|
Items in cart [0] |
Questions? Call 888-624-8373 |
| Copyright © 2009. National Academy of Sciences. All rights reserved. Terms of Use and Privacy Statement |
Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 211
Appendix A
Participation in
Criminal Careers
Christy A. Visher and Jeffrey A. Roth
INTRODUCTION
This appendix is concerned with those
who participate in criminal careers. More
specifically, it reviews estimates of the frac-
tion of the population that commits at least
one crime during some observation penod.
This fraction, called the participation [ever, is
of interest as an indicator ofthe pervasiveness
of delinquents and criminals in society, and
many find the estimates some exceeding 60
percent over the lifetime of urban males
surprisingly high. But more important from
the perspectives of testing theory and devel-
oping policy is an understanding of how par-
ticipation varies across subpopulations, and of
what factors are associated with greater risk of
future participation.
The authors wish to thank Alfred Blumstein for
stimulating the development of this appendix. We
are grateful for helpful comments by many partici-
pants in the Workshop on Criminal Career Re-
search, especially David Farrington and Robert
Gordon. Delbert Elliott, Lyle Shannon, Paul Tracy,
Neil Weiner, and Marvin Wolfgang helped us by
providing special tabulations and interpretations of
their data. We also appreciate the skillful editing by
Jean Shirhall. We are responsible for any remaining
errors in the appendix.
Obstacles to the Understanding and
Measurement of Pervasiveness
Given the importance of measuring par-
ticipation, it is unfortunate that, as noted by
Gordon (1976), substantial confusion and
ambiguity exist in the literature concerning
appropriate measures of pervasiveness and
their definitions. The student of participa-
tion is thus confronted with such terns as
"arrest probability" (Christensen, 1967~;
"offender rate" or "'real' rate of onset"
(Wolfgang, Figlio, and Sellin, 1972: 133~;
"probability of committing at least one of-
fense" (Wolfgang, Figlio, and Sellin:281~;
"prevalence," used in somewhat different
senses by Gordon (1976), Blumstein and
Graddy (1982), and Elliott et al. (1983~;
"incidence," used in two senses by T.
Monahan (1960) and in other senses by
Gordon (1976), Elliott et al. (1983), and
Farrington (1983a); "static prevalence" (Lit-
tle, 19651; "delinquency rate" (Hindelang,
Hirschi, and Weis, 1981~; "hazard rate"
(Gordon and Gleser, 1974~; "age-specific
risk" and "age-specific rate," defined di~er-
ently but used interchangeably (Ball, Ross,
and Simpson, 1964~; and "criminality"
(Hutchings and Mednick, 1975~.
While each of these tens reflects some
277
OCR for page 212
212
aspect of pervasiveness, standardization is
lacking across authors with respect to the
base (e.g., a cohort, a population, surviving
cohort members, surviving cohort members
not already offenders) and the observation
period (e.g., lifetime, lifetime through a
stated age, preceding year, time between
record updates, time not incarcerated be-
tween record updates). Even when the
measures are standardized with respect to
these variables, their values depend on the
domain of"crime" in which participation
occurs (e.g., all offenses, index crimes, felo-
nies, specific crime types) and on the par-
ticipation threshold (self-reported commis-
sion, sel£reported police contact, recorded
police contact, court referral, conviction).
Therefore, comparison of estimated values
across studies is not at all straightforward.
As explained by Gordon (1976), the root
of some of the confusion is imprecise adap-
tation of the epidemiological concepts of
`d ,, ~ do. . ~ ,, ~
preva hence ancl 1nclaence. AS le ex-
plains, prevalence is generally intended to
refer to the fraction of a group currently
experiencing a condition, such as heart dis-
ease. Incidence describes the group's expe-
rience over an interval of time, and the
incidence of heart disease during a year is
the number of contractions of heart disease
divided by the size of the population. Thus,
"prevalence" is a concept counting persons,
ad. . ,. .
Inch ence IS a concept counting occur-
rences, and both concepts use the popula-
tion as a base.
Following Gordon (1976:209), "choosing
a verbal label for the kind of rate that is of
main interest here is a matter of discretion."
However, choosing a consistent set of labels
that permits comparisons across studies is
important for understanding. To avoid mis-
understandings, we refer to the concept as
participation.
The study of participation would be
straightforward if individuals maintained
and made available accurate diaries oftheir
legal and illegal activities, including pre-
cise dates. By using the diaries, it would be
easy to measure the fraction participating in
robbery, or the larger fraction participating
in a broader category of crime, such as "FBI
index offenses," or the still larger fraction
CRIMINAL CAREERS AND CAREER CRIMINALS
that has ever committed a nontragic offense.
At the aggregate level, the participation
fraction could be tabulated across subpop-
ulations of interest. At the individual level,
a binary indicator of participation-nonpar-
ticipation could be analyzed using sophisti-
cated statistical techniques to identify fac-
tors associated with a higher probability of
participation.
Unfortunately, accurate diaries for repre-
sentative samples of individuals do not ex-
ist. In their place two imperfect devices are
commonly used to measure participation-
criminal justice agency records of arrests,
court referrals, or convictions and the self-
reports of survey respondents. Official rec-
ords cover only offenders whose participa-
tion comes to police or court attention at
least once. Presumably, therefore, those of-
fenders commit more serious crimes than
other offenders, and they may be unrep-
resentative in other ways as well (e.g., less
adept at avoiding detection, more often un-
der the influence of drugs or alcohol, more
often from neighborhoods under intensive
police patrol). Also, because official records
are maintained for operational rather than
research purposes, special efforts are
needed to augment them with other perti-
nent information beyond the demographic
data used in police identification. Self-
reports of participation, on the other hand,
may be sought from more representative
samples and may easily be augmented with
information on other variables that are hy-
pothesized to be risk factors. But the results
from the sample that actually responds to
the survey may be biased (e.g., if serious
offenders are less likely than nonoffenders
to cooperate with interviewers). Also, the
self-reports may be distorted by the respon-
dents' failure to recall events accurately, by
their misunderstanding of instructions, and
by their intentional deception.
These and other problems with official-
record and self-report participation mea-
sures are discussed by Weis (Volume II). As
explained in this appendix, however, many
findings obtained by using one of the ap-
proaches can be reconciled with findings
obtained by using the other. Before report-
ing findings, we first discuss the importance
OCR for page 213
APPENDIX A: PARTICIPATION IN CRIMINAL CAREERS
of understanding participation and then
specify a conceptual framework for analyz-
ing it.
Importance of Participation
Studies of participation have long been
recognized as valuable for both intellectual
understanding and policy development (T.
Monahan, 1960; Ball, Ross, and Simpson,
1964~. Intellectually, as noted by Blumstein
and Graddy (1982), specific attention to of-
fender participation is a first step in improv-
ing the understanding of the "causes of
crime." Many studies ofthe causes of crime
have analyzed aggregate crime rates. How-
ever, the aggregate crime rate may mask
variation in participation levels, in individ-
ual offending frequencies, or in the average
duration of criminal careers.
There is no a priori reason to assume that
a common set of causes influences individ-
uals' offending participation, frequency,
and duration. Thus, disaggregating crime
rates into those dimensions is an essential
preliminary step if multiple causal struc-
tures are to be discovered or confirmed
empirically. For example, only by partition-
ing the aggregate crime rate into its multi-
ple dimensions can one design studies that
allow for the possibility that one set of
factors (e.g., peer influences, family stress,
and school problems) is associated with
participation, a second (e.g., economic
needs, situational stress, opportunity) with
the frequency of serious offending, and a
third (e.g., effects of aging, the assumption
of legitimate adult activities) with termina-
tion of the criminal career. While the sepa-
ration of these dimensions by no means
rules out the possibility that some factors
influence all three dimensions, the analyti-
cal separation facilitates testing the hy-
pothesis that different forces are operative
at the three career stages.
If different sets of individual characteris-
tics are associated with the respective ca-
reer dimensions or stages, identifying them
could improve the efficiency and equity of
resource allocations within the criminal jus-
tice system and elsewhere. For example,
subject to ethical constraints, knowledge of
2~3
patterns in participation could suggest strat-
egies for designing community-based pre-
ventive programs that reduce criminal par-
ticipation and for giving high-risk groups of
children special priority in admissions to
the programs. Similarly, if factors that dis-
tinguish offenders from nonoffenders, such
as demographic characteristics, were found
not to distinguish among offenders in terms
of their frequency of serious offending, the
effectiveness of criminal justice decisions
about arrested offenders could be im-
proved.
Consideration of policies aimed at spe-
cific criminal career dimensions rather than
broad-based "crime control" may have dif-
ferential implications for the acceptability
of various policy alternatives. It has been
suggested that correlates of frequency and
duration are more neutral in terms of race or
socioeconomic status than correlates of par-
ticipation (Blumstein and Graddy, 1982~. In
that event, strategies based on career mod-
ification or the incapacitation of high-rate
serious offenders may become more wiclely
accepted than early preventive intervention
strategies. Thus, greater unclerstancling of
the pervasiveness of offenders, their clesis-
tance patterns, and their individual offend-
ing frequencies as separate components of
"the crime problem" is important from both
intellectual and policy standpoints.
A Conceptual Framework for Offender
Participation
This discussion draws on the conceptual
framework of Gordon and Gleser (1974) and
defines terms that are used throughout the
rest of this appendix. For simplicity, the
terminology is introduced with respect to a
single birth cohort.
Suppose that a cohort of N individuals
was born in a single year, and that It of the
cohort members initiate criminal careers by
committing their first crimes at age t. Al-
though 1~ could theoretically be computed
for any age between birth and the age by
which all cohort members have died, only a
negligible number of criminal careers begin
before age 6 or after age 45. One approach
to computing cumulative participation in
OCR for page 214
214
valves the distribution of the age-specific
initiation rate, a`, which is defined by
a'= IJN
for each age.i Thus, a' is simply the propor-
tion of the cohort that initiates a criminal
career by committing a first offense at age t.
Aclding up the initiation rates from zero to
any age of interest t* (18, for example), one
can calculate the cumulative participation
rate, D'*, defined by
t*-1
D'*= ~ a'.
'=o
(2)
Perhaps the most common participation
measure in the literature reviewed here is
Die, cumulative participation by age 18.
Another common measure is cumulative
lifetime participation, obtained by setting t*
to about 45, since few individuals commit
their first crime after that age.
While at is useful in calculating cumula-
tive participation, it does not present a clear
picture of the relationship between initia-
tion of offending and age. The reason is that
a cohort member who began offending by
age 15 is no longer at risk of becoming an
offender at age 16. Thus, a falling-off of
initiation rates at later ages reflects both
-
iAs defined in Equation 1, a~ is the fraction of the
original cohort that commits a first offense at age t.
Because some of the original cohort members will
have died before age t, al understates the probabil-
ity that a surviving cohort member will initiate a
criminal career at age t. That probability is given by:
I'
N - Xt
where Xe is the number of cohort members who
have died before age t. Over the age range of
interest in most studies, say, 16 to 30, X' is negligi-
bly small, and so no mortality adjustment is made.
Of potentially greater importance is sample attri-
tion because of cohort members who refuse to be
interviewed or who leave the jurisdiction. Until the
cohort reaches middle age, attrition from these
causes reduces the denominator more than does
mortality. But because those who refuse or leave
may commit crimes of which the researcher is un-
aware, the numerator is also erroneously reduced,
leaving the net effect on a' uncertain.
CRIMINAL CAREERS AND CAREER CRIMINALS
behavioral patterns of interest to the re-
searcher and artifacts of the declining pop
1 ulation at risk of becoming a first offender.
To isolate the behavioral relationship, it
is common to compute and analyze the
age-specific hazard rate, hi, the conditional
probability of becoming an offender given
that one has not already done so. For age t,
hi is defined by
It (3)
N - Z t-~ '
s=o
where Z is the number of cohort members
who became offenders before some age,
and s and t are indices of age. The term
"hazard rate" is borrowed from reliability
analysis in operations research, in which,
for example. one wants to ignore already-
burnt-out light bulbs in computing the
probability that a bulb will fail in the next
hour of a test. The hazard rate is identical to
the "offender rate" or "real rate of onset"
reported by Wolfgang, Figlio, and Sellin
(1972:132-133, 282~. Because the concept
isolates behavioral patterns from mathemat-
ical artifacts (also see the discussion in Gor-
don and Gleser, 1974), its use in analyses of
criminal career initiation is becoming in-
creasingly common.
Age-specific participation may be com-
puted from hazard rates, according to the
following formulae
t-1 t-1 5-1
Dt= ~ as = ~ he ~ (1 - hr.)'
s=0 s=0 r=0
(4)
where r is an index of age.
Equations 1 Trough 4 are operational
only if the age at the first offense is known-
2Gordon and Gleser (1974) first corrected an error
made by T. Monahan (1960) in simply summing
hazard rates to compute prevalence rates. However,
the "correct" equation (9) of Gordon and Gleser
itself contains a typographical error. The Monahan
error was repeated by Wolfgang, Figlio, and Sellin
(1972) in summing their first-index-offense proba-
bilities (p. 282) to calculate age-specific index-
o~ense probabilities (p. 126), but the magnitude of
the error is negligible.
OCR for page 215
APPENDIX A: PARTICIPATION lN CRIMINAL CAREERS
an unusual circumstance. More commonly,
D! iS measured through the self-reports of
t-year-old respondents when asked if they
have ever committed a crime. If arrest
rather than commission is used as the par-
ticipation threshold, however, the equa-
tions can be applied to official arrest records
recorded by date. To distinguish between
the two thresholds, participation measures
based on committing crimes ("doing"
crime) are denoted D; measures based on
arrests ("busts") are denoted B. For many
purposes, attention is focused on the frac-
tion of a cohort or other sample that is
currently participating in crime, i.e., that
commits at least one crime during an obser-
vation period. This fraction, the current
participation level, is denoted d.
Current participation is related to crimes
per capita, C, through the parameter, A, the
individual offending frequency (i.e., crimes
committed per year). The relationship is
given by:
C = do.
An important implication of Equation 5 is
that variations in aggregate crime measures
can be due to variations in either current
participation, cI, or offending frequency, A.
Research on A is reviewed by Cohen (Ap-
pendix B).
The remainder of this appendix is di-
vided into three sections. The next section
draws on a variety of studies to develop
ranges of estimates for cumulative partici-
pation by age 18, cumulative lifetime par-
ticipation, and current participation. Then,
based on other analyses in the same body of
literature, the second section reviews find-
ings concerning factors associated with
participation. The final section presents
conclusions and suggestions for further re-
search.
ESTIMATES OF CRIMINAL
PARTICIPATION
This section draws on previous research
to provide estimates of participation in of-
fending, according to the measures just de-
fined. The review is organized by type of
participation estimate: participation in of
215
fending by age 18 (D18 and B18), lifetime
participation (BJ ), participation by high
school age (DHS and BHS), and crime-
specific current participation (CI).3 We report
the values associated with each study as
they were reported in published results or,
in some cases, as they were derived from
data presented. Before summarizing the
empirical estimates, we briefly review the
history of attention to participation in of-
fending. We also discuss the scope of our
research review and present an overview of
the types of studies and data sources that are
commonly used in estimating participation
in offending.
Background
Scholarly attention to participation in of-
fending (or "prevalence") became promi-
nent in the 1940s and 1950s with the first
self-report studies of criminal behavior,
which used small local samples (e.g.,
Porterfield, 1946; Wallerstein and Wyle,
1947; Nye, 1958~. As reported by T.
Monahan (1960:67), in the late 1950s, the
Senate Committee on Juvenile Delin-
quency stated that "probably up to 20 per-
cent of the male population coming of age
could easily be expected to have a juvenile
court record if the 1955 rate remains con-
stant" and a member of Me committee
(Thomas C. Hennings) noted that addi-
tional delinquents would have avoided
court contact. As Monahan also noted
(1960:67), the estimation procedure that the
Senate committee used was very approxi-
mate and tenuous, resting on unverified
assumptions about the fraction of adjudi-
cated delinquents who were repeaters. Af-
ter the Senate report was released,
Monahan attempted to calculate precisely
the estimate in which the Senate was inter-
ested, denoted here as Bit. An employee of
the Municipal Court of Philadelphia,
3An arrest-based measure of current participa-
tion, b, could be defined analogously to d, and
crime-based measures for participation by age 18
(Dis) and lifetime participation (D~) are also possi-
ble. However, values for Die, Din, and b do not
generally appear in the published literature.
OCR for page 216
216
Monahan used 1957 data on the age distri-
bution of first offenders referred to Philadel-
phia Juvenile Court, to estimate that about
27 percent of all Philadelphia boys and 8
percent of the girls would appear in juve-
nile court in Philadelphia on a delinquency
charge before reaching age 18. Unfortu-
nately, his method was slightly in error (see
Gordon and Gleser, 1974), but nonetheless
his calculations established the feasibility
of estimating cumulative participation from
a single year's data.
A few years later, a clear appreciation for
the participation (or "prevalence") statistic,
also related to officially recorded delin-
quency, appeared in Ball, Ross, and Simp-
son (1964:90~:
Accurate delineation of the incidence and prev-
alence of juvenile delinquency is an indispens-
able prerequisite to analysis of adolescent behav-
ior; it is important to know whether 2 percent, 20
percent, or 40 percent of the American adoles-
cents appear in court before adulthood.
This important paper included a formal dis-
cussion of the methods, definitions, and
equations for calculating cumulative partic-
ipation from annual statistics, and it also
reported estimates based on first court ap-
pearances in Lexington, Kentucky.
Shortly thereafter, Little (1965) re-
sponded to the work of Ball, Ross, and
Simpson, noting that their techniques of
estimating "prevalence" could result in
misleading estimates. In particular, in any
single year, persons of different ages repre-
sent different birth cohorts. Thus, trends in
participation across successive cohorts (i.e.,
"cohort effects") will cause participation es-
timates based on a single year's data to
misrepresent offender participation among
any single cohort. However, in relatively
stable periods, these estimates are suffi-
ciently close approximations to be useful for
many purposes. This analytic issue is dis-
cussed more fully later in this appendix.
In the 1970s, attention to the concept and
to the estimation of participation in offend-
ing was revived by Gordon (1973, 1976;
Gordon and Gleser, 1974~. In the only re-
view of its kind to date, Gordon (1976)
assembled the results of 10 "prevalence"
CRIMINAL CAREERS AND CAREER CRIMINALS
studies ant] assessed their consistency
across different samples ant! communities.
Only studies that were based on official
records of juvenile delinquency (by age 18)
were included. Gordon adjusted the esti-
mates in different samples for variation in
race composition "in order to enhance for-
mal comparability" (1976:203~. For exam-
ple, for samples that were dominated by
white males, Gordon used data from other
samples on black and female participation
to arrive at an estimate for all males and
females. Gordon concluded that for some
purposes a useful criterion of delinquency
is the juvenile court record, and, using this
definition, he estimated Bit at about 17
percent for white urban males (1976:27
2721.
Scope of Literature Review
In selecting studies for this review of
criminal participation estimates, we used
several criteria. First, the study had to re-
port participation rates occurring in popula-
tions of interest or to provide the data
needed to calculate them. This restriction
eliminated studies designed to achieve par-
ticular rates, by selecting institutionalized
populations (for which participation is 100
percent), matched offender and nonof-
fender samples (for which participation is
designed to be 50 percent), or samples with
other precletermined rates (e.g., Glueck and
Glueck, 1934, 1940, 1950; Empey and
Erickson, 1966; Erickson, 1973; Morash,
1984~.
Second, measures of criminal involve-
ment that reflect both participation and in-
dividual frequency confuse two separate
and distinct elements of a criminal career
and were therefore not useful for the pan-
el's purposes. For this reason, studies using
aggregate data, such as UCR arrest or crime
rates (e.g., Borclua, 1958; Clark and
Wenninger, 1962; Chilton, 1964) are not
included in this review. This consideration
also eliminated studies that reporter! mean
frequencies for specific crimes without sep-
arately tabulating the "zero" or "none" cat-
egory (e.g., Arnold, 1965; Clark and Harvek,
1966; Williams and Gold, 1972; Elliott and
OCR for page 217
APPENDIX A: PARTICIPATION IN CRIMINAL CAREERS
Voss, 1974; Gold and Reimer, 1975; Krohn
et al., 1980; Ensminger et al., 1983) and
reports of other indices of criminal involve-
ment that combined participation with fre-
quency (e.g., "commission of burglary two
or more times"; Havighurst et al., 1962;
Hathaway and Monachesi, 1963; Berger
and Simon, 1974~.
Third, following the emphasis in the pan-
el's report, we focused primarily on studies
~ . . . . .m ·
OI participation in specluc serious crimes,
such as robbery, burglary, and aggravated
assault, or in official-record categories that
included those crimes, such as "arrests for
nontragic offenses," "juvenile court refer-
rals," and "convictions for indictable of-
fenses." This focus further restricted the
scope of our literature review and elimi-
nated self-report studies that tapped only
participation in status offenses or minor de-
linquency, such as underage drinking, van-
dalism, and theft under $2 (e.g., Slocum and
Stone, 1963; Akers, 1964; Gould, 1969;
Hindelang, 1971; Waldo and Chiricos,
1972), or that combined serious and minor
offenses into one scale (e.g., Nye, 1958;
Dentler and Monroe, 1961; Winslow, 1967;
Hirschi, 1969; Gold, 1970; Walberg, Yeh,
and Patton, 1975; Wilkinson, 1980~.
The fourth restriction was that the study
be published in the English language, ei-
ther as a book, journal article, or report to a
research sponsor. Papers presented at pro-
fessional meetings, unpublished data, or
working papers were not included in this
review.
Types of Studies and Data Sources
Participation estimates are available from
four types of studies:
and
· life-table calculations,
· prospective longitudinal studies,
· retrospective official-record searches'
· self-reports of cross-section samples.
The first type of study uses life-table
techniques to derive participation estimates
from annual statistics that describe court or
police activity in a single jurisdiction.
These studies reflect the participation be
2~7
havior of multiple birth cohorts as of a point
in time, and the method is similar to that
used to estimate mortality or morbidity at
successive ages. Although many police de-
partments and courts could compile life-
table participation statistics, only a few such
studies have appeared in the research liter-
ature (i.e., T. Monahan, 1960; Ball, Ross,
and Simpson, 1964; Little, 1965; Far-
rington, 1981~.
If an agency maintains separate age sta-
tistics for first offenses (whether defined by
arrests, referrals, or convictions), it is possi-
ble to compute cumulative participation by
a given age. This life-table method of com-
puting B is useful because it produces a
current "snapshot" of participation patterns
among the birth cohorts at risk at the time of
the analysis. If the age-specific first-offense
rates remain constant across those cohorts,
the results are identical to those from a
longitudinal study of one cohort (Gordon
and Gleser, 1974), and they provide reason-
able approximations of a cohort's behavior if
trends across cohorts are not unduly severe.
Perhaps because the approach does not per-
mit analysis of the relationship between
participation and variables other than the
demographic attributes (age, sex, and race)
recorded at arrest, it is not widely used to
analyze B.
Life-table estimates of age-specific cur-
rent participation, b, in a given year can also
be easily computed by combining justice
agency data on the age distribution of ar-
restees or court referrals during the year
with data for the same year on the age
distribution of the general population. At
any age, b is simply the ratio of offenders to
the total population. Published studies us-
ing life-table approaches have generally
reported cumulative estimates only; annu-
alized estimates rarely appear in the litera-
ture.
A second type of study is the prospective
longitudinal study. Stimulated in part by
interest in the relationships between partic-
ipation and a broad array of characteristics,
researchers have used longitudinal studies
to track one or more cohorts of individuals
over substantial follow-up periods. The
studies are called "prospective" because
OCR for page 218
218
they involve samples of persons not known
in advance to be criminals their criminal-
ity is expected to emerge in the future, if at
all. If the cohort is representative of all
cohorts at risk of offending during the ob-
servation period, the longitudinal approach
does not produce participation estimates
markedly different from those that would be
obtained in a life-table study.
The special strength of the longitudinal
study lies in its relating participation to an
array of variables. Prospective longitudinal
studies enable researchers to relate crimi-
nal career initiation and participation not
only to age, sex, and race but also to char-
acteristics of individuals (e.g., poor school
achievement, early antisocial behavior),
characteristics of their families (e.g., low
socioeconomic status, antisocial behavior in
parents), and life events (e.g., parental dis-
cord and breakup, onset of substance
abuse).
CRIMINAL CAREERS AND CAREER CRIMINALS
few prospective longitudinal studies have
been carried out with large samples. We
report criminal participation estimates from
10 major prospective longitudinal studies
conducted in the United States, Great Brit
ain, or Denmark. Analyses are typically per
fon~led and interim results published be
tween successive waves of data collection.
Because estimates in interim reports fre
quently change as additional data are gath
ered or errors corrected, we tried to obtain
the latest comprehensive reports in prepar
ing this review. Thus, estimates reported
here may differ from published interim es
timates based on the same data set.
One type of prospective longitudinal
study is the two-wave prospective study,
which has been used to test the power of
some indicator (e.g., teacher ratings of be
havior), measured at one point in time, to
predict future participation by some subse
quent age. While a large number of such
A prospective longitudinal study involves studies have attempted to ascertain predic
a sample of individuals selected at a point in five power with respect to various measures
time. Initially, information concerning the of"delinquency" (see Loeber and Dishion,
correlates of interest is gathered from the 1983), this review is limited to those yield
subjects, from their parents or teachers, or ing estimates of participation as previously
from records of school performance or defined.
teacher ratings. The subjects are then The third and fourth types of study in
volve single-wave, cross-section samples,
for which participation is estimated from
either retrospective official records or self
reports. Officially recorded participation,
Ba' is the proportion of age-a members of
the sample who have an official criminal
record (police arrest, court referral, or con
viction) at the time of sample selection. The
samples are usually selected to achieve a
particular distribution of one or more hy
pothesized correlates of participation. How
ever, many of the samples also provide a
basis for estimating participation in a sub
population of interest.
Surveys of cross-section samples can also
provide estimates of self-reported offend
ing. By their nature, studies in this category
are well suited to estimation of crime
specific participation. Depending on how
questions are worded, studies in this cate
gory can yield estimates of either the frac
tion of persons who have participated in
specific offense types at any time before the
tracked over time, and their records are
periodically updated. The updates may
consist of indicators of criminal activity
gathered from official records or sel£reports
and information on the correlates of interest
or life events that might trigger criminal
career initiation. Tracking and record up-
dates begin shortly after the sample is se-
lected but events occurring before sample
selection may also be recorded. In practice,
the samples are sometimes selected (and
their records located) retrospectively (e.g., a
1945 birth cohort selected in 1964), but on
some basis other than known criminal activ-
ity. As long as the sample is still represen-
tative of the cohort (despite mortality, mo-
bility, and other influences), the resulting
data may be analyzed as if the sample had
been selected at the beginning of the obser-
vation period, with the advantage that re-
sults become available sooner.
Because of the effort involved in tracking
chic and 'inflating their records, only a
At, a_. ~,,~ ~,,= ~.~
OCR for page 219
APPENDIX A: PARTICIPATION IN CRIMINAL CAREERS
219
age at time of interview (cumulative partic- nile Aid Division ofthe Philadelphia Police
ipation, Da), or only during the year (or Department on officially recorded police
some other interval) preceding the inter
view (current participation, d). Single-wave
survey studies focusing exclusively on mi
nor offenses have been excluded from this
review, and participation estimates in the
reviewed studies are reported only for the
more serious crime types. The samples in
some of the cross-section studies reviewed
here reflect an age range, such as "high
school age," and the resulting cumulative
participation estimates are denoted DHS.
Estimates of Criminal Participation by
Age 18
Criminal participation by age 18 is per
haps the most frequently reported measure
of participation in crime. The measure re
flects involvement in deviant or criminal
behavior as a juvenile since, in most U.S.
jurisdictions, the authority of the juvenile
court ends at a person's 18th birthday.
Moreover, this measure is easily compared
across multiple samples because of its pre
cise definition. We located 22 U.S. or for
eign studies that reported participation by
age 18 for large birth cohorts representing
an entire urban area, for samples represent
ing some subpopulation of interest, or for
smaller"high-risk" samples (see Table 1~.
In all these studies, estimates were based
on official records of criminal activity re
corded police contact, arrest, juvenile court
referral, or conviction and thus we report
Bit. Self-report estimates of participation
are usually obtained by sampling high
school students of different ages, and so the
number of 18-year-olds in any single study
is too small to support estimates of partici
pation precisely by age 18.
In the following pages, estimates of Bit
are drawn from longitudinal studies of two
Philadelphia cohorts, other U.S. longitudi
nal studies, life-table analyses, analyses of
multiple data bases, and British studies.
The Philadelphia Cohorts
In 1964, Wolfgang and his colleagues
began assembling records from the luve
contacts with a cohort of 9,945 boys born in
1945 who resided in Philadelphia when
they were between the ages of 10 and 18
and were therefore at risk of police contacts
in Philadelphia throughout that entire age
range. As reported by Wolfgang, Figlio, and
Sellin (1972) and shown in Table 1, Bit, as
measured by police contacts for nontragic
offenses (but including status and liquor
violations), reached 34.9 percent for the
entire sample, with levels of 28.7 for whites
and 50.2 for blacks.4
As expected, participation estimates
based on the FBI's Uniform Crime Report
(UCR) index offenses (homicide, rape, rob-
bery, aggravated assault, burglary, larceny,
and auto theft) were lower than estimates
based on all crimes, especially for whites
(13.6 percent for all males, 8.2 for whites,
and 26.8 for blacks). For non-index offenses,
Bit was only slightly lower than that for all
nontragic offenses, which indicates that very
few offenders participated in index crimes
only. Particination in crimes involving in-
jury or theft was much lower than the other
participation rates because of the narrower
definition of criminal behavior-only about
7 percent for an offense causing injury and
10 percent for an offense involving theft.
Wolfgang and his colleagues later se-
lected a second birth cohort for study,
28,338 males and females born in Philadel-
phia in 1958 and residing there through age
17. Based on Tracy, Wolfgang, and Figlio
(1985), comparisons are presented in Table
1 between cohorts I and II. With the iden-
tical domain of"all nontragic offenses" for
the two studies, the overall value of Bit,
based on recorded police contacts for males,
declined slightly, from 34.9 in cohort I to
32.8 in cohort II. However, this small de-
crease represents the net effect of larger
decreases for each race separately (about 20
percent) and an increase in the proportion
4Some of the studies reviewed in this paper use
the race designation "black," while others use
"nonwhite." Because blacks comprise nearly all the
nonwhite samples studied here, we have used the
designation "black" throughout.
OCR for page 220
220
ao
~.
Q
o
.~1
.,'
· -
.,'
S~
C)
o
U]
,'
U]
r~
m
E~
~2
~ C) E~
a,
-
Ll ~
C) ~
q~ ~
° ~m
C 4,
o ~ ~
.,' Ll
~ JJ 4,
.,~ Q. q~
C: ~ C
.,'
q4 C
4, 0 Ll
ao 0
~ 00
~ JJ
-o
o. -
~ q~ q~
1c°n ° ~
l =-
U)
~U)
o: ~
a ~a ~n 4 ~C) C)
u' to ~a, 01 · -· - · -
c e D
~1 a, ~a, ~4 cq~ c q~ c
tu 0 u, ~ ~n ~ . -~ . - m. -
q4 : ~a) ~ q3
O ~O ~U) q~ O
~a ~0 ~c ~c
C) ~ ~_ _ ~ ~O ~O ~O ~
In · - o' q ~n xul · - 00 C ~C C ~ C
_ 4 ~a~_ a ~%4 · - - ~.-
o 1 q4 ~o u' ~on ~ ~Ll
C ~ C ~ ~ ~ ~o ~o ~o
~ ·- ~ ~ -- ~ ~ X ~ E · -,- e ~ ·- q~ o= 1 q~ mm 1 ~ om 1
~ ~ ~== ~ ~ ~ ~ ~1 ~ ~ ~ ~ ~0 ~
c ~·,, 4, c ~·= a, ~ ~e q ~c Lle q ~c ~ c : ~ c ~4J c: ~
0 ~e ~o ~c s C 0 - - 0 0-- o ~u o ~c) o ~ ~=
C C) rl ~C ~· - 4J · - ~ sC ~ q~ ~ q~ ~ q4
C C O H -1 01- ~C 4 ~JJ ~1 ~41 q~ J~
· - 3 : ~- -^ 3 3 X t) ~ ~ ~o: t' ~ ~ ~ · - c O u, c o u, c o ~n
1 1 ~1 1 C) C) ~O O O
04: u~ u ~: co tn ~4: 0 c ~C-)
Ul ~n
JJ
4~
C C
O O
U t)
~a,
.,, . -
OP' 0
a, ~
CO) o0
a,
~:
U)
JJ
4~
~n
4
·- 8 C
o =o e
c
0 ~
C) ' 4)
~ O- -
0 as
.^ .e
C,
U)
·e q~ 00
h4 0 ~
JJ a,
~ h4 ~
0 0-,1
s s
0 0
C) ~ `4
~ ,~
.,'
s ~,
~ e
.Q
u~ ~r
~ a,
·rl _1 a'
n
c)
4~
o
c)
v
·,1
a,
L'
o
4J
~;
·;
0 ~
~_ _
C:
~1- ~
· - C ~ 0 O' CO
· - q4- -
-4 ~_ ~
' - 0 ~_
c~ ~, _ 3- - -
C C
' CO ~O
~s ~:~.= ~C
C ~C) C ~t$4 t
0 ~a - ~
O s
3 E ~g u'
4J
C
1 lU
ra ° c
~ ~ O
.,, ~ · -
·rl _~ ~a
4,
· -
P. 0- -
C,
S
JJ U)
:,
o
U)
o
a,
C)
·,.
o
U]
o
U
s"
a,
·,4
o
P'
::'
c ~a
o
· - q~
O
C
~ O
.,. ~ .,,
C) ~ 4~
· - _~ ~o
P'
~, _
~ `4
Q'
S
~q
o
JJ
·eLl
..HO
H ~I H ~S
O ~ O ~O
4~ ~ U]
~ ~ ~ ~a' ~ JJ
O ~ O ~ O .C: C
S O U] S O 0-,S ~ (D
O S ~ O O S 4J O O Ll ~ O
O C ~ U O C ~ C)- - · -
0 4~ U 4~ ~ U)
c, ~ ~ P~ ' o 04
· S · - ~ · S · - ~ · U1
u~ ~ In ~ 4~ ~n tn u
·'1 L`.. ~ ra ~ I.. ~ =- - a' ~o
:~ · - ~ ~ 3 · - ~ 0, 3 ~ r~ ~
~Q 3 .Q 3 ·o 3~.
' ~ O ' ~ O ' ' O
a~ ~ u~ ~ a~ a~ o, ~ a~ ..
C ~ ~ _' C ~ o -I C H -~
· - 0` ' O · - 01 ' O · - H ~ O
(Ll t~ _~ {~ ~ O H
10
a: ~
C `0
o
·, ~4
O
C
~ ~ O
.,1 ~ · -
C)
·,1 _~ ~
4) p'
· -
~ ~ C)
P4 o. -
C) ~'
S ~
JJ U]
~ ~a P
o
OCR for page 221
221
~r tD
·. ~
CO
t-, ~ ~sls
eo o~ co a~ a
·- ·eU~ oCD U ~· · · ··
£ ~4· .. . 0 0 ~ un~ 0
~ ~r
~o ~ ,4
un ~ ~ ~0
0 ~r~ \D · · . ..
. .· . ~ ~ ~ ~ ~r
cor~ C ~ -
u, ~.. _
a'1 ~r
uS 0 ~r 0 ~ a ~u~ <~ co 0 ~r ~ r~ ~ ~
~C> ~tD ~ U ~ ~ ~ U~ tD ~0 0 U~ u~ tD
u~ ~1 1 ~ ~- c~!
~as _
·. ~r ~r .. .... ....
£ ~a~ £ 04 a' £ ~:~: ~£ 1~
o'
c 0 ~ eq
0 LB O a) 0 ·- 0 ~
~ u' ~u, ~ 0
0 ' ~ C ~a ~ ~
1 C) ~ C ~ ~ C)C)
~ ~ ~ ·^ ~ ~ ·^ _
q ~a~ 0 ~ ~ ~o ~ 0 q ~q~
u~ c 0 0 q~ u ~ ~ ~ ·
O ~41 0 · ~ ~ 04 ~ 4,
' ~ ~c) ~a, ~ eq
O c: · - C) ~c u, 0 c ~ ~
a ~JJ ~ ~ a ~ c~ a ~ c c
O ~ q ~ q ~ O · 0-1
4)~ eq q4 :~ ~a 0 c) ~ q ~ ~ q4 a ~ ~c ~
· - ~O ~ O O ~O O ~ O O _' O ~ ~Dl
0 c G)t) 0 C) v 4 ~C) 4~ D O C ~ 0 ~ 0
q~ 4~ ~ C) C O ~:- ~ 0 ~ ·^ ~ 0 ~O
~ ~ · - ~ O =-~4 ~ C ~ JJ ~ ~ ~ ~ 0 q~ al
U) ~ ~ ~1 ) 1 H4 ~1 C C -~ ~d ~ 0 tu ~1 ~ C C C
. - ~ · - ~ q ~C t1 · - ~ ~ ~·rl ~ :, ~JJ 41
0 :~ c ~c ~3 :> c ~ ~ ~ ~c c) q~
a, L1 0 4, ~ L' a, 0 0 :' ~ 0 ~ Ll ~ ~4' C ~ ~4 ~ q~
0 c) ~ O ~ ~c-= ~ ~O ~ ~ c ~ · - O O ~ O
t ~t ~, t ~,,~
Q,
s . ~c ~
:' ~O ~·-
0 o, a' ~ 0 _ · - 0 ~0
~ ~O ~c
a ~4s, E! a) a ~c ~ ~:>
· - 0 · - ~a, ~ c ~ ~ c 4,
L ~4~ 0 ~ · ~C C oq > 0 ~a ~c
c) ~ O t) ° ~ ° ~ ° ~ ~C cn ~ ~ ~ C 3 ~
C 0 C ~ ~C ~C ~s ~s ~ =; ~ C) s~ ~C
`.' ~as < ~ ~a, ·- a,
O ~ ~ U) ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ." ~ ~
~D1 ~ ~ ~ O ~ 0
·r4 0 "1 ~ O ~ O . - O ~ O ~ ~41 0 :' I ~ ~ - 1 0 0` q~ ~ D
C) ~ ra ~1 ~ C.) ~rl C) '- ~>~ C-) ~4~ ~ ~ .- ~ ~ ~ ~ 16
O O q ~O O S S ~ S C
Pt O ~P ~P P~ ~S P4
e
C 0
C s
·. C)
~n a~ C
·rl
0 ->
-t 4.
Ul
Q, t
C 0
as
:'
~ r~ -
.r4 .r4 D
C Ll O
a) ~o
~s
Y '
C C.) ^
a a
O ~ ~
D 0 D Q
0 ~ ~r
0 ~0 - a~
0
a'
to ~ ~C ~ :>t
S O ~ ~D
~ .:1 ~ ~
Y 04 ~ +
. 0 0? CO
C
.-u ~q~ a
D ~0 ~ 0
c
~ =- -
r~
c t)
-
3 JJ c
0 ~
Q) r,
c 3 ~
0 t- u, ~
c ~c ~ o'
·,~ _I ·- c ~
D - ~ ,~ _
O O
a: ~
o
~·-
0 ~' ' 4~
1 ~ ~ ~to
C C C ' 4
4J ~ e4 a ~ ~ ~L,
C C Id
C _
~4 ~a ~ D 0
u, a,
E ~ c ~ ~ O ~ :>,
t) ~0 u~ 0 ca c,) ~
~ C ~ C 0 q4 C ~Ul C Ul
C.) W ~ W ~ W JJ ~ · - ~ W
~ ~C ~ W W
-~ ~-~ ~w ~ 0 ~, >,, 1
C ~ C C) ~ C ~ S C
,1 .,~ · - 0 ·- O ~ -- ~ ·-
c ~I c) ~a, 4 ~0 E ~a ~ )"
4J w O w ~ ~ ~O ~a w -'l w ~
o ~ o sq~ c um e
C) C)
~: U) U)
.,'
c
o
D
0
W
-~ 1
e N
a,
o
U~
~3
O
~; ~
C .-
a'
~ a'
O -
~ _ _
~D
a'
-
C
~5
C
W O
w co
o ea~
E ~o
~ ~ _
~O
q~
o
o
,.
4~
w ~ c
~ O
c ~ · -
~a
_ ~ 0 _
w w
:> ~=
~r-~ ~ w
~a _
c
O
w C,
~w
c ~ 0
~ .- D w
S ~ _t
o~ ~ ~
C: ~ w r.
O o' N ~ O'
£ ~ :>' c ~
_ _~ ,0 _
.
E"
wc
O c
~o O ~O q ~
· ·' JJ >, w ~a
c w ~ 4J 0
o O O D C O ~t ~
w C . - ~ ~ w ~ O
43 ~ ~ ~ c 0 4J ~ 0 e
C)
c ~ 0 ~ c ~
C ~ ~ W ~ 'rl ~ ~ W C C
~ ~q ~ > P~ ~ ~·rl ~ ~ O
0 0' ~ ~ ~ ~s ~ ~
C,) ~ 0 r~ ~ X ~ 0 ~ ~ 0
w C ~ ~ ~ 0 Q w JJ ~ to 0
~ W ~ ~ ~ ~, ~ ~ 0
4~ ~ X ~ ~ W
W ~ O ~ W ~ O · - · ~
~ r' 0 W C) ~ O ~ C) W U] ~a
to S
:)
· -
~ ~r ~ _
c ~o Q ~o
cr,
~ ~ a
~ _ w ~
0 N _
0 C ~
O O ~ C
W O
U]
a.
OCR for page 281
281
+ + + + + + + +
~ ~ ~ ~ _ ~
o ~ ~ ~loo
1 1 1 1 1 1 1, 1
- 1-
Cal Cal
_ _ _ _ __ _ _
c
+ + o
14 UP
In · -
+ o loo + +
Up
.
Q 0=
o 0
0 V
U] ~ tR O
~0 ~
A, 0 to ~ O
- O ~c
e0 8 o ~
0= ~ X
c)
mm~ an
r, 0
. - ~
x: 03
4,
cn
Q Q
o
S L,
0 C) ~
tn Ce Q
~ P4
· - ~ o
.,, ~ o
14 ~ s
C)
U]
.o
~:
~ ~ ~ 0 ~ a, 0 ~ ~ --~ ~ ~
· ~ · ~ · · · · · · · · ~
mm ~ ~ u~ r~ om ~ ~ -
mm ~ ~ C~ ~ mm ~ mm
· ~
0 a,
~ C~
a) ~ ~ a' c~ ~ ~ 0 U~ un ~ cr, 0
· ~ · ~ ~ · · ~ · · ~ ~ · ~ ·
~ ~ I ~ cr~ ~ d' ~ ~ ~ ~ iD U~
mm ~ ~ ~ d' mm ~ U~= ~ ~ -
O O O
.,l. - ~
~ ~ oo
54 ~ ~
X X X
0 0
.-, . - · -
00 0
~ ~ 0
ZZ Z
x
- _
o
~ 1
O CO
eQ -
~ · -
O
E~
0
~ ~0 ~ `:
U} ~ ::, _
_ ~ ~ ~..
O a,
~n _I Y ~ ~
0 ~ ~ ~ ~O
- ' - 0- - ~ ~1
.Q ~ tu ~ O ~Q. 1
· - ~ ~ ~ ~O
~ ~ m~ ~e ·- ~
E~ ~ `: ~ ~ 0 0 ~O
C) . - O ~q ~ ~n
·~"
c -e 0^o mm m c, a,= e0 ~ ~
_ o. - ~ m. - - 0 ~ ~ _ U] ~ ~
. - ~ ~ 0 ~ 0
~ ~ ~ ~ ~ as ~ ~ ~ 0
.rl _I O ~ ~ ~ ~0 ~ ~ ~ ~ ~ ~ ~ · - ~
Q - ~ ~, _ ~z ~ s_ ~ ~ ~ ~ ~ o
o C) C'
~:
o ~
- o^
- ~ ~
s C.
~ ~· -
·,1 . -
3 i4-
~o ~
-
o
~ - ^ ~
o 1 ~ -I
-~ CO- _
_ _
U]
~q
0
eq · - ~
0 ~ ~n
o _I O
s
~ Q,
·- o
~ ~ :Z
_I
1
~r
_I _I
- 1
U]
~n ~ ~
0 0-
~ · -
U]
~ eq C,) ~
o 0 ~-^
~ ~ o
:n
· -
o
- ~ ~
a,- ~
O CD- -
=- -
0 ~ - ~
~ 0 u] ~
tn ~ ~ u,
~ ~ 0
1
0 ~ m- -
~ o o ~
o ~ ~ ~
OCR for page 282
282
culminating in delinquency, Robins and
Wish (1977) employed an actuarial tech-
nique developed by Robins and Taibleson
(1972~. Designed explicitly to exploit the
power of longitudinal data bases on individ-
uals, the technique searches for chains of
hypothesized causal links, restricting the
search to causes occurring before their hy-
pothesized effects, and adjusting for the
longer exposure period that follows
"causes" that occur at earlier ages. Using
this technique, they found no empirical link
between Bit and alcohol use or sexual ac-
tivity by age 15, failure in elementary
school, or leaving home before age 18. In
separate analyses, they found participation
increased with marijuana use, excessive el-
ementary school absences, and school drop-
out by age 15. However, the effect of school
dropout disappeared when the other behav-
iors were controlled simultaneously.
Using ratings of various characteristics,
Farrington (1983a) compared cumulative
participation by age 25 between the "most
adverse quartile" of the Cambridge sample
and the rest of the sample. Of the character-
istics tested, the following showed the
greatest discriminatory power with respect
to participation: "troublesomeness" at ages
~10, daring at ages ~10, truancy at ages
1~14, aggressiveness at ages 1~14, hostile
attitudes toward police at age 14, and anti-
establishment attitudes at age 18. Of sample
members ranking in the most adverse
quartile on each of these attributes sepa-
rately, more than half were convicted of an
indictable offense by age 25. In similar
comparisons, proaggression and prodrug at-
titudes at age 18 showed less discriminatory
power, and "neurotic extroversion" at ages
10 and 14 showed virtually none. Nervous-
ness showed an inconsistent relationship:
conviction by age 25 was actually less prev-
alent among the most nervous quartile of
8-year-olds than among other 8-year-olds.
However, the relationship was reversed
when measurements of nervousness at age
14 were used.
Wadsworth (1979:95-97) presents a sum-
mary discussion relating early antisocial be-
havior to participation among male youths.
He reported no consistent relationships be
,, . . .
CRIMINAL CAREERS AND CAREER CRIMINALS
tween participation and parental reports of
aggressiveness, bed-wetting, or referral to a
child guidance clinic. He did report a rela-
tionship to the Pintner and Maudsley per-
sonality tests, symptoms such as stammer-
ing and tics at age 15, and truancy. He found
cheating on schoolwork a correlate of par-
ticipation in minor crimes, and he reported
greater participation in sex offenses among
boys experiencing late puberty.
In summary, the studies reviewed here
show that later participation increases with
the emergence of early antisocial behavior
as observed by parents, teachers, and peers.
However, even among the highest risk
groups identified in these studies, participa-
tion in subsequent offending does not ex-
ceed 65 percent. Thus, this literature leaves
unanswered the question of why the antiso-
cial behavior patterns of many children and
adolescents terminate short of officially re-
corded delinquency and adult arrest.
School Performance and Intelligence
Nine of the studies reviewed report that
poor school performance and low intelli-
gence are associated with higher participa-
tion. The results of these studies are sum-
marized in Table 15, except those of Reiss
and Rhodes (1961), which are summarized
in Table 10.
Wolfgang, Figlio, and Sellin (1972:63)
found a nearly monotonic, increasing rela-
tionship between participation and a
school-achievement scale, for both blacks
and whites in Philadelphia cohort I. Among
the two lowest achievement categories, par-
ticipation among black males exceeded 50
percent, and the black-white differential
was large. The race differential nearly dis-
appeared among the highest category of
achievers.
In an early study, Polk, Frease, and Rich-
mond (1974) reported that for sons of both
blue-collar and white-collar families, a bet-
ter high school grade-point average was
related to lower participation, based on ju-
venile court records. Also, Palmore and
Hammond (1964) report lower officially re-
corded participation for both black and
white high school students with averages of
OCR for page 283
283
C,
o
~ o
.,, A'
s
tQ
o ~
·^ V
· -
a,
o
.,,
a,1
v
.,1
a,
~V
a
· -
H
C)
o
P'
o
o
S
U]
m
~:
E~
~n ~ s
~ c_ ~ ~
0 ~ ~me ~0
·- O ~ te `: 3
O
V ~U] ~ ~ ~ H
O ~o
v o ~ - ~ v u'
a) ~ ~ ~ ~ ~0
· - a) ~ ~ ~ s ~S
V
' - S ~ S ~·~- - Q ~ ~- - ~ ~
=- - V ~V S~ ~S ~ V
·~ - u, ~ ~·~ O ·` O a, ~ v
^ 3 ~ ~ - _ u' _ v ~^ ~ · -
+ + + +
_ _ _ _
I C~ ~ ~ ~ ~
· · · · ~
CO
r
mm ~ m
· ~ · ·
u~ ~ ~ kD ~
- - ~ - -
I
· ~ · ·
= -
m mm ~ m~
a)
v
-
CQ
a
~:
a)
Q
.
-
S
a)
V
a,
a,
~ Lr, ~
· · ~
~r ~ a'
>
a)
>
.,'
Sv
~:
·`
u, a)
_
a
a
_ ·,1
S
V
ml
O
O
~ S
Q4 V
o
C~
a~ u~ ~
· · ~
OD C~ ~
C~ d'
o3 a,
~1
~ 3 0
>, ~0 ,,$
U]
U]
a,
U]
a,
·e
S
·,'
S
c
S ~S- -
·,' a,) ·,l a
:~: ~C ~
.,, .,'
O
~S
1
· -
r' ~5 S
_ ~
cn~
ml ~
a, ~
Q] 0
o
c)
o
~ 0
0 s
0 V
s U]
v
tQ
3
O
8 `,
V
tQ ~
a,
-
· -
CO
~ ml
a) ~
o
g
s
v
U]
s
·^
s
s
.,,
s
C
o
U]
a,
o
o
s ~
V c:
u' a)
-
_
a
. -
~ s
o C
V
m
ao u~
· ~
C~
er
ao
· ~
0
U~
C) ~
v ^1
U]
~n .`
_
~n
~ ~ a
-
a
ml~
0) ~ Q
O
~S
~0 V
~ V
o
C~
= - ~'=
O _
U] U]
_ ~_~S
Z E ~ ~
O ~ ~ ~ ~ _ - ~ _
c: ~l ~ ~, ~ eP ~ O ~O O ~
~1 ·' t-~l r_ u~ vom1 ~ ~ ~o
=. - - 0` Q '~ a' ~ ~ m_ O k
~, - ~ ~ ~ ~ ~ - ~ ~ ~
b' C,Q ~ -E ~ ,', - Q tt _, - , - ~ _
O o 0 G5
: ~ ~
o
oo
OCR for page 284
284
o
~ o
.,' .,,
s
o
.,, ~
·rl
P:
-
a)
v
.~
-
U]
A
a:
a)
o
ran
m
A:
EN
· -
s
U)
~ ~ 0
.~' marl
en 0 ~
JJ a)
UB
-
3
O
S
In
~0
-
-
-
-
a)
tQ
S
o
IC0 ood' CO ~-'
· · ~e · ~
In
up Infir ~ ret ~ ~ ~up
~ ~ ~e · · · · · O ~
_' ~ ~ ~ r- ~ a' Lo ~ IS
Cal ~ Cal
. -
o
S
a)
a)
a)
Sit
a)
A:
-
1
to
-
~ ~ fir UP ~ ~
· · · · · ~
mao
Cal Cal
O O
· · · · · ~
~ l~ u~ ~ _I tD
d' ~r d' d~ un d'
a
" ~O ~ O
C
01 ~ O ~ ~ ~~1
U] H ~ ~1 3
a, 0 0 ~ >1 ~ t~
E ~_' ~ 3 ~tJ _ S~
tt ~ N ~ O
-
C o ~, o 3 ~-I Q 1 ~ ~ ~ ~ ~
·,4 O O ~O O ~ ~· - oD _I Q Q ·r1 0
~ O ~ O ~ O ~ O ~ ~ ~ _-~ ~ ~ O
~ ~ I ~ o ~ I ~ ~ 8 ~ ~ ~ o Vo ~ ~
|= ~ | Z V | 0\ ~ | ~ ~ Pt ~l~ H
U] ~1
~S
O
~ ~n a) 0 ~
C) O ~ ~ eq ~ _
· - m ~ O
·' ~ e- -
O · - ~_ ~.- _
Ql ~ o' eq ~ - a' ~ s
_ ~ ~C ~ ~ ~ ~ ~
~0= 0 ~ ~ ~ O
=~1 ~U~-~ ~ m--
ml.- mlo ~n mlo .- ~ o
-ol · - ~ · -
a, ~ ~a, ~ c
c- - ~ ~ ~ · - ~ ~
0 ~ ~ 0 ~ s ~ C
C a
~ ~ ~ _ ~ ~ _. -
0 0 0
C) C) C)
-
~s 0 _
_
C
0 3 ~ · -
4, tQ ~ ~ ~
U] ~ _ ~ _
O
OCR for page 285
285
) U]
~ ·'= ~ · - ~ ·'Q O
Q 03 ~ ~ ~ S Q ~Q
a, 0 0 ~ ~a) 0 0 in
_1 _I ~ 1 _I _ ~ O
- ~ ~ ~ ~ U]
m~ ~ O
lo- - O ~-~1 ~ IS
~ aY 0 ~ in ~ BY O ~ 1
O ~ S- - ~ ~O C) be:- -
O ~ ~ ~
~ ~ ~ ~ ~ ~ ~ ~ ~ O
_ Q n' ~ ~ ~ ~_Q
+ +
_ _
~ ~ us us O ran ao ~
IU] ~ ~ ~ ~U] ~ o ~ Cot
~ · ~ ~ ~ ~ ~ ~ ~
~T o o o o or o o o o
1 1 1 1 1 1 1
~ ~ ~ ~ t9 U~ o
cn ~ 0 d~ C ~U) ~ C~ ~ ~
~ · ~ ~ ~ ~ ~ ~ ~
~r 0=== mr 0===
11 1 1 1 1 1 1
~n en u~ u
a) a
~n ~ ~u, tQ
a
~ ~o ~ ~ a) ~
a, ~ .Y a.,
~ C.~=
·~ ~{~ ~·- m~-
s ~s ~s - 5:
3m~m :~m3 m
8 0
s u' ~ ~-
t) ~ eQ
u,
~ 0 ~
m~ um ~ ~ _
·~1 N m l ~. - u, u, -
· - ~ = -
G' ~ o- ~ ~ ~10
O ~ eQ ~ ~
a, ~ ~a.) ~ ~ u,
0 [Q eq ~ a,
s ~ a) ~ ~ 0
U] ~ ~ ~ 3 · -
~ . - ~ ~ 0 cn~
~ 4' O'~ ~ ~ mlu'
C ~C~
_
00
a'
-
~, .,'
S
V
U] ~
a) s~ . -
=- -
~ :~:
.,'
:n
tQ
·,'
o
~r:
-
.,'
·,'
C)
V
.,'
.,'
S
'CQ
o
a,
a)
·,'
S
o
~Q
o
u]
u]
o
.-l
· -
~:
~1
.
OCR for page 286
286
C or better compared with students with
lower averages. Similar to the Philadelphia
results, black youths in the lower achieve-
ment category had participation rates above
50 percent.
In the study of a high-risk sample of
children referred to a guidance clinic, Rob-
ins, Gyman, and O'Neal (1962) found that
having at least one juvenile court appear-
ance by age 18 was much lower for youths
who graduated from high school (9.1 per-
cent) than for youths who left school before
grade nine (74.2 percent). (For further evi-
dence that school failure may be a precursor
of delinquency, see Havighurst etal., 1962.)
In a larger study, Reiss and Rhodes (1961)
reported an inverse relationship between
measured IQ and officially recorded partic-
ipation among children of white-collar and
blue-collar parents (see Table 10), but they
did not control for school achievement.
Three British studies examine the rela-
tionship between school performance/intel-
ligence and participation, also based on
official records. Wadsworth (1979) reported
that higher teacher ratings of pupil dili-
gence at age 10 were associated with a
lower rate of conviction for an indictable
offense by age 15. In a series of compari-
sons, Farrington (1983a) found participation
by age 25 adversely affected by low IQ,
limited vocabulary, and leaving school at an
early age. Moreover, in his multivariate
analysis of factors predicting at least one
conviction between the ages of 10 and 13,
IQ at ages 8-10 had a significant effect
independently of antisocial behavior and
parental characteristics. Ouston's (1984)
study of another London sample found that
standardized reading and IQ scores at age
10 were clearly related to participation
rates, based on police and court records. As
would be expected, in a similar analysis
with test scores taken at age 14, these same
measures were slightly more predictive of
later delinquency.
Last, Hindelang, Hirschi, and Weis
(1981) examined the association between
officially recorded and self-reported partic-
ipation and grouped values of school grades
and general-knowledge test scores. For
white males and females of both races, they
CRIMINAL CAREERS AND CAREER CRIMINALS
reported gamma statistics demonstrating
the expected relationship. For black males,
the associations were statistically insignifi-
cant and, for the general-knowledge mea-
sure, in the "wrong" direction. However,
the authors expressed strong doubt con-
cerning the validity of the grades and self-
report data for this group.
Taken as a body, the participation litera-
ture indicates that regardless of race and
social class, higher school achievement is
associated with a lower participation level.
(See also discussions in Gordon, 1976;
Hirschi and Hindelang, 1977; Butter and
Giller, 1984.) However, school achieve-
ment and low intelligence appear to be
closely intertwined and further research is
needed to sort out their relationship to par-
ticipation.
Miscellaneous Attributes
Researchers have also examined the rela-
tionship of participation in offending to var-
ious other characteristics, such as legitimate
activities, psychiatric diagnosis, physical at-
tributes, and peer involvement. Because
these relationships have been studied using
a variety of methodologies, the studies sum-
marized in Table 16 should be considered
only a small, possibly unrepresentative,
portion of the relevant research.
As shown in Table 16, Elliott et al. (1983)
found that in the early years of their study,
youths employed full-time reported higher
levels of current participation than nonem-
ployed respondents. However, in later
years, the participation difference between
full-time employed and nonemployed re-
spondents essentially disappeared. Because
the respondents were aged 11-17 at the
time the study began, the data may reflect a
maturation process, as employment is trans-
formed from a source of freedom from pa-
rental controls to an adult activity tying the
individual to society. Both Farrington
(1983a) and Viscusi (1983) reported a strong
positive relationship between participation
and an unstable job record or low job status
in early adulthood (ages 18-19), which is
consistent with the trend in Elliott et al.'s
data. Viscusi studied over 2,000 young
OCR for page 287
APPENDIX a PARTICIPATION IN CRIMINAL CAREERS
black men from Philadelphia, Boston, and
Chicago and found that for a variety of
criminal behaviors, currently unemployed
men reported higher participation rates
than employed men. This relationship per-
sisted even when controls were introduced
for drug use and criminal history variables.
Farrington's unemployment measure was
also a significant predictor of conviction
between ages 21-24, even controlling for
convictions at earlier ages and low family
income during childhood.
Farrington (1983a) also examined the as-
sociation between conviction by age 25 and
involvement with delinquent peers and
drug use. Not surprisingly, Farrington re-
ported greater participation among those
who reported involvement with negative
peers (not further defined) around age 14
(59.2 percent) than among other youths
(25.3 percent). Drug use was also a signifi-
cant factor influencing participation by
young adults in this London sample and in
the study by Viscusi (1983~.
Robins (1966) reported extremely high
paruc~pation rates among adults diagnosed
as sociopaths. However, this finding does
not necessarily indicate a causal relation-
ship because arrest was 1 of 19 factors used
in making the diagnosis and was the third
most common in this sample.
The results of Farrington (1983a) that
bear on the relationship between participa-
tion and physical attributes are also summa-
rized in Table 16. The vast majority of
studies of physical correlates of criminal
activity have been conducted by comparing
samples of incarcerated criminals with sam-
ples of presumed noncriminals in terms of
the attributes of interest. Because such stud-
ies confound relationships involving both
participation and frequency (A), they were
not reviewed here. The reader is referred to
Mednick et al. (1982) and to Wilson and
Herrnstein (1985) for more comprehensive
reviews of that literature.
Farrington (1983a) reported higherpartic-
ipation among boys who were rated clum-
siest, shortest, and lightest in weight at ages
8-10. However, the participation differen-
tials were slight with respect to size mea-
surements taken at that age, and were non
287
existent with respect to remeasurements at
ages 14 and 18. Wadsworth (1979:99) re-
ported similar body-size effects, but ob-
served that they were eliminated when so-
cial class and birth-order effects were
statistically controlled. (Tabular data were
not reported.)
One physical attribute, low pulse rate, is
considered a measurable indicator of an
undersensitive autonomic nervous system,
which has been hypothesized to be associ-
ated with higher rates of participation (see
Wadsworth, 19761. As shown in Table 16.
Farrington reported a weak relationship be-
tween low pulse rates at age 18 and convic-
tions by age 25. Wadsworth (1976:249) ob-
served no difference in pulse rate at age 8 in
a mildly threatening situation among mem-
bers of his sample who were and were not
eventually convicted. However, those con-
victed of violent or sexual offenses exhib-
ited si~nificantlv lower oulse rates an ob
~. ~
V ~ ~ 7
servation consistent with the theory.
CONCLUSION
Perhaps the most striking finding about
criminal participation is the pervasiveness
of involvement in serious crimes. The best
available estimates suggest that 25~35 per-
cent of urban males will be arrested for at
least one index offense in their lives, and 15
percent will be arrested before reaching age
18. There are systematic demographic pat-
terns of participation in serious crime:
males are more widely involved than fe-
males, and blacks more than whites; also,
the majority of criminals begin their careers
before reaching their early 20s. But demo-
graphic participation patterns offer little
policy guidance, because they are too broad
to offer a basis for decision making, because
their interpretations are ambiguous, and be-
cause basic social values would be affronted
by decision rules that invoked demographic
characteristics .
Other family and individual characteris-
tics related to participation are of more in-
terest to scholars and policy makers. The
family influences most consistently found to
be associated with higher levels of partici-
pation in serious crime include:
OCR for page 288
288
o
o
.,' .,'
U]
o ~
· - v
. -
~4
a)
a:
o
. -
. -
v
.,1
Q
.,,
-
a,
v
a)
-
~Q
to
a)
ee Q
O .,,
a,
a, 0
V s
U] ~
.,' a,
~£
At:
EN
U] ·'
to
~ a)
O ~
Q
. -
v
o
U]
U]
en up A
a) ~ ~
a) 54
a
~ a
O
a
· -
U1 a
o
v
a
a
a)
~:
o
s
o
v
v
V
a) al
s~ U]
o
1
-
· -
-
_ ~
O · -
a
O
~ O
O
V ~
o
o
Io o o oo o o
C0 · · ·e ·
C~ ~ _I C~ O 1D
a~ 0 0 00 0 0
· ~ ....
d' ~ ~ O
~ - ~
0o O O O O O O
~ · · · · · e
er ~ d. ~ ~ O
~ O O O 0 00
t- · ~ . ...
d' ~ ~ ~ ~ ~
I_~ _~ ~· · _ ~ _ _
a' ~+ + 1 +
O O O O O Od' ~- ---
~ · ~ . ...
a' ~ u~ ~co
~ C~
· ~
U~
U, ~U] ~ U1
a, ~ ~ ~ m
o -.- o -.- ~ ~o ~ 8 ~
JJ
s ~ ~ ~ V
ms ~a
o
H V
· - _
_ u, ~n _ a'
_ C0 -
~ ~x ~a
Q ~·' ·' a
Q ~e ~u, - ~ - o,
O ~ ~ ~_
~ ~ ·' ~0 ~ ~0 - a,
_ ~_ _ ~m. - ~ _ m. -
a) U] a, ~= a, ~~ ~·- ~ ~
~1 ~=1 ~ ~1 ~ ~ ~1 ~=1 ~ ~Q
a) 0 a) ~ 0 ~ ~ ~ ~ ~ 1
a, ,1
~ ~m~ ~ O ~ ~ O
Ql ~ - ~ V
O o ° O ~ ~O -
C) C) ~C)
O -
. _ ~ ~_
co
O
.,, ,~, ~ ~V
~a) - ~' - u' -
OCR for page 289
289
C~
· -
0
.,,
a~ ~n
o
o
· -
~n ~
· - o
~o
o
o
.. JJ
U]
U] U] o
a
o
· -
s
-
a)
s
o
· -
z
~q
s
o
.,,
o
tn
o o o o
· · · ~
a' ~ c~ ~
Io o o o
· · ~ ~
~ ao ~ ~
c
U]
a
U]
o
£ ~:
~ ~ o ~ ~ o ~ ~
· · · ~ ~ e ·
O
~ ~ ~ ~ O ~ CO
· · ~ · · · ~
o0 ~ ~ M° O
er
O O O
. . _I _
o u ~_ 1
__ ~
~ _ ~ O O ~
- _ 0 ~1 ~ ~ _ _ _
c'` {Q 1 1 ~ ~ ~ a' O
~n s In ~- u~ oo co ,
aJ . - ~ '~ t~ a) -_ _ _ _ _
,-1 U] ~ -
Q O Q' - O · -
~ ~ o ~ ~rn s s s s s s
·'l a,) _I JJ
eq s ~ ~ s ~ ~ ·~ - ~ ~ ·~ -
·- 0 ~ ~ 0 ~ ~ ~ ~ ~ a
O ~ u' o ~o (' :~ :~: ~ :~3
O tQ ~ _
C, O-,l ~n
·' (V ·'
^ ~, tQ ~_ ~ ~
~ ~ ~s
O ~· - o U] ~ ~
U1- - ~ ~·' ~ ~ = - ~ ~ O
a) ~,~ ~ . - ~= ~ ~
m t) ~ml~ ~ u, mlo
·,l >, ~u, ·,l O · - V Q u,
aJ ~ o ~ ~ s ~ ~ ~ ~ ~ · -
O Q. ~ ~ ~ ~ ~ ~ O ~=
V ~ ~ ~ ~ ·~= ~o s ==
O O O
C)
{:
O - O _
·-l ~ ~ ·^
~ ~· -
~, _ Q _ ~, _
O
P~
OCR for page 290
290
· inadequate parenting, in the form of
inconsistent or sporadically violent disci-
pline, poor parent-child communication,
and poor supervision;
· parental delinquency and cr~m~na~ty;
· parental discord and family breakups;
and
· some indicators of low socioeconomic
family status, such as low income and poor
housing.
High rates of delinquency participation
have been found consistently for children
exhibiting the following behaviors at an
early age:
· antisocial behaviors, such as aggres-
siveness, fighting, and lying; and
· poor school performance.
Some studies suggest that certain factors
measured in the mid-teen years are empir-
ically related to participation, although they
do not necessarily precede initiation of the
criminal career. These include:
· association with delinquent peers;
· abuse of hard drugs;
· employment status, which may have
different effects for juveniles and adults;
and
· large family size.
These family and individual characteris-
tics will be familiar to many readers as
"causes of crime" that have been discov-
ered and supported in large bodies of em-
pirical research. However, much of that
research has measured crime in ways that
reduce distinct career dimensions partici-
pation, offending frequency, diversity and
seriousness in crime types, and duration
to a single number. By limiting attention to
research in which the various career dimen-
sions can be partitioned, this review and
that of Cohen (Appendix B) attempt to iso-
late the separate relationships between in-
dividual and family characteristics and the
respective career dimensions.
Even though the associations just listed
emerge consistently in the participation lit-
erature, two limitations should be noted.
First, while the empirical relationships re-
ported here provide prospective indicators
CRIMINaL CAREERS AND CAREER CRIMINALS
of increased participation risk, predictions
based on them will produce substantial er-
ror rates because, generally, at least 40 per-
cent of the individuals presenting any risk
factor do not become offenders, as mea-
sured by arrest before age 18. Some gains in
accuracy could probably be achieved using
scales that combine multiple characteristics
associated with participation risk. But the
magnitude of those gains may be disap-
pointing because the risk factors do not
occur independently of one another. One
relatively unexplored approach to improv-
ing predictive accuracy is the search for
specific stressful events associated with the
initiation of criminal careers, such as school
failure or a family death. These events may
alter base participation rates from the levels
that would have been expected on the basis
of demographic characteristics and other
risk factors.
Second, caution is essential in assuming
that relationships observed in one time and
place are applicable to others. Associations
involving B18 in Philadelphia cohort I, for
example, describe behavior occurring by
1963 at the latest; major changes in
demography, social norms, and social pro-
grams since that time may well have made
those associations inapplicable today. Sim-
ilarly, differences between Britain and the
United States suggest caution in assuming
that relationships observed in one of the
countries would apply in the other. Conse-
quently, there is a clear need to develop
new U.S. data bases that describe individu-
als' career initiation and participation be-
havior as well as their pertinent character-
istics and experiences, as of the present
time.
New data bases would be especially
helpful in resolving unanswered questions
about criminal participation if Hey were
designed to facilitate synthesis of official
records and self-reports of illegal activity
and contacts with the criminal justice sys-
tem. To maintain both the chronological
accuracy and richness of detail needed to
merge information on event sequences,
such data bases should be developed longi-
tudinally. A longitudinal design would in-
clude periodic reinterviews of subjects to
OCR for page 291
APPENDIX A: PARTICIPATION IN CRIMINAL CAREERS
gather incident reports concerning their il-
legal activity and police contacts, as well as
other relevant information. It would also
include recurrent searches for official rec-
orcis of the subjects' criminal and juvenile
justice system contacts. With carefully de-
signed samples, such data bases would be
useful in resolving some contradictions be
29]
tween the official-record and self-report par-
ticipation literatures (especially those con-
cerning the roles of race and social class), in
facilitating more crime-specific participa-
tion research, in clarifying patterns of cur-
rent participation, and in understanding the
influences of specific events and interven-
tions on participation.
Representative terms from entire chapter:
participation rates
