Professional Documents
Culture Documents
J. DAVID HAWKINS
RICHARD F. CATALAN0
KARL G. HILL
RICK KOSTERMAN
University of Washington
* For helpful comments, we are very grateful to Alfred Blumstein, Delbert Elliott,
Marc LeBlanc, Alex Piquero and Terence Thornberry, as well as to the editor and the
referees.
CRIMINOLOGY VOLUME
41 NUMBER3 2003 933
FARRINGTON ET AL.
1987). However, the peak age of offending varies with different types of
crimes, being relatively early for shoplifting and relatively later for drug
offenses (Tarling, 1993:18). This paper compares how the prevalence of
delinquency varies with age according to self-reports and court referrals.
Although the variation in the prevalence of offending with age is well
established, this is not true of the variation in the frequency of offending
with age. The total number of offenses committed at any age can be disag-
gregated into the number of different offenders (prevalence) and the aver-
age number of offenses committed per offender (the individual offending
frequency, termed lambda by Blumstein et al., 1986). The individual
offending frequency is a crucial criminal career parameter with important
theoretical and policy implications (e.g., for estimating the incapacitative
effects of incarceration).
The British and American studies reviewed by Farrington (1986) sug-
gested that the individual offending frequency did not vary systematically
with age, in official records or self-reports. However, based on a large
sample of juvenile court careers in Arizona, Loeber and Snyder (1990:106)
found that it increased steadily from age 9 to age 16. It also increased
between ages 10 and 14 according to self-reports in the Denver, Pitts-
burgh, and Rochester longitudinal surveys (Kelley et al., 1997). Also, in a
systematic comparison of official records in London and Stockholm, Far-
rington and Wikstrom (1994:76) concluded that the individual offending
frequency stayed constant with age in London but increased to a peak at
age 15 (and then decreased) in Stockholm. This paper compares how the
individual offending frequency varies with age in self-reports and court
referrals.
Changes in the individual offending frequency with age have theoretical
significance in criminology. Hirschi and Gottfredson (1983) argued that
prevalence and individual offending frequency both reflected the same
underlying construct of criminal propensity and, hence, that both should
vary similarly with age. On the other hand, Blumstein et al. (1988) con-
tended that prevalence and individual offending frequency were concep-
tually distinct, could have different predictors, and hence could vary
differently with age. It was ironic that the age-crime relationship termed
invariant by Hirschi and Gottfredson (1983) was the source of the most
important difference in criminal careers between the London and Stock-
holm cohorts.
Generally, there is significant continuity between offending in one age
range and offending in another. What this means is that there is relative
stability in offending, so that people who commit relatively many offenses
during one age range tend also to commit relatively many offenses during
a later age range. Establishing the degree of continuity is important in
predicting future criminal careers. In a longitudinal survey of over 400
936 FARRINGTON ET AL.
age 11 data were substituted, and for youths missing at age 13, their age 14
data were substituted.3 This increased the numbers known on self-reports
to 804 at age 12 and 788 at age 13. The numbers known on court referrals
were those known from the survey data to be resident in King County
during the time period. These numbers were increased slightly by assum-
ing that those who were not interviewed at one age (e.g., 15) but were
known to be resident at the next age (e.g., 16) were in fact resident at the
first age (e.g., 15). After these adjustments, information was available for
over 760 youths on self-reports at each age, and for over 700 youths on
court referrals at each age. The number of youths known on both varied
from 795 at age 11 to 663 at age 17 (Table 1).
Self-report information was available for eight types of offenses:
1. Burglary (Broken into a house, store, school, or other building
without the owners permission): available at all ages.
2. Vehicle theft (Taken a motor vehicle, such as a car or motorcycle,
for a ride without the owners permission): available from age 13.
3. Larceny (Taken things worth (a) $5 or less, (b) more than $5 but
less than $50, or (c) more than $50): available at all ages.
4. Robbery (Used a weapon or force to get money or things from
people): available from age 13.
3. The youths missing were not significantly different from those known on self-
reported delinquency. For example, exactly the same fraction of youths missing or
known at age 12 admitted at least one offense at age 11 (30.3% of those missing at age
12; 30.3% of those known at age 12). At age 14, 43.0% of those missing at age 13
admitted at least one offense, compared with 47.2% of those known at age 13 (chi-
squared = 0.41, ns.). Because of a gap in funding, the youths were not interviewed at
age 17. For ease of exposition, the age 18 self-reports in spring 1993 are treated as
though they were age 17 self-reports.
940 FARRINGTON ET AL.
RESULTS
PREVALENCE OF OFFENDING
Table 2 shows the annual prevalence of offending at each age from 11 to
17, according to both court referrals and self-reports.4 The offenses cov-
ered at each age were the same in both; for example, because only bur-
glary, larceny, vandalism, and marijuana use were known in self-reports at
age 11, only court referrals for these four offenses were included at age 11.
All eight offenses were known in both court referrals and self-reports from
age 13 onward. As expected, the prevalence of offending was much higher
in self-reports at all ages. In total, 85.9% of youths admitted committing
at least one of these eight offenses, and 34.0% were referred to court for at
least one offense.
The prevalence of offending increased with age in both the court and
self-report data. Part of the increase between ages 12 and 13 was attribu-
table to the increased number of offenses covered (from four at age 12 to
eight at age 13). This had very little effect on the court data; the preva-
lence of court referrals at age 12 was 2.5% based on all eight offenses,
compared with 2.1% based on four offenses. It had more effect on the
self-report data; the prevalence of self-reported offending at age 13 was
29.1% based on four offenses, compared with 41.5% based on all eight
offenses.
It is plausible to conclude that there was a steady increase in the preva-
lence of self-reported offending with age; the relatively high prevalence at
4. In general, the self-reports covered offending ever and in the last year at
each assessment. Both sources of data were used in estimating prevalence.
COMPARING DELINQUENCY CAREERS 941
age 11 was inflated slightly by the inclusion of ever data. There was a
sharp increase in the prevalence of court referrals between grades 6 and 7,
ages 12 and 13 (for all 8 offenses, from 2.5% to 8.0%; chi-squared = 22.3, p
< .0001). This probably reflects the fact that juveniles aged up to 12 are
handled differently from older juveniles at all stages of juvenile justice,
including detention, referrals, adjudication, and placement (Snyder, 2001).
For example, Washington State law specifies that, up to age 12, the court
has to prove that children knew that they were doing wrong before they
can be charged.
Table 2 also shows the prevalence of offenses of different types accord-
ing to court referrals and self-reports. As expected, the prevalence of
offending was higher in self-reports for all eight types of offenses. The two
942 FARRINGTON ET AL.
prevalence estimates were most similar for vehicle theft (court 23.8%, self-
report 33.1%) and most dissimilar for marijuana use (court l.8%, self-
report 49.1 Yo). In general, the most prevalent offenses according to court
referrals were also the most prevalent according to self-reports. The main
exceptions were vehicle theft (relatively more prevalent in court referrals)
and marijuana use (relatively more prevalent in self-reports).
The eight types of offenses were grouped into three categories: property
(burglary, vehicle theft, and larceny), aggressive (robbery, assault, and
vandalism), and drug offenses (marijuana use and drug selling).S Table 2
shows that self-report prevalences were higher than court referral
prevalences, especially for drug offenses (4.5% and 50.8%, respectively).
Table 3 shows changes with age for categories of offenses for the compara-
ble age ranges 12-13,14-15, and 16-17. The prevalence of court referrals
increased significantly for all categories of offenses between ages 12-13
and 14-15, and then did not change significantly up to age 16-17. The
prevalence of self-reported offenses generally increased both between
ages 12-13 and 14-15 and between ages 14-15 and 16-17.
increased with age, but that this increase was not seen in court referrals
because of limitations on the number of court referrals per offender per
year. The maximum number of court referrals in one year was 14, and
there were only 8 cases (out of a possible 5,156) when a youth had 10 or
more referrals in one year. This is partly because frequent offenders are
detained (and hence are not at risk of committing more offenses) and
partly because prosecutors tend to refer only the more serious offenses of
frequent offenders.
Table 2 also shows individual offending frequencies for the eight differ-
ent types and three categories of offenses. As expected, frequencies were
always higher in self-reports. The difference between court and self-report
data was least for burglary (court 1.6, self-report 3.2) and greatest for drug
offenses (court 1.8, self-report 41.1). Table 3 shows that the frequency of
court referrals did not change significantly among ages 12-13, 14-15, and
16-17 for any of the three categories of offenses. However, for total
offenses, the frequency of court referrals decreased significantly between
ages 14-15 and 16-17. For all categories of offenses, the frequency of self-
reported offending increased between ages 12-13 and 14-15 but then did
not change significantly. However, the frequency for total offenses
increased both between ages 12-13 and 14-15 and between ages 14-15
and 16-17.
CONTINUITY
Continuity in the prevalence of offending between ages 12-13 and
14-15, and between ages 14-15 and 16-17, was investigated separately for
court referrals and self-reports. The odds ratio was used to measure the
strength of each relationship. For example, Table 4 shows that 50% of
youths with a court referral for a property offense at age 12-13 also had a
court referral for a property offense at age 14-15, compared with only
11.7% of youths with no court referral for a property offense at age 12-13
(odds ratio = 7.5, 95% confidence interval 4.3-13.2, p < .05); 59.9% of
youths who self-reported a property offense at age 12-13 also self-
reported a property offense at age 14-15, compared with only 25% of
youths who did not self-report a property offense at age 12-13 (odds ratio
= 4.5, 95% confidence interval 3.3-6.1, p < .05).
The results were very consistent. There was significant continuity for all
types and categories of offenses between all age ranges. Table 4 shows the
results for categories of offenses. Based on odds ratios, continuity was
generally higher for court referrals than for self-reports. This may possibly
reflect bias in official processing (e.g., picking up known offenders), or
alternatively the fact that those who are referred to court are a more
extreme group who are more likely to persist. Continuity was generally
highest for drug offenses and lowest for property offenses.
COMPARING DELINQUENCY CAREERS 945
CHRONIC OFFENDERS
results were obtained for property and aggressive offenses separately. For
property offenses, 22.2% of official chronics (10 out of 45) were among the
32 self-reported chronics, compared with 3.2% of official nonchronics,
odds ratio = 8.6; for aggressive offenses, 39.1% of official chronics (9 out
of 23) were among the 43 self-reported chronics, compared with 4.9% of
official nonchronics, odds ratio = 12.4, p < .05. The numbers of official
chronics in drug offenses were too small to repeat this analysis.
948 FARRINGTON ET AL.
AGE OF ONSET
Table 7 shows that, as expected, an early age of onset predicted a large
number of offenses in both self-reports and court referrals. Youths with a
first court referral at age 11-12 had 11.4 referrals on average, compared
with 5.4 for those with a first court referral at age 13-14 ( t = 2.9, p < .Ol).
Similarly, those with a first self-reported offense at age 11-12 had 65.7
self-reported offenses on average, compared with 47.2 for those with a first
offense at age 13-14 ( t = 2.1, p < .05). Also, youths with an onset of
offending at age 13-14 committed more offenses than those with an onset
at age 15-16.
In general, the first self-reported offense occurred before the first court
referral (average ages 12.7 and 15.1, respectively).6 This was investigated
more systematically in cases where a youth had both a self-report and a
court referral. For all offenses, the self-report came first in 84.9% of cases,
which had an average time interval of 2.9 years to the first court referral.
The figures were similar for property (78.9% and 2.7 years), aggressive
(81.0% and 2.8 years), and drug offenses (82.6% and 2.4 years).
The average number of offenses per year after onset decreased with
increasing age of onset for court referrals, from 1.9 per year for those with
a first referral at age 11-12 to 0.9 per year for those with a first referral at
age 13-14 and 0.5 per year for those with a first referral at age 15-16.
Similar decreases were seen for property, aggressive and drug offenses
separately.
The average number of self-reported offenses per year after onset
decreased only slightly (and nonsignificantly) with increasing age of onset,
from 10.9 per year for those with a first offense at ages 11-12 or 13-14 to
9.3 per year for those with a first offense at age 15-16. The average num-
ber of offenses per year after onset decreased with increasing age of onset
from age 13-14 to age 15-16 only for property offenses. Surprisingly, the
average number of offenses per year after onset did not decrease with
increasing age of onset from age 11-12 to age 13-14 for any category of
offenses; in fact, there was a significant (p < .005) increase for drug
offenses, with later starters being more frequent offenders.
An early age of onset (age 11-12) predicted a high rate of offending
after onset in court referrals, but not in self-reports. This may be a conse-
quence of the reluctance to refer youths aged 11-12 (in grades 5 and 6) to
court after offending. Those who were referred may have been a rela-
tively extreme group with a high frequency of offending.
YO Per Offender
Age
11 2.2 0.8
12 3.8 0.8
13 11.5 2.3
14 15.3 2.7
15 22.3 4.2
16 21.0 2.4
17 15.3 1.7
Total 36.7 3.5
Offense Type
Burglary 17.7 6.9
Vehicle theft 36.6 9.9
Larceny 30.6 4.9
Robbery 12.5 2.4
Assault 18.0 3.6
Vandalism 14.3 3.3
Marijuana use 3.4 0.1
Drug selling 13.6 0.6
Property 31.7 7.2
Aggressive 22.5 3.9
Drug 8.2 0.4
each age. The probability of a court referral per offender and per offense
increased sharply between ages 12 and 13 (from 3.8% to 11.5% of offend-
ers, and from 0.8% to 2.3% of offenses). Both probabilities increased to a
peak at age 15 and then decreased. In total, 36.7% of self-reported
offenders were referred to court, but only 3.5% of self-reported offenses
led to a court referral.7
Table 9 shows that the probability of a court referral increased signifi-
cantly between ages 11-12 and 13-15. As mentioned earlier, there is a
reluctance to refer juveniles under age 13 to the court. Another possibility
is that younger children over-report trivial acts, but this would not explain
the sharp increase between ages 12 and 13. The probability of a court
referral per offender decreased between ages 13-15 and 16-17 for prop-
erty offenses but did not change significantly for aggressive, drug, or total
7. Only students with at least one self-reported offense were included in these
analyses.
COMPARING DELINQUENCY CAREERS 95 1
CONCLUSIONS
Like most research, this study is based on one particular sample in one
particular place at one particular time and, hence, it is unclear how far its
results can be generalized. The analyses are based only on eight types of
offenses, and they cover only the juvenile years from age 11 to age 17.
Nevertheless, the study is based on a large urban sample, uniquely con-
tains comparable annual data on court referrals and self-reports over the
juvenile age range, and provides the most extensive existing comparisons
of delinquency career findings according to court records versus self-
reports. There were important similarities and important differences
between results obtained from the two sources.
As expected, the prevalence of offending was greater in self-reports
than in court referrals, especially for drug offenses. In both court referrals
and self-reports, the prevalence of offending generally increased with age.
Whereas there was a steady increase in the prevalence of self-reported
offending, there was a sharp increase in the prevalence of court referrals
COMPARING DELINQUENCY CAREERS 953
between ages 12 and 13 (grades 6 and 7), probably reflecting the reluc-
tance to refer very young offenders to the juvenile court. There was signif-
icant continuity in offending from one age range to the next in both self-
reports and court referrals, but the continuity in court referrals was
greater.
Court referrals and self-reports clearly differed in their conclusions
about whether the individual offending frequency varied or stayed con-
stant with age. This frequency stayed constant in court referrals but
increased in self-reports. In agreement with Hirschi and Gottfredsons
(1983) argument, prevalence and frequency varied similarly with age in
self-reports. It is possible that the frequency of offending truly increased,
but that this increase was not seen in court referrals because of limitations
on the number of referrals per offender per year. The individual offending
frequency was much greater in self-reports than in court referrals, espe-
cially for drug offenses.
In both court records and self-reports, a small fraction of the sample
(the chronic offenders) accounted for half of all the offenses. However,
the concentration of offending, and hence the importance of chronic
offenders, was greater in self-reports, again because of limitations on the
number of court referrals per offender per year. There was a significant
overlap between chronic offenders identified in court referrals and chronic
offenders identified in self-reports. Therefore, to a considerable extent,
self-reports and court referrals identified the same people as the worst
offenders. Further research is needed to investigate characteristics of self-
reported chronics who are not official chronics (possibly the most success-
ful offenders) and of official chronics who are not self-reported chronics
(possibly the least successful offenders).
As expected, an early age of onset predicted a large number of offenses
in total. However, in self-reports, this was largely because early onset
offenders had more time at risk after onset. An early age of onset (age
11-12) predicted a high rate of offending per year after onset only in court
referrals, not in self-reports. It is possible that the reluctance to refer
youths aged 11-12 (in grades 5 and 6) meant that those who were referred
were relatively extreme in their persisting criminal potential.
The first self-reported offense was, on average, 2.4 years before the first
court referral (Table 7). Given the average self-reported rate of offending
after early onset of about 11 offenses per year, it can be estimated that the
average offender committed about 26 offenses before the first court refer-
ral. Research is needed to investigate the characteristics of youth who
commit large numbers of offenses before their first court referral, and how
far in advance they can be predicted. Prevention efforts before the age of
first court referral have the potential to reduce crime significantly.
The criminal justice system was not very efficient in bringing offenders
FARRINGTON ET AL.
REFERENCES
Blumstein, Alfred, Jacqueline Cohen, and David P. Farrington
1988 Criminal career research: Its value for criminology. Criminology 26:l-35.
Blumstein, Alfred, Jacqueline Cohen, Jeffrey A. Roth, and Christy A. Visher (eds.)
1986 Criminal Careers and Career Criminals. Washington, D.C.: National
Academy Press.
Dunford, Franklyn W. and Delbert S. Elliott
1984 Identifying career offenders using self-reported data. Journal of Research
in Crime and Delinquency 21:57-86.
Elliott, Delbert S.
1995 Lies, damned lies and arrest statistics. Sutherland Award presentation to
the American Society of Criminology. Boston.
Elliott, Delbert S. and Harwin L. Voss
1974 Delinquency and Dropout. Lexington, Mass.: Lexington Books.
Elliott, Delbert S., David Huizinga, and Scott Menard
1989 Multiple Problem Youth: Delinquency, Substance Use and Mental Health
Problems. New York: Springer-Verlag.
FARRINGTON ET AL.
Farrington, David P.
1986 Age and crime. In Michael Tonry and Norval Morris (eds.), Crime and
Justice. Vol. 7. Chicago: University of Chicago Press.
1989 Self-reported and official offending from adolescence to adulthood. In
Malcolm W. Klein (ed.). Cross-National Research in Self-Reported Crime
and Delinquency. Dordrecht. The Netherlands: Kluwer.
1990 Age, period. cohort, and offending. In Don M. Gottfredson and Ronald
V. Clarke (eds.), Policy and Theory in Criminal Justice: Contributions in
Honor of Leslie T. Wilkins. Aldershot, England: Avebury.
1992 Criminal career research in the United Kingdom. British Journal of
Criminology 32521-536.
1997 Human development and criminal careers. In Mike Maguire. Rod
Morgan, and Robert Reiner (eds.), The Oxford Handbook of Criminol-
ogy. 2d ed. Oxford: Clarendon Press.
Farrington, David P. and Donald J. West
1993 Criminal, penal and life histories of chronic offenders: Risk and protective
factors and early identification. Criminal Behavior and Mental Health
31492-503.
Farrington, David P. and Per-Olof H. Wikstrijm
1994 Criminal careers in London and Stockholm: A cross-national comparative
study. In Elmar G. M. Weitekamp and Hans-Jurgen Kerner (eds.), Cross-
National Longitudinal Research on Human Development and Criminal
Behavior. Dordrecht, The Netherlands: Kluwer.
Farrington, David P., Sandra Lambert, and Donald J. West
1998 Criminal careers of two generations of family members in the Cambridge
Study in Delinquent Development. Studies on Crime and Crime
Prevention 7:85-106.
Fox, James A. and Paul E. Tracy
1988 A measure ol skewness in offense distributions. Journal of Quantitative
Criminology 4:259-274.
Hamparian, Donna M.. Richard Schuster, Simon Dinitz, and John P. Conrad
1978 The Violent Few. Lexington, Mass.: Heath.
Hawkins, J. David, Brian H. Smith, Karl G. Hill, Rick Kosterman. Richard F.
Catalano, and Robert D. Abbott
2003 Understanding and preventing crime and violence: Findings from the
Seattle Social Development Project. In Terence P. Thornberry and
Marvin D. Krohn (eds.), Taking Stock o f Delinquency: An Overview of
Findings from Contemporary Longitudinal Studies. New York: Kluwer/
Plenum.
Hindelang, Michael J., Travis Hirschi. and Joseph G . Weis
1981 Measuring Delinquency. Beverly Hills, Calif.: Sage.
Hirschi, Travis and Michael R. Gottfredson
1983 Age and the explanation of crime. American Journal of Sociology
89552-584.
Homey. Julie and Ineke Haen Marshall
1991 Measuring lambda through self-reports. Criminology 29:471-495.
COMPARING DELINQUENCY CAREERS 957
Kelley, Barbara T., David Huizinga, Terence P. Thornberry, and Rolf Loeber
1997 Epidemiology of serious violence. Washington, D.C.: Office of Juvenile
Justice and Delinquency Prevention.
Krohn, Marvin D., Terence P. Thornberry, Craig Rivera, and Marc LeBlanc
2001 Later delinquency careers. In Rolf Loeber and David P. Farrington
(eds.), Child Delinquents: Development, Intervention, and Service Needs.
Thousand Oaks, Calif.: Sage.
LeBlanc, Marc and Marcel Frechette
1989 Male Criminal Activity from Childhood through Youth. New York:
Springer-Verlag.
Loeber, Rolf and Howard N. Snyder
1990 Rate of offending in juvenile careers: Findings of constancy and change in
lambda. Criminology 28:97-109.
Moffitt, Terrie E.
1993 Adolescence-limited and life-course-persistent antisocial behavior: A
developmental taxonomy. Psychological Review 100:674-701.
Nagin, Daniel S. and David P. Farrington
1992 The onset and persistence of offending. Criminology 30:501-523.
Piquero, Alex R.
2000 Assessing the relationships between gender, chronicity, seriousness, and
offense skewness in criminal offending. Journal of Criminal Justice
28:103-1 15.
Piquero, Alex R., David P. Farrington, and Alfred Blumstein
2003 The criminal career paradigm. In Michael Tonry (ed.), Crime and Justice.
Vol. 30. Chicago: University of Chicago Press.
Snyder, Howard N.
2001 Epidemiology of official offending. In Rolf Loeber and David P.
Farrington (eds.), Child Delinquents: Development, Intervention, and
Service Needs. Thousand Oaks, Calif.: Sage.
Tarling, Roger
1993 Analyzing Offending: Data, Models and Interpretations. London: Her
Majestys Stationery Office.
Tolan, Patrick H. and Peter Thomas
1995 Implications of age of onset for delinquency risk. 11: Longitudinal data.
Journal of Abnormal Child Psychology 23:157-181.
West, Donald J. and David P. Farrington
1977 The Delinquent Way of Life. London: Heinemann.
Wolfgang, Marvin E.. Robert M. Figlio, and Thorsten Sellin
1972 Delinquency in a Birth Cohort. Chicago: University of Chicago Press.
Wolfgang, Marvin E., Terence P. Thornberry, and Robert M. Figlio
1987 From Boy to Man, from Delinquency to Crime. Chicago: University of
Chicago Press.