Gender differences in the transition to early parenthood

Article (PDF Available)inDevelopment and Psychopathology 18(1):275-94 · February 2006with27 Reads
DOI: 10.1017/S0954579406060159 · Source: PubMed
Abstract
Data gathered over the course of a 25-year longitudinal study of 1,055 young people was used to examine gender differences in the onset of early parenthood and the developmental processes that place males and females at risk of becoming a young parent. Results revealed clear gender differences in the timing of early parenthood, with females being twice as likely as males to become a parent between the ages of 16 and 25 years. In contrast, the risk factors and life course processes that placed males and females at risk of an early transition to parenthood were very similar. Two exceptions were a gender-specific effect for maternal age and exposure to parental change, suggesting that having been raised by a younger mother and having experienced parental changes in your family of origin increased risks of early parenthood for females but not males. These findings contribute to our understanding of the effects of gender on life course development.

Figures

Figure

Full-text (PDF)

Available from: David M Fergusson, Jan 22, 2016
Gender differences in the transition
to early parenthood
LIANNE J. WOODWARD,
a
DAVID M. FERGUSSON,
b
and L. JOHN HORWOOD
b
a
University of Canterbury & Christchurch School of Medicine; and
b
Christchurch School of Medicine
Abstract
Data gathered over the course of a 25-year longitudinal study of 1,055 young people was used to examine gender
differences in the onset of early parenthood and the developmental processes that place males and females at risk of
becoming a young parent. Results revealed clear gender differences in the timing of early parenthood, with females
being twice as likely as males to become a parent between the ages of 16 and 25 years. In contrast, the risk factors
and life course processes that placed males and females at risk of an early transition to parenthood were very
similar. Two exceptions were a gender-specific effect for maternal age and exposure to parental change, suggesting
that having been raised by a younger mother and having experienced parental changes in your family of origin
increased risks of early parenthood for females but not males. These findings contribute to our understanding of the
effects of gender on life course development.
In recent years, there has been considerable
theoretical and empirical interest in the impor-
tance of gender for child development and
psychopathology across a range of areas, in-
cluding crime and antisocial behavior ~e.g.,
Keenan, Loeber, & Green, 1999; Silverthorn
& Frick, 1999!, mental health ~e.g., Crick &
Zahn-Waxler, 2003; Nolen-Hoeksema & Gir-
gus, 1994!, physical health ~e.g., Fillingim,
2003!, pregnancy0childbearing ~e.g., Coley &
Chase-Lansdale, 1998!, and suicidal behavior
~e.g., Beautrais, 2002!. Of particular interest
has been the extent to which there are gender
differences and similarities in the developmen-
tal trajectories and causal processes that shape
male and female development, with some ar-
guing that different developmental models need
to be developed for males and females ~e.g.,
Keenan & Shaw, 1997; Silverthorn & Frick,
1999!, and others suggesting that substantial
overlap may exist in the risk factors and life
course processes that influence male and fe-
male development ~e.g., Moffitt, Caspi, Rut-
ter, & Silva, 2002!.
One area of development in which gender
differences are perhaps most likely to mani-
fest is in the transition to early parenthood
given the differing biological and social roles
that males and females have in the conception
and care of children. For example, compared
to their early parenting male peers, younger
mothers are much more likely to assume the
role of primary caregiver, and in many cases,
may be the sole parent. Consistent with this is
research showing that an early transition to
parenthood has far-reaching physical, social,
and emotional consequences for females. These
include educational underachievement ~Klep-
inger, Lundberg, & Plotnick, 1995; Marini,
1984!; socioeconomic disadvantage ~Fursten-
berg, Brooks-Gunn, & Morgan, 1987; Wil-
This research was funded by grants from the Health Re-
search Council of New Zealand, the National Child Health
Research Foundation, the Canterbury Medical Research
Foundation and the New Zealand Lottery Grants Board.
Address correspondence and reprint requests to: Li-
anne Woodward, Child Development Research Group,
Department of Psychology, University of Canterbury, Pri-
vate Bag 4800, Christchurch, New Zealand; E-mail: lianne.
woodward@canterbury.ac.nz.
Development and Psychopathology 18 ~2006!, 275–294
Copyright © 2006 Cambridge University Press
Printed in the United States of America
DOI: 10.10170S0954579406060159
275
Advertisement:
liams, McGee, Olaman, & Knight, 1997 !,
welfare dependence ~Moore, Morrison, & Glei,
1995!, reduced marital opportunities ~Furs-
tenberg et al., 1987 !, maternal depression
~Colletta, 1983; Williams et al., 1997!, and
parenting problems, including increased rates
of physical punishment and child abuse ~Has-
kett, Johnson, & Miller, 1994; Woodward &
Fergusson, 2002!. Adding to these concerns
are recent findings suggesting that early moth-
erhood may now be an even more deviant and
disadvantaging life event than in previous de-
cades ~Butler, 1992; Maughan & Lindelow,
1997; Moffitt & Team, 2002!, with many con-
temporary young mothers now raising their
children in mother headed households, with
little or no financial or emotional support from
a partner, and in a socioeconomic climate of
declining benefit levels, qualification infla-
tion, and increased labor market demands
~Hotz & Williams-McElroy, 1997!.
Given recent debates about the importance
of gender for life course development and the
differing biological and social roles of youn-
ger mothers and fathers, it is surprising that
there have been no systematic evaluations of
the extent to which gender differences exist in
the prevalence and developmental processes
associated with an early transition to parent-
hood. Until recently, almost all studies con-
cerned with early parenthood have been based
on samples of younger mothers and their chil-
dren ~Brooks-Gunn & Chase-Lansdale, 1995;
Coley & Chase-Lansdale, 1998; Furstenberg,
Brooks-Gunn, & Chase-Lansdale, 1989!. Con-
sequently, very little is known about the full
extent of early fatherhood or the developmen-
tal processes that place young men at risk of
fathering a child and0or becoming a parent at
an early age. Of the handful of recent studies
that have examined developmental issues re-
lating to early fatherhood, virtually all have
been based on all male samples ~Dearden, Hale,
& Blankson, 1994; Fagot, Pears, Capaldi,
Crosby, & Leve, 1998; Jaffee, Caspi, & Mof-
fitt, 2001; Pears, Pierce, Kim, Capaldi, &
Owen, in press; Thornberry, Smith, & How-
ard, 1997 !. Furthermore, many of the samples
used in these studies consist of selected sam-
ples of high risk young males recruited during
middle childhood or early adolescence ~Fagot
et al., 1998; Pears et al., in press; Stouthamer-
Loeber & Wei, 1998; Thornberry et al., 1997!.
Studies concerned with the developmental
processes associated with an early transition
to motherhood have linked a wide range of
antecedent selection factors to an early transi-
tion to parental roles and responsibilities for
women ~Brooks-Gunn & Chase-Lansdale,
1995; Chase-Lansdale & Brooks-Gunn, 1994;
Furstenberg et al., 1989; Jaffee, Caspi, Mof-
fitt, Belsky, & Silva, 2001; Miller-Johnson
et al., 1999; Woodward, Fergusson, & Hor-
wood, 2001!. These factors span an individual’s
childhood social background ~poverty, wel-
fare dependence, large family size!, family
experiences ~family status, maternal age, pa-
rental educational underachievement; family
conflict, parent–child relations!, behavioral ad-
justment ~conduct problems!, educational
achievement ~IQ, school performance!, ado-
lescent peer interactions ~rejection, involve-
ment with deviant peers! and risk taking
behavior ~early sexual behavior, drug use; Bar-
done, Moffitt, Caspi, Dickson, & Silva, 1996;
Geronimus & Korenman, 1992; Manlove,
1997; McCormick & Brooks-Gunn, 1989;
Morgan, Chapar, & Fisher, 1995; Serbin et al.,
1998; Serbin, Moskowitz, Schwartzman, &
Ledingham, 1991; Udry, 1979; Woodward, Fer-
gusson, et al., 2001!.
Although much less research has been done
with males, findings from studies examining
the risk factors and life course processes asso-
ciated with early fatherhood generally suggest
marked similarities between the psychosocial
profiles of males and females becoming par-
ents at a young age. Consistent with the above,
studies of young fathers suggest that both fam-
ily and individual factors play an important
role in the prediction of early fatherhood. An-
tecedent factors linked with an early transi-
tion to fatherhood include social background
~socioeconomic disadvantage!, family life ~ma-
ternal age, parental change, presence of older
siblings in home!, behavioral adjustment ~de-
linquency!, educational ~poor school engage-
ment, educational underachievement!,peer
factors ~deviant peer involvement!, and risk-
taking behavior ~early onset sexual activity,
substance use0abuse; Dearden et al., 1994;
Fagot et al., 1998; Jaffee et al., 2001; Pears
276
L. J. Woodward, D. M. Fergusson, and L. J. Horwood
et al., in press; Stouthamer-Loeber & Wei,
1998; Thornberry et al., 1997; Xie, Cairns, &
Cairns, 2001!.
Adding to these findings from single gen-
der studies are recent comparative results from
the Carolina Longitudinal Study ~Gest, Ma-
honey, & Cairns, 1999; Xie et al., 2001! show-
ing that school aged males and females who
were aggressive, poor school achievers, un-
popular with their peers, and who came from
low socioeconomic status ~SES! backgrounds
were most at risk of becoming parents early
~Gest et al., 1999!. This risk profile character-
ized those becoming parents by age 20 as well
as those becoming parents by age 23.5 years.
A second study from the same group examin-
ing a broader range of antecedent risk factors
also demonstrated clear similarities in the risk
profiles of teen mothers and fathers, with early
parenthood being linked with individual ~high
aggression, low academic ability!, family ~low
SES!, and peer factors ~low popularity; Xie
et al., 2001!. However, some gender specific
effects were also found, with African Ameri-
can ethnicity, socioeconomic disadvantage, and
peer affiliations being more predictive of teen
motherhood than teen fatherhood.
Taken together, these findings suggest a
number of key points. First, findings suggest
that for both males and females, the transition
to parenthood represents a highly selective pro-
cess whereby those individuals raised in fam-
ilies characterized by socioeconomic adversity,
parental instability, and maternal role models
of young and single motherhood, as well as
those who have limited educational and per-
sonal resources are the most likely to become
parents at an early age. Second, these findings
highlight the possibility that there may be a
substantial overlap in the antecedent risk fac-
tors associated with early motherhood and early
fatherhood. Third, as might be expected given
the differing biological and social roles of
males and females in the processes of concep-
tion and early parenting, findings suggest that
it is possible that some gender differences
might exist in the developmental processes
that place males and females at risk of an
early transition to parenthood.
Against this general background, the goal
of this paper is to extend our understanding of
the role of gender in the transition to parent-
hood, and in par ticular, the extent to which
gender differences exist in the onset of parent-
hood by age 25 and the developmental pro-
cesses placing young people at risk of early
parenthood. As noted above, most previous
studies have tended to focus exclusively on
teenage parenthood, and especially teenage
motherhood. However, in the last 40 years
there has been a dramatic shift in the age of
entry to parenthood, with most contemporary
young people now deferring childbearing and
fatherhood until their late 20s or early 30s.
This population change, evident across most
industrialized countries, has increased mark-
edly the disparity between teenage and older
~cohort typical! parenthood. It also suggests
that, in addition to teenage parenthood, some
consideration should also be given to those
young people making an early or substantially
off-time entry to parenthood relative to the
norm. There is some suppor t for extending
analyses of early parenthood to include both
teenage and younger parents ~,25 years!. Spe-
cifically, there is growing evidence to suggest
that the developmental risks associated with
early motherhood may not be unique to the
offspring of teenage parents, but may, in fact,
represent a special case of a more general trend
for childhood risks to increase with increasing
youth of the mother ~Fergusson & Woodward,
1999; Ketterlinus, Henderson, & Lamb, 1991;
Ragozin, Basham, Crnic, Greenberg, & Rob-
inson, 1981!. For example, results from the
Christchurch Health and Development Study
have shown clear and pervasive linear associ-
ations between maternal age at first childbirth
and children’s later educational and psycho-
logical functioning, with decreasing maternal
age being associated with increasing rates of
later educational, psychosocial, and mental
health problems up to the age of 18 ~Fergus-
son & Woodward, 1999!. Thus, for the pur-
poses of this study early or off-time parenthood
was defined as having become a parent by
age 25.
To examine gender differences in the tran-
sition to parenthood, prospective longitudinal
data collected on a large representative birth
cohort is used to develop parallel models for
males and females of ~a! the timing of early
Gender and early parenthood 277
parenthood and ~b! the antecedent risk factors
and life course processes associated with early
parenthood ~by age 25!. Antecedent risk fac-
tors span a wide range of developmental
domains ~family background, structure and
functioning; individual characteristics; peer re-
lations! and periods of development ~child-
hood, adolescence!. This approach not only
has the advantage of allowing for direct gen-
der comparisons, but also helps to overcome
many of the methodological problems that have
hindered research with younger fathers. These
problems include difficulties associated with
the assessment of paternity from medical
records and0or maternal report, the need to
recruit representative samples of young fa-
thers that encompass both involved and un-
involved fathers, and problems relating to
sample retention due to the higher rates of
repartnering and blended families among youn-
ger parents.
The specific aims of the study were as
follows:
1. To examine gender differences in the rate
of onset of early parenthood for males and
females within the cohort over the period
from 16 to 25 years.
2. To compare the psychosocial profiles of
male and female cohort members who did
and did not become parents by age 25 years.
A wide range of prospectively measured
childhood and adolescent characteristics
were examined. Childhood factors included
measures of family social background,
structure, family functioning and the qual-
ity of family relations, and child charac-
teristics such as intellectual ability and
behavioral adjustment. Adolescent factors
included measures of deviant peer involve-
ment and risk taking, including early sex-
ual intercourse, drug use, novelty seeking
and deviant peer relations.
3. To develop multivariate models of the risk
factors and life course processes associ-
ated with early parenthood risk for males
and females, and to test for gender differ-
ences in the developmental processes asso-
ciated with early motherhood and early
fatherhood.
Methods
Sample
Participants were members of an unselected
birth cohort that has been extensively studied
as part of the Christchurch Health and Devel-
opment Study. This study consists of a longi-
tudinal study of a birth cohort of 1,265 children
~635 males, 630 females! born in Christ-
church, New Zealand. Christchurch is the sec-
ond largest city in New Zealand ~NZ!, with a
current population of approximately 320,000
~84% NZ European, 7% NZ Maori, 2% Pa-
cific Island, 4% Asian, 4% other!. The sample
for this study was recruited over a 4-month
period during 1977 by contacting mothers of
all live-born children giving birth in public
and private maternity hospitals within the
Christchurch urban region. Of the 1,310 moth-
ers giving birth during this time, 97% agreed
to participate. These children and their fami-
lies have now been studied at birth, 4 months,
1 year, at annual intervals to age 16, and again
at ages 18, 21, and 25 years. Based on a multi-
informant model, data collection spans a range
of sources, including parent interviews, teacher
assessments, self-reports, standardized psy-
chometric tests, and medical and official
records. An overview of the study design has
been given previously ~Fergusson, Horwood,
Shannon, & Lawton, 1989!.
Measures
Early parenthood. At age 25, sample mem-
bers were asked if they had ever given birth
to, or fathered a child. For each child identi-
fied, they were then asked to provide details
of the child’s age, gender, living arrange-
ments, as well as their current involvement
with the child. Comparison with parental sta-
tus information collected during earlier assess-
ment phases showed high ~100%! agreement
between earlier and age 25 reports of parent-
hood status. For the purpose of this study,
early parenthood was defined as having be-
come a biological parent by age 25 years. Both
custodial and noncustodial parents were in-
cluded. By age 25, 220 ~21.5%! sample mem-
bers had become parents. Only 17% of sample
278
L. J. Woodward, D. M. Fergusson, and L. J. Horwood
members becoming parents by age 25 had been
or were currently legally married to the other
biological parent of their child.
The description of the timing of these young
people’s transition to parenthood as early is
justified on the grounds that ~a! the majority
~78%! of the cohort had yet to become parents
at age 25, and ~b! national demographic data
indicated that, for the same period, the median
age of first parenthood for males and females
was 28.3 and 30.1 years, respectively ~per-
sonal communication, Statistics New Zea-
land !. We would also note that on the basis of
recent OECD international comparisons, New
Zealand has one of the highest rates of teen-
age childbearing ~27.701,000! in the devel-
oped world, falling behind the United States
~45.9! and England and Wales ~29.2! for the
same period ~2001; Boddington, Khawaja, &
Didham, 2003!.
Prospective childhood measures. To examine
the extent to which the timing of early parent-
hood was influenced by an individual’s earlier
childhood and adolescent experiences, a se-
ries of measures of social background, family
structure, family functioning, behavioral ad-
justment, cognitive ability, and adolescent func-
tioning were identified for inclusion in this
analysis. These measures were identified on
the basis of previous research and theory link-
ing these factors to an early transition to par-
enthood for males or females. Furthermore,
with the exception of ethnic identification and
child abuse exposure, all childhood measures
were assessed prior to the first cohort member
becoming a parent. A description of each of
these measures is provided below.
Social background. Three measures of each
sample members family social background
were included. First, family SES at the time of
the sample members birth was assessed using
the Elley and Irving ~1976! scale of SES for
New Zealand. This scale categorizes families
into six classes on the basis of paternal occu-
pation. For the purposes of this analysis, this
scale was collapsed into three levels: pro-
fessional0managerial, clerical0technical0
skilled, and semiskilled 0unskilled0unem-
ployed.An account of the construction and val-
idation of this scale is given by Elley and Irv-
ing ~ 1976!. Second, maternal education was
scored according to the highest academic qual-
ification repor ted by the sample member ’s
mother at the time of their birth and coded as
1 no formal qualifications,2 high school
qualifications,or3 tertiary level qualifi-
cations. The third measure was the young
person’s ethnicity. This was classified as either
New Zealand Maori or non-Maori based on the
individual’s self-reported ethnicity at age 25.
Family structure. Three measures of fam-
ily structure were included. Age of the sample
members mother at the time of first childbirth
and the type of family ~single0two parent! en-
tered by each respondent at birth were the
first two measures. Third, on the basis of com-
prehensive life history data collected annually
from birth to 15 years, a composite measure
of the total number of parental changes due to
separation0divorce, death, remarriage, and rec-
onciliation was formed.
Family functioning. Four measures describ-
ing sample members childhood family rela-
tionships were included. The first measure
consisted of the extent to which each sample
members family was characterized by inter-
parental conflict. Interparental conflict was
assessed on the basis of three items describing
the quality of marital relations during the pre-
ceding 12 months. These questions were asked
annually from birth until age 10, and included
~a! whether the parents had engaged in pro-
longed arguments, ~b! whether the child’s
mother reported an assault by her partner, and
~c! whether the child’s mother reported hav-
ing experienced sexual difficulties. These items
were then summed to produce a scale measure
of interparental conflict during the first 10
years of the sample members life. Second, at
age 18 and 21, sample members were inter-
viewed about the extent to which their mother
and father had used physical punishment dur-
ing their childhood years ~birth to age 16!.
Ratings of maternal and paternal behavior were
then combined into a composite 4-point scale
based on the highest level of exposure to phys-
ical punishment reported at either age 18 or
21 ~Fergusson & Lynskey, 1997!. This classi-
Gender and early parenthood 279
fication consisted of parents never used phys-
ical punishment, parents seldom used physical
punishment, at least one parent regularly used
physical punishment, and at least one parent
used physical punishment too often or too
severely. Third, at age 18 and 21, sample mem-
bers were also interviewed about their experi-
ence of childhood sexual abuse prior to the
age of 16 years. Respondents who reported
having been sexually abused by a parent or
another person were asked a further series of
questions relating to the extent and nature of
their abuse experiences ~Fergusson, Lynskey,
& Horwood, 1996!. On the basis of this ques-
tioning, sample members were classified into
four groups based on the most severe form of
sexual abuse reported at either age 18 or 21.
These groups were no sexual abuse; noncon-
tact sexual abuse; contact sexual abuse not
involving attempted or completed intercourse;
and sexual abuse involving attempted or com-
pleted oral, anal, or vaginal intercourse. The
final measure of childhood family relations
consisted of sample members’ perceived at-
tachment to their parents. This was assessed
at age 15 using the 28-item Inventory of Par-
ent and Peer Attachment ~Armsden & Green-
berg, 1987!. A total parental attachment score
was computed by summing the trust and com-
munication subscale scores and then subtract-
ing the alienation subscale score from the total.
Examples of items from the parental attach-
ment scale include, “I tell my parents about
my problems and troubles” and “My parents
help me to understand myself better.” The co-
efficient a for the parental attachment scale
was .87.
Child characteristics. The earlier behavior
and school achievement of sample members
was described using four measures. The first
two measures assessed the child’s tendencies
to conduct and attentional problems in middle
childhood. At each year from age 7–13, parent
and teacher reports of these behaviors were
obtained using instruments that combined items
from the Rutter ~Rutter, Tizard, & Whitmore,
1970! and Conners ~1969, 1970! behavior rat-
ing scales. Conduct problem items spanned a
range of behaviors relating to disobedience
and defiance of authority, fits of temper and
irritability, aggression or cruelty to others, de-
struction of property, lying, stealing, and sim-
ilar behaviors. Attentional problem items
spanned a range of behaviors relating to inat-
tention, poor concentration, short attention
span, distractibility, restlessness, and hyper-
activity ~Fergusson, Horwood, & Lloyd, 1990!.
All items were scored on a 3-point scale rang-
ing from 1 ~not at all ! to 3 ~a great deal !. For
the purposes of the present analysis the parent
and teacher item scores were summed for each
year and then averaged over an interval of
7–13 years to create two global measures of
the child’s tendencies to conduct and atten-
tional problems in middle childhood. This pro-
cedure was used to reduce the effects of
situational and rater bias. The resulting scales
were of excellent reliability with coefficient a
values of .97 for the scale of conduct prob-
lems and .93 for attentional problems. The
third measure consisted of the Test of Scho-
lastic Abilities, which was administered at age
13 to assess young people’s scholastic ability.
This is a general purpose test designed to as-
sess “verbal and numerical reasoning abilities
deemed to be prerequisites for success in ac-
ademic aspects of the school curriculum”
~Reid, Jackson, Gilmore, & Croft, 1981, p. 4!.
An account of the construction and validation
of this measure has been provided by Reid
et al. ~1981!. The reliability of this test, as-
sessed using coefficient a, was .95. The final
measure was a teacher rated assessment of
each sample members school grade point av-
erage. When sample members were aged 11,
12, and 13 years, their class teacher was asked
to rate their performance in five curriculum
areas: reading, written expression, handwrit-
ing, spelling, and mathematics. Ratings were
made on a 5-point scale that ranged from very
good to very poor. For this analysis, teacher
ratings were summed over the five curriculum
areas and then averaged over the 3-year
period to provide a robust measure of the
individual’s school performance over the in-
terval from 11 to 13 years. The reliability of
this measure, assessed using coefficient a,
was .96.
Adolescent functioning. Four measures of
adolescent risk taking and behavior spanning
280
L. J. Woodward, D. M. Fergusson, and L. J. Horwood
early substance use, sexual behavior, novelty
seeking, and deviant peer involvement were
included. Early substance use was assessed
by interview at age 15. Those reporting using
cigarettes, alcohol, or cannabis at least monthly
in the last year were classified as early regular
substance users. Early sexual intercourse was
assessed at ages 15 and 16 years. Sample mem-
bers were questioned about whether they had
ever engaged in consensual sexual intercourse,
with those responding positively being asked
to provide an estimate of the age at which they
initiated consensual intercourse. Those report-
ing having consensual sexual intercourse be-
fore age 16 were classified as having early
sexual intercourse. Novelty seeking was as-
sessed at age 16 using novelty seeking items
from the Tridimensional Personality Question-
naire ~Cloninger, 1987!. This scale was found
to be of moderate reliability ~a .76!. Fi-
nally, to provide a measure of the extent to
which sample members were involved with
delinquent or substance using peers at age 15,
an index of deviant peer affiliations was con-
structed. This index was based on sample
members reports of the extent to which their
best friend and other friends used tobacco,
alcohol, and cannabis, truanted, or broke the
law. These items were summed to produce a
scale measure of the extent to which the
sample member reported affiliating with de-
linquent or substance using peers. The con-
struction of this scale has been described
previously ~Fergusson & Horwood, 1996! , and
was of moderate reliability, with an a coeffi-
cient of .76.
Statistical analysis
The analysis was conducted in four stages. In
the first stage!, life table methods were used
to estimate the accumulative rates of parent-
hood for males and females over the interval
from 16 to 25 years, with a log-rank test used
to compare the rate of parenthood onset in the
two groups.
In the second stage the associations be-
tween rates of parenthood by age 25 and a
series of risk factors were examined sepa-
rately for males and females. For the purposes
of data display all continuously scaled risk
factors were classified into a series of broad
class intervals to permit an examination of the
extent to which variations in risk were associ-
ated with changes in the rate of parenthood.
Life table estimates of the accumulative rate
of parenthood by age 25 were calculated for
each level of each risk factor, and for each
gender group. The strength of association be-
tween the risk of parenthood and the risk fac-
tor was assessed using the log-rank test. The
analysis was then extended to test for gender
differences in the strength of risk factor asso-
ciations. This was achieved by fitting a nested
proportional hazards regression model for each
risk factor of the form:
h~t; Xi ! ho~t !exp@B0 B1Gi
B2
m
Xi
m
B2
f
Xi
f
#,
where h~t; Xi ! is the hazard or instantaneous
risk of transition to parenthood at time t for
participant i; ho~t ! is an arbitrary baseline
hazard function; Gi is the individual’s gender;
and variables Xi
m
and Xi
f
represent the risk
factor Xi for males ~m! and females ~f !, re-
spectively. Specifically, Xi
m
was a variable
that took the value Xi if the subject was male,
and zero otherwise. Conversely, Xi
f
was a vari-
able that took the value Xi if the subject was
female, and zero otherwise. The coefficient
B1 represents the gender difference in the rate
of parenthood. The coefficients B2
m
and B2
f
represent the impact of the risk factor Xi on
the rate of parenthood for males and females,
respectively. A test of the equivalence of the
risk factor association for males and females
may be obtained from a Wald chi square test
of the null hypothesis that B2
m
B2
f
.
In the third stage of the analysis, the nested
model above was extended to develop a multi-
variate model of the significant predictors of
parenthood. The general model fitted was of
the form:
h~t; Xi !
ho~t !exp
B0 B1Gi
(
B2j
m
Xij
m
(
B2j
f
Xij
f
,
Gender and early parenthood 281
where h~t; Xi !, and ho~t ! were defined as
above; Gi was the individual’s gender; and
Xij
m
and Xij
f
were the set of predictors Xij for
males and females, respectively. In fitting this
model all variables were scaled in their origi-
nal metric as described above, rather than in
class intervals. Model fitting was conducted
using both forwards and backwards methods
of variable selection to identify the best fitting
and most parsimonious set of predictors
~Greenland, 1989!. A test of the significance
of each risk factor in the fitted model was
obtained from a Wald chi square test of the
joint hypothesis B2j
m
B2j
f
0. As previ-
ously, tests of the equivalence of the effect of
each risk factor for males and females were
derived using Wald chi square tests of the hy-
pothesis B2j
m
B2j
f
for each risk factor. In
the final fitted model, risk factor effects that
were not significantly different between males
and females were constrained to be equal.
In the final stage of the analysis, the results
of the fitted proportional hazards model were
used to construct a risk factor index for each
sample member by computing a weighted sum
of the significant predictors for each sample
member with each variable weighted by its
regression coefficient. Separate scores were
calculated for males and females. The result-
ing scores were then classified into a series of
class intervals and life table estimates of the
rate of parenthood for males and females were
calculated for each level of risk.
Sample size and sample bias
The present analysis is based on a sample of
1,055 participants ~522 males, 533 females!
for whom information was available on par-
enthood up to age 25. This sample represented
83% of the initial cohort of 1,265 partici-
pants. However, as a result of sample attrition
and missing observations on some variables
the sample N values vary somewhat between
analyses and the final prediction model is based
on a sample of 949 participants. To examine
the impact of sample loss on analysis con-
clusions, regression imputation of missing
observations on the predictor variables was
conducted using the impute function of Stata
~StataCorp, 1999!, and the regression analy-
ses repeated with missing values replaced by
their imputed estimates. This analysis pro-
duced essentially identical conclusions to the
results reported here, suggesting that the ef-
fects of missing data and possible sample se-
lection bias on the results were minimal.
Results
Rates of early parenthood in the cohort
Figure 1 shows a plot of the cumulative per-
centages of males and females who became
biological parents over the interval from 16 to
25 years. The earliest age of entry to parent-
hood for both males and females was 16 years,
with six ~1.1%! females and three ~0.6%! males
becoming parents at age 16. This early gender
difference in the rate of parenting onset per-
sisted throughout the period from 16 to 25
years, with the rate of early motherhood con-
sistently being 1.5 to 2 times the rate of early
fatherhood. By age 25 years, 26.2% of fe-
males and 16.6% of males had become par-
ents. A log-rank test showed this difference to
be highly significant, x
2
~1! 14.9, p , .0001.
Family of origin and individual predictors
of parenthood by age 25
Although the above results confirm the pres-
ence of strong gender differences in the rate of
onset of early parenthood, the extent to which
differences also exist in the developmental pro-
cesses that place young men and women at
risk of an early transition to parenthood re-
mains unclear. This issue is examined in
Table 1, which shows the associations be-
tween a series of childhood and adolescent
risk factors describing the sample members
life history and experiences prior to age 16
and rates of parenthood by age 25. Childhood
risk factors reported in this table include
measures of young people’s family of origin
~family social background, structure, and func-
tioning! and measures of their earlier personal
characteristics ~behavioral adjustment, scho-
lastic ability, and grade point average!. Ado-
lescent risk factors spanned a range of risk
taking measures, including early sexual risk
taking, substance use0experimentation, nov-
282
L. J. Woodward, D. M. Fergusson, and L. J. Horwood
Figure 1. The proportion of young men and women becoming parents by age 25.
283
Table 1. Cumulative rates (%) of parenthood by age 25 for males and females
by measures of childhood and adolescent risk factors
Males Females
Measure n % n %
Sociodemographic background
Socioeconomic status ~birth!
Professional, managerial 98 8.5 119 13.8
Clerical, technical, skilled 284 13.4 291 23.0
Semiskilled, unskilled, unemployed 140 27.4 123 46.1
p ,.0001 ,.0001
Maternal education ~birth!
No qualifications 256 22.1 268 38.0
High school qualifications 169 12.7 150 18.5
Tertiary qualifications 97 8.7 115 8.9
p ,.001 ,.0001
Ethnic identification ~25 years!
Mãori 64 38.2 70 48.9
NZ European0other 446 13.3 456 22.8
p ,.0001 ,.0001
Family structure
Maternal age at first childbirth
,20 years 84 27.5 94 57.1
20–24 years 210 22.9 217 24.1
25 years 228 6.8 222 12.0
p ,.0001 ,.0001
Family placement at birth
One-parent family 34 38.2 34 68.3
Two-parent family 488 15.1 499 23.4
p ,.001 ,.0001
Changes of parents ~0–15 years!
None 346 11.0 336 15.9
1–2 changes 94 24.8 94 32.8
3 changes 82 30.3 103 54.2
p ,.0001 ,.0001
Family functioning
Interparental conflict ~0–10 years!
Lowest 50% 267 9.0 289 14.5
51–75% 94 25.2 80 43.4
Highest 25% 100 24.2 102 44.3
p ,.0001 ,.0001
Parental use of physical punishment ~0–16 years!
Never0seldom 421 13.2 447 21.0
Regular 75 25.6 43 37.0
Severe0harsh 25 40.5 42 69.0
p ,.0001 ,.0001
Childhood sexual abuse ~0–16 years!
None0noncontact CSA 496 15.4 437 22.0
Contact not intercourse 12 25.0 42 41.7
Intercourse 13 38.5 53 47.2
p ,.01 ,.0001
Attachment to parents ~15 years!
Lowest 25% 113 24.6 124 35.5
Middle 50% 248 14.8 230 25.6
Highest 25% 110 9.5 131 17.0
p ,.005 ,.001
284 L. J. Woodward, D. M. Fergusson, and L. J. Horwood
elty seeking, and deviant peer involvement.
The associations are shown separately for
males and females, and for each comparison,
the strength of association has been tested using
the log-rank test. Examination of the table leads
to the following conclusions:
1. For both males and females, all risk factors
in the table were significantly ~ p , .05!
related to the risk of early parenthood.
Strong associations between early parent-
hood risk and childhood social disadvan-
tage, parental role models of young single
Table 1. ~cont.!
Males Females
Measure n % n %
Child behavior
Conduct problems ~7–13 years!
Lowest 25% 101 6.2 193 13.9
Middle 50% 247 12.9 244 28.0
Highest 25% 161 27.2 84 46.3
p ,.0001 ,.0001
Attentional problems ~7–13 years!
Lowest 25% 88 5.9 210 13.7
Middle 50% 238 11.0 242 31.3
Highest 25% 183 27.4 69 41.8
p ,.0001 ,.0001
School achievement
Scholastic ability ~13 years!
Lowest 25% 113 19.5 81 37.7
Middle 50% 175 14.9 209 25.4
Highest 25% 88 2.4 102 14.8
p ,.001 ,.001
Grade point average ~11–13 years!
Lowest 25% 164 26.4 82 35.8
Middle 50% 243 13.4 256 29.6
Highest 25% 82 5.1 166 15.8
p ,.0001 ,.001
Adolescent factors
Early regular substance use ~15 years!
No 311 13.4 336 21.0
Yes 160 20.9 149 36.3
p ,.05 ,.001
Early sexual intercourse ~,16 years!
No 382 11.6 385 17.0
Yes 123 31.1 135 49.7
p ,.0001 ,.0001
Novelty seeking ~16 years!
Lowest 25% 104 7.0 114 18.6
Middle 50% 269 17.4 258 22.4
Highest 25% 94 21.0 106 41.0
p ,.01 ,.001
Deviant peer affiliations ~15 years!
Lowest 50% 234 10.2 208 13.7
Middle 51–75% 114 13.8 128 30.3
Highest 25% 123 28.8 149 38.8
p ,.0001 ,.0001
Gender and early parenthood 285
parenthood, family dysfunction, educa-
tional underachievement, behavioral mal-
adjustment, and adolescent risk taking
clearly demonstrate the highly selective
nature of the timing of parenting onset
for both males and females. Furthermore,
across both gender groups there were clear
and pervasive tendencies for rates of early
parenthood to increase with increasing lev-
els of exposure to earlier social, family and
individual adversity.
2. Despite the lower rate of early parenthood
for males, the associations between ante-
cedent risk factors and rates of early par-
enthood appear to be similar across gender
groups. This is illustrated by a comparison
of the profiles of the relative risk of parent-
hood in the highest risk group compared to
the lowest risk group for each risk factor
across males and females. For both gender
groups these relative risks typically ranged
between 2 and 5, and for both groups the
median relative risk was 2.9. The similar-
ity in the psychosocial backgrounds of
young males and females becoming par-
ents by age 25 years was further confirmed
by fitting a series of nested proportional
hazards regression models to test the equal-
ity of regression slopes across males and
females for each risk factor ~see the Meth-
ods section!. These analyses failed to re-
veal any significant gender differences in
the regression slopes.
Multivariate model of earlier life course
predictors of early parenthood: Do gender
differences exist?
The above results suggest that the risk of an
early transition to parenthood was related to a
wide array of childhood and adolescent risk
factors, and that although there were clear gen-
der differences in the rate of parenthood, the
various risk factors appeared to act in a simi-
lar way for both males and females. To exam-
ine the multivariate properties of the data, a
nested proportional hazards regression model
was fitted using the risk factors in Table 1 as
predictors of parenthood. In this analysis all
continuous variables were scaled in their nat-
ural metric as described in the Methods sec-
tion, rather than in the arbitrary categories used
in Table 1. For each factor having a signifi-
cant net effect in the fitted model, tests of the
equality of the model parameters between
males and females were also conducted. The
results of the final fitted model are given in
Table 2, which shows the fitted model param-
eters, standard errors and test of significance
of each factor in the model, together with the
results of the tests of gender equivalence of
the model parameters for males and females.
Table 2 shows the following:
1. The analysis identified nine factors that
made significant net contributions to the
prediction of early parenthood in the co-
hort. These factors included: gender ~ p ,
.01!; family SES ~ p .02!; Maori ethnic-
ity ~ p , .005!; age of respondent’s mother
at first childbirth ~ p , .0001!; changes of
parents ~ p .04!; parental use of physical
punishment ~ p , .01!; childhood conduct
problems ~ p .01!; teacher rated grade
point average ~ p , .001!; and early onset
sexual intercourse ~ p , .0001!.
2. In addition, there was evidence of a signif-
icant gender difference in the regression
slopes for males and females on two of the
nine predictor variables. For the age of
the respondent’s mother at first childbirth,
the estimated coefficient for females was
large and highly significant ~B ⫽⫺.120,
SE .028, p , .0001! compared to the
coefficient for males that was smaller and
statistically nonsignificant ~B ⫽⫺.044,
SE .032, p .17!. A similar pattern of
results was found for the effects of expo-
sure to parental changes during childhood,
with the estimated coefficient being signif-
icant for females ~B .076, SE .028,
p , .01! but not for males ~B ⫽⫺.012,
SE .044, p .79!. These results indicate
that having a mother who herself first gave
birth at a younger age and having been
exposed to multiple caregiver changes were
more influential in predicting early parent-
hood for females than for males. However,
there was no evidence that the regression
parameters for any other risk factor varied
significantly with gender.
286
L. J. Woodward, D. M. Fergusson, and L. J. Horwood
More generally, these findings suggest that
the onset of parenthood was influenced by a
wide array of risk factors, and that to a large
extent, the risk factors and life course pro-
cesses associated with early motherhood were
similar to the risk factors and life processes
associated with early fatherhood. Those most
at risk of an early transition to parenthood
were Maori females from low SES back-
grounds, who were born to mothers who them-
selves were young at first childbirth, who
experienced multiple family changes, who ex-
perienced frequent or severe physical punish-
ment in childhood, who exhibited early-onset
conduct problems, who did less well at school,
and who engaged in early-onset sexual behav-
ior. Conversely, those least at risk were non-
Maori males from high SES backgrounds, who
grew up in stable family environments, who
were not subject to physical punishment,
who did not develop conduct problems in child-
hood, who did well in school, and who exhib-
ited later onset of sexual intercourse.
Although results from the proportional haz-
ards regression analysis identified the signifi-
cant predictors of parenthood, the model pa-
rameters by themselves do not give a clear
indication as to how the different risk factors
combine to influence the accumulative risk of
parenthood up to age 25. This issue is exam-
ined in Figure 2. In this figure, the regression
parameters in Table 2 were used to estimate a
weighted risk factor score for males and fe-
males. The resulting score was then classified
into a series of class intervals and the proba-
bility of parenthood by age 25 calculated for
each group. The figure shows the observed
rates of parenthood for males and females for
each level of the risk factor score. For both
males and females, there was evidence of sub-
stantial variability in the rate of early parent-
hood with increasing levels of risk exposure.
For those in the lowest quartile of the risk
score, the rate of parenthood was 6% for males
and 2.4% for females. By comparison, in the
highest decile of risk scores, rates of parent-
hood had increased to 52% for males and 71%
for females. At all but the very lowest level of
risk, the observed rate of parenthood was con-
sistently higher among females than males,
Table 2. Final fitted proportional hazards regression model of the childhood
and adolescent predictors of early parenthood
Test of
Gender
Equiv. of
Regression
Slopes
Risk Factor B ~SE !
Factor
Signif.
p x
2
~1! p
Gender 2.357 ~.929! .01
Socioeconomic status .298 ~.123! .02 1.20 .27
Mãori ethnicity .551 ~.17!,.005 0.00 .98
Age of participant’s mother at first childbirth ,.0001 6.20 .01
Females .120 ~.028!,.0001
Males .044 ~.032! .17
Change of parents ~0–15 years! .04 5.32 .02
Females .076 ~.028!,.01
Males .012 ~.044! .79
Parental use of physical punishment ~0–16 years! .272 ~.100!,.01 2.30 .13
Childhood conduct problems ~7–13 years! .022 ~.009! .01 0.09 .76
Grade point average ~11–13 years! .332 ~.089!,.001 0.00 .97
Early sexual intercourse ~,16 years! .830 ~.152!,.0001 2.39 .12
Note: N 949.
Gender and early parenthood 287
reflecting the higher rate of transition to par-
enthood among females.
1
Supplementary analysis
The above results are based on an examina-
tion of the transition to parenthood up to age
25. However, as noted previously, descrip-
tions of early parenthood are commonly asso-
ciated with teenage parenthood, and it could
be suggested that the risk factors and life pro-
cesses associated with transitions to parent-
hood in the teenage years may be different to,
or contribute in a different way, to parenthood
onset in the early 20s. To examine this issue, a
sensitivity analysis was conducted in which
the end point of follow-up for the propor-
tional hazards model was varied over the age
range from 18 to 25 years. In general, these
analyses identified a consistent set of predic-
tors of parenthood for all end points, and the
parameter estimates for these risk factors were
very similar across all models. However, the
precision of estimation was reduced for mod-
els fitted at the youngest ages, reflecting the
lower accumulative rates of parenthood in the
teenage years. As a result, tests of gender dif-
ferences in the effect of different risk factors
were typically nonsignificant for models fit-
ted for teenage parenthood, despite the fact
that parameter estimates for these models sug-
gested similar gender differences in effect sizes
to those reported in Table 2 for the measures
of changes of parents and age of respondent’s
mother at first childbirth. These findings sug-
gest that, in general, the same set of risk pro-
cesses were operating in a similar fashion for
both males and females to predict both teen-
age and later parenthood.
Discussion
In this study we used prospective data col-
lected over the course of a 25-year longitudi-
nal study to examine gender differences in the
timing and early life course experiences that
place young people at risk of an early transi-
tion to parenthood. This study contributes to a
growing body of research concerned with gen-
1. It has been suggested that the use of regression weights
to calculate a risk factor index may capitalize on sam-
ple specific properties that may not generalize to other
samples and that a risk score calculated using unit
weights may be more appropriate. We would note that
in this sample a unit weight risk score performs just as
well as the regression-based risk index. For example, a
simple risk score can be calculated by categorizing the
significant predictor variables as shown in Table 1,
assigning a value of 0 to the lowest risk category for
each predictor, a value of 1 for the next highest risk
category, and a value of 2 for the highest risk category
~if applicable! and then summing the significant pre-
dictors for males and females separately. When tabu-
lated against the rate of parenthood by age 25, this
index produced an almost identical pattern of results
to those seen in Figure 2.
Figure 2. The rates ~%! of parenthood to age 25 for males and females by proportional hazards risk
score.
288 L. J. Woodward, D. M. Fergusson, and L. J. Horwood
der and life course development by compar-
ing the risks of early parenthood for males
and females, as well as the psychosocial pro-
files of younger mothers and fathers. Of par-
ticular interest was the extent to which the
childhood and adolescent risk factors and life
course experiences associated with early par-
enthood were the same or different for males
and females. The study’s large representative
sample of males and females, its prospective
longitudinal design, repeated measures over
time, and multimethod approach make it
uniquely suited to addressing these important
questions. The major study findings and their
theoretical and applied implications are dis-
cussed below.
Gender differences in the onset
of early parenthood
Consistent with previous research, there were
clear differences in the risk of early parent-
hood for males and females in this cohor t.
Although both gender groups began the tran-
sition to parenthood during their 16th year,
the rate of parenting onset among females was
approximately double that for males, with this
difference persisting throughout the teenage
years and early 20s. By age 25, just over a
fifth of the cohort had become parents, with
around 60% of these being women. This is
consistent with rates of early parenthood from
US samples, which show that by age 25, 22.6%
of young people have become a biological
parent, with rates being three times higher in
females ~33.6%! than males ~11.3%; Cohen,
Kasen, Chen, Hartmark, & Gordon, 2003!.
Although these findings suggest that a sub-
stantial number of young people become par-
ents by age 25, they also highlight that the
transition to parenthood for this early parent-
ing group is off-time relative to their peers
given that nearly 75% of females and 85% of
males in the cohort were still yet to become
parents. Findings also confirm a significant
shift in the onset of parenting for contempo-
rary youth, and are in line with recent national
data showing a shift toward the deferment of
parenthood until the early 30s ~personal com-
munication, Statistics New Zealand, 2003!.
Given clear demographic shifts toward de-
layed parenthood and the growing disparity
between the life course experiences of youn-
ger and older parents, there is perhaps a need
to reconsider our notions of early or off-time
parenthood to include not just teenage par-
ents, but also those individuals who make the
transition to parenthood in their early 20s, at
an age that is now much earlier than most of
their peers.
Gender differences in the antecedent risk
factors associated with early parenthood
In contrast to strong gender differences in the
prevalence of early parenthood, findings sug-
gested that the psychosocial profiles of early
parenting males and females were highly sim-
ilar. Irrespective of gender, bivariate analyses
showed that early parenthood was correlated
with a wide range of childhood factors span-
ning socioeconomic disadvantage, maternal
role models of young single parenthood, inter-
parental conflict and family breakdown,
punitive parenting, sexual abuse, weak parent–
child bonds, Maori ethnicity, childhood
behavior problems, and educational under-
achievement. Furthermore, a series of adoles-
cent lifestyle factors were also linked with
early parenthood, suggesting that adolescents
with novelty seeking tendencies who initiated
sexual intercourse early ~,16 years!, engaged
in early drug use and who socialized with de-
viant peers were at increased risk of becoming
a young parent. Also of note was the finding
that for males and females, rates of early par-
enthood increased with increasing levels of
earlier psychosocial adversity.
The strong overlap in the psychosocial pro-
files of younger mothers and fathers was fur-
ther confirmed when multivariate modeling
methods were applied to the data. These analy-
ses sought to identify the key childhood and
adolescent predictors of early motherhood and
early fatherhood, and to test for possible gen-
der specific effects that might imply that one
gender was more vulnerable to the effects of a
risk factor than the other. Results revealed
marked similarities in the risk factors and life
processes placing both males and females at
risk of a becoming a young parent. In partic-
ular, the nested regression model identified
Gender and early parenthood 289
eight significant risk factors for early parent-
hood ~aside from gender!. For six of the eight
risk factors, no significant gender differences
in model coefficients were found, suggesting
that, for the most part, the same childhood and
adolescent risk factors were predictive of both
early fatherhood and early motherhood and
contributed to parenthood risk in a similar man-
ner for both gender groups. The two excep-
tions to this were the effects of age of the
respondent’s mother at first childbirth and ex-
posure to parental changes during childhood
~birth to 15 years!, with results suggesting
that early exposure to maternal role models of
young parenthood and parental changes may
be more salient for girls than for boys in pre-
dicting a subsequent early entry to parenthood.
An examination of the relationship be-
tween risk exposure and subsequent parent-
hood risk showed that the effects of these
earlier life course adversities and disadvan-
tages were cumulative. At low levels of risk
exposure ~25th percentile! rates of parent-
hood remained low. However, with increasing
levels of risk exposure ~both in terms of the
severity and diversity of risk factors! a steady
and rapid increase in rates of early parenthood
was evident for both males and females. Fe-
males in the highest risk decile of the cohort
were approximately 30 times more likely to
become a parent, whereas males in this decile
were almost 9 times more likely to become a
parent than their lowest risk peers.
The key independent predictors of early
parenthood risk identified in this analysis are
in general agreement with other studies of
younger mothers and fathers. In particular,
childhood conduct problems and0or anti-
social behavior have been shown to be one of
the most robust predictors of pregnancy and
early childbearing0fatherhood ~Bardone et al.,
1996; Fagot et al., 1998; Jaffee et al., 2001;
Serbin et al., 1991; Stouthamer-Loeber & Wei,
1998; Woodward & Fergusson, 1999!. For ex-
ample, results from the Dunedin longitudinal
study showed that among the 77 study mem-
bers who had become parents before age 21,
58% had a prior diagnosis of conduct disorder
~Moffitt et al., 2002!. Furthermore, educa-
tional underachievement has also been previ-
ously linked with an early onset of parenthood
among females ~Fergusson & Woodward,
2000b; Hoffman, Foster, & Furstenberg, 1993;
Hotz & Williams-McElroy, 1997; Klepinger
et al., 1995; Upchurch & McCarthy, 1990!.
Although there has been some debate as to
whether an early transition to parenthood serves
to disrupts schooling or whether in fact, poor
school achievement makes early parenthood
more likely, converging evidence now exists
to suggest that for females at least, the latter
life sequence is more frequently the case ~Fer-
gusson & Woodward, 2000b; Maynard, 1995;
Upchurch & McCarthy, 1990!. Extending on
this, the present study suggests that education-
ally underachieving young males are also more
likely to father a child than their higher achiev-
ing male peers. The extent to which these at
risk young fathers will remain a part of their
children’s lives will be of interest.
In conjunction with individual tendencies
to early onset conduct problems and lower
educational attainment, it is clear from the
present study that young people’s experiences
in their family of origin play an important role
in determining parenthood timing. Similar to
the findings of Gest et al. ~1999!, family SES
was predictive for both males and females even
after taking into account more proximal child
and family factors. This could reflect class-
related differences in attitudes toward family
formation and parenthood timing, and0or in-
creased economic barriers to alternative life
paths such as going to university or fur ther
career training ~Fergusson & Woodward,
2000a!. Maori ethnicity was also predictive of
an earlier transition to parenthood for males
and females. Although the reasons for this are
not clear, previous international studies have
also demonstrated cultural differences in par-
enthood timing. There is also some suggestion
from previous analyses of this cohort that
young Maori women who became pregnant
were less likely to seek a termination than
other women ~Woodward, Horwood, & Fer-
gusson, 2001!.
In addition to social demographic factors,
findings from this study also highlight the im-
portance of young people’s experiences in their
family of origin in determining the timing of
their own transition to adult parental roles.
Three family process factors made significant
290
L. J. Woodward, D. M. Fergusson, and L. J. Horwood
and unique contributions to early parenthood
risk. First, there was evidence to suggest that
for both males and females, having been raised
in a family environment characterized by high
rates of physical punishment was associated
with an increased risk of an early transition to
parenthood. One possible explanation for this
finding could be that parental negativity may
hasten young people’s exit from the family,
thus weakening family of origin ties as well as
parental monitoring of activities and decision
making. To date, few studies have examined
linkages between specific parental practices
and parenthood risk. However, given con-
cerns about the intergenerational transmission
of negative parenting practices, developing a
better understanding of the process by which
childhood exposure to punitive parenting in-
creases risk of early parenthood would seem
an important issue for future study.
Second, a family factor linked with early
parenthood risk was the age of the respondent’s
own mother at first childbirth, with findings
showing that having been raised by a mother
who was herself a young parent contributed to
early parenthood risk. However, this associa-
tion was only evident for females, suggesting
that exposure to maternal role models of early
motherhood is more likely to encourage an
accelerated entry to parenthood among daugh-
ters than among sons. An important question
that needs to be addressed by future research
is the extent to which the age of a child’s
father influences later parenthood timing. It
could be hypothesized on the basis of gender
role modeling theory that just as childhood
exposure to young motherhood appears to be
more salient for daughters, the age of the
fathers own entry to fatherhood may be more
predictive for sons.
Third, a gender difference was also evident
for the effects of exposure to parental separa-
tion on later risks of early parenthood. The
number of parental transitions experienced by
young people in their family of origin made a
unique contribution to early parenthood risk
for females, but not for males. Because most
family breakdowns ~approximately 90%! oc-
curring within the cohort resulted in children
living with their mothers, these results could
reasonably be interpreted as suggesting that
spending time in a mother headed household
may place daughters, but not sons, at in-
creased risk of themselves becoming young,
and in many cases, single mothers. Several
mechanisms could account for this associa-
tion. One explanation could be that children
subject to multiple parental transitions are more
likely than children from intact families to be
exposed to parental dating behaviors and con-
versations that serve to heighten their sexual
awareness and acceptance of sexual relations
~Maccoby, 1991!. A second explanation is that
disrupted family relations during childhood
may impair parental monitoring of girl’s ac-
tivities and behaviors. A third explanation could
be that family breakdown may potentially
weaken family ties and increase the likelihood
that young people will seek out intimacy and
new family connections through early partner-
ship formation and parenting. Finally, these
findings may reflect differences in expecta-
tions, either conscious or unconscious, of par-
enthood as a result of earlier experiences of
family life patterns. For example, from an evo-
lutionary perspective, early stressful family
experiences involving fewer economic re-
sources, punitive parental care, and high lev-
els of family disruption ~especially among
girls! may well incline young people toward a
more quantity-oriented, rather than quality-
oriented, reproduction strategy ~Belsky, Stein-
berg, & Draper, 1991!. Such a strategy would
involve investing in early sexual partnering
and more children, as opposed to the achieve-
ment of a stable partnership and the mutual
investment in fewer children. Fur ther re-
search should help to clarify the relative role
of psychosocial, evolutionary, and also impor-
tantly genetic processes in explaining the clear
link between early rearing context and the de-
velopmental timing of young people’s transi-
tion to parenthood.
Collectively, these findings suggest that the
transition to early parenthood represents a
highly selective process. Early parenting young
men and women share a diverse range of psy-
chosocial disadvantages encompassing behav-
ioral, intellectual, familial, and socioeconomic
adversity that likely not only conspire to in-
crease their risk of an early transition to par-
enthood, but that would also be expected to
Gender and early parenthood 291
affect their ability to provide an optimal chil-
drearing environment for their developing in-
fant. In addition to the diverse difficulties of
early parenting young men and women, find-
ings also highlight the need to distinguish be-
tween gender differences in prevalence and
underlying etiological processes. Such find-
ings add to our understanding of the effects of
gender on life course development and raise a
number of important questions and issues for
further consideration.
The first issue concerns the mechanisms by
which strong gender differences in rates of
early parenthood emerge. Several possibilities
exist. First, the elevated rates of teen and young
parenthood among females in the cohort may
reflect differences in the onset of sexual de-
velopment and behavior because girls tend to
mature and initiate sexual intercourse earlier
than boys. Second, and related, these gender
differences might also reflect general popula-
tion trends for females to be more likely to
become romantically involved with older and
more sexually active partners compared to their
male peers.
The second issue concerns the mechanisms
that account for the striking similarities in the
social backgrounds of younger parents as well
as the marked overlap in life course processes
that place young males and females at risk of
an early transition to parenthood. One inter-
pretation of these findings is that individuals
from similar family backgrounds and with sim-
ilar cognitive abilities and behavioral tenden-
cies are more likely to meet each other and to
form an attraction than individuals who do not
share a common psychosocial profile ~Kan-
del, Davies, & Baydar, 1990; Krueger, Mof-
fitt, Caspi, Bleske, & Silva, 1998!. Such
assortative pairing between similar individu-
als is likely to reinforce commonalities and
give rise to similar life course outcomes, in-
cluding an earlier onset of parenthood.
If this is the case, it raises important issues
regarding the intergenerational consequences
of these partnerships on both social and ge-
netic grounds. Children born as a conse-
quence of these partnerships are likely to be at
substantial risk, not only because of their par-
ents’ immaturity but also because, due to as-
sortative pairing, both contributing par tners
are more likely to be characterized by com-
mon problematic developmental histories and
genetic tendencies. Early data from the Dune-
din longitudinal study provides some support
for this possibility. Specifically, findings sug-
gest that children born to early parenting sam-
ple members may be at unusually high risk for
antisocial behavior by virtue of having been
born to two parents who are both young and
inexperienced, and who have higher rates of
antisocial behavior than their peers ~Moffitt
et al., 2002!. Follow-up evaluations of these
high risk individuals and their new families
should help to shed light on these processes
and their implications for the children of these
younger parents.
References
Armsden, G. C., & Greenberg, M. T. ~1987!. The in-
ventory of parent and peer attachment: Individual
differences and their relationship to psychological well-
being in adolescence. Journal of Youth and Adoles-
cence, 16, 427–454.
Bardone, A. M., Moffitt, T. E., Caspi, A., Dickson, N., &
Silva, P. ~1996!. Adult mental health and social out-
comes of adolescent girls with depression and con-
duct disorder. Development and Psychopathology, 8,
811–829.
Beautrais, A. ~2002!. Gender issues in youth suicidal be-
haviour. Emergency Medicine, 14, 35 42.
Belsky, J., Steinberg, L., & Draper, P. ~1991!. Childhood
experience, interpersonal development, and reproduc-
tive strategy: An evolutionary theory of socialization.
Child Development, 62, 647–670.
Boddington, B., Khawaja, M., & Didham, R. ~2003!. Teen-
age fertility in New Zealand. Wellington, NZ: Statis-
tics New Zealand.
Brooks-Gunn, J., & Chase-Lansdale, P. L. ~1995!. Ado-
lescent parenthood. In M. Bornstein ~Ed.! , Handbook
on parenting: Vol 3. Who is the parent? ~pp. 113 –
149!. Mahwah, NJ: Erlbaum.
Butler, A. C. ~1992!. The changing economic conse-
quences of teenage childbearing. Social Science Re-
view, March, 1–31.
Chase-Lansdale, P. L., & Brooks-Gunn, J. ~1994!. Corre-
lates of adolescent pregnancy and parenthood. In C. B.
Fish & R. M. Lerner ~Eds.!, Applied developmental
psychology ~pp. 207–236!. New York: McGraw–Hill.
Cloninger, C. R. ~1987!. A