University of Calgary
PRISM: University of Calgary's Digital Repository
Graduate Studies Legacy Theses
1997
The conditional deterrent effect of incarceration on a
sample of Canadian young offenders: a survival
analysis
Smith, Nicole
Smith, N. (1997). The conditional deterrent effect of incarceration on a sample of Canadian
young offenders: a survival analysis (Unpublished master's thesis). University of Calgary,
Calgary, AB. doi:10.11575/PRISM/14768
http://hdl.handle.net/1880/26861
master thesis
University of Calgary graduate students retain copyright ownership and moral rights for their
thesis. You may use this material in any way that is permitted by the Copyright Act or through
licensing that has been assigned to the document. For uses that are not allowable under
copyright legislation or licensing, you are required to seek permission.
Downloaded from PRISM: https://prism.ucalgary.ca
THE UNIVERSITY OF CALGARY
The Conditional Deterrent Effect of Incarceration on a Sarnple of
Canadian Young Offenders: A Survival Analysis
by
Nicole Smith
A THESIS
SUBMITTED TO THE FACULTY OF GRADUATE STUDIES
iN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF MASTER OF ARTS
DEPARTMENT OF SOCIOLOGY
CALGARY, ALBERTA
SEPTEMBER 1997
O Nicole C. Smith 1997
National Library 1*1 of Canada Bibliothèque nationale du Canada
Acquisitions and Acquisitions et Bibliographie Services services bibliographiques
395 Wellington Street 395. tue Wellington Ottawa ON K l A ON4 Ottawa ON K1A ON4 Canada Canada
The author has granted a non- L'auteur a accordé une licence non exclusive licence allowing the exclusive permettant à la National Library of Canada to Bibliothèque nationale du Canada de reproduce, loan, distribute or sell reproduire, prêter, distribuer ou copies of this thesis in microfonn, vendre des copies de cette thèse sous paper or electronic fonmats. la forme de microfiche/^ de
reproduction sur papier ou sur format électronique.
The author retains ownership of the L'auteur conserve la propriété du copyright in this îhesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts fiom it Ni la thèse ni des e;un-aits substantiels may be printed or otherwise de celle-ci ne doivent être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation.
ABSTRACT
This study examines the effects of incapacitation upon the onset and persistence of
young offenders' criminal careers. Deterrence research suggests that incapacitation can result
in partial or complete desistence fiom criminal activities. In contrast, Iabelling theory
indicates that increased recidivism results fiom incarceration. Police correctional data was
used in the d y s i s . The data set inchdes the exact dates of police contact, admission and
release dates fiom a secured facility and time fiee in the community to examine these
conflicting propositions. Using survivd statisticai modeIs, this analysis estimates whether
or not an offender will re-offend, and if so, the duration between criminal occurrences.
Findings suggest that criminal career research and longitudinal data are useful for
understanding the conditional impact of incapacitation on different types of young offenders.
This thesis would not have been possible without the help of several individuals. The
first of whom is Dr. Bruce Arnold. Without your patience, guidance, and support 1 know I
could never have completed this research. 1 cannot begin to express my gratitude for
everything you have done.
1 aiso wish to express th& to Dr. Louis Holscher and Dr. Alan Smart for sitting on
rny cornmittee. 1 greatly appreciate your time and advice.
Additionally 1 am grateful to the Caigaq PoIice Service, the Solicitor General
(Ottawa), the Social Sciences and Humanities Research CounciI of Canada (Grant #4 190-96-
1165), the Research Unit for Socio-Legal Studies, and the Department of Sociology,
University of Calgary for their support of this project.
Finally, 1 wish to thank my famiIy and fiiends for their continued support and
encouragement,
TABLE OF CONTENTS
. . ApprovalPage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i l
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tableofcontents v
Chapter 1 : Conceptual Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.0 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I 1.1 Deterrence Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.1 Punishrnent and Social Order . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.1.2 Deterrence Doctrine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2 Labelling Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . - 7 1.2.1 Reintegrative Shaming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3 Conditional Deterrence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 1.4 General Theory of Crime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 1.5 Life Course and Developmental Perspectives . . . . . . . . . . . . . . . . . . . . 17 1.6 Criminal Careers and the Persistence of Offending . . . . . . . . . . . . . . . . . 21 1.7 Criminal Careers and the Dynamic Mode1 . . . . . . . . . . . . . . . . . . . . . . . 25
Chapter 2: Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 2.0 Longitudinal Data in Studying Criminal Careers . . . . . . . . . . . . . . . . . . 31 2.1 OfficialData . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Sampling 36 2.3 Survival Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
Chapter 3: Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 3.0 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 3.1 Table 1 : Characteristics of the Sarnple . . . . . . . . . . . . . . . . . . . . . . . . . . 41 3.2 Table 2: Survival Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 3.3 Control Variables and Conditional Deterrence . . . . . . . . . . . . . . . . . . . . 53
3.3.1 Table 3: Age At First Police Contact . . . . . . . . . . . . . . . . . . . . . . 53 3.3.2 Table 4: Age At First Incarceration . . . . . . . . . . . . . . . . . . . . . . . 57 3.3.3 Table 5: PoIice Contacts Prior to First Incarceration . . . . . . . . . . 61
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 4: Discussion and Conclusion 67 4.0 Theoretical Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 4.1 Persistence Or Desistence of Offending . . . . . . . . . . . . . . . . . . . . . . . . . 69 4.2 Criminal Justice System Policy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Conclusion 74
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References 75
1
THE CONDITIONAL DETERRENT EFFECT OF INCARCERATION ON A
SAMlPLE OF CANADIAN YOUNG OFFENDERS: A SURVIVAL ANALYSIS
CHAPTER 1: CQNCEPTUAL FRAMEWORK
1 .O MTRODUCTION -
Many youth wilI be involved in minor and iirnited offending during adolescence, but
there is concern that some young offenders are becoming increasingIy violent high rate
offenders. This is apparent in newspaper headlines such as "Killer teen gets 20 months"
(Calgary Herald, August 3, 1996). The judge who sentenced the young offender in this case
claimed to be lenient because "he feIt there was a very good chance of rehabilitation"
(Calgary Herald, August 3 1, 1996). The 15 year old girl received nine months closed
custody and 1 1 months open custody. Imprisonment of young offenders is often believed
to have an effect on behaviour. But, it seems that for sorne young offenders incarceration
does little to change their offending behaviour. We therefore need to examine and test,
theoretically and empirically, the differential impact that formai legal sanctioning can have
on the criminal careers of young offenders, so that policy decisions regarding incarceration
can be evaluated and if necessary modified.
Incarceration is an attempt to modiQ crirninal behaviour and "frequency, seriousness,
and career length are thought to be afTected most directly by career modification"
(Gottfiedson and Gottfredson, 1994: 445). "Participation perhaps may be affected most
directly by prevention or very early intervention" (GotdÏedson and Gottfiedson, 1994: 445).
This research therefore looks at incarceration early in the course of offending as a possible
2
catdyst for change in criminal activity among young offenders. In addition to retribution,
cornmitting juveniles to a secured custody facility is intended to reduce criminal activity and
possibly prevent future crime. In Canada some scholars maintain that punishment through
incarceration shodd only be used sparingly (St-Amand and Greenberg, 1996). Yet, in
Canada in 1994-1995 there was a significant increase in the use of imprisonment
(approximately 15%) (St-Amand and Greenberg, 1996) of young offenders since 1990- 199 1
(the sarnpling date of this study). Of the youth who were found guilty in court, 34% were
given custody orders. Slightly less than half (47%) of these were given secured custody
dispositions (St-Amand and Greenberg, 1996).
As the use of incarceration for juveniles is increasing it becomes increasingly
necessary to study its effectiveness in controlling criminal activity. Many studies (for
examples see Gottfredson and Gottfredson. 1994; Keane, Gillis and Hagan, 1989; Sherman
and Berk, 1984; Visher, Lartimore and Linster, 1 99 1 ; Walker, Farrington and Tucker, 1 98 1 )
have tested whether punishment, through formai sanctioning, can change offending patterns.
Results of these studies indicate several possibiIities. Some research agrees that
imprisonment does prevent future criminal involvement, but most stipulate that this is
dependent on several individual and situation specific variables (see Sampson and Laub,
1993). This research examines ifand when young offenders return to custody following a
period of incarceration. It is expected that the probability of recidivism wiIl depend on
factors such as the age of the offender and previous involvement in criminai activity and with
the criminal justice system. Two general groups have been defined for study. One group
consists of serious chronic or habituai offenders, while the other p u p is composed of young
3
offenders whose participation in crime is limited and transient (see section 2.1). By
definition these two groups should exhibit different histories of cnrninal behaviour, but upon
further examination 1 expect to not only find differences between these two groups but within
each of them as well.
1.1 D E T E W N C E THEORY -
1 1 PUNISHMENT AND SOCIAL ORDER
Society constructs social norms defining the d e s and behaviours to be accepted and
practised by its members. If these are violated, individuals and groups will react in an
attempt to re-establish social order. Punishrnent thus becomes a consequence of crime in an
effort ta realign behaviours of the community or society. Negative sanctions are also
imposed by individuals or groups to instill a desired response among the population.
Irnplementation of negative sanctions may often, but not necessarily, take the form of legal
punishment. Our laws define mle breaking or illegal behaviour and give specific members
of society the power to enforce conformity to these laws. The threat of legal punishment, as
"prrscribed and adrninistered in accordance with the law" (Gibbs, 1975: 325). is intended
to establish the perception of consequences among the population; consequences that will
deter individuals fiom breaking laws. Gibbs (1975) defines punishment as an action that is
intended to inflict pain or the fear of pain, but additionally punishrnent must be perceived by
society's members as painhl or as threatening the infliction of pain.
In Canada, the Young Offenders Act (YOA) was implemented in order to further
regulate behaviour among our youth population. This legislation provides the legal
framework through which adolescent behaviour can be monitored and modified. It defines
4
illegal behaviour and gives the means through which to control and punish any behaviour
that may deviate fiom what the population has defined as legal. The YOA legally constmcts
behavioral noms for adolescents. The threat and administration of legal sanctions is a
socially constructed mechanisrn through which conformity is sought. When adolescents do
not conform, legislation allows society to punish behaviour that is deemed to be
unacceptable. Additionally, knowledge arnong juveniles that they can be punished further
encourages them to conform.
1.1.2 DETERRENCE DOCTRINE
Punishment enables an individual to offer retribution for their offence. But, in
addition to retribution, punishrnent and the threat of punishrnent is intended to prevent
further cnminal activity among both punished offenders and the larger community.
Punishment, and the threat of punishment, are intended to maintain social order through
deterrence. The deterrence doctrine States that detemence "refers to any instance in which
an individual contemplates a criminal act but refrains entirely fiom or curtails the
commission of such an act because he or she perceives some risk of legal punishrnent and
fears the consequence" (Gibbs, 1975: 326). Detemence is thereby the perception of risk and
a subsequent change in behaviour. Therefore punishment and the threat of punishrnent, as
mandated through the YOA, present a means through which deterrence may occur.
Underlying the deterrence doctrine is the conceptuaiization of individuals as rational
beings. Deterrence is based on the principle that actors perceive the risks and benefits
associated with criminal activity and determine their course of action by weighing the
outcome of crime. This proposition was explored by Ehrlich's (1973) economic study of
5
illegitimate activity. Ehrlich (1973) claims that even if individuais do have a general
preference for crime, they should still respond to incentives. Deterrence thus becomes a
decision making process whereby the costs and benefits of both legitimate and illegitimate
activities are evaluated, rather than considering oniy the negative impact of sanctioning.
Based on perceptions of both reward and risk, individds c m then decide on a course of
action.
Deterrence theory generally divides deterrence into two categories: specific
deterrence and general deterrence. Traditionaily, specific deterrence has referred to the
effects of punishment or sanctioning on the subsequent behaviour of a punished offender,
whereas general deterrence refers to the impact of sanctioning on potentiai offenders
(Paternoster and Piquero, 1995). Specific deterrence deters through actuaI punishrnent. In
contrast, general deterrence involves the perception of punishment by the generaI public
through the example of others who have been punished. Some researchers have suggested
that these two distinctions do not comprehensively defrne deterrence. For example, Stafford
and Warr (1 993) suggest that specific and general deterrence should be reconceptualized to
include punishment that offenders were able to avoid. They suggest that deterrence is
dependent not only on the crimes for which individuals are punished but the crimes for
which punishment is avoided.
Gibbs (1975) aiso suggests that the conceptualization of deterrence needs to be
broadened. In addition to the categanzation of specific and general deterrence, a distinction
needs to be made between absolute and restrictive deterrence. Absolute deterrence refers to
entirely refiaining from crirninaf activity based on fear associated with apprehension for that
6
crime. Restrictive deterrence, in contrast, encompasses contemplation of crime and while
the individual does not refraui entirely frorn crime they do curtail commissions of that crime
based on the fear of punishment (Gibbs, 1975). The distinction between completely
terminahg criminal activity and restricting criminal activity following punishrnent, allows
research to broaden the range of behaviour that is associated with deterrence. Sherman,
Schmidt, Rogan, Gartin, Cohn. Collins and Bacich (1991) conducted a study on domestic
violence that suggests the consequences of arrest are variable. This finding would suggest
therefore that we need to consider a range of behaviours following incarceration if we are to
get a better understanding of deterrence.
Other researchers, such as Wilson (1 985) and Murray and Cox (1 979, further support
the need to encompass a range of behaviours in the definition of deterrence. These
researchers suggest that if only absolute deterrence is included in the definition of deterrence,
we will not get a compiete picture of behaviours that result fiom punishment or the threat of
punishment. Measuring oniy absolute deterrence results in a passlfail categorization,
whereby a reduction in criminal behaviour will go unrecognized. Therefore in this study
deterrence has been conceptualized as not returning to custody during the penod of
observation or as returning to custody but after a significant period of time free in the
community. Including both possibilities in the definition ailows us to expand the use of
deterrence research beyond a yesho type of question. Cornparing degrees of change in
criminal behaviour provides a detailed observation of how punishment effects crime.
Restricting criminal behaviour may be indicative of changing perceptions of the threat or risk
associated with criminai activity. Reduction in offending arnong offenders is therefore
7
representative of deterrence. Consequently this research includes the time it takes for an
offender to return to custody in its mode1 of deterrence.
Measuring both restriction and termination of crirninai activity provides a descriptive
look at the behaviours that follow formal legai sanctioning. This research studies the effect
of legal sanctioning on future behaviour in six month intervals throughout the observation
period. Therefore short term variability rnay be observed (Horney, Osgood and Marshall,
1995). This smdy consequently describes not only whether or not an individual returns to
custody but how long it may take for them to do so. Individuals who return to custody may
represent deterrence even though they do not terminate their criminai careers.
1.2 LABELLiNG THEORY -
In contrast to deterrence theory is the possibility that punishment confirms a
delinquent identity and thus perpetuates, rather than deters, criminal activity. Research in
this vein is suppo~zed by labelling theory. Labelling theory purports that punishment labels
an offender and thus maintains delinquent behaviour.
Central to labelling theory is the idea that legal sanctioning may stigmatize the
offender and move them fiom primary to secondary deviance, as defined by Lemert (Hagan,
1991). Primary deviance consists of behaviour that may cause an individual to be labelled
deviant while secondary deviance is behaviour that is a result of internaiizing a deviant label
or role (Hagan, 1991). Primary deviance is behaviour which is generally considered to be
atypical of a person's character (Rogers and Mays, 1 987). Deviant behaviow is ofien thought
to be situational or impulsive and is therefore usually tolerated (Rogers and Mays, 1987).
But, the initial criminal acts can lead to a cornmitment to crime and to a deviant self-concept
8
thus promoting the individual into secondary deviance (Lemert. 1972). Through labelling.
secondary deviance becomes a consequence of sanctioning and enforcement rather than an
action in and of itseif (Rogers and Mays, 1987). Criminal behaviour therefore becomes the
result of being identified and labelled as a criminal. "Labelling theorists maintain that
punishrnent causes the offender to be labelled a deviant by others" (Tittle, 1975: 400) and
thus may perpetuate a criminal identity and thereby criminal activity.
"From the Iabelling perspective. a deviant career does not ernerge direct!y from an
initial act of deviance" (Rausch. 1983:40). Many juveniles will be involved in criminal
activity at one point in their adolescence, but this does not mean that their criminal behaviour
will necessarily progress into long term criminal careers. Occasional delinquent acts may
be random events with no systematic explanation (Smith and Brame. 1994). However. the
reactions of others to the initial acts of delinquency may affect future criminal behaviour
(Smith and Brame. 1994). Deviance is therefore influenced by how people respond to
delinquency. In this study it is the juvenile justice system that is responding to the
behaviour. Crime. under the labelling perspective. becomes an interaction between the
offender and the agency who defines the offence. I t becomes a sequential process of
response and counter-response (Regoli, Poole and Esbensen. 1985). Therefore delinquency
can be "uncertain, situational. flexible, changeable and subject to selective perception and
response" whereby "the making of a confirnied deviant or chronic law violator is a
continuing process, not simply a single happening" (Rogers and Mays. 1987: 1 17).
Consequently. the behaviour of the juveniles in this study rnay vary considerably following
punishment. Some offenders may intemalize the delinquent label and alter their self concept
9
while others rnay dernonstrate a different response to incarceration by tenninating their
cnminal involvement.
It is further specuiated that Iabelling closes off legitimate options and thereby
perpetuates the deviant identity. Labelling establishes a pattern of deviant behaviour as it
is thought to "lead to further deviance because the labelee is faced with dwindling
nondeviant alternatives, because deviance begins to converge with Fisher] changing
concept of self' (Tittle, 1975: 400). Therefore IabeIling theory impiies that the experience
of punishrnent rnay alter the individual perceptions of self and thereby explain why
imprisonment rnay not deter future criminal involvement, but instead lead to the continuation
of criminal involvement.
According to Becker (1 963), the experience of being labelled rnay also represent a
crucial step in building a stable pattern of deviant behaviour. For instance, Klernke (1 W8),
in a study of shoplifting, suggests that legal sanctioning rnay affirm the delinquent identity
for a new offender thereby sustaining or even possibly ampli@ing behaviour. Consequently
ordering a juveniie to secured custody rnay produce further cnminal behaviour rather than
prevent it. However, Thorsel1 and Klemke (1 972), indicate that the ability of an offender to
reject a deviant label also varies. Therefore imprisonment rnay not have the same effect on
al1 young offenders.
Punishent rnay now have two theoretically and behaviourally distinct outcornes:
it cm instill fear in those who have been punished and thus reduce cnrninal activity or it can
invoke identification with a deviant identity thereby rnaintaining criminal activity. We know
that some offenders are recidivists. Regardless of the punishrnent incurred they will still
1 O
continue to engage in d e breaking behaviour. Yet. we also know that some offenders wiIl
discontinue criminal involvement. Like deterrence theory, labelling theorists recognize that
individuak will have different responses to punishment. They explain different behaviours
as possibly varying according to the social setting, the type of deviant behaviour and who
does the labelling. "There is evidence to suggest that the labelling process apparently can
function either as a negative, socially disintegrative force or as a positive, socially integrative
force, depending upon the social setting and the interpersonal circurnstances" (Thorsell and
Klemke, 1972: 396). It is expected that some offenders in ths study will return to custody
following punishment, while others will not. Those offenders who return to custody after
a short period of time fiee in the community may represent offenders who have internaiized
the label of delinquent during the process of punishment.
1.2.1 REINTEGRATIVE SHAMiNG
Braithwaite (1989) proposes a theory that uses the concept of labelling or
stigmatization and incorporates or integrates it into a process of shaming. Braithwaite's
theory of sharne and reintegration is an artempt to theoretically integrate legai and extra-legai
controls as it implies that the individual mu t be suffIcientIy tied to society in order for legal
sanctioning to be effective. The ability to deter offenders is dependent on the individuai's
ries to society and consequently society's ability to shame the offender but also to reintegrate
them back into the community. "Individuals with more social bonding are more likely to
receive reintegrative shaming and thus less likely to commit crime" (Bernard and Snipes,
1996: 3 17). Braithwaite (1989) explains that some offenders are Iabelled by punishment
while others reintegrate themselves into the community and discontinue criminal
11
involvement. Contrasting outcornes are presented as a system of sharne and reintegration
whereby some individual offenders will reintegrate themselves into the cornmunity and
others will internalize the shame and stigma of their actions and proceed in criminal roles
and identities.
If an individual decides to commit a criminal act then members of society will
attempt to re-establish compliance of that individual. The process of sharning seeks to re-
establish compliance through the moralizing qualities of social control, but the individual is
kee to accept or reject an attempt to persuade îbrough social disapproval (Braithwaite, 1989).
An individual is deterred fiom subsequent criminal behaviow if they are persuaded through
social disapproval, but in contrast an individual is also fiee to resist sharning and will thus
continue to be involved in criminal activity.
It appears that in order to provide a comprehensive description of offending
behaviour following incarceration a range of behaviours needs to be considered. Young
offenders may seek reintegration into the cornrnunity following incarceration or they may
resist what has been defined to be acceptable behaviour and engage in further delinquency.
Like deterrence and labelling theory, Braithwaite's theory of sharning and reintegration
suggests that individuals are uniikely to respond to punishrnent in a uniform and predictable
manner. The question then becomes how to identifj who will respond to punishrnent, how
they may respond and what conditions lead them to respond as they did. It becomes apparent
that many cntena need to be considered and evaluated in order to determine the outcome of
punishrnent. Offending needs to be exarnined as a process or as a sequence of events,
whereby young offenders and their behaviows are examined as they develop and evolve.
12
Therefore this study looks at criminai histories of young offenders as ordered events which
may influence fùture events. By using longitudinal data to evaluate criminal behaviour, this
research begirs to identifi possible variables that may be associated with persistent
O ffending following punishment.
1.3 - CONDITTONAL DETERRENCE
Throughout this discussion, it has been suggested that different behaviours can result
fiom imprisonrnent. An individual can either be deterred or labelled by punishrnent. Yet,
neither of these theories explains why or under what specific conditions some individuals
will be labelled while others will be deterred. Along with the criminal act itself, conditionai
deterrence suggests that individual characteristics and circurnstances need to be recognized
as part of a process of crime and crime prevention.
It has been suggested that legal sanctioning, a forma1 control mechanism, impacts
behaviour of offenders (Gottfiedson and Gottfredson, 1994). Other theorists. for example
Gottfiedson and Hirschi (1 990) contend that extra-Iegd or informal control factors may be
the determinants of criminal behaviour. as opposed to legal control. Debate has occurred
over which plays a more significant role in predicting behavioral outcomes: legal or extra-
legal controls. In considering only legal or forma1 controls, only effects of legai sanctioning
or the threat of legai sanctioning are determined to have an influence on criminai activity.
In cornparison, some theories (Gottfredson and Hirschi, 1990) rely primarily on extra-legal
factors, such as self control and effective parenting, to describe criminal behaviour.
According to Sherman, Smith, Schmidt and Rogan (1992), the debate between legal
and extra-legai controls ignores the possibility that the explanation of criminai behaviour Iies
13
in how the formai and informal controls interact. "Such interactions may be far more
powerful predictors of future crime than either legai or extralegal controk viewed in
isolation" (Sherman, Smith, Schmidt and Rogan, 1992: 363). They M e r suggest that this
is important in research that addresses questions of specific deterrence. If such an approach
is considered then it precludes the question of simply whether or not imprisonment can
change criminai behaviours. Instead this approach asks for whom, in which situations, and
at what point in their criminal careers might imprisonment change the criminal behaviours
of young offenders.
One way to begin to integrate Iegal and extra-legal controis is to consider the
conditional deterrence hypothesis. This hypothesis claims that "legai threats deter oniy those
potential offenders who are sufficiently tied to conventionai society to s a e r from its adverse
reaction to legai sanctioning" {Sherman, Smith, Schmidt and Rogan, 1992: 363). M i l e this
study does not encompass many of the extra-legal control variables that wodd be needed to
give a complete description of the effect of irnprisonment, it does provide some key factors
useiùt in describe offending behaviours. Incorporating both legal and extra-legd controls
into a mode1 of delinquency integrates centrai arguments from both perspectives.
Conditional deterrence estimates the effect of incarceration given specific individuai
and situationai characteristics unique to the offender. The conditional deterrence hypothesis
thereby ailows us to study juvenile delinquents as being distinct from other juveniles and
fiom other adult offenders. Additionally, within the classification of juvenile delinquent,
subgroups of offenders can be studied. This is significant as we cm now address the
differentiai impact of punishment on different Ends ofjuvenile offenders. For exarnple, this
14
study addresses how age and prior exposure to formal social control may influence the
relationship between punishment and deterrence. It is expected that characteristics of the
offender, such as the age at first police contact. wilI impact the eficacy of incarceration.
1.4 - GENERAL THEORY OF CRIME
To gain insight into the impact of formal sanctioning on deterrence, it appears that
a wide array of offenses, offenders, and theoreticai perspectives must be evaluated. Sherman
and Berk, suggest that a "careful accumulation of findings fiom different settings will help
us differentiate the variables which are crime- or situation- specific and those which apply
across settings" (1984:262). Along this line of reasoning, this project seeks to describe
variables as they relate to the incarceration experience of offenders who are between the ages
of 12 and 1 8. The research describes whether or not different patterns of incarceration and
police contact can explain variation in the behaviour of individuais following incarceration.
As well, it explores the impact of age on the probability of reincarceration.
Conditional deterrence offers one plausible explanation as to why offenders differ in
their criminal behaviours. Gottfredson and Hirschi (1 990) offer another theory of crime that
challenges this and other theoretical perspectives. Essentially, their general theory of crime
proposes that criminal propensity is a stable and enduring phenornenon which is developed
at a very young age. Therefore, individual cnminal tendencies or propensities are persistent
over time. This theory is based on the premise that ail criminal behaviour, whether it be
white collar crime, murder or shoplifling, can be explained through a limited set of variables
correlated with self control (Dean, Brame and Piquero, 1996; Paternoster and Brame, 1 997).
Gottfiedson and Hirschi, build on ideas based in classical theory. They agree that the
15
cost of crime depends on the "individual's current location in or bond to society"
(Gottfredson and Hirschi, 1990: 87). This is supplemented by individual self control or
restraint. As a result, Iow self control becomes the explanatory factor behind crime. For
people with Iow seif control, satisfying their wants becomes a priority that may not be
controlled by extemal bonds to society or the individual's current location. Criminal acts
rnay thus come to be a quick route to filfilling their wants. People who lack in self control
tend to be "impulsive, insensitive, physical (as opposed to mental), risk-taking, short-sighted,
and nonverbal, and they will tend therefore to engage in criminal and analogous acts"
(Gottf?edson and Hirschi, 1990: 90). n e s e traits subsequently can be used to identi@ people
who are IikeIy to be involved in criminai activity. Recognizing these traits as being the
source of crime supplies a theory that identifies criminai propensity at a young age.
Additionally, "since these traits tend to persist through life, it seems reasonable to consider
them as comprising a stable construct usefùl in the explanation of crime" (Gottfiedson and
Hirschi. i 990: 90).
However, low self control does not always nor necessarily result in crime. "No
specific act, type of crime, or form of deviance is uniquely required by the absence of self-
control" (Gottfiedson and Hirschi, 1990: 91). This suggests that the degree and range of
criminal behaviour rnay Vary among individuals. Propensity therefore, is not always directly
proportionate to crime. "Self control suggests that people differ in the extent to whch they
are restrained fiorn criminai acts" (Gortfredson and Hirschi, p.88, 1990). From one prernise,
that crime is the result of low self control, this theory can explain a wide range of unique and
variable behaviours and behavioral patterns. Due to varying degrees of self control, we cm
16
therefore expect to see considerable variation among some offenders in their cnminal
offending.
Criminal behaviour is a composition of low self control in children and the inability
of parents to control the traits or behaviours associated with low self control, or
sociologically speaking ineffective socialization (Gottfredson and Hirschi, 1990). Therefore
al1 criminal behaviour can be explained through low self control and ineffective socialization.
While credited with being a parsirnonious mode1 some would disagree with its ability to
completely predict a range of behaviows. For exarnple, the life course perspective uses some
of the elements of Gottfkedson and Hirschi's theory, but incorporate the idea that life events
can change individual criminal tendencies. They suggest that crime is dynarnic and therefore
camot be explained solely on the presence or absence of certain traits (Sampson and Laub,
1993).
If crime is in fact a dynamic process then factors such as age may play a significant
role in its development and possibly termination. Age is a key, yet controversial, component
in testing different theoretical perspectives. Age is a salient factor in many theories as it is
one of the underlying principles as to how crime and criminality differ fiom theory to theory
(Bartusch, Lynam, Moffitt and Silva, 1997; Nagin, Fanington and Moffitt, 1995; Nagin and
Paternoster, 1991). In contrast, general theorists conclude that there exists an invariant age-
crime curve (Gottfiedson and Hirschi, 1990). This curve suggests a relationship whereby
crime peaks in middle to late adolescence and rapidly declines thereafter. And, while most
research supports this finding, the reasoning behind its existence differentiates rnodels of
offending. Gottfiedson and Hirschi (1983: 554) propose that the "age distribution of crime
17
is invariant across social and cultural conditions" and addibonally that the "identification of
the causes of crime at any age may suffice to identifj them at other ages as well". In other
words crirninal propensity is stable and enduring, so the correlates of antisocial behaviour
at one point in tirne shouid be the sarne at the next measurement due to weak social controls
and tow self esteem. Criminal propensity is thus malleable early in the life course, up until
around age 8 or 10, and therefore may fluctuate (Patemoster and Brame. 1997). But once
fixed, generai theorists expect behaviours to be maintained and constant. Enduring criminai
propensities are established at a very young age and are consistent over time, therefore age
is not a determining factor of criminality beyond the early formative years. This study
examines the possibility that age does influence the probabiIity of deterrence, or that criminal
behaviour is not fxed. Offenders who are first in contact with the poke at a young age may
respond differently to incarceration than those offenders who are older at first police contact.
The same may hold true for the age at which offenders are first incarcerated.
1.5 LIFE COURSE AND DEVELOPMENTAL PERSPECTIVES -
A different approach to studying criminal behaviour is necessary, if in fact variables
relating to crirninal behaviour are specific to the crime and situation. In contrast to
Gottfredson and Hirschi, Sarnpson and Laub (1 990) theorize that although most criminal
behaviours are enduring, individuals do deinonstrate instances of change in criminai
propensity (Patemoster and Brame, 1997). Sarnpson and Laub add a dynarnic element to the
life course mode1 by relaxing Gottfiedson and Hirschi's assumption of stable criminal
propensities. They suggest that certain events, such as mariage and employment, can
greatly influence criminal tendencies. This premise allows thcm to suggest, as did
18
conditionai deterrence theory, that the effectiveness of legal sanctioning in controlling
criminal behaviour c m be dependent on other factors.
Similar to conditional deterrence, Sarnpson and Laub maintain that order is
maintained through social control whereby social ties prevent individuais from engaging in
d e breaking behaviour. Crime and deviance become more likely when an individual's bond
co society is weakened or broken (Sampson and Laub, 1993). So, like conditional deterrence
they contend that formal and informai social control are necessary to maintain social order,
but they apply this premise across the life course and hence develop a more dynamic mode1
of offending. Centrai to their argument lies the concept that these controls can be time
dependent.
The life course perspective is based on the principle that events and life pathways are
instrumentai in shaping criminaiity (Sampson and Laub, 1993). It suggests that criminal
behaviour is a fluid concept and thus evolves over tirne. Offending patterns are developed
through a series of events. Consequently, crime becomes a dynamic process. It is further
specuiated that the dynamic elements of the mode1 are partly contingent on age (Bartusch,
Lynam, Moffitt and Silva, 1997). This means that the ability of both informal and formal
social controls to prevent crime may be dependent on the âge of the individuai. Life course
thus becomes defined as "pathways through the age differentiated life span where age
differentiation is manifested in expectations and options that impinge on decision processes
and the course of events that give shape to life stages, transitions, and turning points" (Elder
in Sarnpson and Laub, 1990: 8).
Within the life course perspective there are two centrai concepts. These concepts infer
19
the dynamic components of offending. The first is trajectory. A trajectory refers to a
pathway or line of development over tirne. They are long tem patterns of behaviour that are
devetoped through a sequence of life events (Sampson and Laub, 1990). Transitions, the
second concept, are specific life events. These may alter trajectories. For exarnple, the
experience of incarceration may change an individuai's offending pattern by decreasing their
criminal activity. I t is the interaction of ~ajectories and transitions that aIlow offending
patterns to change. An important concept in the life course framework is the "dynamic
process whereby the interiocking nature of trajectories and transitions generate rurningpoints
or a change in the Iife course" (Elder in Sampson and Laub, 1993: 304).
The idea that transitions and trajectories interact creates the notion of a dynamic
model yet it does not preclude the possibility that sorne behaviours are stable and enduring.
This has been influentid in research on criminal behaviour as it piovides the opportunity to
look at both continuity, a stable element of offending, and within individuai change over
tirne, a dynamic cornponent of offending (Sampson and Laub, 1993).
Individuai change is descnbed through exarnining individual trajectories and
measuring crime at different stages of the life course. Recall that Gottfredson and Hirschi
daim this is unnecessary because even though c ime itself may Vary, criminal propensity
does not and consequently studying individual pathways does not generate any new
information. Sampson and Laub disagree and consequentIy suggest that crirninal behaviour
may Vary by age. As do Gottfiedson and Hirschi, Sampson and Laub recognize a constant
aggregate relationship between crime and age, but do not attribute it to the same causes.
Instead they believe that a declining crime rate in late adoiescence is due to social control
newly instigated through events such as employment and marriage.
Loeber and LeBlanc (1990), following similar conceptual propositions as Sampson
and Laub, also propose a dynamic mode1 of offending that incorporates within individual
change. They define developmental criminology as "the study of the development and
dynarnics of problem behaviours and offending with age" as well as the "identification of
explanatory or causal factors that predate, or CO-occur with, the behavioral development and
have an impact on its course" (Loeber and LeBlanc, 1990: 377). Like Sarnpson and Laub
they look for events that change behaviow and additionaily suggest that these changes may
be predominantly found in adolescence.
To explain problem or antisocial behaviour we seek to identify explanatory factors.
For Gottfredson and Hirschi, these remain constant throughout the life course, but for
proponents of developmental theory explanatory factors may change. "While differing in
their specific features, developmental theories al1 assume that different explanatory factors
affect crimes occurring at different points in life" (Loeber and LeBlanc in Paternoster and
Brame, 1997: 52). Consequently, we need to look longitudinally at causal factors over a
period of offending to establish whether there are in fact different causal factors based on the
age of participation.
Developmental theories and life course perspectives contend that certain events may
have a behaviorai impact on the "evolution of crime and deviance from childhood through
adulthood" (Nagin, Farrington and Mofitt, 1995: 1 1 1 ). Accordingly, childhood or early
events may have a significant impact in creating and maintaining behaviours, criminal- or
otherwise. According to Granovetter's study of occupational mobility, "careers are not made
2 1
up of random jurnps fiom one job to another" but progress and develop systematicaily (1982:
85). Granovetter (1985) concIudes that initial events are influentid in shaping a career.
Consequently, the first interruption in the offending pattern may prove instrumental in either
terminating or stabilizing criminal behaviours. This research therefore focusses prirnarily
on the first 3 incarceration periods as they pertain to the development of criminal patterns
of behaviour. In considering the first 3 events, it is expected that the firçt incarceration period
rnay have a different effect on individual behaviour than might either of the 2 successive
incarcerations.
The h three events are also of interest because we want to look at individual level
changes that may occur in a relatively short period of time. Horney, Osgood and Marshall
suggest that there have been "few attempts to look at within-individual variability in
offending over relatively short periods of time" (1 995: 658). The effects of incarceration on
young offenders may only be short term or may occur early in the life course. By observing
behaviour in six month intervals following the first three incarceration periods we may be
better able to understand how criminal behaviours either progress or terminate early in the
course of offending.
fi CRIMINAL CAREERS AND THE PERSISTENCF. OF OFFENDING
Whether crime should be studied as a stable underlying propensity or as a dynarnic
and malleable pathway has been an ongoing debate in the literature. Theories such as
Gottfiedson and Hirschi's general theory argue 'persistent heterogeneity' which maintains
that criminal behaviour is derived fiom an underlying stable criminal tendency, In contrast,
life course and developmental theories daim 'state dependence' which explores the idea that
22
crime is a dynarnic concept dependent on variable individual and social factors (Nagin and
Paternoster, 199 1 ).
Advocates of persistent heterogeneity suggest that incarceration will not impact the
criminal propensity of individuals. Gottfiedson and Hirschi, state that "because low-self
control arises in the absence of the powerfùl inhibiting forces of early childhood, it is highfy
resistant to the powerful inhibiting forces of later life, especially the relatively weak forces
of the criminai justice system" (1 990: 255). Therefore deterrence and rehabilitation are not
likely outcornes of incarceration. Nor do Gottfiedson and Hirschi maintain that incarceration
is an effective mechanism of crime control. Offenders are often sentenced based on past
records of offenses therefore a substantial amount of crime rnay have already occurred prior
to incarceration. As a result criminal justice intervention of criminal behaviour is ineffective
and is not likely to produce a long term reduction in cnminality.
In contrast to persistent heterogeneity, proponents of state dependence theory indicate
that incarceration rnay alter life trajectories. Transitions, such as incarceration, are embedded
in life trajectories and rnay generate change (Sampson and Laub, 1990). Incarceration, for
some individuais, rnay cause a change in the life course and thus criminai behaviour as it
rnay either label or deter offenders.
Both state dependent and persistent heterogeneity perspectives do not preclude the
possibility that individuais rnay Vary quite drasticaily in their frequency of offending.
Therefore, these two perspectives are not mutually exclusive. Consequently, in either
perspective, it is possible that heterogeneous subgroups of offenders rnay emerge. In the
current study, offenders have been classified as either serious habituai offenders (SHOs) or
23
nonserious habitua1 offenders (nonSHOs) thereby creating two subgroups with distinctive
offending patterns. 1 expect to fmd variation between SHOs and nonSHOs, but the analysis
will aiso address the possibility that subgroups of offenders exist within each of the SHO and
nonSHO population. It is therefore possible that offenders withui each category will respond
differently to punishment. Additionally, the categories of SHO and nonSHO should not
predehe behaviour. Although certain behaviours are more commonly associated with the
label of SHO or nonSHO, this does not preclude the possibility that nonSHOs will act in a
manner similar to SHOs or vice-versa.
While some youth rnay demonstrate patterns of offending that appear consistent with
the notion of a stable chinal propensity, others rnay not. Some youth may continue in risk
taking behaviour following incarceration, or even possibly escalate in their ofending
behaviour. In contrat, other young offenders may be deterred by the expenence of
incarceration and remain out of custody for the rest of the observation penod. Additionally,
it is possible that a subgroup of offenders exists, who have a higher fi-equency of offending
than do other offenders. Some studies, for exarnple Visher, Lattimore and Linster's ( 199 I )
research on serious youthful offenders, suggest that a subgroup of offenders may be
disproportionately responsible for criminai activity. I t would therefore be expected that this
group of offenders would respond to incarceration differently than would other offenders.
These offenders are likely to retum to custody and may do so quickly.
As noted earlier, age is an important concept in criminai career research. Whether
or not offending is considered to be an age-graded phenornenon is dependent once again on
the perspective of the research. Persistent heterogeneity suggests that criminal propensity
24
is not dependent on the age at which criminal behaviour begins. Whereas theories that use
a state dependent perspective claim that age of participation plays an important role in
establishing offending behaviours throughout the life course. For example. Moflitt's
developmental theory -'proposes that antisocial behaviour beginning during the peak age of
participation at mid-adolescence is qualitatively different in origin from antisocial behaviour
that begins during childhood (Bartusch. Lynarn. Mofitt and Silva 1997: 14). This suggests
that the age of participation may define the offending characteristics and trajectories of a
particular offender. Therefore. it is expected that offenders who begin their criminal careers
at a young age. during childhood. are more Iikely to show persistent criminal behaviours
throughout the life course than are offenders whose first police contact occurs later in
adolescence.
Age at first incarceration may effect offending patterns in a similar way as does the
age at first police contact. The age at which young otyenders are first incarcerated usually
represents one of the earliest experiences with public legal IabeILing (Farrington. 1977). or
the first time at which a deviant action is recognized as severe enough to warrant a
disposition to secured custody by the criminal justice system. That is. the tïrst incarceration
experience represents a potentially stiçmatizing or labelling public event tliat may
significantly weaken an individual's social bonds. -'One of the most crucial steps in the
process of building a stable pattern of deviant behaviour is Iikely to be the experience of
being caught and publicly labelled as a deviant" (Becker. I 963 : 3 1 ). 1 t is expec ted therefore
that offenders who are first incarcerated at a young age are more Iikely to be persistent.
serious offenders than are young offenders who are first incarcerated in later adolescence.
25
Although, if no variation is found between offenders who participate in crime at various ages
then we have strong support for Gottfredson and Hirschi's general theory and other theories
that may suggest persistent heterogeneity.
One of the few established relationships in criminology is that past behaviour is
indicative of future criminal behaviour (Fanington, 1977; Nagin and Fanington, 1995).
Both persistent heterogeneity and state dependence agree with this finding. Persistent
heterogeneity suggests it is due to the stable and enduring criminal propensities. However,
state dependent theories such as the life course perspective or deveiopmentd theories,
believe this is the result of changing structures of behaviours and opportunities. As
individuals commit delinquent acts, they possibly weaken tneir ties to society and others
thereby Iimiting the constraints of social control. Or, recall labelling theory, it too is a state
dependent theory as it suggests that an act of labelling c m alter life trajectories by
stigrnatizing individuals and identiQing them as crirninals. This study examines how events
leading up to incarceration may be predictive of events following incarceration. Or more
specifically, I expect that offenders who have a high number of police contacts pnor to their
first incarceration will continue to engage in risk taking behaviour following release from
incarceration whereas offenders with a low number of police contacts prior to incarceration
may show evidence of deterrence.
1.7 - CRiMWAL CAREERS AND THE DYNAMIC MODEL
Differences between persistent heterogeneity and state dependence research extends
to how criminal behaviour c m be M e r studied. General theorists use a cross sectional
research design. If propensity is constant then it can be measured at any point in time and
26
provide a reliable measure. Therefore Gottfredson and Hirschi in particular, cite the use of
longitudinal research as unjustified and further suggest that it will not provide any more
information than would cross sectional research (1988, 1990). Based on the premise of
criminal propensity it would be hard to justify the utility of longitudinal data. But, if
criminal behaviour is dependent on a series of events then longitudinal data is very usehl
in describing criminal activity. Researchers who advocate the use of longitudinal research
often conduct their studies using a criminal career model or components of it.
Developmental theories base their entire defirution of offending on the sequences or
patterns of offenses as they occur over time. Consequently these models need longitudinal
data, Therefore much developmental research draws on at least a few concepts or
components of the cnminal career mode[. Whether or not the use of longitudinal research
is justified is only one part of the debate concerning the use of criminal careers.
Criminal career research provides a fiamework which "describes the sequence of
offenses during some part of an individual's lifetirne" (Fanington, 1992: 52 1 ). The crirninal
career model further defines the beginning (onset) of a career, the end (desistance or
termination) of a career, and the career length (duration) (Blumstein, Cohen and Farrington,
l988a; Farrington, 1992). Additionally, criminai career research establishes the number of
offenses committed by an individual. This is called lambda.
Lambda is useful in developmental research as it identifies distinct features of
offending as well as possible subgroups of offenders. Lambda is an individual crime rate or
"the nurnber of crimes an active offender corn i t s in a unit of tirne" (Blumstein, Cohen and
Farrington, 1988b: 58). As well as describing offending rates, lambda is useful for
27
identifjing serious habituai offenders. But, Gottfkedson and Hirschi daim that the use of
lambda to identifi chronic offenders is unjustified as even the most active offenders will
decrease their criminal involvement with age (1 988). Therefore, according to Gottfiedson
and Hirschi al1 theoretical and policy implications of lambda are unwarranted. This research
seeks to examine the possibility that a subgroup of offenders is responsible for a large
proportion of crime therefore a concept such as lambda is usefiil not only in detecting change
but in recognizing heterogeneous groups of offenders. Although the fiequency of offending
will not be measured in this research, changes in the nmber of offenders who are
incarcerated will be evaiuated.
Using a criminal career frarnework, correlates of behaviour Ieading to incarceration
cari be measured at various and distinct points of an individuai offender's criminal career.
It is therefore possible to determine whether or not certain factors, such as age and prior
formal sanctioning, can alter criminal behaviours. This ailows us to research the possibility
that certain events may cause individuals to teminate their careers or possibly even increase
their rate of offending (escalation) thus affecthg their probability of reincarceration.
Changes in behaviour can only be explored through a mode1 that examines a continuum of
behaviour over a period of time.
Loeber and LeBlanc (1 990), take some of the features listed above and develop hem
into dynarnic concepts. They describe onset, terrnination and duration as boundary concepts.
These concepts define the parameters of offending. Loeber and LeBlanc extend these
concepts to describe offending as part of a dynamic career. The processes are grouped into
activation, aggravation and desistance (Loeber and LeBlanc, 1990). They suggest that
2 8
boundary concepts are cornmon elements of longitudinal research, but the developmental
perspective is underdeveloped (Loeber and LeBlanc, 1990).
In terms of incarceration, 1 am primarily interested in how impnsonment can be
described througti dynamic terms. This utilizes the criminai career framework yet
incorporates elements of other dynamic theoretical perspectives. Specifically, incarceration
can be viewed as a life event within a career that has a definable beginning, middle and end.
We can examine the effect of incarceration, as a 'passive7 element (Loeber and LeBlanc,
1990) of offending, on subsequent behaviour.
According to Loeber and LeBlanc, activation "refers to the way the development of
criminal activities, once begun, is stirnulated and the way continuity, tiequency, and diversity
is assured (1990: 382). They are attempting to describe how criminal irajectories are
established following the initial criminal act. One subprocess of activation is stabilization,
or the process by which the continuity of offending over time is increased. LabelIing theory
would tell us that crirninal behaviours are maintained through movement fiorn primary to
secondary deviance. Caspi, Bem and Elder Jr. suggest that stabikation ofien occurs because
an "individual's dispositions systematically select him or her into particular environments,
environments that, in turn, might reinforce and sustain those dispositions" (1987: 308).
Others may suggest that social bonds are weakened and thus criminai behaviour cannot be
effectively controlled. Regardless of the perspective, through longitudinal examination of
the data we cm describe whether or not incarceration acts as part of the process of
stabilization. This research explores the possibility that incarceration may sustain delinquent
behaviour thereby leading to further periods of incarceration.
29
If incarceration does not stabilize offending behaviours then it may be a component
in the processes involved in desistance. Gottfi-edson and Hirschi (1 990), suggest desistance
will occur as offenders age and therefore incarceration will not increase the likelihood of
desistance. Others, following a deterrence perspective, rnay conclude that if the punishrnent
was severe, swift and certain then it may succeed in deterring subsequent criminai
involvement. Blurnstein, Cohen and Farrîngton (1988a) argue that specific deterrence
policies are designed to intervene in criminai careers such that behavioral changes are
maintained or termination occurs early in the criminal career. If incarceration can produce
a long term change in criminal behaviour then it is important that we recognize when and
how this change is most likely to occur.
Criminal career models allow us to explore the nature of incarceration and future
offending. Each perspective discussed above provides insightful arguments and evidence
into the explanation behind criminal behaviour, its development, and its relationship with
age. M i l e the data set in the present study does not contain a11 the necessary variables to
test the validity of each of the proposed theories, it does present the opportunity to examine
offending as a sequence of events. In particular, it examines the sequential unfolding of
events. From this data set it is possibIe to discuss the potential effects forma1 legal
sanctioning may have on subsequent juveniIe behaviour. Specifically, repeated exposure to
incarceration and police contact is likely to be associated with a type of offending behaviour
and offender that is conceptually different than the offender who has only a fleeting or
limited offending history. 1 expect that offenders who are not deterred by the first
incarceration expenence are unlikely to be deterred by subsequent custody periods. As well,
3 O
the higher the nurnber of police contacts an individual has pnor to incarceration the less
Iikely that individuai is to be deterred by incarceration. Additionaily, the relationship
between incarceration and deterrence may be further rnediated by the age of the offender at
first police contact (onset) and the first incarceration experience. If either or both of these
events occur at a relatively young age. in comparison to their peers, then it is unlikeIy that
incarceration will deter these offenders fiom continuing in risk taking behaviour.
Some offenders wiI1 not return to custody f i e r the initial period of incarceration but
others will. Theoretically we can explain the different reactions to forma1 legal sanctioning.
Aspects from different theories provide different models as to why sorne offenders will
continue to engage in risk taking behaviour and others will desist. Empirically we can begin
to offer an explanation by looking at individuai juvenile criminal careers. This research
begins to explore crime as a process whereby different correlates can be examined as
possible causes of either desistance or stabilization of crime.
CHAPTER 2: METHODOLOGY
2.0 LONGITUDNAL DATA M STUDYTNG CRI - MINAL CAREERS
R e d from Chapter 1 ba t crirninal careers consist of a beginning (onset), an end
(termination) and a series of events between. Therefore the study of crirninal careers implies
the necessity of Longitudinal data. However, the need for longitudinal data has been debated
in the literature. In particular Blumstein, Cohen and Farrington (1 988) and Gottfiedson and
Hirschi ( 1986, 1988) have been debating the utility of longitudinal data in criminal research.
Blumstein, Cohen, Roth and Visher (1986) argue that longitudinal data is necessary
to study a phenomenon such as criminal careers simply because criminai careers are
longitudinal by definition. Beyond that, it is the most accurate and comprehensive way to
examine the nurnber, timing and sequencing of events as they occur over the course of an
individual's criminai career. Gottfredson and Hirschi disagree with this. They claim, as
docurnented by Esbensen and Menard (1 990), that longitudinal data has been over-rated, it
does not answer key questions in explaining criminal behaviour, and any findings could been
more easily estabiished through cross-sectional research.
The conditionai effects of incarceration on subsequent criminal behaviour implies
that criminal behaviour may change or develop over time. This data set contains repeated
observations of the same individuals thereby exarnining how cnminal careers progress and
are possibly modified. According to Wall and Williams (1 970), only longitudinal research
can provide the observations necessary to illustrate patterns of change in an individual.
Blumstein, Cohen and Farington M e r support the use of longitudinal research by stating
that the main advantage of this type of research lies in its ability "to provide detailed
32
information about the natural history and course of development of offending" (1988b: 67).
Stattin, Magnusson and Reichei, explain that "the general consensus as to the understanding
of how crime evolves in individuais stresses the need for such longitudinal data" (1 989: 368).
Additionally, Wall and WilIiarns suggest that "only the longitudinal method can give a true
picture of cause and effect relationships over time" (1 970: 8). The strength of longitudinal
research is that we can clearly identi& the timing and ordering of events which in turn allows
us to identiSl and test causal relationships.
2.1 - OFFICIAL DATA
The data set used in this study contains the exact dates of police contact and of
admission and release fiom a secured custody facility for a selected sample of 386 young
offenders. These criminal histories were collected by the Solicitor General's Office (Ottawa)
and were analyzed using survivai anaiysis. Police contact indicates the fiequency at which
police have an encounter with an individual that results in a report being w-ritten. Charges
need not be laid for an incident to be recorded. This allows for inclusion of police contact
before age twelve. Admission and release dates fiom secured custody provide information
on who was incarcerated, for how long and at what stage in their criminal careers. Official
data of this nature looks at the criminal careers of young offenders Iongitudinally thereby
exploring delinquency as a dynamic and variable process.
Official data, as recorded by police and corrections agencies, indicate the "extent to
which, and method whereby, the pubiic agencies of social control are dealing with deviance
they define" (Zay in Hagan, 1991 : 34). Police contact and corrections information illustrate
offending as it is defined and recorded by formal social control agencies. Consequently, the
3 3
data was not recorded for socid science research and thus produces certain limitations, one
of which is that official data only provide part of the offending history. Officiai agencies are
unable to observe and record dl criminai events, thus much criminal activity may go
undetected (Blumstein, Cohen, Roth and Visher, 1986; Hagan, 199 1 ).
This research helps alleviate underestirnation of the actual crime rate by including
police contact as part of individuai criminal histones. Although some criminal behaviour
will still go unrecodzed, these data present a detailed representation of the fiequency and
timing of police contact with juveniles. It thereby increases our knowledge of the potential
number of crimes occurring. This information is vaiuable as the conditions through which
incarceration may deter or label offenders cm be analyzed through the events preceding
incarceration.
Many factors may contribute to the progression or termination of criminal careers.
Things such as aicohol, drug use, education and socio-economic status may al1 play a
significant role in our understanding of criminal behaviour. Yet none of these variables are
available in this data set. Data pertaining to variables such as those described is ofien best
obtained through self-report data. However self-report data has limitations. Offenders may
not be abIe to accurately recall criminai acts thus affecting validity and reliability.
Additionaity young offenders can be a difficult population to identie and follow up. While
official data does not consider some potential sources or correlates of delinquency it does
provide important data that otherwise would be unavailable. This data set contains exact
dates of police contact and incarceration. This ailows us to calculate the probability of an
offender retuming to custody and fürther we can estimate when they are most likely to be
reincarcerated. So even though there are other variables that may be related to delinquency,
official data provide important information that allows for an assessment of risk and the
timing of reincarceration.
A concem with official data. or more generaliy any secondary data source, lies in the
quality of the data (Kiecolt and Nathan, 1985). For instance, there exists the potential for
miscoding or nonrecording of data. In some instances errors are observable and cm be deait
with appropriately, However other errors are not apparent and must be assurned to be
random therefore not significantly affecting validity or reliability.
Some cases in this data set contained dates that were either out of range or illogical.
A d2te could be out of range either because it occurred after the end of the observation period
or because the offender was not between the ages of 12 and 1 8 ' at the time of incarceration.
In either instance the observation was removed from that individual's incarceration history.?
Other histories contained dates that did not logically fit into the offender's incarceration
records. For these observations, miscoded data was recoded to set the incarceration length
to one day. This preserved the number of incarcerations and minimized error in regards to
length of the incarceration period.
Benefits of this data source far outweigh any possible limitations. Juvenile
Offenders who tum 18 while in custody at a youth facility can only be moved to an adult facility through application to the courts (Hudson, Hornick and Burrows, 1988). Therefore if an individual was under 18 at the time of admission, but over 18 at the tirne of release, the data was included in the analysis.
While this usually resulted in the rernoval of an observation, in 2 cases entire histories had to be deleted. This reduced the total nurnber of cases from 222 to 220.
3 5
dehquents can not be identified under the Young Offender's Act (YOA). It is therefore
difficult to generate a sample of young offenders -- especially a sample containing such a
large number of SHOs. This is a difficult population to gain access to, therefore this data set
provides the opportunity to explore questions that may otherwise not be testable. In
obtaining the criminal histones of this group of offenders we can continue to search for
regularities within criminal behaviour.
According to H. H. Hymm, secondary data potentidly "expands the types and
number of observations to cover more adequately a wider array of social conditions,
measurement procedures, and variables than c m usually be studied by pnmary surveys".
subsequentiy producing "a more comprehensive and defrnitive ernpirical study of the
problerns the investigator has formulateci" (1 972: 1 1). Information on the large nurnber of
variables and individuals in this data set, inctuding the exact dates of police contact and
incarceration, would be difficuIt to obtain using primary research especidly if we had to rely
on the mal1 of subjects. To collect data on patterns of criminai activity over long periods
of time by surveying individuals on past behaviour is a dificult task. We can attempt to
"reconstruct the p s t by retrospective questioning in a current survey, but then rnemory errors
may cause siippage between measurement and concept" (Hyman, 1972: 12).
Finally in researching how crirninal activity changes over time we need to be able to
study long penods of tirne (Hyman, 1972). To col tect this data first hand researchers either
must rely on the memory of subjects, and as already stated this c m be problematic, or they
must commit thernselves to studying a large group of individuals over a long period of time.
This requires a large cornmitment on the part of the researcher and is often not feasible. B y
using official data this research avoids both of these problems.
This data set provides a rich and detailed history of official offending arnong
juveniles. We know that past behaviour is a good indicator of future behaviour (Nagin and
Paternoster, 1991) and that certain events can alter that behaviour. Yet few Canadian studies
have the data to demonstrate this arnong the youth population. This data set contains a wide
array of information on the key variables in identiîying criminal patterns and developmental
changes and is thus a valuable source of information
22 SAMPJJNG
Originally the sample consisted of the criminal histories of 4, 565 youth who had
police contact during the month of December 1991. Of this sample 200 of the young
offenders had been identified as serious habitua1 offenders (SHOs) while 4,365 had not been
identified as such. or are considered to be average or typical young offenders. Designation
of the SHO label is based on a point system as developed by the policing agency fiom whom
the data was obtained. Points, ranging from 1 to 3, are given in accordance with the
seriousness of the crime cornrnitted. Admittance into the SHO program is generally based
on both a high rate of offending as well as the commission of serious offenses. Afier
accumulation of a specified nurnber of points, usually 50, an offender becomes a SHO.
In order to obtain statistically comparable numbers of SHOs and nonSHOs, a
disproportionate stratified random sample without replacement was drawn3. From this
3
Few females were selected through this sampling procedure therefore gender was not included in the analysis.
37
procedure the criminal careers of 19 1 SHOs and 195 nonSHOs were selected. Mso included
in these criminal histones was a one year follow up period beginning h m the sarnpling date
of December 3 1, 199 1. These retrospective histories provide exact dates of events thereby
producing a data set that can be used to qnalyze the exact timing and sequencing of cnminal
occurrences. Longitudinal data of this nature and detail alIow for description and
explmation of the conditions under which incapacitation effects criminai acrivity.
Additionally, in constnicting two groups. SHOs and nonSHOs, a control group is
established. A possible weakness of studies on serious habiniaI offenders is that they do not
establish a control group. A control group illustrates any difference in patterns of offending.
. By studying nonSHOs as well as SHOs, we c m identify whether or not there exists a
subgroup of serious habitua1 offenders wherein lambda is hi& and remains high throughout
their crirninai careers. Control groups allow us to test whether or not criminal behaviour
varies between groups or is in fact as Gotdiedson and Hirschi suggest, simply a function of
age.
2.3 SURVIVAL ANALYSIS -
Survival analysis ailows us to analyze social processes. Change and development are
processes that require a different type of analysis than many other sociological studies. I t is
the timing of events that defines the concepts in t h s study, not simply whether or not they
occur. We recognize that some offenders may desist from criminai activity following
incarceration, but others may not. This is what makes survival analysis applicable to this
study, it allows us to include both those individuals who reoffend and those who do not.
Survival analysis measures and models the patterns and correlates of the occurrence
38
of events (Yamaguchi, 1991). It examines whether an event occurs and if so when. "By
definition, occurrence of an event assumes a preceding tirne interval that represents its
nonoccurrence (Yamaguchi, 199 1 : 1). The penod of nonoccurrence equates to a period
where the individual offender is at risk of reoffending. More specifically, once an individuai
is released f?om custody there will be a penod when the individual is at risk of retuming to I
secured custody.
Nonoccurrence signifies that the event under study did not occur during the period
of observation. Many statistical procedures, such as multiple regression,. code nonoccurrence
of an event as missing data. This causes the loss of vaiuable information and may bias the
results. Survival models are able to include this information in the analysis through the use
of censored observations (Allison, 1984; Singer and Willet, 199 1 ; Yamaguchi, 199 1 ).
Censoring means that the event did not occur during the penod of data collection, but it
incorporates the possibiiity that the event occurred outside this time fiarnework.
Specificaily, "censoring exists when incomplete information is available about the duration
of the nsk period because of a limited observation period' (Yamaguchi. 199 1 : 3). The ability
of survival analysis to deai with censoring makes it possible to include a range of behaviour
that may not be observable through other statistical analyses.
The probability of reincarceration is measured using hazard rates. S w i v a l analysis
models hazard rates which express "the instantaneous nsk of having the event at time t. given
that the event did not occur before rime t" (Yamaguchi, 1991 : 9). The hazard rate is the
unobserved dependent variable that measures the occurrence of an event as well as its timing
(Allison, 1984). I t measures the failure rate at a given time as a proportion of the part of the
3 9
popuIation that hm survived up until this tirne (Schmidt and Witte, 1988). Survival analysis
was fmt used in medicine, ofien in studying disease. in studies of cancer, hazard rates were
studied to determine the probability of death at a given time based on some treatment.
Patients who had been given a treatment were studied to see how long they survived
following th s treatment. From this the probability of death at a given time could be
determined.
In this study the hazard rate rneasures the probability that an offender will be
reincarcerated during a given 6 month interval. Hazard rates measure the "risk attached to
a person at a given time" (Yamaguchi, 1 99 1 : 10). During a specific time interval some
individuals at risk will return to custody while others will not. Given that an individual has
not been reincarcerated up until the beginning of the 6 month period (or has survived), we
can estirnate the likelihood that they will reoffend during this interval. Hazard rates represent
the proportion of offenders that return to custody d u ~ g a time penod as a proportion of the
sarnple that has not been reincarcerated up until this time. They measure the probability of
failure. Hazard rates are assurned to change as a function of time, which is represented as
time since the last event (AIIison, 1984). Therefore a change in the hazard rate dernonstrates
different probabilities of retuming to custody directly related to the time elapsed since their
release frorn custody.
Finally, data are available on several subsequent observations of incarceration.
Therefore we can measure successive or repeated events, for example the third and fourth
incarcerations, using the same method. By comparing survival times, the early careers of
offenders c m be compared to determine if beginning events unfold for SHOs in the sarne
40
way they do for nonSHOs and M e r possible ciifferences within each group c m be
examined.
CHAPTER 3: RESULTS
3.0 iNTRODUCTION -
Recall fiom chapter 1 that the aim of this study is to examine the effect of
incarceration on subsequent juvenile behaviour. It was suggested that young offenders will
either be deterred or labelled by the experience of incarceration. Deterrence has been
conceptualized as a decrease in the rate at which juveniles are reincarcerated foliowing a
period of incarceration. Labelling, in contrat, has been defined as continuation or even
possibly escalation of the rate at which young offenders r e m to custody following a period
of incarceration. Overall the results suggest that the likelihood of deterrence or labelling
varies according to whether the young offender was a SHO or a nonSHO. But, additionally
deterrence and labelling are related to several conditions that occur over the life course. For
example, the age of the offender appears to be associated with the probabiIity of deterrence
as does the amount of previous involvement the young offender has experienced with formal
sanctioning. After controlling for these conditions it seems that not only are there
differences between SHOs and nonSHOs, but that different responses to incarceration can
be found within each of the two groups.
3.1 TABLE 1 : CHARACTERISTICS OF THE SAMPLE -
Table 1 describes the incarceration histories of the SHOs and nonSWOs sarnpled. It
is apparent that a higher percentage of SHOs than nonSHOs were incarcerated during the
period of observation. Of the 19 1 SHOs in the analysis, oniy 6% ( 1 1 out of 19 1 ) were not
incarcerated compared to a substantially higher nurnber of nonSHOs who were never
imprisoned, 79% (155 out of 195). Throughout the first four incarceration periods, a
significantly higher percentage of SHOs than nonSHOs were incarcerated. Additionally. we
can determine the number of offenders incarcerated decreases for each successive custody
period for both SHOs and nonSHOs.
- --
It appears that SHOs were incarcerated more times over their adolescent careers than
were nonSHOs, an average of 6.3 tirnes compared to only 3.3 tirnes respectively.
Furthemore, the totai arnount of tirne that SHOs and nonSHOs were incarcerated between
the ages of 12 and 18 varies considerably. SHOs, on average, were incarcerated for a total
of 273 while nonSHOs were incarcerated for about half of that time, 140 days. Also, each
incarceration period, on average, was shorter for nonSHOs than for nonSHOs. It is evident
that SHOs and nonSHOs have different incarceration histories. This is not sq r i s ing as the
very definition of SHO suggests a pattern of serious and fiequent offending with which
incarceration would be consistent. But it may also suggest that SHOs are less IikeIy to
change their behaviour following incarceration as they are each incarcerated an average of
6.3 tirnes. With this understanding we can now begin to explore how these histories may
also Vary in regards to the timing of events. Finally this discussion will allow theoretical
TABLE 1: MCARCERATION HISTORIES FOR JUVENILES BETWEEN THE AGES OF 12 AND 18
SHO
NonSHO
l 166i386 198/386 1 7,"" 1 (51%) 1,236 1 1:4'0%)6 1 l 249 1 12 1 (43%) 1 (41%)
nümber of individuals held in custody Average number of
rirncs incarccrared
6.3
3.3
never
111191 (6%)
1 551195 (79%)
Total days each
individual is
incarccrated (Avg.1
273
140
1 time
1801191 (94%)
401195 (21%)
Average length of
cach incarceration
pcriod (in days)
45
26
2 times
1731191 (9 1 Oh)
2.51195 (13%)
3 times
1561191 (82%)
181195 (9%)
4 times
143/191 (75%)
121195 (6%)
propositions to be examined and evaluated.
3.2 TABLE 2: SUR - VIVAL ANALYSE
The following tables examine conditional deterrence hypotheses in the early stages
of a sample of young offenders' cnminal careers. Specifically the tables test the timing of
and conditions under which incarceration of young offenders may or may not occur. Under
different conditions, or more the specifically based on previous experience. some offenders
will be deterred from future criminal behaviour while others will intemalize the label or role
of deviant and persist in defiant risk taking behaviour. This research examines how
punishrnent through incarceration in a secured custody facility may change the criminal
careers ofjuvenile offenders at various times in the criminal career. In order to evaluate how
legal sanctioning may change criminal identities and behaviours a number of contingency
variables were controlled for (see chapter 1). Tables 2A through 2C address how the
classifications of SHO and nonSHO may effect the likelihood of deterrence following the
fist, second, and third incarcerations. Additionaliy, these tables examine the possibility that
a subgroup of young offenders exists. Incarceration, as a form of legal sanctioning, is
unlikely to deter this group of offenders. Tables 4 through 7 study the effects, on the
probability and timing of events, of additional control variables such as age at first police
contact, number of police contacts pior to first incarceration, age at first incarceration and
finalty the length of first incarceration.
in order to evaluate the conditional deterrence hypotheses, specific entries within the
tables are observed. First we examine the number of individuais who were censored, or did
not r e m to custody, to assess whether or not and when incarceration deters young offenders
44
from continued criminai behaviour. Recail that the definition of deterrence includes both
desistance From criminai activity as well as a possible increase in the Iength of time between
criminal events (Wilson, 1985: 171). Therefore, we also consider the maximum possible
length of time it took for al1 individuais to be reincarcerated for their next offence.
Additiondly, the median srnival times are discussed as they represent the 50th percentile
(Lee, 1992) or the time at which 50% of the sample has reoffended. Median survival times
allow us to examine differences between SHOs and nonSHOs as well as identifi a possible
subgroup of offenders who are unlikely to be deterred by incarceration. Finaily we look at
time specific occurrences within Tables 2A through 2C by studying the hazard rates
(instantaneous failure rates) of SHOs and nonSHOs. Hazard rates measure the probability
of failure, in the population, as a finction of time (Lee, 1992: 11). This is comrnonly
referred to as the risk of failure. Specificaily in this study hazard rates represent the
probability that young offenders in the population will fail, by returning to custody, during
my given 6 month interval'.
Tabie 2A estimates the length of time between the first and second incarceration.
This is the t h e at which a young offender is at risk of returning to secured custody. More
specificaily, it measures the time between the first release date from custody and if and when
a second admission date to custody occurs. We also controI for whether the young offender
was a SHO or a nonSHO in order to measure similarities and differences between the two
Six months was chosen as it allows for sufficient time for variation to be s h o w between individual reincarceration times while summarizing the data in a usefuI and readable manner.
groups and within each group's response to incarceration.
Findings strongly indicate that SHOs and nonSHOs respond differently to the
punishment effects of incarceration (see Table 2A). To begin with, 94% (1 80 of a possible
191 ) of SHOs were incarcerated at least once, while only 2 1% (40 of a possible 195) of
nonSHOs were incarcerated at least once. The difference in the number of SHOs and
nonSHOs who did not r e m to custody a second time indicates that deterrence is not
uniform. in other words, detemence is conditional on whether the young offender was a SHO
or a nonSHO. Of the SHOs who were incarcerated, only 4% (7) were censored as they did
not commit an act that caused them to be returned to custody for a second stay. In
cornparison, a larger group, 38% (1 5), of nonSHOs were censored during the same period
of observation. Therefore, more nonSHOs than SHOs appear to be deterred f?om risk taking
behaviour following the first incarceration.
It appears the maximum possible length of time it takes for al1 SHOs to return to
custody is longer than that of nonSHOs. Variation in the arnount of time it takes for young
offenders to retum to custody suggests that those nith longer times, although not terminating
their careers, are nonetheless deterred. For instance, it took some SHOs up to 36 months to
return to custody, whereas al1 nonSHOs who were not censored returned to custody within
a 24 month period.
Variation in the length of time it takes for a youth to return to custody is evident not
only between the SHOs and nonSHOs but within each of the two groups. Regardless of the
maximum time it takes for either SHOs or nonSHOs to be reincarcerated, it appears there
exists a subgroup of SHOs that return to custody quite quickly. This is evidenced by the
TABLE 2: LENGTH OF TlME BETWEEN SUCCESSIVE MCARCERATIONS FOR SHOS AND NONSHOS h
TABLE ZA: TIME BETWEEN FIRST RELEASE AND RETURN TO SECURED CUSTODY 1 I 1 lnterval ~tart 1 SHOs 1 nonSHOs
Time 1 I 1 1 1 1 I
1 1 median survival time = 4.38 1 rnedian survival t h e = 8.09
(rnonths) number entering inicwal (numbcr
ccnsored)
- - - ... . .
TABLE 28: TIME BETWEEN SECOND RELEASE AND RETURN TO SECURED CUSTODY
1 1 median survival time = 3.90 1 median survival time = 4.90
Interval Start Time (rnonths)
O
6
12
18
24
3 O
1 TABLE 2C: TIME BETWEM THIRD RELEASE AND RETURN TO SECURED CUSTODY
cumulative proponion surviving
SHOs 1 nonSHOs
Time (months)
hazard rate
number cnicring intervai
(number censored)
173 (6)
Interval Start
numixr entering intcrval
(number censored)
25 (1)
9 (1)
6 (2)
4 (2)
1 (-1
SHOs 1 nonSHOs
I 1 median survival cime = 3.50 1 median survival t h e = 4.25 *note: these hazard rates are high (0.3333) due to the small number of individuals entering the last interval
numbcr encering intervai
(numbcr censored)
cumulative proportion surviving
0.23 17
hazard rate
0.2079
cumulative proportion surviving
0.3878
0.2965
0.2965
O. 1977
O. 1977
0.1977
numbcr entering interval
(number censored)
156 (4)
20 ( 2 )
10 (6)
1( I )
cumulative proportion surviving
0.0864
0.0972
0.11 1 1
0.1 1 1 1
0.3333'
hazard rate
0.147 1
0.0444
0.0000
0.0667
0.0000
0.0000
37 (6)
17 (3)
7 (2)
2 (-1
1 (-1
number entering intervai
(numbcr censored)
18 (2)
4 (-1
4 (4)
hazard rate
0.1363
0.0747
0.0374
0.0 187
0.0000
cumulative proportion surviving
O. 1429
0.0827
0.0473
0.0473
hazard rate
0.25 19
0.0889
0.0909
0.0000
cumuIadve proponion suwiving
0.294 1
0.294 1
0.294 1
hward raie
0.1818
0.0000
0.0000
4 7
median survival times for SHOs and nonSHOs. The median survival time was 4.38 months
for SHOs and 8.09 months for nonSHOs. Difference in the average length of time between
events for SHOs and nonSHOs indicates that a subgroup of SHOs r e m to custody quickly.
This suggests that the experience of incarceration did not effectively deter these young
offenders fiom continuing in their criminal careers
To examine evidence of deterrence ancilor labeiling in the population, we can also
compare the hazard rates (failure rates) of SHOs and nonSHOs within specific intervals. As
expected the hazard rates agree with the previous findings for Table 2A. But, additionally
we find that hazard rates were highest in the first 6 months for both SHOs F(t) .1737] and
nonSHOs [h(t) .0944]. High failure rates during this first interval indicate effects of
labelling as individuals are returned to custody quickly. Quick return to custody may
indicate that incarceration did not deter hture criminal behaviour but served to label the
youth as a criminal. This indicates a young offender who is persistent in their criminal
career.
Throughout Table 2A we see that SHOs have higher failure rates than nonSHOs for
each 6 month interval. Therefore, SHOs are more likely to return to custody and they appear
to do so more quickiy than do nonSHOs. Overall, data indicate that it was the nonSHOs who
were more likely to avoid behaviour that would put hem at risk of reincarceration. But, we
also find evidence that some SHOs were deterred while others were not. Therefore it appears
that deterrence varies between SHOs and nonSHOs, as well as within each of the SHO and
nonSHO groups. The SHOs who retum to custody quickly may be representative of a
subgroup of offenders who are largely responsible for crirninal activity. And, it appears that
these offenders are not deterred by incarceration.
By def i t ion SHOs and nonSHOs represent different groups of offenders. ïherefore
it was expected that the two groups may respond differently to punishrnent. However
significant variation in the risk of returning to custody was found within each of the two
subpopulations. SHOs did not respond to punishment in a uniforrn fashion nor did
nonSHOs. Additionally it appears that some nonSHOs returned to custody very quickly.
These nonSHOs responded to punishment in a manner similar to many SHOs. Some SHOs.
on the other hand, did not return to custody. This behaviour is more strongly associated with
the nonSHO group. Therefore the definition of SHOs and nonSHOs did not force or
predetennine the actions following incarceration. The data indicate that sorne offenders will
present offending characteristics that are more comrnonly associated with a different history
of offending or classification type.
In Table 2B we further explore conditiond deterrence by exarnining the length of
time between the second and third incarceration. This illustrates the effect of repeated
incarceration on the probability of desistance and on the timing of any subsequent
incarceration should it occur. According to Braithwaite, continued punishment "is a denial
of confidence in the morality of the offender by reclucing nom cornpliance to a crude cost-
benefit calculation" (1989: 72). Table 2B examines whether repeated incarceration has
similar or different effects on offenders when compared to the first incarceration.
As in Table 2A, we found varying responçes to punishrnent by SHOs and nonSHOs.
Again it appears that nonSHOs are less Iikely to return to custody than are SHOs. Following
a second incarceration only 10% (1 7) of SHOs were censored, whereas significantly more
49
nonSHOs, 28% (7), were censored, This indicates that nonSHOs are more likely to be
deterred fiorn risk taking behaviour than are SHOs.
The data also suggest that the effects of conditional deterrence are less after the
second incarceration for both SHOs and nonSHOs. For nonSHOs this is demonstrated
through the substantially smailer percentage censored after the second incarceration, 28%
compared to 38%. Additionally, for SHOs we cm illustrate the decreasing effects of
deterrence following the second incarceration, by exarnining the timing of the next
incarceration. In table 2B, it took some SHOs up to 30 rnonths to retum to custody, whereas
it took slightly longer d e r the first incarceration, 36 months. This decrease in the maximum
possible time to r e m to custody represents a decrease in deterrence as young offenders are
returning to custody more quickly then they were d e r the first incarceration.
Consistent with findings following the first incarceration. we find significant
variation behveen the SHOs and nonSHOs in regard to the average time between the second
and third incarceration. It appears that the median survival time for SHOs is shorter than that
of nonSHOs. This indicates that nonSHOs are more likely to be deterred fiom returning to
custody than are SHOs. NonSHOs had a median survivd time of 4.90 months, while SHOs
had a median survivai t h e of 3.92 months, approximately 1 month shorter. This indicates
SHOs and nonSHOs respond differently to incarceration. It also suggests evidence of a
subgroup of SHOs who were not deterred by repeated incarceration but return to custody
quickl y.
Although the median survival tirne for nonSHOs is still longer than that of SHOs, it
is significantly shorter than it was after the first incarceration. Notice that SHOs' median
5 O
survival tirne decreased by ody a couple of weeks, falling fiom 4.38 months to 3.92 months,
indicating that most SHOs continue in behaviours established early in their ofiending
careers. In contrast, nonSHOs showed a large decrease in their median survival time,
dropping tiom 8.09 months to 4.90 months. This suggests that a second incarceration greatiy
reduces the deterrent effect of punishrnent for nonSHOs.
Failure rates again show (as in Table 2A) the same difference in incarceration
patterns of SHOs and nonSHOs in the population following the second incarceration. SHOs
and nonSHOs who were incarcerated a second time, still appear to be at the greatest risk
during the first 6 month interval. But again, as in the previous table, SHOs are stiIl at a
greater risk than are nonSHOs. SHOs and nonSHOs had hazard rates of h(t) .2019 and h(t)
.1471 respectively during the first 6 months. Once again this provides evidence that
nonSHOs may be more likely to be deterred following incarceration than are SHOs. Afler
the initial 6 month intervai we still find that hazard rates are higher for SHOs than nonSHOs,
but are lower than during the first 6 months.
Recall that after the first incarceration, SHOs were at a greater risk of returning to
secured custody than were nonSHOs and that both were at greatest risk of failure during the
first 6 months. These findings are comparable to the finding indicated after the second
incarceration. But. in addition we fuid that f i e r repeated exposure to punishrnent both SHOs
and nonSHOs remain less likely to be deterred then after the first incarceration.
It appears that the second incarceration decreases the likelihood of deterrence for
both SHOs and nonSHOs. Substantially fewer individuals desist fiom risk taking behaviour
following the second incarceration than did afler the first incarceration. This may be because
5 1
afler the first incarceration some individuals may discover that the legai threat is overstated
and that punishment is quite tolerable (Zimring and Hawkins, 1973) and therefore the initial
legal punishrnent may have a different effect on behaviour than the second exposure to legal
punishment (Cameron in Sherman, Smith, Schmidt and Rogan, 1992).
Data also suggest that a subgroup of SHOs demonstrate persistent criminal behaviour
foltowing their second incarceration that is consistent with the labclling process. This
subgroup of SHOs are likely to r e m to custody after their second incarceration and do so
quickly. Overail, both SHOs and nonSHOs return to custody more quickly than they did
following their first incarceration, although the difference in the length of time between
events showed more change in the nonSHO sample. Therefore it appears that nonSHOs are
still more likely to be deterred than are SHOs and M e r that a subsarnple of SHOs are the
most persistent offenders.
To further explore conditional deterrence in young offenders' early criminal careers,
we examine one more incarceration period. The timing between the third and possible fourth
incarceration should aid in establishing the role of incarceration early in the life course.
Therefore we looked at whether or not a fourth incarceration was Iikely following the third
reiease from custody and if so, the timing of when it was likely to occur.
Consistent with findings in Tables 2A and 2B, more SHOs than nonSHOs were
censored or did not return to secured custody during the period of observation. Of the
nonSHOs who were incarcerated three times, 33% (6) were censored, in comparison only
8% ( 1 3)of SHOs censored. This indicates that even after repeated punishment SHOs are
more likely than nonSHOs to persist in risk taking behaviour which may lead to
incarceration.
Although we find patterns of desistance sirnilar to previous tables, we additionally
find that both SHOs and nonSHOs are less likely to be deterred following a third
incarceration than they were d e r the first or second incarceration. This is demonstrated
through the timing of the fourth incarceration. The maximum time SHOs and nonSHOs are
at risk of returning to custody is significantly shorter than in previous findings. SHOs in
Table 2C only took up to 12 rnonths to be reinstitutionalized and nonSHOs were
reincarcerated within 6 months. Both times show considerably less variation than in
previous tables. This finding suggests that repeated incarceration early in the life course may
label the offender thus causing persistent deviant behaviour.
We further examine the timing of a possible fourth incarceration by considering the
median survival tirnes. Once again we find SHOs have a shorter rnedian survival time than
do nonSHOs. but we also find that each group has a shorter median survival time than it did
f i e r the first or second incarceration. SHOs had a median survivd time of 3.50 months f i e r
the third incarceration while nonSHOs had a rnedian survival time of 4.25 months. So again
we find that SHOs are less likely to be deterred than are nonSHOs, but the ciifference in tirne
between them is less than f i e r the first or second incarceration. Recalf that after the first
incarceration, nonSHOsl rnedian survival tirne was almost 4 months longer than that of
SHOs, and after the second incarceration it dropped to approximately 1 month longer.
Finally d e r the third incarceration the difference falls to about 3 weeks.
Findings in tables 2A through 2C indicate that SHOs and nonSHOs respond
difTerently to punishrnent. Specifically we find that SHOs are more persistent in their
53
criminal behaviours than are nonSHOs. Yet, even though we ùiitially find greater evidence
of deterrence arnong the nonSHO population, we see that SHOs and nonSHOs both
experience labelling through repeated punishrnent, as indicated by decreased median survival
times. We see indications that a subgroup of SHOs exist that are unlikely to be deterred
through incarceration. These offenders were reincarcerated quickly d e r each release from
custody indicating that they are persistent in their criminal behaviours.
3.3 - CONTROL VARIABLES AND CONDITIONAL DETERRENCE
3.3.1 TABLE 3: AGE AT FIRST POLICE CONTACT
The previous tables appear to indicate that public labelling through incarceration may
stabilize offending behaviours in some individuals. Yet, we find variation in which
offenders were the rnost likely to continue with risk taking behaviour and which offenders
show the greatest likelihood of specific deterrence. According to Farrington's study on the
effects of public labelling, he finds that public labelling leads to increased deviance, but
further that "repeated labelling leads to greater deviance amplification" meaning that
repeated labelling may have a cumulative effect on deviant behaviour ( 1 977: 1 1 8). Results
support Fanington's conclusions, but it was aiso found that SHOs and nonSHOs behaved
differently even after the same exposure to repeated labeliing. Conditional deterrence may
explain some of these differences as well as differences that arose within the SHO group
itself. In order to examine conditional deterrence 3 more variabIes or conditions under which
deterrence or labelling may occur were controlled for.
The first variable considered was the age of the individual at first police contact (see
Table 3). Recail that this variable permits us to examine deviant behaviours that occur prior
54
to the age of 12. Therefore this represents the earliest experience of being caught and
publicly labelled as a deviant. Therefore, we want to see if the age at which labelling first
occurs effects the likelihood of deterrence following incarceration.
Age at first police contact is correlated with onset, which is an important factor in the
early development of crirninal careers. Specifically, research suggests that early antisocial
behaviour is indicative of later antisocial behaviour, leading to the proposition that early
criminal experiences will partly determine future criminai trajectories (Farrington, 1977;
Nagin and Fanington, 1992; Sarnpson and Laub, 1992). Therefore age at first police contact
may be useful in descnbing and eventually predicting whether or not incarceration deters
young offenders and further whether it can explain variation in behaviour between SHOs and
nonSHOs as well as within the SHO subsarnple.
To separate SHOs and nonSHOs into age categones we used the median age at which
individuals had their first police contact. The median age was used rather than the mean as
the mean tends to be skewed by extreme cases (see Lee, 1992). Using the rnedian age we
divided the sample into 4 categones, individuals whose first police contact was before age
14 and individuals whose first police contact was between 14 and 18, for both SHOs and
nonSHOs. This allows us to differentiate between early and late police contact. This
distinction allows us to examine "the different implications that early onset has for the
subsequent course of the behaviour than does later onset" (Loeber and LeBlanc, 1990: 391)
or in this study how the age of first police contact effects the probability of deterrence.
It appears in Table 3, that both SHOs and nonSHOs who were first in contact with
the police early are less likely to be deterred by incarceration. This is suggested by the Iarge
5 6
number of SHOs less than 14 years old who return to custody following their first
incarceration. Oniy 2% (3 out of 124) of these SHOs were censored. A slightly larger
percentage, 7% (4 out of 56), of SHOs who were 14 or older at first police contact did not
retum to custody. NonSHOs' demonstrated similar patterns of behaviour. Of the nonSHOs
who were under 14 at the tirne of their first police contact, 15% (2 out of 13) did not return
to custody. However aimost half, 48% (13 out of 27), of nonSHOs who were 14 or older
at police contact were not reincarcerated. So, it appears that based on censoring, those
who are first in contact with the police early in the life course are more likely to continue
with risk taking behaviour. This rnay be the result of social bonds being broken or possibly
of labelling but regardless it seems that young offenders who are involved in criminal
activity at a young age are more likely to persist in risk taking behaviour than are those who
are not in contact with the police until their later teens.
The timing of the second incarceration (if the individual was not censored) again
shows that eariy onset is consistent with persistent deIinquent behaviour following
incarceration. Some SHOs with an early first police contact show indications of deterrence
by remaining out of custody for longer time periods. For example, some SHOs took up to
36 months to r e t m to custody. But, most return quite quickiy. This is dernonstrated
through a rnedian survival time of only 4.12 months. SHOs who had a late first police
contact survive slightiy longer. Their median survival time was 5.09 rnonths. The biggest
difference is found within the nonSHO group. NonSHOs who were iess than 14 at the time
of their first police contact had a median survivai time of 5.36 months. This is not
significandy different than the SHO tirnes, but it is substantialIy shorter than the nonSHOs
57
whose fint police contact was at 14 or older. These nonSHOs had a median survival time
of 10.22 months. So once again, we find that police contact at an early age may label the
offender thus perpetuating their criminal careers.
Finally, the hazard rates for each of the four groups, SHOs whose first police contact
was at less than 14 years of age, SHOs 14 to 18 years old, nonSHOs whose first police
contact was at less than 14 years of age. and nonSHOs 14 to 18 years old, indicate that
offenders are at the highest risk of being reincarcerated during the fmt six months following
release from custody. After that the hazard rates drop, indicating a lower probability of
reincarceration in successive intervals. But once again, the hazard rates indicate that early
first police contact increases the likelihood of reincarceration.
3.3.2 TABLE 4: AGE AT FIRST INCARCERATION
Gottfredson and Gottfredson (1 994) tell us that incarceration is an attempt at career
modification. The age at first incarceration represents formal legai sanctioning that rnay lead
to stigmatization of the young offender and therefore alter their criminal careee. Ofien
sentencing to secured custody does not occur until the offender has already had several
police contacts. Therefore the first incarceration signifies an event whereby the offending
rnay have assumed a violent or repetitive nature. The offender rnay thus be committed to the
deviant role. Recall that incarceration may cause shaming which in turn rnay effect the
probability of deterrence. It is thus possible that official actions, such as incarceration, can
invoke societal or informal reactions that effect future criminal behaviour (Braithwaite, 1989;
Zimring and Hawkins, 1973). However, the extent to which formal and informal sanctioning
rnay change behaviour rnay depend on the age of the offender when they are first
58
incarcerated. It has been suggested that early intervention rnay have a significant direct
effect on participation in criminal activity (Gottfiedson and Gottfiedson, 1994). Therefore
it is possible that intervention, through incarceration, at an earlier age may have a different
impact on offending than does incarceration later in the life course.
Age is an imporîant factor in the early development of a young offender's criminal
career. Recall the previous discussion on the importance of earIy behaviour on fùture
behaviour. Therefore we next examine the effect that age at first incarceration has on the
probability of returning to secured custody. We again used the median age to divide the
sarnple into 4 categories, 12 to 14 years old and 15 to 18 years old, for both SHOs and
nonSHOs.
The percentages of individuais censored in each category indicate that age at first
incarceration does influence the likelihood of deterrence. It appears that individuais who
were between 12 and 14 years old when they were first incarcerated are more likely to be
reinstitutionalized than are individuals who were between 15 and 18 years old at the tirne of
first incarceration. SHOs who were younger at their first incarceration are at a high rîsk of
returning to custody during their adolescent crirninal careers, in fact, less than 1 % ( 1 out of
106) did not retum to secured custody compared to 8% (6 out of 74) of SHOs 15 to 18 years
old who were not reincarcerated. NonSHOs show a sirnilar pattern of c e n s o ~ g . Those who
were 12 to 14 years old were more likely to retum to custody than were offenders who were
older at first incarceration. Only 20% (4 out of 20) of nonSHOs who were younger at their
first incarceration did not return to custody cornpared to 55% (1 1 out of 20) of the nonSHOs
who were 15 to 18 years old. It appears therefore that young offenders who are oider when
60
they are first incarcerated are more susceptible to deterrence through legal sanctioning.
Possibly, if an offender is older when they are îïrst incarcerated, they are better able to resist
the criminal label. Or perhaps incarceration is able to deter older offenders a s they are berter
able to recognize the future costs of continued offending and therefore are deterred by
punishrnent.
Recall that deterrence has been defmed as to inciude a tirne distinction. This allows
for the possibility that while the offender may not terminate their criminal career, they can
take a longer penod of time to be reinstitutionaiized. Consistent with the findings in
previous tables, we find that some SHOs rnay take a considerable amount of time to return
to custody (up to 36 rnonths), but most return very quickly. SHOs who are older when they
are first incarcerated take slightly longer to retum to custody than do the younger SHOs, 4.79
months compared to 4.13 months. The difference in swiva l times for nonSHOs is
significantly larger, 5.85 months for nonSHOs 12 to 14 compared to more than 24 months
for nonSHOs 15 to 18. Therefore, timing of the second incarceration for both SHOs and
nonSHOs indicates that age at first incarceration affects the likelihood of deterrence. Young
offenders who are incarcerated early in the life course are Iess likely to be deterred by formai
and possibly informal sanctioning than are those who are incarcerated at an older age. It
appears that at an early age they are assigned a role of criminal or deviant and do not have
the ability or means to refute or change this identity.
Consistent with findings in the previous tables, the hazard rates indicate that juveniles
are at the highest risk of reinstitutionalization during the first 6 rnonths foilowing release
from custody. The hazard rates for SHO under 15 years of age at first incarceration was
6 1
h(t)= 0.190 1 while SHOs 15 to 18 had a hazard rate of h(t) = 0.1 149. NonSHOs had lower
hazard rates during the first 6 month interval. NonSHOs under 15 years old had a hazard rate
of h(t) = 0.15 18 and nonSHOs 15 to 18 had a hazard rate of only h(t) = 0.0753. However,
within each of the four categories we find different patterns of risk of reincarceration afier
the first 6 months. SHOs who are 12 to 14 years old at first incarceration showed a pattern
of increasing and decreasing risk, the hazard rates appear to rise and fail without pattern. In
contrast, SHOs who were 15 to 18 years old showed a steady decreasing risk of
reincarceration in each successive 6 month interval. This suggests that over tirne SHOs 15
to 1 8 are less likely to be reinstitutionalized than are SHOs 12 to 14 years old.
3.3.3 TABLE 5: POLICE CONTACTS PWOR TO FlRST MCARCERATION
Whiie we found that age of first police contact can explain part of the variation
among young offenders' delinquency, it is likely that other factors may influence the
relationship between punishrnent and deterrence. It has been suggested that public labelling
through legal means may alter the behaviour of an individual, specificalIy by either deterring
or labelling the offender. Labelling theory hypothesizes that legai involvement will &rm
the delinquent identity for a new offender thereby perpetuating or even ampliQing behaviour
(Klernke, 1978). Additionally, labelling theonsts have argued that the further one is pushed
into the legal labelling process the greater the likelihood of future deviance (Klemke, 1978).
Therefore, in Table 5 we examine whether a high number of police contacts prior to
incarceration rnay push the individual further into the legal labelling process consequently
affecting the likelihood of deterrence foliowing incarceration,
We expected that individuals, both SHOs and nonSHOs, who have a higher than
62
average number of police contacts prior to first incarceration wili demonstrate a stable
pattern of deviant behaviour. It is therefore possible that public labelling through police
contact may lead to stabilization rather than deterrence of a criminal career. Again the
median number of police contacts rather than the mean was used to separate offenders into
categories. The mean number of police contacts was skewed as there are a few cases with
high values that cause the average to be significantly higher than the median. Therefore the
median is a better representation of the sample. The median therefore allows us to break the
sample into those with 7 or less police contacts prior to first incarceration and those with 8
or more police contacts, for both SHOs and nonSHOs. This once again splits the sample into
4 groups.
When looking at censoring it appears that the hypothesis is supported by the
nonSHOs but not by the SHOs. Every SHO who had 7 or less police contacts eventually
returned to custody meaning that none of these SHOs were censored. However, 8% (7 out
of 93) of SHOs who had 8 or more police contacts were censored. Although this appears
to be in conflict with the hypothesis, other observations within the table suggest that a high
nurnber of poIice contacts does decrease the IikeIihood of deterrence. For instance, support
for the hypothesis is shown when we examine censoring in the nonSHO sample. We found
that 40% (14 out of 35) of nonSHOs who had 7 or less poIice contacts prior to first
incarceration did not retum to custody, whereas only 20% (1 out of 5) nonSHOs who had 8
or more police contacts were censored. It appears therefore that a high number of police
contacts may label the offender thus perpetuating the delinquent identity. By being in
contact with the police a large number of times offenders may lessen their ability to
64
reintegrate themselves into society. They may weaken their social ties to the cornrnunity
while increasing their ties to a criminal role.
Lending additional support to the idea that repeated legal encouriters early in the life
course may lead to persistent deviant behaviour is found through examining the timing of
a possible second incarceration. SHOs who had fewer police contacts prior to incarceration
did take longer to return to custody than did SHOs with more police contacts. A few SHOs
who had 7 or less police contacts managed to survive 36 months without returning to
custody, which is alrnost 12 months longer than SHOs who had 8 or more police contacts
prior to incarceration. But, the difference in median survival times between the SHO
subgroups was minimal, 4.50 months for those with 7 or less police contacts and 4.27
months for SHOs with 8 or more police contacts. While the median survival tirne for SHOs
with fewer police contacts is longer than for SHOs with more police contacts, the largest
difference between deterrence and labelling effects was found within the nonSHO sample.
NonSHOs who had 7 or less police contacts prior to first incarceration dernonstrated strong
signs of deterrence indicated by a median survival time of 9.81 months. In cornparison,
nonSHOs with 8 or more police contacts indicated evidence of deviance amplification. On
average these nonSHOs returned to custody very quickly, shown by a median survival tirne
of only 1.10 months.
Hazard rates give evidence consistent with the above discussion. They reinforce the
concept that behaviour following incarceration varies according to the conditions or events
that occur prior to incarceration. Once again the hazard rates indicate that there are
differences in the probability of reincarceration between SHOs and nonSHOs, as well as
65
within each of the SHO and nonSHO populations. And, as seen in the previous tables the
hazard rates suggest that young offenders are at the highest risk of reincarceration during the
first 6 rnonths following release from custody, d e r which the risk of reincarceration
decreases. During the first interval SHOs with less than 7 police contacts had a hazard rate
of h(t) = 0.1667 and SHOs with 8 or more police contacts had a hazard rate of h(t) = 0.1806.
indicating that SHOs with more police contacts are less likely to return to custody than are
SHOs with fewer police contacts. This relationship rnay be mediated by some other factor,
such as age or the seriousness of the offenses, but at this point more research wouid need to
be done to cIariQ this relationship. NonSHOs who had less than 7 police contacts during the
first 6 months following release fiom custody had a low hazard rate at oniy h(t) = 0.0788,
but those with 8 or more police contacts had a significantly higher hazard rate at h(t) =
0.57 14 within oniy the first month.
Recall that Klemke's analysis of shoplifting suggests that youths who were exposed
to the poiice increase or ampli@ their involvement in shoplifiing, therefore labelling through
police interaction "explains at least in part, the high percentage of youth who continue to
shoplifl after being apprehended" (Klemke, 1978: 40 1). The data in Table 5 agree with
Kiernke's findings. The early experience of labelhg rnay cause the young offenders to move
to secondary deviance. Or, public labelling rnay begin to dissolve social bonds and thus limit
opportunities and resources for the youth. Regardless of why, it appears that young
offenders rnay have substantial experience with formal social controt before they are
incarcerated. Events such as police contact rnay serve to alter life trajectories before the
youth is ever incarcerated. Hence, incarceration rnay be another transition event along a
previously established cnmind trajectory, in which case deterrence may be unlikely.
It seems there are some general patterns or processes associated with juvenile
offending. An offender whose first police contact is at a young age may establish a pattern
of offending that is difficult to change even with incarceration. The sarne holds for an
offender who is fmt incarcerated at a young age. It therefore appears that age may effect the
probability of detenence. Offenders who have a hgh nurnber of police contacts pnor to
incarceration also appear to be persistent in their criminal careers. Hazard rates throughout
the analysis support the idea that certain offenders are at a greater risk of recidivism than are
others. Conditional deterrence and life course theories explain the varying risk as a result
of individual characteristics and events. It appears therefore that offending patterns rnay
unfold over time as an accumulation of experiences and perceptions.
CHAPTER 4: DISCUSSION AND CONCLUSION
4.0 THEORETICAL OVERVIEW -
The cnminal justice system punishes, through incarceration, young offenders in an
attempt to control and modifj criminal behaviours. This study has looked at the eariy
incarceration histories of young offenders as a possible tirne when criminai behaviours may
be in their developmental stages. Adolescence is a penod of change and growth and
therefore criminal behaviours rnay change in a short period of tirne.
Within a criminai career framework this research has investigated whether pattems
of offending are fixed or if they can be modified through incarceration. Conditional
deterrence theory suggests that individuais may either sustain or change behaviours
according to events and individual characteristics. Therefore specific conditions, the age of
the offender at first police contact, the nurnber of police contacts prior to incarceration, and
the age of the offender at first incarceration, were studied as events which may rnediate and
possibly change the course of offending. It was found that early contact with the police, a
high number of police contacts prior to incarceration, and early incarceration are al1 related
to persistent juvenile offending and thus repeated incarceration experiences.
One explanation as to why incarceration may change behavioral pattems was
fomulated in deterrence theory. [ndividual (specific) deterrence refers to the preventative
effects of punishment on the individuais who have been punished (Cook 1980; Paternoster
and Piquero 1995). Deterrence theory is based on the principle that if punishment is swifi,
certain, and severe then criminal behaviour can be restricted and recidivism reduced (Gibbs
1975; Hagan 199 1). Individuafs are believed to be rational beings from the deterrence
68
perspective (Cook 1980; Ehrlich 1973). They calculate the risk and potential benefits
associated with criminal activity and decide on a course of action. If punishment is thought
to be severe, certain, and swifi then the cost of crime wiIl be higher and rnay not outweigh
the gains. Although deterrence theory rnay explain sorne behaviour, it does not account for
the variable responses juveniles rnay have to incarceration.
Some offenders rnay persist in criminal activity even after punishment. A second
theory, discussed in section 1.2. suggests that continued offending rnay be the result of
labelling. Labelling theory suggests that deviance "is itself a product of reactions by a social
audience or by social control agents" (Tittle, 1975:399). Rather than detemng criminal
behaviour, labelling theory suggests that punishment c m perpetuate criminal activity.
Therefore, an individual who breaks society's d e s rnay be labelled as someone who cm not
conform to society's expectations and noms (Becker, 1963). They rnay not be able to resist
the deviant label and will consequently persist in their criminal behaviour regardless of
punishment or the threat of punishment. However, like deterrence theory, labelling theory
does not adequately account for the full s p e c t m of behaviour that rnay follow a period of
incarceration, therefore a broader theoretical perspective was dso considered.
Braithwaite (1989) presents a theory which integrates pnnciples from both labelling
and deterrence theory. Braithwaite (1 989) uses concepts of stigmatization, labelling, and
incorporates them into a mode1 of extra-legal controls. lf an individual is suEciently tied
or bonded to society then legal sanctioning rnay initially shame the offender, but eventually
it will reintegrate them into the community. Consequently, offending will decrease.
However, if the bonds to society are weak then punishment rnay resdt in an enduring
stigmatization and the offender may continue in risk taking behaviour.
Braithwaite's theory (1989) of reintegrative shaming offers an expianation for a wide
range of behavioral outcornes. But, yet another theoretical perspective attempts to
investigate and explain the offending process. Sherman and Berk (1984) propose a theory
based on conditional deterrence. They claim that the likelihood of deterrence is dependent
on several events and conditions other than sotely on the experience of incarceration. Tt thus
becomes possible that severd factors may either mitigate or enhance the deterrent effect of
incarceration. By including the age of the offender at first police contact and at first
incarceration and the nurnber of police contacts prior to incarceration this research was able
to address for whorn, under what conditions, and at what point in their criminal careers might
incarceration or punishrnent change the behaviours of juvenile offenders.
4.1 PERSISTENCE OR DESISTENCE OF OFFENDING -
In Chapter 3, findings fiom tables 2 through 5, indicate that the effect of incarceration
on the offending patterns of young offenders is complex. Some juvenile offenders'
involvement in risk taking behaviour will stop or decrease foIlowing incarceration, while
other young offenders will persist or possibly acceierate their cnminal involvement. This
relationship is dificuit to study without considering several factors. The findings suggest
that whether the offender was a SHO or a nonSHO, the nurnber of incarcerations, age at first
police contact, age at first incarceration, and the number of police contacts prior to first
incarceration al1 affect the probability of future incarceration during adolescence. For
exarnple, SHOs were more likely to return to custody, as were offenders whose first police
contact was at an early age.
70
In order to examine the effect of each of these variables on the likelihood of
incarceration it was necessary to examine events chronologicaily. In other words, the
ordering of events needed to be established so that their impact on incarceration could be
evaluated. Therefore a criminal career framework was used. According to Farrington,
"criminal career research requires exact information about the timing of offenses" ( 1992:
523). Consequently. this research used sunival analysis in order to examine the exact timing
of events. This strengthened the ability to describe the likelihood of reincarceration for the
selected sample of young offenders.
Defining deterrence as the length of time until an offender returned to custody
provided measures that readily illustrate some important differences not only between SHOs
and nonSHOs but aiso within each of the two subgroups. By definition it was expected that
SHOs and nonSHOs wouid respond differently to incarceration. By looking at longitudinal
individual level data these differences could be explored in greater detail. Further it allowed
the opportunity to generate some support theoretically as to which factors preceding
incarceration may be indicative of long term versus transient offending.
Developmental theory explains persistent criminal behaviour as sternrning from
various explanatory factors that affect crime differently at various points over the life course
(Paternoster and Brame 1997). In this case, repeated incarceration may weaken social bonds
and restrict opportunities thereby changing the life trajectories of individuals. Or, it is
possibie that once punishrnent has been experienced, the cost of getting apprehended for a
crime is less then had previously been calculated and thus criminal behaviour can continue.
W61e the continuation of criminal activity may be explained by criminal propensity
7 1
or low self controi (Gottfiedson and Hirschi 1990), the termination of crimind activity
following incarceration suggests a developmental sequence of behaviours. Criminal
propensity or low self control refers to the lack of ability individuals have to restrain
themselves fiom cornmitting criminal acts (Gottfiedson and Hirschi 1990). Offenders
therefore mature out of crime. Desisting fiom criminal activity or not returning to custody
rnay reflect this rather than a change in criminai propensity. While this is possible,
developmental and life course theones suggest that incarceration rnay alter pathways and
thereby result in a change in behaviour. Incarceration rnay deter subsequent offending in
certain circumstances, for example when the offender is older or has fewer poiice contacts,
as social bonds rnay have a greater influence on the behaviour these individuais following
incarceration
One weakness of this study lies in its inability to measure criminal propensity.
Therefore the existence of a stable and enduring propensity c m neither be supported nor
critiqued. Continuation of crirninal behaviour following punishrnent rnay !end support to the
concept of persistent heterogeneity but this is a measure of crime not crirninal propensity.
Therefore crimind propensity rnay not be adequately evaiuated through examining criminal
behaviour or a change in criminal behaviour.
It was expected that variation would be found between SHOs and nonSHOs in part
due to the sampling process, but the variation within each of the subgroups suggests that
behaviour is dependent on conditions other than this. While criminal propensity may explain
continuation of risk taking behaviour and a change in criminai behaviour for offenders as
they mature, it does not explain some of the other variation as well as devehpmental models
72
might. Developmental models expect and predict variability baçed on individual and
situational variables. fhey theorkc that age will effect the probability of deterrence, as will
the number of times incarcerated. Therefore the developmental perspective provides a
framework through which heterogeneous groups of offenders can be studied.
Concepts in both &te dependent and persistent heterogeneity perspectives are able
to explain the relationships found in this analysis. However, neither is completely
satisfactory on its own. Sarnpson and Laub (1 990) attempt to find a middle gound between
general theory and developmental theones in their life course model. This research too
suggests that this may be the best way to explain the relationships found between
incarceration and subsequent crirninal behaviour. The explanation of crime in Gottfiedson
and Hirschi's (1 990) general theory of crime is partially substantiated in this research. Some
criminal behaviours endure despite attempts to modiS and control them. This quite possibly
is the result of low self control. But, in contrast a high degree of variability is represented
in this data. The length of time it takes offenders to return to custody varied considerably.
Additionally, reincarceration \vas dependent on the nurnber of times the offender had been
incarcerated. This expenence is better explained through developmental models. Therefore
it is possible that components fiom each perspective are at work. Theoretically each expiains
part of young offenders' crirninal behaviours. So, perhaps it is possible to integrate
components of these two perspectives in order to best explain the processes involved in
j uvenile delinquency .
Findings of this study indicate that incarceration probably will not deter the subgroup
of young offenders who are largely responsible for criminal activity. After repeated exposure
73
to secured custody, they continue to engage in illegal or risk taking activity and possibly
even escdate their involvement. But, this is not to suggest that inmeration wilI never deter
offenders. Some young offenders were not incarcerated a second time, thus suggesting a
change in behaviour associated with incarceraiion. It appears necessaiy therefore to study
offenders using a criminal career framework in order to gain insight into a developmental
sequence of offending that is variable and conditional on characteristics that are specific to
the individual and environment.
4.2 - CRIMINAL KJSTICE SYSTEM POLICY
Of importance is what the relationship between incarceration and future offending
says about our criminal justice system's ability to prevent, control and punish juvenile
crirninal behaviour. "Knowledge of the etiology of offending is essential for the
development of preventative programs since the modification of etiological factors is the
backbone of prevention" (Lorion in Loeber and LeBlanc, 1990: 456). Gottfredson and
Kirschi argue that crime control is not to be found through the criminal justice system, but
instead through effective parenting. Other developmental theorists suggest that incarceration
policies need to focus on types of offenders (Bartusch, Lynam Moffitt and Silva, 1 997).
What may prove successfül for one type ofjuvenile delinquent is unlikely to be effective for
a different category of offender. The recommendations of deterrence theory on its own seem
unable to account for the variability of offending patterns. Providing swifi, certain and
severe punishment in today's society has proven difficuit to achieve. Possibly the integration
of deterrence with other theories would give it more explanatory power. Until more is
known about the causes of crirninal behaviour we cm do relatively little to change it. "Given
74
our imperfect state of knowledge of stable and variable causes of crime, conclusions about
optimal intervention are inherently premanire" (Loeber and LeBlanc, 1990).
4.3 CONCLUSION
Although this research begins to address sorne of the issues that have been debated
in crirninologicai studies on offending, it does not offer any definitive conc1usions. It was
found that some criminal behaviours were persistent over time while others changed,
possibly as a result of incarceration. But, the predictive and explanatory causes of stability
and change still require a more in-depth examination of the data.
It may also be necessary to include measures of informa1 social control or extra legal
factors in evaluating the utility of incarceration on criminal career modification. These
factors were not available in this study but should be considered in future studies.
Overall, this research provides the opportunity to continue the discussion of crime
and criminality. It looked at how different theoretical perspectives may best explain change
and stability in offending arnong young offenders. Additionally, it used the timing of the
next incarceration as a descriptive and sequenced event in the course of offending and
punishrnent. However, more research is needed to begin to corne to some agreement as to
why crime occurs and how to best control or modi@ it. Findings from this research need to
be validated using other types of methods and data. For example, interviewing juvenile
offenders would be beneficial to understanding the early stages of a criminal career. Also,
this research looked at statistical models to draw conclusions about a population of offenders,
but we should additionally explore more of the informal controls associated with juvenile
offending.
REFERENCES
Allison, Paul 1984 Event History Analysis: Regression for Longitudinal Event Data. Newbury Park:
Sage Publications.
Bartusch, Dawn, R., Donald R. Lynam, Teme E. Moffitt and Phil A. Silva 1997 1s Age Important? Testing A General Verus A Developmental Theory of Antisocial
Behaviour. Criminology 35 (1): 13 - 47.
Becker, Howard S. 1963 Outsiders: Studies in the Sociology of Deviance. New York: The Free Press.
Bernard, Thomas J. and Jeffery B. Snipes 1996 Theoretical [ntegration in Criminology. In Michael Tonry and Norval Morris (eds.).
Crime and Justice. Vol. 20, Chicago: University of Chicago Press.
Blurnstein, Alfied, Jacqueline Cohen and David P. Farrington 1988 Criminal Career Research: Its Value for Crirninology. Criminology 26(1): 1 - 35. 1988 Longitudinal and Criminal Career Research: Further Ciarifications. Criminology
26(1): 57 - 74.
Blumstein, Alfred, Jacqueline Cohen, and Daniel Nagin (eds.) 1978 Deterrence and Incapacitation: Estimating the Effects of Criminai Sanction on Crime
Rates. Washington, D.C.: National Academy of Sciences.
Blumstein, Alfred, Jacqueline Cohen, Jeffery A. Roth and Christy Visher (eds.) 1986 Criminal Careers and "Career Criminals". Vol. 1. Washington, D.C.: National
Academy Press.
Braithwaite, J. 1989 Crime, Shame, and Reintegration. Cambridge: University Press.
Caspi, Avshalom, Daryl J. Bem and GIen H. Elder, Jr. 1987 Moving Against the World: Life-Course Patterns of Explosive Children.
Developrnentai Psychology 23(2): 308 - 3 13.
Dean, Charles W., Robert Brame and Alex R. Piquero 1996 Criminal Propensities, Discrete Groups of Offenders and Persistence in Crime.
Criminology 34(4): 547- 574.
Ebensen, Fim and Scott Menard 1990 1s Longitudinal Research Worth the Price? The Criminologist: Officiai Newsletter
of the American Society of Crirninology. March-April 15(2).
Ehrlich, Isaac 1973 Participation in Illegitimate Activities: A Theoretical And Empirical Investigation.
Journai of Political Economy 8 1 : 52 1 - 565.
Farrington, David P. 1977 The Effects of Public Labelling. British Journal of Cnminology 17(2): 1 12 - 125. 1992 Criminal Career Research in the United Kingdom. British Journal of Crirninology
32(4): 521 - 536.
Gibbs, Jack P. 1975 Punishrnent and Deterrence: Theory, Research, and Pend Policy. In Leon Lipson and
Stanton Wheeler (eds.) Law and the Social Sciences, 3 19 - 368, New York: Russell Sage Foundation.
Geerken. Michael R. and Walter R. Gove 1975 Deterrence: Some Theoretical Consideratiom. Law and Society (Spring): 497 - 5 13.
Gonfredson, Stephen D. and Don M. Gottfredson 1994 Behavioral Prediction and the Problem of Incapacitation. Criminology 32(3): 441 -
474.
Gottfiedson, MichaeL and Travis Hirschi 1 986 The Tme Value of Lambda Would Appear to be Zero: an Essay on Career Criminais.
Criminal Careers, Selective Incapacitation, Cohort Studies, and Related Topics. Criminology 24(2): 213 - 234.
1988 Science, Public Policy, and the Career Paradigm. Criminology 26(1): 37 - 55. 1990 A General Theory of Crime. Stanford California: Stanford University Press.
Granovetter, Mark S. 1982 Getting a Job: A Study of Contacts and Careers. Cambridge, Massachusetts: Harvard
University Press.
Hagan, John 199 1 The Disreputable Pleasures: Crime and Deviance in Canada. Toronto: McGraw-Hill
Ryerson Lirnited.
Hirschi, Travis and Michael Gottfiedson 1983 Age and the Explanation of Crime. Arnerican Jounial of Sociology 89(1): 552 - 583.
Horney, Julie, D. Wayne Osgood and Ineke Haen Marshall 1995 Criminal Careers in the Short-Term: Intra-Individuai Variability in Crime and its
Relation to Local Life Circumstances. h e r i c a n Sociological Review 60: 655 - 673.
Hudson, Joe, Joseph P. Hornick and Barbara A. Burrows (Eds.) 1988 Justice and the Young Offender in Canada. Toronto: Wall & Thompson.
Hyman, Herbert H. 1972 Secondary analysis of sarnple surveys: principles, procedures, and potentialities. New
York: Wiley.
Keane. Carl, A. R. Gillis and John Hagan 1989 Deterrence and Amplification of Juvenile Delinquency by Police Contact. British
Journal of Crirninology 29(4): 336 - 352.
Kiecolt, Ji11 K. and Laura E. Nathan 1985 Secondary anaiysis of s w e y data. Beveriy Hills: Sage.
Klemke, Lloyd W. 1978 Does Apprehension for Shoplifting Amplifj or Terminate Shoplifting Activity? Law
and Society 12: 39 1 - 403.
Laub, John H. and Robert J. Sampson 1993 Turning Points in the Life Course: Why Change Matters to the Study of Crime.
Criminology 3 l(3): 30 1 - 325.
Le Blanc, Marc, Gilles Cote and Rolf Loeber 199 1 Temporal paths in delinquency: Stability, regression, and progression analyzed with
panel data from an adolescent and a delinquent male sample. Canadian Journal of Criminology 33: 23 - 44.
Lee, Elisa T. 1992 Statistical Methods for Survival Data Analysis. John Wiley and Sons: New York.
Lemert, Edwin 1972 Human Deviance, Social Problems and Social Control. 2nd edition Englewood Cliffs,
N.J.: Prentice-Hall.
Loeber, Rolf and Marc Le Blanc 1990 Toward a developmental criminology. In Michael Tonry and Norval Moms (eds.).
Crime and Justice. Vol. 12., Chicago: University of Chicago Press.
Mofitt, Teme E. 1993 Adolescence-limited and life-course persistence antisocial behaviour: a
developmental taxonomy . Psychological Review 1 00:674 - 70 1.
Murray, Charles A. and Louis A. Cox, Jr. 1979 Beyond Probation: Juvenile Corrections and the Chronic Delinquent. Beverly Hills:
Sage.
Nagin, Daniel S. and David P. Farrington 1992 The Onset and Persistence of Offending. Crirninoiogy 30(4): 50 1 - 523.
Nagin, Daniel S., David P. Farrington and Teme E. Moffitt 1995 Life-Course Trajectories of Différent Types of Offenders. Criminology 33(1): I 1 1 -
139.
Nagin, Daniel S. and Raymond Paternoster 199 1 On the relationship of past to future delinquency. Criminology 29: 163 - 189.
Paternoster, Raymond and Robert Brame 1997 Multiple Routes to Delinquency? A Test of Developmental and General Theories of
Crime. Criminology 35(1): 49 - 80.
Paternoster, Raymond and Alex Piquero 1995 Reconceptuaiizing Deterrence: An Empiricai Test of Personal and Vicarious
Experiences. Journal of Research in Crime and Delinquency 32(3): 25 1 - 286.
Rausch, Sharla 1983 Court Processing Versus Diversion of Status Offenders: A Test of Deterrence and
LabeIing Theories. Journal of Research in Crime and Delinquency January :39-5 3.
Regoli, Robert M., Eric D. Poole and Finn Esbensen 1985 Labeling Deviance: Another Look. Sociological Focus l8(l): 19-28.
Rogers, Joseph W. and G. Lamy Mays 1987 Juvenile Delinquency and Juvenile Justice. New York: John Wiley & Sons.
Sampson, Robert J. and John H. Laub 1990 Crime and Deviance Over the Life Course: The Salience of Adult Social Bonds.
Amencan Sociological Review 55: 609 - 627. 1993 Crime in the Making: Pathways and Turning Points Through Life. Cambridge:
Harvard University Press.
Schmidt, Peter and A. D. Witte 1988 Predicting Recidivism Using Survival Models. New York: Springer-Verlag.
Sherman, Lawrence W. 1992 Policing Domestic Violence: Experiments and Dilemmas. New York: The Free
Press.
Sherman, Lawrence W. and Richard A. Berk 1984 The Specific Deterrent Effects of Arrest for Domestic Assault. American
Sociological Review 49: 261 - 272.
Sherman, Lawrence W., Janell D. Schmidt, Dennis P. Rogan, Patrick R. Gartin, Ellen G. Cohn, Dean J. Collins and Anthony R. Bacich 199 1 From Initiai Deterrence to Long-Tem Escaiation: Short-Custody Arrest for Poverty
Ghetto Domestic Violence. Criminology 29 (4):821 - 849.
Sherman, Lawrence W., Douglas A. Smith, Janell D. Schmidt, and Dennis P. Rogan 1992 Crime, Punishment and Stake in Conformity: Legal and Extralegal Control of
Domestic Violence. Amencan Sociological Review 57(4).
Slade, Daryl 1996 Killer teen gets 20 montfis. The Calgary Heraid, August 3 1, pages B 1, B2
Smith, Douglas A., and Robert Brame 1994 On the Initiation and Continuarion of Delinquency. Crirninology 4(32): 607 - 629.
St-Amand, Carol and Peter Greenberg 1996 Youth Custody and Probation in Canada 1994-95. Juristat: Canadian Center for
Justice Statistics 16(5).
Stafford, Mark C. and Mark Warr 1993 A Reconceptualization of General and Specific Deterrence. Journal of Research in
Crime and Delinquency 30(2): 123 - 135.
Stattin, Hakan, David Magnusson, and Howard Reichel 1989 Criminal Activity at Different Ages: A Study Based on a Swedish Longitudinal
Research Population. British Journal of Cnminology 29(4): 368 - 385.
Thorsell, Bernard A. and Lloyd W. Klemke 1972 The Labelling Process: Reinforcement and Deterrent? Law and Society Review
(February): 393 - 403.
Tittle. Charles R. 1975 Deterrents or Labelling? Social Forces 53 : 399 - 4 10.
Visher. Christy A.. Parnela K. Lattimore and Richark L. Linster 1991 Predicting the Recidivism of Serious Youthfül Offenders Using Survival Models.
Criminology 29(3): 329 - 366.
Walker. Nigel, David P. Famngton and Gillian Tucker 1982 Reconviction Rates of Adult Males Afier Different Sentences. British Journal of
Criminology 2 l(4): 357 - 360.
Wall, W. D. and H. L. Williams 1970 Longitudinal Studies and the Social Sciences. London: Heinemann.
Williams. Kirk R. and Richard Hawkins 1986 Perceptual Research on General Deterrence: A Critical Review. Law and Society
Review 20(4): 545 - 570.
Wilson. James Q. 1985 Thinking About Crime. New York: Vintage Books.
Yamaguchi. Kazuo 1991 Event History .4nalysis. Newbury Park: Sage Publications.
Zimring. Franklin E. and Gordon J. Hawkins 1973 Deterrence: The Legal Threat in Crime Control. Chicago: The University of Chicago
Press.
TEST TARGET (QA-3)
APPLIED IMAGE. Inc 1653 East Main Street - - . Rochester, NY 14609 USA -- --= Phone: 71 6/48243W --
, Fax: 71-88-5989
O t 993. Applied Image. lx., Ail Rights R e s e ~ e d