|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
AN EXAMINATION OF ADULT ONSET OFFENDING
A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ARTS
UNIVERSITY OF FLORIDA
This document is dedicated to my husband, Larry D. Smith, and to my parents, Alvaro
Gomez and Routa Kroumovitch-Gomez
I would like to thank Dr. Alex Piquero, my committee chair and mentor, for his
guidance, availability, and support. I would also like to thank Dr. Lonn Lanza-Kaduce
and Dr. Terry L. Mills, my committee members, for their highly constructive comments
and suggestions that greatly enhanced the quality of my thesis.
I am very appreciative to the Department of Criminology and Law, the faculty,
staff, and graduate students. The collaborative and supportive environment has
significantly contributed to my graduate experience.
I have to show my gratitude to the "girls," for sharing the unique experiences in
graduate school, but, most of all, for all of the fun we have shared.
Finally, I would like to show my greatest appreciation to my husband, Larry D.
Smith, to my parents, Routa Kroumovitch-Gomez and Alvaro Gomez, and to my sister,
Gaida Gomez. It is through their love, support, and encouragement that I have been able
to become the person that I am.
TABLE OF CONTENTS
ACKNOWLEDGMENT S .............. .................... iv
LI ST OF T ABLE S ................. ................. vi......... ...
AB STRAC T ................ .............. vii
INTRODUCTION .............. ...............1.....
LIFE-COURSE PERSPECTIVE AND THE CRIMINAL CAREER ................. ...............3
ONSET OF ADULT OFFENDING .............. ...............6.....
D A TA .............. ...............13....
V ariabl es ................ ................. 14..............
Analytic Plan .............. ...............16....
RE SULT S .............. ...............18....
M odel s .............. ...............18....
Discussion ................. ...............24.................
LIST OF REFERENCES ................. ...............29........... ....
BIOGRAPHICAL SKETCH .............. ...............33....
LIST OF TABLES
1. Adult Onset Compared to All Other Sample Members ................ .......................20
2. Adult Onset Compared to Non-offenders ................. ...............22...............
3. Adult Onset Compared to Desisters- .........._ ..... ._ ....__ ...............23
Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Arts
AN EXAMINATION OF ADULT ONSET OFFENDING
Chair: Alex Piquero
Major Department: Criminology and Law
Criminal onset has been a highly researched area in life-course criminology;
however, the majority of research has focused on juvenile offenders. As a consequence,
adult onset offending has not received adequate attention in the Hield, and the factors
related to this type of criminal initiator are understudied and neglected. In this study, we
first focus on whether or not adult onset exists, and, if so, what factors account for adult
onset. Secondly, by applying a variety of independent variables, this study examines
whether correlates of adult onset can be established, and, if so, if these correlates are the
same or similar for other categories of criminal initiators such as nonoffenders, desisters,
and persisters. Three key findings emerged from our analysis. First, females are less
likely than males to be adult onset offenders. Second, participants who had mothers that
smoked cigarettes during pregnancy were more likely to be adult onset offenders than
nonoffenders. Third, participants who had higher scores on the total battery score of the
California Achievement Test (CAT) were less likely to be adult onset offenders. Future
directions are outlined.
Research on adult criminal onset is frequently regarded as unnecessary due to the
long established belief that adult criminals are most often grown juvenile delinquents. In
general, both to researchers and to policymakers, it seems as though crime is associated
with younger persons (Aday 2003). This belief is most likely rooted in two pieces of
time-honored criminological knowledge. The first is that adolescents are
disproportionately accountable for crime (Farrington 1986). The second is the largely
accepted finding concerning aggregate patterns of the age-crime curve, which describe
crime rates as reaching their highest points in the late adolescent years and subsequently
declining with age (Sampson & Laub 1992). Consequently, the majority of research
available on both the onset of criminal careers and broader crime is focused on younger
populations. Research on adult offending, and, more specifically, adult criminal onset, is
therefore less common and under-explored in the field of criminology (Sampson & Laub
1992). However, this is not to say that adults offend less often than juveniles, or that
adult criminal onset has not been recognized in empirical research. Adult offenders are
consistently responsible for the maj ority of arrests in the United States (Federal Bureau of
Investigation 2000,2001,2002), and adult criminal onset has indeed been identified in
several criminological longitudinal studies (Eggleston & Laub 2002).
Nonetheless, the results of these studies do not come without debate. The
occurrence of adult criminal onset is often blamed on the use of official records or is
disregarded as an infrequent phenomenon, and described as "negligible" or "rare"
(Wolfgang, Thornberry, & Figlio 1987; Moffitt et al. 2001). For example, Terrie Moffitt
and colleagues (2001) state that "onset of antisocial behaviour after adolescence is
extremely rare," and believe that the studies that have found significant amounts of adult
onset are faulted by their use of official records instead of self-reports. The argument is
that criminal or deviant behaviors taking place in the adolescent or childhood years that
go undetected and are not officially documented will not be accounted for in the studies
that use official records instead of self-report interviews or questionnaires. Hence, while
adult onset has only been examined in few studies, the research Eindings are conflicting,
leaving an area of life course criminology understudied and not well understood
(Eggleston & Laub 2002).
The purpose of this thesis is to examine the adult onset phenomenon and the
correlates associated with adult onset. Before data on this issue are presented, a brief
overview of the life-course perspective is presented, followed by an overview of the
Endings regarding adult onset.
LIFE-COURSE PERSPECTIVE AND THE CRIMINAL CAREER
A person's life experience is shaped and defined by the multitude of events and
changes that continuously occur in the lifetime (Elder 1985). The life-course perspective
takes into account that a person's present condition is preceded by a lifetime of
complexity and richness. In order to fully understand a single phenomenon, the past must
be explored and unraveled (Elder 1985). For this reason, the two central concepts in the
life-course perspective are traj ectories and transitions. Traj ectories are the long-term
patterns and sequences in an individual's life. These are the life pathways such as
marriage, parenthood, careers, self-esteem, and criminal or non-criminal behaviors.
Transitions, on the other hand, occur within traj ectories and are single events that are
often age-graded, such as changes in societal roles or status (Elder 1985; Sampson &
Laub 1992). They can include graduation, divorce, retiring, an arrest, etc. These specific
life events can be so abrupt and influential that they can transform or modify life
traj ectories. Therefore, there is a sequence of life traj ectories, transitions, and
adaptations. It is the adaptation to the transitions that occur in a lifetime that can lead to
changes in life traj ectories. This interlocked nature of traj ectories and transitions leads to
the broadly accepted viewpoint of the life-course perspective that an individual's
childhood is connected to adulthood experiences. From a general standpoint, life-course
perspectives are concerned with the duration, timing, and ordering of transition events
and their effects on long-term social development and traj ectories (Sampson & Laub
1992). Therefore, this perspective attempts to explain discontinuity as well as the
permanence of behaviors throughout the life-course (Simons, Johnson, Conger, & Elder
A criminal career can be examined by using the life-course perspective. This
approach is often referred to as the criminal career paradigm. According to Blumstein and
colleagues (1986: 12),"A criminal career is the characterization of the longitudinal
sequence of crimes committed by an individual offender." Criminal acts, even when
organized groups commit them, are ultimately the responsibility of individuals. The
criminal career paradigm, instead of focusing on aggregate rates, focuses on individuals
and attempts to investigate the origins of criminal behavior. One of the goals of this
perspective is to aid in developing crime control policies that would focus on interrupting
or altering criminal careers (Blumstein, Cohen, Roth, & Visher 1986).
There are four main components to the criminal career paradigm: participation,
frequency, seriousness, and career length (Blumstein, Cohen, Roth, & Visher 1986).
Participation separates those who have a criminal career to those who do not, frequency
indicates the rate of criminal activity for an active offender, seriousness describes the
severity of offenses, and career length describes how long the offender is actively
committing crimes (Blumstein, Cohen, Roth, & Visher 1986; Piquero, Farrington, &
Blumstein 2003). As previously mentioned, the life-course perspective is concerned with
duration, timing, and ordering of transition events. As such, when examining the criminal
career, there is focus on the timing and reason behind the onset of criminal activity, the
duration or persistence of the offending, if and why the criminal activity may escalate,
and why and when offenders desist from a life traj ectory of crime (Piquero, Farrington, &
Blumstein 2003; Farrington & Maughan 1999). Each aspect of the criminal career is
important to empirically study, and can provide insight into the criminal career. However,
criminal initiation, or onset, is particularly critical in that it can offer an understanding of
what behaviors may generate criminality (Chu 2002). Empirical research has suggested
that some of the predictors of early onset are low family income, poor child rearing skills,
psychomotor impulsivity, and low verbal intelligence (Farrington et al. 1990). Research
has also suggested that marital disharmony, large family size, parental criminality, and
poor academic achievement or school failure are predictors or future offending
(Farrington et al. 1990). Given the importance of criminal onset, this particular aspect of
the criminal career paradigm has received a vast amount of interest by researchers
(Piquero et al. 2003). However, most studies have focused on juvenile criminal initiation,
and little research exists that pays attention to the onset of adult offenders (Eggleston &
ONSET OF ADULT OFFENDING
Although there is not a great deal of empirical research available that has
exclusively focused on adult criminal onset, the phenomenon has been identified in a
number of studies (Eggleston & Laub 2002). Blumstein, Cohen, Roth, and Visher' s
(1986) Criminal Careers and' "Career Criminals, an in-depth summary of longitudinal
studies with available data, indicated the presence of adult criminal onset. They noticed
that most of the nondelinquent juveniles did not become adult offenders, however, a
considerable amount of adult offenders had not been juvenile delinquents. This result
seems to be contradictive of Robin' s well-known statement, "adult antisocial behavior
virtually requires childhood antisocial behavior" (1978:611). In fact, about half of adult
offenders included in the Blumstein and colleagues (1986) summary did not have
Even though juvenile delinquents are far more likely than nondelinquents to
become adult offenders, 40 to 50 percent of adult offenders do not have records of
juvenile police contacts: because nondelinquent juveniles greatly outnumber
delinquent juveniles, even though a smaller fraction of the nondelinquents become
adult offenders, their great numbers lead to a substantial contribution of adult
offenders." (Blumstein, Cohen, Roth, & Visher 1986:88)
In a follow up study to Delinquency in a Birth Cohort (1972), From Boy to Man,
From Delinquency to Crime (1987), Wolfgang, Thornberry, and Figlio found that 24.2
percent of the sample that had committed a criminal offense were adult offenders only,
with no records as juveniles. In describing the offenses of the sample, the researchers
In terms of the characteristics of offenses committed, we have seen that the
juvenile-only and adult-only offenders committed approximately the same number
of offenses, but that the persistent offenders committed far more. In terms of
seriousness, though, the most harmful offenses were those committed by the adult
group, followed in descending order by those committed by the persistent offenders
and then the juvenile offenders." (Wolfgang, Thornberry, & Figlio 1987: 32)
When the researchers examined the distributions of ages of initiation to
delinquency and to criminal behavior, they found that 16 years was the modal age for
criminal onset. However, they did discover onset in the age range froml8-22, although
less than in the younger age categories, and "a negligible number of subj ects entering the
criminal population after age 22" (Wolfgang, Thornberry, & Figlio 1987: 37).
Nonetheless, while the researchers report a small and "negligible" quantity of individuals
whose onset to criminal activity and delinquency occurred between ages 18-22 and after
the age of 22, these subj ects with adult onset still comprise 24 percent of those who were
arrested in the sample.
A study in Sweden (Stattin, Magnusson, & Reichel 1989), using official record data
that followed a subj ect population from age 10 to age 30, set out to provide basic
descriptive information about the criminal activity of the sample at different ages. They
found that the peak age for criminal onset for the male subj ects in the sample was 15
years of age. However, one in four of the men who had a criminal record in the sample
were first registered for an offense after the age of 20 years. This same study found that
the female subjects in the sample tended to first engage in criminal activity later in life
than the male subj ects in the study. They found that the peak age for criminal onset for
the female subj ects in the sample was 21-23 years of age.
In an effort to compare criminal career characteristics of two male cohorts born in
different time periods in the same area of South London, Farrington and Maughan (1999)
used the Cambridge Study in Delinquent Development and the Inner London Study. The
Cambridge Study provided a male cohort sample born between September 1952 and
August 1954, and the Inner London Study offered a male cohort sample born between
September 1959 and August 1960. When examining the distribution of age at first
conviction within the samples, this study detected significant adult onset. For the
Cambridge cohort, Farrington and Maughan report that 20 per 100 subj ects obtained their
first conviction in the age bracket 10-16, 10 per 100 in ages 17-20, and seven per 100 in
ages 21-33. For the Inner London cohort, they report that 15 per 100 of the male subj ects
obtained their first conviction in the age brackets of 10-16 and 17-20, and like the
Cambridge cohort, seven per 100 acquired their first conviction during the ages of 21-33.
An exploratory study of Chinese male prisoners conducted by Chu (2002)
concentrates on investigating the life experiences and criminal initiations, or onset, of the
prisoners. This study is unique for two reasons. The first is that is uses self-report
questionnaires, and the second is that it applies the criminal career paradigm to a setting
outside of western nations. The results suggest that criminal onset for the sample of
Chinese male prisoners is later in life than that for prisoners in western nations. The peak
age of criminal initiation was shown to be between the ages of 18 and 19 years. Chinese
prisoners in the sample who were incarcerated for violent crimes and were then currently
in prison for a minimum of a second prison term had an average age of onset of 22.23
years, and also had a peak age of onset between 18 and 19 years of age.
Another study that specifically focuses on adult onset offenders is that of Eggleston
and Laub (2002). Motivated by the understudied topical area of adult criminal onset, they
first set out to explore if an adult onset offender actually exists. They first closely
examine the existing literature, and from that made the observation "that the adult onset
offender should be systematically studied as a population among adult offenders in
general" (Eggleston & Laub 2002: 605). Using the 1942 and 1949 birth cohort data from
Racine, WI, the researchers investigated whether the correlates of crime are different for
adult offenders with juvenile records than for those adult offenders who do not have
juvenile records. The sample consisted of both male and female subjects, and the data
was collected using both records of official police contacts as well as the original
interviews. The sample was comprised of 61.2 percent non-offenders, 14.3 percent
juvenile only offenders, 11.3 percent adult only offenders, and 13.1 percent persistent
offenders. The researchers found that, "adult onset was not a rare event" (Eggleston &
Laub 2002: 612) in the Racine, WI sample. In general, this study suggests that the
predictor variables for adult offenders with a juvenile onset and those with an adult onset
are quite similar. Race, gender, socioeconomic status, and family size all had significant
effects on adult offending for those with and without a delinquent past. On the other
hand, single parent households and parental crime did not have significant effects on
adult offending for either group. Employment was the only variable in this study that
affected the probability of offending differently for juvenile onset offenders and adult
onset offenders. The variable had no effect on adult offending for those with juvenile
onset; however, continuous employment increased the probability of offending for those
with late onset. Although Eggleston and Laub's research results seem to "imply that late
onset offenders may not need to be analyzed separately from persistent offenders" (2002:
613), the researchers encourage further investigation in this neglected research area:
these Eindings should not preclude further investigation of adult onset offenders.
The adult onset population is a neglected dimension of criminological research and
to remedy this fact, future research should compare early versus late onset
offenders using the definition of late onset offending after the peak offending years
of adolescence." (Eggleston & Laub 2002:613)
Some of the suggestions for future research presented in this study include using
higher level of variation in the independent variables, a sample from a larger city "such as
Philadelphia" (Eggleston & Laub 2002: 613), and a more current population sample.
While there are a few exceptions, it is important to note that the maj ority of
research that examines the criminal career uses official indicators when establishing age
of criminal initiation, or onset. This methodological approach has both advantages as well
as disadvantages. While official records can provide the advantage of higher accuracy in
regards to timing and the type of offense, they also have the disadvantage of not detecting
those individuals who have offended but were not caught and documented by official
sources (Eggleston & Laub 2002). This particular disadvantage of using onfcial
indicators for establishing criminal initiation is at the root of the controversy regarding
the existence of adult onset as well as the value of empirically examining the
phenomenon (Eggleston & Laub 2002).
Research on criminal initiation, or onset, has received extensive empirical
consideration (Piquero, Farrington, & Blumstein 2003). However, the maj ority of
research currently available is directed towards juvenile onset offenders and, for the most
part, disregards adult onset. This offender category continues to be discounted or labeled
insignificant, although adult onset offenders have been empirically recognized in
numerous studies. Some research even suggests that adult onset offenders account for
approximately 50 percent of all adult offenders (Eggleston & Laub 2002; Blumstein,
Cohen, Roth, & Visher 1986). While conflicting views regarding the existence of adult
onset offenders as well the value of studying this population exist, the number of adult
onset offenders appears to be non-negligible. As such, the empirical study of adult
onset, though a neglected area of criminology, needs to be further investigated and
developed, as encouraged and suggested by Eggleston and Laub (2002).
In this study, we first focus on whether or not adult onset exists, and if so, what
factors account for adult onset. In addition, we also explore whether there are gender
differences in the prevalence and predictors of late onset. For example, prior research has
suggested that females are more likely than males to have late criminal onset (Kratzer &
Hodgins 1999). Secondly, by applying a variety of independent variables, this study
examines whether correlates of crime with adult onset can be established, and if so, if
these correlates are the same or similar for other categories of criminal initiators such as
nonoffenders, desisters, and persisters. As such, we explore whether we can predict adult
criminal onset, and whether the factors that relate to adult onset among males are similar
to or different from the factors relating to adult onset among females.
The current study has several points of distinction, and fulfills many of the research
suggestions for investigating adult onset offenders presented by Eggleston and Laub
(2002). The researchers point out that, "among the methods to increase the power in a
study is to have a high level of variation in the independent variables" (Eggleston & Laub
2002: 613). The current study will use an array of independent variables, and these
measures contain adequate variability. In describing a limitation in their own study,
Eggleston and Laub (2002:613) state the following:
.. the Racine data set is skewed toward less serious offenses compared to larger
cities, such as Philadelphia.... Therefore, an analysis of different cities with higher
percentages of more serious offenses may produce different results from those
found in the present study.
The data used in the current study is from the Philadelphia portion of the National
Collaborative Perinatal Proj ect (Niswander & Gordon 1972), which is not only based in a
larger city than Racine, WI, it is also the precise city recommended by Eggleston and
Laub. The data in the current research consists of an all Black sample, and has both
female and male subj ects. Eggleston and Laub's (2002) research on adult onset offenders
did not find any distinction in the effect of gender on adult offending for those with
juvenile onset or adult onset. However, they note, "that this population deserves to be
systematically investigated" (2002:614). The current research effort will examine if there
is a gender effect in the existence of adult onset as well as for the predictive variables of
adult onset. Finally, Eggleston and Laub remark that, "More contemporary samples may
also provide insight into the adult onset phenomenon" (2002: 614). The present study
may be able to fulfill this recommendation, because it uses a sample born a few years
later than the sample in the Racine, WI, data set and followed (in adult official records)
up through 1998.
This research effort utilizes data from the Philadelphia portion of the National
Collaborative Perinatal Proj ect (NCPP) (Niswander & Gordon 1972). The NCPP was a
large-scale medical proj ect carried out at a number of university-related hospitals with an
aim to collect information on birth defects, neurological conditions, birth complications,
and familial and socioeconomic conditions. The participants were infants of mothers who
took part in the NCPP and, therefore, reflect the characteristics of families who would be
interested in receiving low-cost maternity care provided by a public clinic at
Pennsylvania Hospital (Denno 1990). In Philadelphia, the sample included nearly 10,000
pregnant patients who delivered their children at Pennsylvania Hospital between the
years 1959 and 1965. The sample on which the present research is based includes 987
participants who were chosen from the first four cohorts (1959-1962) of 2,958 African
American mothers who took part in the NCPP. It is these 987 participants who were
followed by the original research team at the University of Pennsylvania. Comparisons
between the sample of 987 participants and the excluded sample indicated no significant
differences in key variables (Denno 1990).
A plethora of information related to mother and the infant participant were
collected by teams of doctors, psychologists, and public health officials across all NCPP
sites. The information included several measures associated with the mother' s pregnancy,
information relating to the birth of the child, as well as important psychological and
sociological indicators collected through ages 7/8. In the early 1980's, a research team
from the University of Pennsylvania collected information related to school functioning
and criminal history information, including all police contacts through age 17. Numerous
studies have been performed with the age 17 criminal history data (Denno 1990; Gibson,
Piquero, & Tibbetts 2000, 2001; Piquero 2000a, 2000b, 2001; Piquero & Chung 2001;
Piquero & Tibbetts 1999; Tibbetts & Piquero 1999).
In the summer of 1998, a criminal history follow-up was completed for all 987
participants of the Philadelphia sample (Piquero & White 2003). The Philadelphia Police
Department served as the data source for offenses committed by the cohort after reaching
the legal adult status of age 18. Adult criminal history data, in the form of conviction, are
available through July 1998 or through age 36 for those sample members born into the
1962 cohort and age 39 for those born into the 1959 cohort. Detailed data were collected
on the date and type of each conviction. The Philadelphia Police Department also has an
extensive alias-tracking system that enables the cross listing of names and social security
Both maternal age at childbirth (0 = 18 and older, 1 = less than 18) and maternal
marital status at childbirth (0 = married, 1 = single or unmarried) were retrieved from
medical records. Socioeconomic status (SES) was measured across all NCPP sites by a
single-item general SES score that is a multipart measure of three indicators collected at 7
years of age for each participating child: education of head of household, income of head
of household, and occupation of head of household (Myrianthopoulos & French 1968).
Maternal cigarette smoking was assessed during pregnancy. Mothers were asked to self-
report the average number of cigarettes they smoked each day. Although, originally, this
variable was coded continuously, for this research effort, we code this variable
dichotomously (0=mother did not smoke, 1=mother smoked). The variable for mother' s
changes in marital status is coded continuously.
The participant' s sex was coded as 1 = male, 2 = female. The NCPP includes
several different measures of cognitive abilities, and two are included for this analysis.
The first is the digit span component of the WISC verbal scale, measured at age
seven/eight. The WISC, because it measures both intellective and non-intellective traits,
was described as one of the most psychometrically trustworthy measures of intellectual
performance in children (Caspi, Harkness, Moffitt, & Silva 1996) and was one of the
most popular tests of executive deficits (Lynam, Moffitt, & Stouthamer-Loeber 1993).
For the NCPP, the shorter form of the WISC was utilized and Lezak (1983) had shown
that the longer and shorter forms of the WISC were functionally comparable. The digit
span test of the WISC was designed to assess attention span and short-term memory, two
of the maj or features of neuropsychological functioning/cognitive abilities (Moffitt
1997). Several scholars argued, and some demonstrated, that the WISC was actually
more a measure of neuropsychological variation than it was of intelligence or IQ (Moffitt
1997; Moffitt et al. 1994; Piquero 2001; Seguin, Phil, Harden, Tremblay, & Boulerice
The second measure of cognitive abilities is the total battery score of the California
Achievement Test (CAT). The CAT was designed for the measurement, evaluation, and
analysis of school achievement, and its emphasis was upon content and obj ectives in the
basic curricular areas of reading, mathematics, and language (Tiegs & Clark 1970).
Originally, this test was used in California schools to assess academic achievement and
produced total scores in the academic areas of reading, mathematics, language, and
overall academic achievement. The total battery score, which was used as the second
measure of cognitive abilities, reflected a student' s standing in terms of total achievement
level. In the Philadelphia NCPP, the level four CAT for grades seven and eight (ages
twelve to fourteen) was used. The CAT, in general, was highly praised in terms of its
validity, comprehensive test and interpretive materials, reliability, and standardization
procedures (Denno 1990).
For offending through age 17, a research team from the University of Pennsylvania
collected information related to criminal history, including all police contacts (Denno
1990). For data beginning at age 18 through the follow-up period, data in the form of
convictions were obtained from the Philadelphia Police Department (Piquero, Gibson,
Tibbetts, Turner, & Katz 2002; Piquero & White 2003). Individuals who did not exhibit a
police contact prior to age 18, but did have at least one conviction after age 18 were
considered adult onset offenders.
The current research is motivated by two specific questions: Does adult onset exist,
and if so, can it be explained by several individual and familial indicators that have been
found in prior research to be related to criminal offending. The first portion of the
analysis examines the data to determine if adult onset emerges in the Philadelphia
sample, and if so, whether there are gender differences in this population. Then, using
the constellation of familial and individual variables described earlier, we apply those
indicators to predict adult offending. In addition to predicting adult offender status, we
also examine the total number of adult offenses in order to distinguish between one-time
adult offenders and recidivistic adult offenders. Four models will be estimated; one will
compare adult onset offenders to everyone else in the sample, the second will compare
adult onset offenders to nonoffenders, the third model with will compare adult onset
offenders to those who offended as juveniles and then desisted, and the fourth model will
compare adult onset offenders to those who offended both as juveniles and adults.
We begin by examining whether adult criminal onset exists. The sample population
was organized into four different groups of offenders, based on whether their age of first
police contact occurred during the juvenile years or as an adult. Specifically, those
individuals who never offended were categorized as nonoffenders (n = 689). Individuals
who do not have police contact in the adolescent years, but do have a conviction in the
adult years are categorized as adult onset offenders (n = 78). Those individuals who had a
police contact as juvenile, but did not have a conviction in the adult years are categorized
as desisters (n = 144). Finally, those individuals who had a police contact as a juvenile
and an adult conviction were categorized as persisters (n = 76). In sum, there appears to
be a non negligible amount of adult onset in the Philadelphia NCPP data. In the next
section, we examine the correlates associated with this adult onset, and compare adult
onsetters to different types of offenders.
The first model compares adult onset offenders to everyone else in the sample.
Because we are interested in discriminating between two groups, logistic regression is
employed in this model, as well as for the other four models. This first full model is
shown in Table 1. Three variables in the full model were statistically significant; the
participant' s sex, maternal cigarette smoking, and the total battery score of the California
Achievement Test (CAT). The effect for sex was negative and significant, indicating that
female participants were less likely to be adult onset offenders than male participants. In
terms of maternal cigarette smoking, those participants who had mothers that smoked
during their pregnancy were more likely to have an adult criminal onset than those
participants whose mothers did not smoke. The Einal significant variable in the full
model, the CAT score, indicates that those individual's who had higher total battery
scores on the California Achievement Test were less likely to be adult onset offenders
than those who had lower scores. None of the other variables in the full model were
Male and female estimates predicting adult onset offenders compared to everyone
else in the sample are also shown in Table 1. As can be seen, two coefficients are
significant among males, but there are no significant coefficients among females.
Specifically, males whose mothers smoked during pregnancy are more likely to incur an
adult criminal onset than those male participants whose mother did not smoke during
pregnancy. Also, males who had higher total battery scores on the California
Achievement Test were less likely to be adult onset offenders than those male
participants who had lower scores. Among females, there were no statistically significant
For the second model, adult onset offenders were compared to nonoffenders. This
second full model is shown in Table 2. The statistically significant variables in the full
model are the same significant variables observed in the first full model. For sex, male
participants were more likely to be adult onset offenders than female participants. As in
the first full model, those participants who had mothers that smoked during their
Table 1. Adult Onset Compared to All O her Sample Members
Variables B SE Wald Exp Variables B SE Wald Exp Variables B SE Wald Exp
Full Model (B) (B) Male (B) (B) Female(B()
Sex -1.404 .294 22.74* .246
SES score .006 .007 .617 1.006 SES score .004 .008 .275 1.004 SES score .009 .017 .275 1.009
Mother's -.137 .388 .125 .872 Mother' s .021 .413 .002 1.021 Mother's -6.561 21.39 .094 .001
Marital .036 .283 .017 1.037 Marital .222 .316 .493 1.248 Marital -.960 .807 1.417 .383
Status Status Status
Changes in .004 .006 .568 1.004 Changes in .002 1.0071 .050 1.002 Changes in .018 .012 2.144 1.018
Marital Marital Marital
Status Status Status
Smoke .504 .250 4.070* 1.655 Smoke .785 .290 7.307* 2.193 Smoke -.418 .522 .641 .659
WISC .014 .013 1.241 1.014 WISC .020 .015 1.929 1.021 WISC -.005 .029 .033 .995
CAT -.015 .007 4.485* .985 CAT -.017 .008 4.122* .983 CAT -.010 .014 .544 .990
Constant 1-2.025 1.256 2.598 .132 Constant -4.143 1.36 9.284 .016 Constant 1-2.543 2.538 1.004 .079
Chi-square/ 43.88/ Chi-square/ 14.56/ 7 Chi-square/ 8.653/
df 8 df df 7
more likely to be adult onset offenders than those participants whose mothers did not
smoke during the pregnancy. This model also indicated that those individuals who had a
higher total battery score on the California Achievement Test were less likely to be adult
onset offenders. Similar to the first full model, none of the other variables in the second
full model were statistically significant.
Male and female estimates predicting adult onset offenders compared to
nonoffenders are also shown in Table 2. As can be seen, two coefficients are significant
among males, but there are no significant coefficients among females. Specifically, males
whose mothers smoked during pregnancy are more likely to incur an adult criminal onset
than those male participants whose mother did not smoke during pregnancy. Also, males
who had higher total battery scores on the California Achievement Test were less likely
to be adult onset offenders than those male participants who had lower scores. Among
females, there were no statistically significant variables.
The third model compared adult onset offenders to those who offended as
juveniles, but not as adults. The full model is shown in Table 3. In this estimate, only the
variable sex was significant. Female participants were less likely to be adult onset
offenders, and more likely to have only offended as juveniles.
Male and female estimates predicting adult onset offenders compared to desisters
are also shown in Table 3. As can be seen, one coefficient is significant among males,
and no coefficients were shown to be significant among females. Males whose mother
smoked during pregnancy were more likely to acquire adult criminal onset than those
male participants whose mothers did not smoke during pregnancy.
Table 2. Adult Onset Compared to Non-offenders
Variables B SE Wald Exp Variables B SE Wald Exp Variables B SE Wald Exp
Full Model (B) (B) Male (B) (B) Female(B()
Sex -1.681 .298 31.91* .186
SES score .()(5 .()(7 .5()4 1.()(5 SES score .()(4 .()(8 .194 1.()(4 SES score .(1() 1 .()17 .354 1.)1()
Mother's -.247 .4()3 .374 .782 Mother's -.12() .436 .()75 .887 Mother's -6.664 23.()4 .()84 .()(1
Marital .159 .296 .289 1.172 Marital .392 .338 1.345 1.479 Marital -1.()41 .831 1.568 .353
Status Status Status
Changes in .()(4 .()(6 .357 1.()(4 Changes in .))( 1.()(71 .()()( 1.()()( Changes in .(23 .()13 3.13() 1.()23
Marital Marital Marital
Status Status Status
Smoke .52() .258 4.()79* 1.683 Smoke .822 .3()3 7.387* 2.276 Smoke -.477 .528 .816 .62()
WISC .()19 .(13 2.()82 1.()19 WISC .(25 .()15 2.623 1.()25 WISC -.()(8 .()29 .()84 .992
CAT -.()21 .(13 8.223* .979 CAT -.()24 .()(9 7.265* .977 CAT -.()12 .()14 .814 .988
Constant 1-1.6()2 1.267 1.599 .2()1 Constant -3.954 1.39 8.()69 .()19 Constant 1-2.1()6 2.463 .731 .122
Chi-square/ 62.25/ Chi-square/ 2().228/ Chi-square/ 9.5()5/
df 8 df 7 df 7
* (p < .05)
Table 3. Adult Onset Compared to Desisters
Variables B SE Wald Exp Variables B SE Wald Exp Variables B SE Wald Exp
Full Model (B) (B) Male (B) (B) Female(B()
Sex -.913 .345 6.992* .4()1
SES score .()(7 .()(8 .6()4 1.()(7 SES score .())( .)1() .()()( 1.()()( SES score .()21 .()18 1.4()7 1.()22
Mother's -.179 .478 .141 .836 Mother' s .(14 .5451 .()(1 1.()14 Mother's -6.577 24.33 .()73 .()(1
Marital -.28() .345 .656 .756 Marital -.15() .399 .142 .861 Marital -.758 .853 .79() .468
Status Status Status
Changes in .()13 .()(8 2.277 1.()13 Changes in .(14 .()13 1.294 1.()14 Changes in .()(7 .()14 .223 1.()(7
Marital Marital Marital
Status Status Status
Smoke .435 .3()2 2.()73 1.545 Smoke .726 .354 4.2()7* 2.()67 Smoke -.193 .628 .()95 .824
WISC .()1() .17 .356 1.)1() WISC .(14 .()19 .55() 1.()14 WISC .(14 .()41 .124 1.()15
CAT -.()(7 .()(8 .734 .993 CAT -.()14 .)1() 1.972 .986 CAT .()()1 .()17 .()(6 1.()(1
Constant -.593 1.65() .129 .553 Constant -1.735 1.73 1.()(3 .176 Constant 1-3.()32 3.614 .7()4 .()48
Chi-square/ 15.()46 Chi-square/ 8.592/ 7 Chi-square/ 7.317/
df / 8 df df 7
* (p < .05)
Model four compared adult onset offenders to those who offended both as juveniles
and adults. No variables were found statistically discriminate among these two groups of
offenders in either the full model estimates or in the male and female estimates.
Model Hyve compared adult onset offenders with one conviction to those adult onset
offenders with more than one conviction. No variables were found statistically significant
among these two groups in either the full model estimates or in the male and female
In this study, we explored the adult criminal onset phenomenon. We first examined
whether adult onset exists in the Philadelphia NCPP data, as well as which correlates of
criminal activity were associated with adult onset. Further, we sought to investigate
whether these correlates are the same or similar for other categories of individuals,
nonoffenders and offenders, including desisters and persisters.
Three key findings emerged from our analysis. First, females are less likely than
males to be adult onset offenders. This result is interesting in that it is inconsistent with
prior research that suggests females are more likely than males to be adult onset offenders
(Kratzer & Hodgins 1999). Second, participants who had mothers that smoked cigarettes
during pregnancy were more likely to be adult onset offenders than nonoffenders. Prior
research has suggested a relationship between maternal cigarette smoking and persistent
offending (Piquero et al. 2002), but this is the first study that we are aware of that links
maternal cigarette smoking to offspring criminality after the age of eighteen (without
evidence of juvenile criminality). It is possible that the variable, maternal cigarette
smoking is actually a proxy for other factors that may contribute to criminality. For
example, one possibility is that the variable could be a stand-in for the mother's lifestyle,
socioeconomic status, or a poor care indicator. Another possibility is that maternal
cigarette smoking is a proxy for neurological problems. Thirdly, participants who had
higher scores on the total battery score of the California Achievement Test (CAT) were
less likely to be adult onset offenders. This particular finding shows the preventative
effects of cognitive abilities in deterring criminal activity. While sex and the total battery
score of the California Achievement Test (CAT) were statistically significant in most
models, it is important to emphasize that sex is more substantively significant than the
In total, our results show that it is important for criminal career researchers to
continue to delve into the area of adult onset offenders, and to persist in the exploration
of the correlates of crime that may predict this type of criminal initiator. Specifically,
results from our effort suggest that there are factors that may predict adult criminal onset,
and that these correlates may be similar to those predictors of other categories of criminal
offenders. In the end, our results corroborate recent research by Eggleston and Laub
(2002) regarding the importance of studying adult onset and contradict contentions from
other researchers that adult onset is a rare phenomenon, one that is ill deserving of
research attention (see Gottfredson and Hirschi, 1990; Moffitt et al.,2001).
To be sure, our research effort has limitations that need to be acknowledged, and
that lead us to outline several important directions for future research. First, our study
used an entirely African American sample. While there is some existing research on adult
criminal onset that explores racial differences, future research should more critically
examine whether adult onset patterns differ across race/ethnicity, and whether the
correlates associated with adult onset differ according to race as well. Second, our study,
as well as other research on adult onset, used data collected from official records. Focus
should be placed on expanding the understanding of adult criminal onset by analyzing the
phenomenon using both onfcial records as well as self-report data (Eggleston & Laub
2002). Third, the age range available in the data limited our current effort, as we only had
information up to ages 33 through 39. Future research should continue to investigate late
onset using data sources with information on older participants. More specifically, future
research may want to explore adult criminal onset with data containing information on
offenders over the age of 55 years, as prior studies have suggested that a significant
portion of senior offenders have no prior criminal records (Aday 2003). Finally, our study
was limited in that we did not have available variables on negative adult transitions.
Adult drug use and/or dependencies, troubled marriages or divorce, and other adult
transitions may be factors to consider when examining adult criminal onset.
The area of adult criminal onset in criminology might also be further illuminated by
the insights of Ferraro's gerontological imagination, in which he emphasizes the
importance of chronological age (Ferraro 1997). He also calls attention to the many
components of the ageing process that is filled with transitions, advantages, and
disadvantages. According to Ferraro, the age of an individual is a key variable to consider
because it identifies a cohort location. This is important, in that a cohort can be used as an
indication of the shared experiences of people born within the same time period (Mills &
Research on adult criminal onset might also be informed by Arnett' s (2000)
concept of 'emerging adulthood'. This new theory of development argues that in modern
industrialized societies, individuals between the ages of 18 to 25 are not adolescents or
young adults. Instead, Arnett believes that this time period of life is a distinct
developmental period, with distinctive life transitions and experiences for those that fall
within this category (Arnett 2000). It may be that life transitions and experiences incurred
during emerging adulthood relate in unique ways to adult onset.
This research effort, as well as preceding research, has indicated that adult onset
offenders do exist. This, in itself, has policy implications. Our society is currently aging,
and the adult and senior populations will continue to grow (Aday 2003). Understanding
adult onset offending may inevitably become an important aspect of both crime control
policies as well as the correctional system. Life-course transitions can be age-graded and
accompany the aging process and adulthood. Certain transitions in the life-course are
linked to behaviors that could incorporate criminality. These transitions are often
changes in societal roles and status. For example, as a senior, an individual may have a
sudden need for healthcare and prescription medicines. Transitions can often be abrupt
and influential, changing the behaviors of the individual. For this reason, providing
societal cushions for adult life transitions, such as national healthcare and prescription
benefits, could be a strategy used to control and alleviate the onset of negative adult
behaviors, such as adult onset offending.
In sum, while criminal onset is a highly researched area in life-course criminology,
the traditional focus on "onset" has been on juvenile offenders. As a consequence, adult
onset offending has not received adequate attention in the field, and the factors related to
this type of criminal initiator are understudied and neglected. Since the current study, as
well as previous empirical research, has identified factors that may be associated with
adult criminal onset, continuing research in the area of adult onset offending can only
enhance our knowledge of the behaviors that may produce criminality over the life-
LIST OF REFERENCES
Aday, Ronald H. (2003). Aging prisoners: crisis in American corrections. Westport, CT:
Arnett, J.J. (2000). Emerging adulthood: a theory of development from the late teens
through the twenties. American Psychologist, 55, 469-480.
Blumstein, A., Cohen, J., Roth, J., & Visher, C. (1986). Criminal careers and "career
criminals. Washington, DC: National Academy Press.
Caspi, A., Harkness, A.R., Moffitt, T.E., & Silva, P.A. (1996). Intellectual performance:
continuity and change. In P.A. Silva & W.R. Stanton (Eds.), From child to adult:
The Dunedin M\~ultidisciplinazy Health and Development Study (pp. 59-74).
Oxford, UK: Oxford University Press.
Chu, M.M. (2002). Incarcerated Chinese men: their life experiences and criminal onset.
British Journal of Criminology, Delinquency, and Deviant Social Behavior, 42,
Denno, D. (1990). Biology and violence: fions birth to adulthood. Cambrid ge, UK:
Cambridge University Press.
Eggleston, E.P., & Laub, J.H. (2002). The onset of adult offending: a neglected
dimension of the criminal career. Journal of Criminal Justice, 30, 603-622.
Elder, G.H., Jr. (1985). Time, human agency, and social change: perspectives on the life
course. Social Psychology Quarterly, 57, 4-15.
Farrington, D.P., Loeber, R., Elliott, D.S., Hawkins, J.D., Kandel, D.B., Klein, M.W.,
McCord, J., Rowe, D.C., & Tremblay, R.E. (1990). Advancing knowledge about
the onset of delinquency and crime. In Lahey, B.B., & Kazdin, A.E. (ed.),
Advances in Clinical Child Psychology, 13 (pp. 283-332). New York: Plenum
Ferraro, K.F. (1997). Gerontology perspectives and Issues, 2nd edition. New York:
Farrington, D.P. (1986).Age and crime. In M. Tonry & N. Morris (Eds.), Crime and
justice: An annual review ofresearch, vohlnze 7 (pp. 189-250). Chicago: University
of Chicago Press.
Farrington, D.P, & Maughan, B. (1999). Criminal careers of two London cohorts.
Criminal Behaviour and Mental Health, 9 (1), 91-106.
Federal Bureau of Investigation. (2000). Uniform crime reports. Washington, DC: U.S.
Government Printing Office.
Federal Bureau of Investigation. (2001). Uniform crime reports. Washington, DC: U.S.
Government Printing Office.
Federal Bureau of Investigation. (2002). Uniform crime reports. Washington, DC: U.S.
Government Printing Office.
Gibson, C.L., Piquero, A.R., & Tibbetts, S.G. (2000). Assessing the relationship between
maternal cigarette smoking during pregnancy and age at first police contact. Justice
Quarterly, 17, 519-542.
Gibson, C.L., Piquero, A.R., & Tibbetts, S.G. (2001). The contribution of family
adversity and verbal IQ to criminal behavior. International Journal of Offender
Therapy and Comparative Criminology, 45, 574-592.
Gottfredson, M.R. & T. Hirschi. (1990). A general theory of crime. Stanford, CA:
Stanford University Press.
Kratzer, L., & Hodgins, S. (1999). A typology of offenders: A test of Moffitt's theory
among males and females from childhood to age 30. Criminal Behaviour and
M~entalHealth, 9, 57-73.
Lezak, M.D. (1983). Neuropsychological assessment. New York: Oxford University
Lynam, D., Moffitt, T.E., & Stouthamer-Loeber, M. (1993). Explaining the relationship
between IQ and delinquency: Class, race, test motivation, school failure and self-
control. Journal ofAbnormal Psychology, 102, 187-196.
Mills, T.L., & Edwards, C.D.A. (2002). A critical review of research on the mental health
status of older African-Americans. Aging & Society, 22, 273-304.
Moffitt, T.E. (1997). Neuropsychology, antisocial behavior, and neighborhood context. In
J. McCord (Ed.), Violence and childhood in the inner city (pp. 1 16-170). New
York: Cambridge University Press.
Moff itt, T.E., Caspi, A., Rutter, M. and Silva, P.A. (2001). Sex differences in antisocial
behaviour: Conduct disorder, delinquency, and violence in the Dunedin
Longitudinal Study. Cambridge: Cambridge University Press.
Moffitt, T.E., Lynam, D.R., & Sylva, P.A. (1994). Neuropsychological tests predicting
persistent male delinquency. Criminology, 32, 277-300.
Myrianthopoulos, N.C., & French, K.S. (1968). An application of the U.S. bureau of the
census socioeconomic index to a large, diversified patient population. Social
Science and Medicine, 2, 283-299.
Niswander, K., & Gordon, M. (1972). The women and thelir pregnancies. Washington,
DC: U.S. Department of Health, Education, and Welfare.
Piquero, A.R. (2000a). Assessing the relationships between gender, chronicity,
seriousness, and offense skewness in criminal offending. Journal of Criminal
Justice, 28, 103-116.
Piquero, A.R. (2000b). Frequency, specialization and violence in offending careers.
Journal ofResearch in Crime and Delinquency, 37, 392-418.
Piquero, A.R. (2001). Testing Moffitt' s neuropsychological variation hypothesis for the
prediction of life-course persistent offending. Psychology, Crime, and Law, 7, 193-
Piquero, A. R., & Chung, H.L. (2001). On the relationships between gender, early onset,
and the seriousness of offending. Journal of Criminal Justice, 29, 1 89-206.
Piquero, A. R., Farrington, D.P., & Blumstein, A. (2003). The criminal career paradigm:
background and recent developments. Crime and Justice: A Review ofResearch,
30, Edited by Michael Tonry, 359-506.
Piquero, A.R., Gibson, C.L., Tibbetts, S.G., Turner, M.G., & Katz, S.H. (2002). Maternal
cigarette smoking during pregnancy and life-course-persistent offending.
International Journal of Offender Therapy and Comparative Criminology, 46, 23 1 -
Piquero, A.R., & Tibbetts, S.G. (1999). The impact of pre/perinatal disturbances and
disadvantaged familial environment in predicting criminal offending. Studies on
Crime and Crime Prevention, 8, 52-70.
Piquero, A.R., & White, N.A. (2003). On the relationship between cognitive abilities and
life-course-persi stent offending among a sample of African Americans: A
longitudinal test of Moffitt' s hypothesis. Journal of Criminal Justice, 31i, 399- 409.
Robins, L. (1978). Sturdy childhood predictors of adult antisocial behaviour: replications
from longitudinal studies. Psychological M~edicine, 8, 611-627.
Sampson, R.J., & Laub, J.H. (1992). Crime and deviance in the life course. Annual
Review ofSociology, 18, 63-84.
Seguin, J.R., Phil, R.O., Harden, P.W., Tremblay, R.E., & Boulerice, B. (1995).
Cognitive and neuropsychological characteristics of physically aggressive boys.
Journal ofAbnormal Psychology, 104, 614-624.
Simons, R.L., Johnson, C., Conger, R.D., & Elder, G., Jr. (1998). A test of latent trait
versus life-course perspectives on the stability of adolescent antisocial behavior.
Criminology, 36 (2), 217-242.
Stattin, H., Magnusson, D., & Reichel, H. (1989). Criminal activity at different ages: a
study based on a Swedish longitudinal research population. British Journal of
Criminology, 29, 368-385.
Tibbetts, S.G., & Piquero, A.R. (1999). The influence of gender, low birth weight, and
disadvantaged environment in predicting early onset of offending: A test of
Moffitt's interactional hypothesis. Criminology, 37, 843-877.
Tiegs, E.W., & Clark, W.W. (1970). Examiner 's manual and test coordinator 's
handbook: Cahifornia achievement tests. New York: McGraw-Hill.
Wolfgang, M.E., Figlio, R.M., & Sellin, T. (1972). Delinquency in a birth cohort.
Chicago: University of Chicago Press.
Wolfgang, M.E., Thornberry, T.P., & Figlio, R.M. (1987). From boy to man, Jfrom
delinquency to crime. Chicago: University of Chicago Press.
Zenta Gomez-Smith was born in Santiago, Chile, in 1982. Her family moved to the
United States in 1988, and eventually settled in central Florida. Zenta graduated from
high school in Deland, Florida, in 1999. In 2002, Zenta received her Bachelor of Arts
degree in sociology from Stetson University. Zenta was married to Larry David Smith on
June 22, 2002. Soon after, they relocated to Gainesville, Florida, in order for Zenta to
pursue a Ph.D. at the University of Florida. She completed her Master of Arts in
criminology and law in May 2004. Currently, Zenta continues to pursue a Ph.D. at the
University of Florida, and is living in Gainesville, Florida, with her husband, Larry, and
their three dogs, Maj or, Ito, and Junior.