correla t e d wi t h the cr iteria . Spe cifi c ally, the Se cur it yRatingAs s e s s ment total scores were cor re l a tedwi th th e dichotomous andco n tinuo u s me a s ures

(47)

35 of recidivism.The sameanalyseswererepeated usingth e InmateRisk xsseeeeene total scores .

The fi nalanalysis tha t was conductedascertaine d the deg r ee of overla p betwe en theSe c ur i t y Ratingand Inma te Risk Ass essme nt s . The cor r el ationbetween the totalri skand ne ed scoreswascalculated.

(48)

ClIAPTER4 RQsults

(49)

37 The numbersof inmate s at each institution wh o compl e tedone orboth assessme ntsar eshown inTa b l e 4.1.

App ro xi ma t e l y threeti me s as many securityRating As s e s s me nt s were obtai ned in compa riso n toInmate Risk As s e s s men t s.Recall thatthe Secu rity.~ati ngAs s e s s men t wa s completedonevery inmate except in cas esof parole re voca t i on s, int er mitt ent sente nces ,andrema nds. In contrast,the Inmate Risk As s e s s men t was completedon ly on inmates whore c ei ve d a sentenceof 93 days or longe r,thu s excludingthose serving intermitt entsente ncesand thos e incarceratedfor parolerevocations.

Thesmall es t sa mpl e sizewas obtained at the Labrado r CorrectionalCentre. In comparisonto the West Coast correctional Ce nt r e and He r Majesty 's Pe nitent iary , the La b r ador Correc tionalCentredoesnot havea hightur nove r rateandtheins ti tut i on cannothou s e as many inmates .

4. 1Re cidiv jsmRate s Overall, 132 (18.57%) ofthe 711 inma t e s were r-ec IdlvLa t.s , Therec idivists 'numberof viol ation-fr e e days ranged from1to181 days (t hemaximum poss i bl e Was 182

(50)

if

~ ~

. "

e

.

£ ~ :

~

~~

, '"

~,

;i :

~

!j

E

~~

!

~

.

~ !

oe

.

f

~ ! i .

s

~

~

(51)

39 days) wi th a ..ea nof 91day s (SO..55.7 3)1.

Table4.2 sho ws there ci d iv i smratefo r each ins t itutionseparate ly. The recidivismrates we r e sign i fica nt lydifferent across !rlstit uti ons, x~(2 ,li=z 711) ..13.66, 12<.05 . Thethreefollow·upcomp ar i s o n s were als o signifi c ant.The re cid i vismrateat th eWe st coas t cor r e c t i o nal Cent rewas lowerthantherateat eitherthe Labrador Co r re ctionalCent r e, XZ(l,N" 302) ..13.66, g<.05 , or Her Maj esty's peni te ntiar y , X:( l,H..622 ) ..

6.45 , Rc .05.There cidivism rateat HerMajesty's penit entia r ywas lower thanthat at the Labrado r Cor re ctiona l Cen t re, X:(l , H= 498) - 3.84,P<.05.

Themeannumber ofvi ol ati on-fre eda ys for the recidivists was100 days at theWest Coa stCorrect i on al Cen t r e,81daysat the Labrado r Cor rec tionalCe nt re, and 92 da ys atHer Majesty's Penit e ntia r y. Thesemeans were not signi f i c a nt lYdifferent,[(2, 131) ""0.79, R> .05.

1. A score of 18 3signifiedthat an inmatewa s a non-recidivi st.Therefore, the maximum numberof viol a ti on-f r e e daysa recidivist couldha vewas 18 2da y s.

(52)

Ta b l e 4.2

Perc ent ageof Re cidiYi:! t sat each lostitiltjon

TotalNumber

Inst itution of Inmates

Wes t Coa stCorrectionalCentre 213 Labrad orcorrectionalCentre 89 HerMajes ty'sPen ite ntiary 409

••

, of Rec i di v i sts

11.74 29.21 19.80

(53)

"

4.1.1Re ci diylsmRatesAcrQ sssecurit y RllttngI.e y e 15 Despite thelac k of varia bilit y in the null.ber of inmates at eachse c urity leve l across inst i tutions as weU as wit h ininstituti ons , chi-squ are testswerecon duc t ed to dete rmi ne if therecidi vi sllrate sdIf f ered across the se cur i t ylevels.

Ta ble4.J sho ws the number of recid ivistsacross ins ti t u t ions andwi thin ins ti t u t io ns fo r eac hse c u rity ratingle vel. Acr o ssall instituti ons,therewere fewer recid ivists at the mini mum se c ur i ty level (I?8\:) than at theme d i umsecurit y level (32.4%:),X.~( l ,Ii.,127) =z4.92, J:l:<.05.Althougha gr e a t er number of mediumsec uri ty inmat e s were re c id i vists, the ir mean nmabe r of viol a t i on-fre edays was slightlyhig h e r tha nit wasfor min imumsecurity in1lates, 96 VB.90daysres pect ively . Thi s wasnot asignific a nt dif f erence,f(l, 126) ..0. 1 11 ,

;g>.0 5.

Recidi visllrate sacross the secu r it y leve l s we re als o exe afn e d for eachins t itut i on. At the West Coast Correct ionalCe ntre a comparisonwas notpossi ble because the25 recidiv ists were all in theminimum secur itylevel.

At the LabradorCor r e c tio n alCen t re , ther e werefewer recidivistsat the mi ni mum securityle vel (23 . 38\) than at the med iumsecuritylevel (66.67\1,X=(l, H. 26) ...9.41, 12<.05 .

(54)

Table 4.3

RecidivIs m Bntes foteachse c urity Ieye] rpr each

security Le ve l

Minb:lulll Med i um Maxilium

Institut i on D

Westcoast Correct i ona l

Cent r e 25 11. 85

Labr a dor Correctlona1

Centre 18 23.38 66.67

Her Ma jesty's

Penl te ntiary 72 20.17 17.39

Total 115 17.83 12 32.43

~No inmateha d a raaxieu msecuri tyra t ingat the We s t coa st cor r ectiona l Cen t reor Labr ador Co r rect i ona lCe n tre and one inmate wasmaximumsecu rity at Her Maj e s t y' s Pen!tenti ar y.

42

(55)

43 At Her Majestyls Penitentiary,72 of the 357 inmates (20.17\) inthe minimumsecurity level and 4 of the23 inma t e s (17.39%) in the mediumsecurity level wer e recidivi sts. Thiswas not a si g nif i ca nt difference , X"( l,N '"76) ". 0.1 0, I:!>.0 5 .

Itsh o u l d be note dthat. the small number of maximum security inmatesdoes not reflecta deficiencyin the sampling but rather, the natureof the institutionsthat were studied.

4.1.2 Re c id i y ismRate:'! AcTQ55 Inmate Risk leyels

To determinewhethe rthere were significant differences in recidivismacross risk levels within .tnstitut !ons and combinedsamples, chi-squaretests were conducted.

TablCl4.4 sha w5:thenumber of rec i di v i s ts across inst itutionsand within institutionsfor each inma t e risk level. Across institutions , nolow-riskinmates, 15of the 87 (17.24%) medium-riskinmates, and27 of the114 (23.68%) hi g h- riskinmates were recidivists.Thiswas a significant difference,X~( 2,H""233)..9. 5 4, J2< .05. Two of the fo l low- upcomp a ri s o n s wer esi gni f i c a nt . The rec idivi smrates were significantlydifferent between thelow andme d i um and low and high-risk levels , X~( l,l:i '"119) '"6.31,11<.05 and X.'(l, N· 146) = 9.3 0 , g<.05 respectively . However, the recidivism rates at the mediumand high-risk categorieswere

(56)

Table 4.4

BflCld iylsmEate s fo r eachRiSk lev el for ea ch In stl t u tign 44

Risk Leve l

Low Med ium High

Inst i t ution n

West Coast Correctiona 1

Centre 10. 0 0 12 19 .05

Labrado r Correctional

Cen tre 55.56 J1. 03

HerMa j es t y' s

Pen!ten t i a r y 22.22 27.27

Total 15 17.24 27 2].68

(57)

45 not significantlydifferent, X)(l , H· 201)..1.2.4,~ >.05.

Themeannumber ofviolation- f reedaysfor the medlWll and high-riskre c i di vi sts was 75 an d96d15ysrespectively.

Thiswasnotasi g ni f ica ntdif f er enc 19 , f( l ,41) . 1. 2 9,

s> .05.

Recid ivismrates across the inmate riskle v e l swe r e alsoexaminedfor each inst itution separately. AttheWest coas t Cor r ectiona lCen t r e,6 ofthe 60inma t e s (10\ ) at the med i u m-risk level and 12of the63 inmates(19 . OS%)at the hig h -rlsk le vel were recidiv ists.ThiswasII si gnif icant diff erence , X1{2. Ii. 14 4)- 5.81, g ( .0 5 . Fol low·u p comparisons showe d that the signi f icantdi f f e r enceis attributa b letothe recidivismrates ofthe lowand high -riskcategories, Xl(I,H..84) - 4.6 7 . R< .05. There wer e no significa n tdifferencesbetweenth elowand w.edium-r i s k categories andth eEedi u. and high- risk ca t e g orie s, ~:·(1. H. 811..2.27, g >.05 andx:';:Cl,H- 123 ) ..2. 0 1 , 11> .05res pect iv el y .

At the LabradorCorrec t ional Centre . 5of the9 inmates (55. 56\) inthe medium- risk leval and9of the 29inmates Pl. oJ\) inthe high-ri s klevelwer e rec :td i vi sts.This was not asigni f i cantdi f f erence,x1(2. l:I" 144)- 5.15, n> .05. Low,med i um, andhigh- ri sk Ineaeee wer eequ a lly like lyto re cid ivate .

At Her Majesty's Penitentiary. 4of the18 Inea tiee

(58)

46 (22.22\ ) inthe medium -riskleve l and6ofthe22inmat e s

(2 7.27\) inthe hi gh-r i s k le velwe r e re c id iv ists.Th is was

nota si g nifica n t dif fere n ce, X.'(2,H=45)=1.75, R>.05 . Low, me diu m, andhi gh-r i s k inmateswere equally likelytorecidivate .

4.2Rela t ions h i p Betw e e n th esecurityRating an d Inma teRisk

The corre lation be t we e n sco resontheSecuri t yRati ng Assessmentand the Inmate Ri sk Assessmentshowedth a t there wa s si g n i f ica nt overla pbe t ween thetwo ins t r u me nts, r( 2 0J )= .53,Q <.05. In add ition , there wereno signifi c antdi ffere ncesinthis re ga r d among thethr ee institu tions, X"(2,M= 203)=2.11 ,a> .0 5.

4.3 securityRatingleyels

Tab le4.5 ehovs that across thethree inst i t u tio ns , the ma j ority of inma tes(9 4 . 4 4 \ ) were ratedasminimum securit y.

Security ra ting scoresrangedtram0to25 wi th a mean of B (.s..D=4.16).An examinatio nof theinstitutions separate ly showed that 99. 06',86.52%, and 93.70\of the inmat eswere minimum secu r i t y at the We s t CoastCorrect i on a l Ce n t re , Labrado r Cor rec tiona 1 Cent r e , an dHerMa jestyIspenitentio r y resp ectively. Si g nif icant lymore inma t e s were mini mum securi t y than medium securi ty,

xra.

H= 682)

=

19.91,

(59)

Tab le 4. 5

NlImbe r gfImil1"esateaCh Inst itutipnat theMlnimY.m...

Medium andMaximum se curityl ey elasDeter m i n e d bythe Secu r i tyBatingAssessment fN ..683)

47

securityLe v e l

Minimum Med i u m Maximum

Inst itution n n

West Coast Correc tional

Centre 211 99.06 <1

Labrador Cor rec t i o na l

Ce nt r e 77 86.52 12 13.48

Her Majesty 's

Penitentiary 35 7 93.70 23 6.04 <1

Total 645 94.44 37 5.42 <1

(60)

48 R<.05.

The mean seor.eon the securi tyRatingAssessme n t ....as 6.69atthe West Coast Cor rectio nalcent re, 8.39at Her Majes ty's penitentiary ,and 9.88 at the Labrado r Correcti o nalCe n tre.Thiswas a sign ifi cantdiffere nce , 1:(2,680 ) .22.72, g< .0 5. The th re efol l ow-u p corrtpa risons we r esign ificant.The mean sc ore was significant ly large r at theLabrador Corre c t ionalCentre th antheWe st Coas t Correctiona l centre, t(680) '"-6.27,g <.05,andHe r Majesty' s pen i tentiary , t(680) .. 3.13.g <.05. As wel l , the meanscoreat Her Maj e s ty'sPen i t e ntia r y was significa ntly larger tha n at the WestCo a st Correctiona l Cent re , t(6BO) .. -4.9 5, gc .05.

4.4 Inmat e Risk[eyeh

Ta ble4.6shows tha t across the three institu tions, approx imatelyhalf of the inmates (48.93\)wereatth e high-ris k le vel followed by the mediuJI-r isk (37.34\) and low-risk le vel (13.73\).The inmate ris k sco res rang e d from oto45 witha me an of19 (S.D:= 9.69).

Withineachinsti tu t i on, the same pa tter nemerged . At theWestCoast Cor r ect i o nalCent re , 43.75\of the inma t e s we r e at th e hi g h-risk le ve l. Th e riskscores rang e d from 0 to 39witha me an of 18(.s.D= 8.90).At the Labrador Cor re c tio nal Cent r e, 65.91\ ofth e in ma t e s wer e inthe

(61)

Table 4.6

NumberofInma t e s ateac h Institntionat thelow Medjum and High-Rlsk[,eyel as DeterminedbytheIn ma t eRisk Assessment CN ,,1)

Inmate Risk Level

49

Institution D

Low Medium

D High

WestCoa st correcti onal

Centre 21 14.58 60 41.67 63 43.7 5

Labrador Correctional

Centre 13.64 20. 4 5 29 65.91

Her Majesty's

Penitentiary 11.11 18 40.00 22 48.8 9

Total 32 13 . 7 3 87 37.3 4 114 48.93

(62)

so hi g h-ri s k category.The scores rangedfrolll 0 to45 wi tha me anof 22 (SD... 11. 63).At Her Majesty'speniten tiary, 48.89\ ofthe inmateswere at the high-risklevel. At this inst i tu t i o n thesc ore s rangedfrom 0 to 40 withillmean of 19

(so.'" 9.29).There wasno signiti cantdifference amo n g the

inst itutio nsin thenumber of inmates ineach riskle vel , x]C4, H'" 233) - 7.8 5 , '">.0 5.

There wasillsign if i ca n t diffe rence betwe enthe me an sco r e s on theassessment acros sthe ins titut i ons, £(2,230) 4.56 , R<.05.Fo llo w- up co mpa ri s o n s sho wed th atthis sig n ificant differenceis attribu table toth e diffe re nce bet....eenthe West CoastCorrectiona lCentre and Labrador corre ctionalCentre,,t(230 )=-3.01.l)<.05.There were no significantdifferences betweenthe West CoastCorrectional Cent reandHe r Majesty'spenitentiary. andthe Labrador Cor r ectionalCentre and Her Majesty·sPenitentiary, J;(2J O)• -0.93,II:>.0 5 and.t.(2301 =z1.70,R;o..05 respecti vely .

4.5 Reliabi l i tyand va lidit yAna lyses Before proc e ed i ng wit hthe reliabi lityand va lidi ty analys e s , itshoul dbe notedthatthro ugh outthe an al ys es, the correlationsobtainedat the thr e e ins ti tut i ons ....ere con verted to a-a cor-eaandtests for sig nifi cant di fferences ....ere co ndu cted (Hays,19 8 8). In casos ....he r e sig n i fica nt differenc e s were not found , asingle correla tionwa s

(63)

calculated using the combined samples.

4.5.1Re.ilibi1tt yofth e Secur;tyBnti09 Asse ssment 4.5.1.1 All sacples combined

Table4.7 showsthe inte rcorrelatio nsamong the items on the SecurityRatingAssessment.Tho valueof Cronbach's alphawas .39 (H=429).

4.5.1.2pi ff eren c e s amo n g the insti tu t ions

Table 4.8 shows the sig nifican t differences among th e thr e e institutions.Althoughte nof the36 chi-squares (27.78 t) weresignificant, the r e isno overallconnection among the m. Therefore, no compellingin t e r p r e t a t i o n ofthe differencescan beoffered.

4.5.2Va ljdity pfthe securityRatingAssessment 4.5 .2.1Allsamples combined

Unlike thedichotomouscri terion , there wereno sig nifican t differences among the institutions using the cont inuo us criterion, number of violat i o n-freedays , X~(2, H::: 683) '"3.19,II.>.05.Howeve r, thecorrelation betwee n thetotalsc o reandtheco nt i nuouscr i terion usi ng the combinedsamp leswas signif ica nt, r( 68 3 )- -.19, R< .05. Inmateswho had highersecurityratingscoreshad fe wer violation-freedays.

51

(64)

Table 4.7

IntercprreJatlgospfthe Iterns poth e Security Rating Assessmentfor theThree Sa mp]es

52

01 02 OJ 0' 05

Itelll (n.. 683 )(n ..693)Iu..683)(n..68 2 )tn..683 )

01.Nature of outstanding cha rg e s/cr o....n

appeals .02 .0 2 .0 9 .0 6

02.Severity of' present

offe nc e .02 .34 .12' .08

OJ. Length

of sentence .0 2 .34 .IS· .09'

0' · Natur e of

pr i o r af!ances .0 9' .12' .I S· .08 05.Recordof

es ca pe s or

.09' .OS'

at t e mpts .0 6 .OS' 06.Historyof

vio l e nc e .02 .07 .0 9 ' .32' .0 1

07 .Ag e .01 .17' .U' .13' .0'

09.Pre-trial

sta tus .0 ' .2 1' .13 .14' .0 1

PSY. psychiatric

stab il i t y -.05 -. 01 .0' -.01 -.0 3

(tablecontinues)

(65)

Ta ble4.7 (continUed )

Int e r c orre l a t l on s of theItems 00 the security Rating Assess ment fo r theThreesarnn]es

53

Q6 Q7 Q' PS,

Item (n""683) (n" 683) (n" 683) (n .. 429)

Q7. Age -.1 5 .OS' -. 0 9

Q9. Pre-trial

status .03 .0 8 -.10'

PSY. Psychi atric

stability -.08 -.0 9 -.10'

n' .0 5 .

(66)

Tab le 4.8

Oi t f er e n cps Among theTD!it i t ll ti gD S go theSec u ri t yRatingTnt ercpr re Ja tio os 5'

WastCoast Lab ra d o r Her

Cor recti o na l co rrecti onal Maj estyts

Items Cont re centre Penit e n ti ary x'

Natureof outstanding

char q8s/ c rownappe a l s .22' -.06 .01 8.030'

lind re co rdof esc apes

or attempts In..213) (n .. 89) (n"" 381 )

Severit yof pres ent

.' 2

.47' .28' 7.674'

offen ce and leng t h

ofsent e nc e (n .. 213) tn..89) (n.. 38 1)

Severity ofpres e nt .30' .0' .03 11.96 9

of fen c e an dnat ure

otprior of f ence s (n" 2nj tn..89) (n.. 381)

Sever i tyotpresent -.05 -.05 .19' 9.44 5"

of fenceand re c o r d

or es c ape s or atte mpts

rn ..

213) (D.. 89) (n...381) (tab l e continues)

Références

Documents relatifs

high school libraries is being conducted under the supervision.of Or.. FizzaI'd, Division of

we~e Qot _available.. The subjecis 'were fnstj.~cied to observe two people performlng.. Subjects 1n this .study .were led to belle,; they ~were ,part1tipattng fn an unpaid' study

The settlement, though, was more than an economic ideal; the move to mission land demanded of the Eskimos not only a change in their food-gathering habits,

c- Classify these words according to their sound: steamed- added- helped- walked?. You are a Journalist and you have to write an interview with

Local political leaders (the descendams of the Subba and Koirala families) extended tlle canal mainly to secure votes in the local election, and for financial

Food costing of items from the NNFB was undertaken at eight food stores in towns throughout the Bonne Bay region between August 17 and 28 2009, including: two stores in Trout

They are huge snakes or lizards with wings and terrifying claws.. They breathe

demonstrated that countries with a higher level of freedom in terms of political rights or civil liberties have greater resilience to maintain health and well-being after the impact