Investigating the Relationship between

Teenage Childbearing and Psychological

Distress Using Longitudinal Evidence*
STEFANIE MOLLBORN
ELIZABETH MORNINGSTAR
University of Colorado at Boulder

Journal of Health and Social Behavior 2009, Vol 50 (September):310–326

The high levels of depression among teenage mothers have received consider-
able research attention in smaller targeted samples, but a large-scale examina-
tion of the complex relationship between adolescent childbearing and psycho-
logical distress that explores bidirectional causality is needed. Using the
National Longitudinal Study of Adolescent Health (Add Health) and the Early
Childhood Longitudinal Study—Birth Cohort, we found that teenage mothers
had higher levels of distress than their childless adolescent peers and adult
mothers, but the experience of teenage childbearing did not appear to be the
cause. Rather, teenage mothers’ distress levels were already higher than their
peers before they became pregnant, and they remained higher after childbear-
ing and into early and middle adulthood. We also found that distress did not in-
crease the likelihood of adolescent childbearing except among poor teenagers.
In this group, experiencing high levels of distress markedly increased the prob-
ability of becoming a teenage mother. Among nonpoor teenage girls, the rela-
tionship between distress and subsequent teenage childbearing was spurious.

Teenage childbearing continues to be framed in this particularly vulnerable group could have
as an important social and public health prob- negative consequences for both mother and
lem in America (Furstenberg 2003). Activists, child. Yet few of the many studies addressing
pundits, and researchers have paid particular this topic have analyzed nationally representa-
attention to the economic and social conse- tive, longitudinal data.
quences of adolescent parenthood (Hoffman To improve our understanding of the effects
1998). There has also been a growing interdis- of teenage childbearing, we used two national
ciplinary body of research on teenage mothers’ longitudinal surveys to assess whether adoles-
mental health. Most of this work is grounded in cent mothers are more distressed than both
the apprehension that high levels of depression their childless teenage peers and older mothers

* This study was supported by a Council on ment is due Ronald R. Rindfuss and Barbara
Research and Creative Work grant at the University Entwisle for assistance in the original design.
of Colorado at Boulder, and by the Department of Persons interested in obtaining data files from Add
Health and Human Services, Office of Public Health should contact Add Health, Carolina
Health Service grant 1 APRPA006015-01-00. This Population Center, 123 W. Franklin Street, Chapel
research uses data from the National Longitudinal Hill, NC 27516-2524 ([email protected]). No di-
Study of Adolescent Health (a.k.a., Add Health), a rect support was received from grant P01-
program project designed by J. Richard Udry, Peter HD31921 for this analysis. The authors thank Jane
S. Bearman, and Kathleen Mullan Harris, and Menken for her helpful comments. Address corre-
funded by grant P01-HD31921 from the Eunice spondence to Stefanie Mollborn, University of
Kennedy Shriver National Institute of Child Health Colorado at Boulder, Institute of Behavioral
and Human Development, with cooperative fund- Science, 483 UCB, Boulder, CO 80309-0483 (e-
ing from 17 other agencies. Special acknowledg- mail: [email protected]).
Downloaded from hsb.sagepub.com at University of Strathclyde on March 20, 2015
310
TEENAGE CHILDBEARING AND DISTRESS 311
(research question 1).1 We also explored throughout their life course (see Settersten
sources of distress in teenage mothers, which 2003). Social norms about the appropriate tim-
have received little direct empirical attention in ing and ordering of transitions are expected to
the past (for important exceptions, see Falci regulate whether such consequences are posi-
and Mortimer 2007; Kalil and Kunz 2002). In tive or negative for an individual (Neugarten,
particular, we asked whether the experience of Moore, and Lowe 1965). However, recent evi-
teenage parenting itself, the selection of dis- dence suggests that more diversity in ordering
tressed girls into motherhood, or a combination and timing life transitions is being tolerated
of the two accounted for subsequently higher now than in the past (Rindfuss, Swicegood, and
levels of distress among adolescent mothers Rosenfeld 1987). Evidence suggests that
(research question 2). This study also investi- teenage childbearing is one “off-time” transi-
gated whether distress increased the likelihood tion that may still be associated with reason-
of subsequent teenage childbearing, an idea ably strong societal norms and negative sanc-
that has received some empirical support but tions for the individuals who violate them
garnered less attention than the opposite causal (Mollborn 2009). These sanctions can be inter-
direction. We examined whether psychological personal, such as facing explicit disapproval
distress itself influenced adolescent childbear- from others, or institutional, such as not re-
ing, or whether any relationship between the ceiving maternity leave from high school.
two was a spurious artifact of underlying fac- Regardless of the sanctions experienced, it
tors such as socioeconomic status, race, or eth- seems possible that teenage parents who vio-
nicity (research question 3). Finally, we looked late societal norms about appropriate transi-
for variation in the relationships between teen tions to parenthood might suffer mental health
childbearing and distress and vice versa across consequences. Just as experiencing discrimina-
racial-ethnic, socioeconomic, and age groups tion based on race or sexual orientation in
(research question 4). Our review of previous everyday life has been linked to poorer mental
research and findings are organized on the ba- health (Mays and Cochran 2001; Williams,
sis of these four research questions. Neighbors, and Jackson 2003), experiencing
Finding preliminary answers to these ques- everyday interpersonal or institutional sanc-
tions using longitudinal, nationally representa- tions may cause psychological distress among
tive data is important because of the conse- teenage parents. Violating social norms has
quences that psychological distress can have been shown to increase negative emotions such
for teenage parents and their families. as embarrassment or shame (Wooten 2006),
Understanding more about the dynamics of the which could have implications for mental
potentially bidirectional relationship between health. Withdrawing social support as a nega-
distress and teenage childbearing would allow tive sanction may also increase distress.
for more effective policies targeting vulnerable While life course theory has anticipated neg-
groups and provide a clearer sense of appro- ative long-term consequences of violating tran-
priate research priorities. For example, if be- sition norms, research has rarely documented
coming a teenage mother increases the proba- these consequences empirically. This lack of
bility of psychological distress, then research substantiation of the theory has led to influen-
and interventions targeted at treating adoles- tial criticisms of the life course perspective
cent parents may improve the outcomes of both (Marini 1984) and a turn away from transition
mother and child. If, instead, being distressed norms toward the idea of an individualized life
makes a teenage girl more likely to have a ba- course in research (Shanahan 2000).
by, then a suitable focus for research and poli- Examining the causal links between teenage
cy would be on identifying and reducing dis- childbearing and distress in this study can pro-
tress among adolescent girls. vide a new opportunity to assess this central
tenet of life course theory by examining
BACKGROUND AND QUESTIONS whether teenage mothers, who may have vio-
Studies of teenage childbearing often draw lated norms about the transition to parenthood,
insight from life course theory (see Elder and suffer short- and/or long-term mental health
Shanahan 2006 for an overview). This theory consequences.
posits that the timing and ordering of life tran- Despite the compelling theoretical justifica-
sitions, such as the transition to parenthood, tion for teenage childbearing being associated
have long-term consequences for individuals with elevated depression and the many empiri-

Downloaded from hsb.sagepub.com at University of Strathclyde on March 20, 2015

312 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR
cal studies supporting it (reviewed below), it is cioeconomic status has been a key task for re-
important to remember that the relationship be- searchers on adolescent parenting; we argue
tween adolescent parenthood and psychologi- that reaching a similar understanding about
cal distress may be bidirectional. Other related mental health should be another.
areas of research have explored bidirectional
relationships in health more thoroughly. For Question 1: Are Teenage Mothers More
example, social scientists have worked to Distressed?
demonstrate the various ways in which social Before trying to disentangle the causal path-
factors can cause disease (e.g., Link and ways between psychological distress and
Phelan 1995), but these researchers and others teenage childbearing, we must first investigate
have also acknowledged that the opposite the extent to which these phenomena are relat-
causal relationship may exist. Sociological re- ed. Maternal distress is an important concern
searchers are often interested in psychological for researchers because it has been linked to
distress (often framed as depression) as both a problematic outcomes for mothers, including
cause and an effect of social phenomena. repeat pregnancies, low parenting competency,
Extensive research has focused on the links be- and poor educational outcomes (Eshbaugh
tween socioeconomic status (SES) and mental 2006; Holub et al. 2007). In addition, maternal
health to examine whether low SES causes de- distress has detrimental effects for children’s
pression, or whether those who are depressed outcomes, including language development,
are more likely to end up with low SES. While coping skills, social adjustment, and physical
both arguments make intuitive sense, the for- health (Hubbs-Tait et al. 1994; Mowbray et al.
mer relationship has received stronger support 2002; Sills et al. 2007; Smith 2004).
than the latter (see Lorant et al. 2003 for a re- The prevalence and severity of adolescent
view). However, Lorant et al. (2003) also rec- mothers’ distress have spawned a considerable
ognize that these two processes frequently research enterprise examining its correlates
work concurrently and are difficult to disen- and effects (e.g., Brown et al. 2007; Eshbaugh
tangle. 2006; Hudson, Elek, and Campbell-Grossman
In the specific realm of research on teenage 2000; Milan et al. 2004; Turner, Sorenson, and
childbearing, acknowledgement of the bidirec- Turner 2000). Teenage mothers’ psychological
tional nature of the relationship between men- distress levels have been found to be high in
tal health and adolescent motherhood has been samples of varying size and generalizability
slower to blossom. Instead, studies of socio- (e.g., Schmidt et al. 2006), sometimes with
economic consequences of teenage childbear- more than half of teenage mothers meeting the
ing have led the way on this issue. Selection criteria for clinical depression (Reis 1988).
bias has been a persistent concern in research Past research on teenage mothers and dis-
on the socioeconomic effects of teenage par- tress has frequently presented information spe-
enthood because many of the perceived prob- cific to adolescent mothers without introduc-
lematic “consequences” of adolescent child- ing comparisons with other groups such as
bearing are largely a function of problems that adult moms or nonpregnant teens (e.g.,
plagued these teens before they had children, Eshbaugh 2006; Schmidt et al. 2006).
increasing their likelihood of becoming young However, some research has introduced a com-
parents. Studies of teenage parents’ socioeco- parison group of older mothers (Deal and Holt
nomic outcomes are a prominent example for 1998). Other studies have compared pregnant
which research has worked to disentangle se- teens to nonpregnant peers from similar back-
lection effects from teenage childbearing ef- grounds and have found no significant differ-
fects (e.g., Geronimus and Korenman 1992; ences in depression (e.g., Milan et al. 2004;
Grogger and Bronars 1993; Hotz, McElroy, Troutman and Cutrona 1990). Research com-
and Sanders 2005). While the degree of selec- paring teenage mothers to peers from a variety
tion is debated, research has suggested that of social locations is rare, with inconsistent re-
much or most of the difference between sults. Vicary and Corneal (2001) found no dif-
teenage parents’ and childless adolescents’ so- ferences in depression between teenage moth-
cioeconomic outcomes is the result of selection ers and older mothers, while Falci and
rather than childbearing (see Hoffman 1998 for Mortimer (2007) found that teenage mothers
a review). Understanding how much the expe- had higher levels of depression than other
rience of adolescent childbearing affects so- young women who had children later if at all.

Downloaded from hsb.sagepub.com at University of Strathclyde on March 20, 2015

TEENAGE CHILDBEARING AND DISTRESS 313
Kalil and Kunz (2002) found that teenage distressed girls into adolescent motherhood ac-
mothers were more depressed than adult moth- counted for most of their higher distress levels
ers and nonmothers, but these differences dis- compared to peers. Our study compares
appeared with background factors and marital teenage mothers with both childless adoles-
status controlled. Our study estimates psycho- cents and a “natural” comparison group of fu-
logical distress levels among teenage mothers ture adolescent mothers to explore whether
compared to both older mothers and childless teenage mothers’ elevated distress levels are a
peers. result of experiencing early childbearing, or
whether they result from the selection of dis-
Question 2: Does Teenage Childbearing tressed girls into adolescent childbearing.
Lead to Distress?
Most research measuring the effect of ado- Question 3: Does Distress Lead to Teenage
lescent childbearing on depression has relied Childbearing?
on relatively small and/or nonrepresentative Several studies have suggested the opposite
samples (e.g., Falci and Mortimer 2007; causal direction, that pre-existing depression
Hudson et al. 2000; Milan et al. 2004; Vicary could increase the likelihood of women be-
and Corneal 2001). Some research that has coming teen parents (Deal and Holt 1998;
sought to improve generalizability has been Eshbaugh, Lempers, and Luze 2006; Milan et
limited by an inability to measure depression al. 2004; Schmidt et al. 2006). Few studies
prior to pregnancy (e.g., Deal and Holt 1998; have examined the causal relationship between
Eshbaugh 2006; Schmidt et al. 2006), which adolescent parenthood and depression empiri-
hinders such studies from establishing causali- cally, but existing research suggests that there
ty between adolescent childbearing and de- may be a link. Kessler et al. (1997) identified
pression. Milan et al. (2004) included compar- increased risk for teenage childbearing among
isons between pregnant teens and their peers, those with early onset psychiatric disorders us-
but data collection started in the third trimester ing retrospective data. Adolescent girls who
of pregnancy. One exception is Booth, had ever met the DSM-III-R assessment for af-
Rustenbach, and McHale (2008), which used fective disorders were two times more likely
the Add Health survey to analyze change in subsequently to become teenage mothers than
depressive symptoms before and after early their peers. Miller-Johnson et al. (1999) found
family transitions. However, this study exam- that parental reports of adolescents’ depression
ined becoming a parent before the age of about predicted the timing of parenthood, but adoles-
23 rather than teenage childbearing. They cents’ reports were not significant. In contrast,
found that young parents’ changes in depres- Yamaguchi and Kandel’s (1987) retrospective
sive symptoms did not differ from those of oth- study of premarital pregnancy reported that de-
ers but did not examine causal linkages be- pressive symptomatology did not affect pre-
tween teenage childbearing and depressive marital pregnancy, but it lowered adolescents’
symptoms. likelihood of having an abortion.
Despite a paucity of data concerning women’s It is also possible that distress itself does not
mental health status prior to pregnancy, the affect the subsequent likelihood of becoming a
general trend in the literature has been to as- teenage mother, but rather that underlying fac-
sume this causal pathway but to mention the tors that increase teenagers’ chances of both
lack of established causality between teenage becoming distressed and bearing children cre-
childbearing and distress as a limitation. ate a spurious relationship. For example, high-
Despite the intuitive appeal of the idea that er depression levels have been reported among
teenage childbearing causes distress, available African Americans, Hispanics, and mothers
evidence suggests that much of this relation- with lower socioeconomic status (Belle and
ship is instead an artifact of the selection of Doucet 2003; Howell et al. 2005). These so-
distressed girls into teenage motherhood. In ciodemographic characteristics are also linked
other words, experiencing adolescent child- to higher adolescent pregnancy and childbear-
bearing does not make teenage girls more dis- ing rates (Furstenberg 2003). Similarly, previ-
tressed, but rather teenage girls who were al- ous sexual experience in adolescence has been
ready distressed more often become teenage linked both to depression, at least for some
mothers. Falci and Mortimer (2007) and Vicary groups (Hallfors et al. 2005; Meier 2007), and
and Corneal (2001) found that the selection of to teenage pregnancy (Kaufmann et al. 1998).

Downloaded from hsb.sagepub.com at University of Strathclyde on March 20, 2015

314 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR
Any or all of these background factors may ac- parent (typically the mother) was also asked to
count for an observed relationship between complete an interview. The Add Health sample
teens’ distress and subsequent childbearing. was representative of U.S. schools with respect
We estimate the association between these to urbanicity, region of country, school type,
variables, as well as the degree to which their ethnicity, and school size. Some student popu-
relationship is spurious. lations were oversampled, and dropouts were
not interviewed, but probability weights al-
Question 4: Do Relationships between lowed researchers to represent the national
Distress and Teen Childbearing Vary across population of adolescents in grades 7 to 12.
Groups? Response rates for the three waves ranged from
It is not only possible that background fac- 77 percent to 88 percent.
tors such as race, ethnicity, and socioeconomic For analyses of psychological distress, 5,429
status account for the relationship between dis- female respondents who participated in waves
tress and teenage childbearing and vice versa, two and three and were not missing weight, re-
but also that sociodemographic groups may ex- gion, or school identifier were eligible for
perience these associations differently. For ex- analysis. Approximately 3 percent of these cas-
ample, Mollborn (2009) found that African es were deleted listwise because of missing in-
Americans and people with lower education formation. For analyses of teenage childbear-
levels perceived weaker norms against teenage ing, 7,163 female respondents who participat-
pregnancy than whites and those with more ed- ed in waves two and three and who were not
ucation. According to life course theory, indi- missing weight, region, or school identifier
viduals who face weaker norms should suffer were eligible for inclusion. Just under 6 percent
less severe consequences for violating them. of these cases were deleted listwise because of
This implies that there could be a stronger link missing information on independent variables.
between teenage childbearing and subsequent This analysis sample was larger than the for-
depression for white adolescents and those mer one, mostly because respondents who
from high-SES families than for black or low- were high school seniors at wave one were not
SES adolescents. Similarly, younger teenage included at wave two, but were re-interviewed
parents may violate stronger age norms against at wave three.
early childbearing, so younger teenage parents The secondary data source for this study was
may be more distressed than those who are old- the Early Childhood Longitudinal Study—
er. We focus on race-ethnicity, socioeconomic Birth Cohort (ECLS-B) (National Center for
status, and age when investigating whether the Education Statistics 2006), sponsored and con-
effect of teen childbearing on distress and the ducted by the U.S. Department of Education’s
effect of distress on teen childbearing differ National Center for Education Statistics. This
across groups. survey collected detailed data on both children
and parents and is representative of the ap-
METHOD proximately 4 million children born in the
United States in 2001. We used information
Data from the first wave of data, collected when the
This study used data from two national sur- infants were about nine months old, to examine
veys. The primary data source addressing each teenage mothers’ distress levels compared to
of the research questions was the National those of former teenage mothers and mothers
Longitudinal Study of Adolescent Health (Add who never gave birth as adolescents. This pro-
Health), a nationally representative survey of vided two different comparison groups that the
students that began in the mid-1990s Add Health did not (adult mothers and former
(Bearman, Jones, and Udry 1997). Investi- teenage mothers giving birth as adults), as well
gators sampled 80 U.S. high schools and 52 as allowing us to map teenage mothers’ distress
middle schools. More than two-thirds of levels into middle adulthood, contributing im-
schools agreed to participate in the study, and portant answers to research question 1. The pri-
those who refused were replaced with schools mary parent (overwhelmingly the mother)
from the same community. For each school, a completed an in-home interview and a self-ad-
subsample of students completed an extensive ministered questionnaire about sensitive topics
in-home interview at wave one that was fol- such as psychological distress. Some analyses
lowed up one and six years later. The primary in this study compared mothers under age 20

Downloaded from hsb.sagepub.com at University of Strathclyde on March 20, 2015

TEENAGE CHILDBEARING AND DISTRESS 315
(N ~ 1,050) to older mothers (N ~ 8,300), and lescent moms-to-be reported high distress lev-
others split the latter category into those who els as others (p < .01).
were teenagers at first birth (N ~ 1,700) and Teenage childbearing. Analyses included in-
those who were age 20 or older (N ~ 6,600).2 dicators at different time points of teenage
The psychological distress scale used in childbearing, defined as a respondent having a
ECLS-B is similar to the Add Health version. child before the month of her twentieth birth-
The variables described below are all from our day. At wave three, respondents reported
primary data set, Add Health. whether pregnancies had occurred within any
ongoing or new romantic and sexual relation-
Variables ships they had experienced since 1995 (wave
Psychological distress. The measure of psy- one). The month and year of each pregnancy’s
chological distress from both data sets used a end were used to calculate the mother’s age at
subset of questions from the Center for her first live birth. We compared the month and
Epidemiologic Studies Depression Scale year of this birth to the month and year of each
(CES-D; Radloff 1977). Both scales asked re- interview to calculate whether respondents be-
came teenage mothers before or after waves
spondents to report the frequency of experi-
one and two. Booth et al. (2008) pointed out in-
encing specific symptoms in the last week,
consistencies between reports of biological
ranging from “never or rarely” to “most or all
children in the roster of household members
of the time.” Item responses are typically
and reports of past births within the context of
summed. However, to create a consistent mea-
relationships. To address this issue, we identi-
sure that could be used across scales with dif-
fied cases where the respondent reported that
ferent numbers of items, we instead calculated
her biological child was living with her at wave
the mean of the response items, ranging from 0 three. If the oldest biological child’s age sub-
to 3. Waves one and two of the Add Health sur- tracted from the respondent’s age at wave three
vey relied on a 19-item modified scale.3 Wave equaled 19 or less, we considered the respon-
three of the Add Health data used a nine-item dent a teenage mother. Including information
version of the scale. ECLS-B used a 12-item from the household roster increased the count
version that focused on the negative experi- of teenage mothers by about 10 percent in
ences of distress. For the Add Health data, Table 1. When we were interested in those who
Cronbach’s alpha for the scale was .91 in wave became pregnant after a certain wave, we ex-
one, .89 in wave two, and .81 in wave three; for cluded respondents who gave birth 10 months
ECLS-B it was .87. or less after the month of interview (for
In some analyses, we also relied on cut- teenage mothers identified in the household
points to distinguish between discrete levels of roster, the date of birth was an estimate in
psychological distress. Following the approach whole years). There were 808 respondents, or
outlined by other researchers working with about 13 percent, who were teenage mothers
Add Health data, we relied on a cut point of 23 who got pregnant after wave one but before
(for the summed scale, which corresponds to wave three, when respondents were 18 to 27
1.21 in our mean scale) to designate high lev- years old. The nonmothers included those who
els of distress for girls (Lehrer et al. 2006; had experienced teenage pregnancies that did
Shrier et al. 2001). Those who scored below 23 not end in live births because of abortion, mis-
but at least 11 (.58 in the mean scale) were carriage, or stillbirth, and girls who never ex-
classified as reporting moderate distress, and perienced a teenage pregnancy.4
those who were below 11 were classified as Control variables. Independent variables, all
low-distress. Table 1 reports means for these measured at wave one, were chosen because of
and other variables, as well as differences be- possible relationships to teenage childbearing
tween those who would become teenage moth- or distress. Family socioeconomic status was
ers by wave three and others. Future teenage measured as parents’ education in years (M =
mothers reported a mean distress level of .73 at 13.25) and parent-reported household income.5
wave one (equivalent to moderate distress), Following Cubbin et al. (2005), we created
compared to .58 for others (p < .01). Just 44 dummy variables representing income as a per-
percent of future teenage mothers reported low centage of 1994 federal poverty thresholds that
levels of distress compared to 60 percent of controlled for the number of people in the
others (p < .01), and almost twice as many ado- household (0 to 100%, 101 to 200, 201 to 300,

Downloaded from hsb.sagepub.com at University of Strathclyde on March 20, 2015

316 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR
TABLE 1. Weighted Means for Variables Used in Analyzing Distress and Teenage Motherhood
Not Teen Mom by Teen Mom
All Respondents Wave 3 Wave 1–3
Variables Measured at Wave 1 (N = 6,391) (N = 5,583) (N = 808)
Teen mom between Waves 1–3 (1 = yes) 00.13 (.01)
Psychological distress scale (0–3) 00.60 (.01) 00.58 (.01) 00.73** (.02)
Psychological distress categories
—Low 00.58 (.01) 00.60 (.01) 00.44** (.02)
—Moderate 00.34 (.01) 00.33 (.01) 00.43** (.02)
—High 00.08 (.00) 00.07 (.00) 00.13** (.01)
Parents’ mean education (years) 13.25 (.12) 13.41 (.13) 12.12** (.10)
Household poverty status (% FPL)
—000–100a 00.13 (.01) 00.11 (.01) 00.23** (.02)
—101–200 00.17 (.01) 00.16 (.01) 00.24** (.03)
—201–300 00.18 (.01) 00.18 (.01) 00.16** (.02)
—301–400 00.12 (.01) 00.13 (.01) 00.05** (.01)
—> 400 00.19 (.01) 00.21 (.01) 00.08** (.02)
—Missing information 00.21 (.01) 00.20 (.01) 00.24** (.02)
Respondent’s race-ethnicity
—Non-Latino Whitea 00.69 (.03) 00.71 (.03) 00.55** (.04)
—Non-Latino Black 00.15 (.02) 00.13 (.02) 00.25** (.04)
—Latino 00.11 (.02) 00.10 (.02) 00.17** (.02)
—Asian/Pacific Islander 00.04 (.01) 00.04 (.01) 00.02** (.01)
—Native American 00.01 (.00) 00.01 (.00) 00.01** (.00)
—Other race 00.01 (.00) 00.01 (.00) 00.01** (.00)
Grade point average
—1–3 00.30 (.01) 00.28 (.01) 00.42** (.02)
—3–3.49 00.23 (.01) 00.24 (.01) 00.17** (.02)
—3.5–4a 00.47 (.01) 00.48 (.01) 00.41** (.03)
—Missing information 00.14 (.01) 00.12 (.01) 00.27** (.02)
School attachment scale (1–5) 04.64 (.03) 04.67 (.03) 04.42** (.06)
Age (years) 15.73 (.12) 15.80 (.12) 15.22** (.14)
Family structure
—2 biological parents a 00.59 (.01) 00.62 (.01) 00.38** (.02)
—2 parents (other types) 00.17 (.01) 00.16 (.01) 00.25** (.02)
—Single mother 00.19 (.01) 00.18 (.01) 00.29** (.02)
—Single father 00.02 (.00) 00.02 (.00) 00.03** (.01)
—Other family structures 00.03 (.00) 00.02 (.00) 00.05** (.01)
Church attendance
—Never/no religion 00.21 (.01) 00.21 (.01) 00.22** (.02)
—Less than once a month 00.18 (.01) 00.18 (.01) 00.19** (.02)
—≥ Once/month < once/week 00.19 (.01) 00.19 (.01) 00.23** (.02)
—At least once a weeka 00.42 (.01) 00.43 (.01) 00.36** (.02)
Born in the United States (1 = yes) 00.94 (.01) 00.94 (.01) 00.96** (.01)
Satisfaction with parent relationship (1– 5) 04.16 (.02) 04.19 (.02) 04.00** (.05)
History of vaginal intercourse (1 = yes) 00.32 (.02) 00.29 (.02) 00.48** (.03)
In a romantic relationship (1 = yes) 00.37 (.01) 00.36 (.01) 00.45** (.03)
Source: National Longitudinal Study of Adolescent Health (1995)
Notes: a Reference category. Numbers in parentheses are standard errors for weighted means.
Weighted means account for sample design effects (stratification and clustering).
* p < .05; ** p < .01; adjusted Wald tests comparing means for teen mothers between Waves 1–3 to other female re-
spondents.

301 to 400, and > 400), as well as an indicator er = 1) and recoded into a series of indicator
for missing income information. Teen respon- variables (1 to 3, 3 to 3.49, and 3.5 to 4), with
dents reported their race-ethnicity (Latino and an additional indicator for missing GPA infor-
non-Latino white, African American, Asian/ mation. A school attachment scale averaged
Pacific Islander, Native American/American
five items about feeling close to people at one’s
Indian, and “other” race).
Two variables measured school-related fac- school, feeling part of one’s school, feeling
tors. Student-reported grade point average happy to be at school, feeling that teachers treat
(GPA) included four subjects, averaged into a students fairly, and feeling safe at school. The
four-point scale (A = 4, B = 3, C = 2, D or low- scale ranged from 1 (weak attachment) to 5

Downloaded from hsb.sagepub.com at University of Strathclyde on March 20, 2015

TEENAGE CHILDBEARING AND DISTRESS 317
(strong attachment), with a Cronbach’s alpha graphic groups. All analyses used Stata’s com-
of .84. plex survey design commands.
Respondents’ age in months and years at
wave one was recoded as a fraction of years. RESULTS
Respondents’ family structure was coded into a Question 1: Are Teenage Mothers More
series of dummy variables after Harris (1999): Distressed?
living with both biological parents, other types
of two-parent families, a single mother, a sin- Comparing teenage mothers’ distress to oth-
gle father, and other family structures. ers’ distress. Bivariate analyses from two data
Religious service attendance was represented sets addressed our first research question. We
by variables indicating frequency, with a com- first examined whether teenage mothers were
bined group of “never” or “no religion” as the more distressed than their childless teenage
reference. Respondents’ status as being born in peers (results not shown in tables). Echoing
the United States was reported by the parent (if past research, evidence from the Add Health
available), and, if not reported by the parent, survey suggested that they were. The mean dis-
those data were obtained from adolescent in- tress level at wave two of teenagers who were
terviews. Respondents reported satisfaction not teenage mothers by wave three was .59,
with their parent-child relationships through compared to .77 for their peers who had be-
their agreement with the statement, “Overall, come adolescent mothers by wave two (p <
you are satisfied with your relationship with .01). This difference represents 7 percent of the
your mother,” ranging from 1 to 5 (with higher observed range of psychological distress
numbers indicating more agreement). When scores. Grouping distress levels into three cat-
the primary parent was the father, or if a re- egories instead, we see that only 43 percent of
sponse about the mother was missing, we sub- teenage mothers by wave two were classified
stituted responses to the same question about as having low levels of distress at wave two,
the father-child relationship. Finally, respon- compared to 60 percent of their childless peers
dents’ reports of having ever had penile-vagi- (p < .01). In contrast, teen mothers were se-
nal intercourse by wave one were included. verely distressed more than twice as often as
their childless peers (p < .05).
Analysis Plan We then asked whether teenage mothers are
First, we used several comparison groups more distressed than older mothers (see Figure
across both data sources to present important 1). Analyses from the ECLS-B that used a
information about differences between teenage modified version of the CES-D to measure dis-
mothers’ distress levels and those of other tress revealed similar comparisons between
women, as well as trends in their distress lev- teenage mothers and older mothers to those be-
els from before pregnancy through middle tween teenage mothers and their childless
adulthood. Add Health data covered prepreg- peers. Teenage mothers’ mean distress score at
nancy through young adulthood, and ECLS-B about nine months postpartum was .56, com-
covered the postpartum period through middle pared to .38 among mothers who had their first
adulthood. Subsequent analyses focused only birth at ages 20 and older (p < .01). This dif-
on Add Health. Multivariate analyses control- ference corresponds to 6 percent of the ob-
ling for age compared girls who became served range of distress scores in the sample.
teenage mothers before and after wave two (be- Assessing differences over time in teenage
cause there were too few teenage mothers by mothers’ distress. Do adolescent mothers’ ele-
wave one) to examine the association between vated levels of distress compared to childless
experiencing teenage childbearing and subse- peers and older mothers continue into adult-
quent psychological distress. Next, we estimat- hood? Wave three distress levels from Add
ed logistic regression models of the relation- Health measured five years after wave two (not
ship between distress at wave one and subse- shown in tables) revealed that although all
quent teenage childbearing by wave three, al- groups became less distressed as they aged,
lowing for a larger sample of future teenage teenagers who became mothers both before
mothers than examining wave two to three and after wave two still had higher continuous-
would. Finally, we introduced interactions to ly measured distress scores in young adulthood
multivariate models to investigate if the rela- than their peers who did not become adolescent
tionships we found differed across sociodemo- mothers (p < .01).

Downloaded from hsb.sagepub.com at University of Strathclyde on March 20, 2015

318 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR
FIGURE 1. Mean Psychological Distress at about 9 Months Postpartum, by Teen Childbearing
Status

Source: Early Childhood Longitudinal Study-Birth Cohort (2001)

Notes: “Current teen mom” was ≤ 19 years old at birth of study’s focal child, “prior teen mom” birth was > 19 years
old at birth of focal child but ≤ 19 years old at first birth, and “never teen mom” was > 19 years old at first birth

Data from ECLS-B assessed teenage moth- Question 2: Does Teenage Childbearing
ers’ levels of distress well into adulthood (see Lead to Distress?
Figure 1). We identified about 1,700 new Past research has shown that comparing
mothers who were at least age 20 at the study’s teenage parents to their childless peers and old-
focal birth but who first gave birth as er parents as we did above frequently results in
teenagers. While distress levels among moth- selection bias, i.e., teenagers who bear children
ers who were age 20 or older at first birth de- are very different on average than those who do
creased steadily with age from .48 at ages not even before they get pregnant. Because of
20–24 to .34 at age 35–39, postpartum distress selection, observed differences in distress lev-
levels at the same ages among respondents who els between teenage mothers and childless
peers or older mothers could be the result of
were once teenage mothers actually increased
underlying factors rather than a result of the
from .52 to .57. Former teenage mothers had experience of teenage motherhood itself. For
significantly higher distress levels than adult this reason, comparing teenage mothers with a
first-time mothers for all age categories from “control” group of equivalent respondents
20 to 39 (p < .01). When we compared average would most accurately estimate the association
distress levels for current versus all former between the experience of teenage motherhood
teenage mothers, they did not differ signifi- and distress. The longitudinal design of the
cantly (.56 and .51, respectively). This finding Add Health survey allows for a natural com-
provides further evidence that adolescent parison group of future teenage mothers, which
childbearing was not associated with a spike in can help establish the degree to which selection
distress levels, but rather teenage mothers’ ele- into adolescent childbearing accounts for ob-
vated psychological distress was long-term. served differences in distress between teenage
mothers and childless teenagers. Regression
In sum, evidence showed that teenage moth-
analyses reported in Table 2 compared the
ers were more distressed on average than both
wave two distress levels of girls who were
their childless teenage peers and older mothers. teenage mothers by wave two (N = 239) with
Teenage mothers’ higher distress levels contin- those who were not yet pregnant by wave two
ued throughout adolescence, young adulthood, (they gave birth at least 10 months after the
and middle adulthood, suggesting that their month of interview) but who would become
distress may be long-term rather than short- teenage mothers by wave three, five years lat-
term in nature. er (N = 558). In other words, the former group

Downloaded from hsb.sagepub.com at University of Strathclyde on March 20, 2015

TEENAGE CHILDBEARING AND DISTRESS 319
TABLE 2. Linear Regression of Distress on Teen Childbearing and Age (N = 5,261)
Variables Wave 2 Distress Wave 3 Distress b
Teen childbearing status
—Teen mom by Wave 2 (reference)
—Teen mom pregnant after Wave 2 a .02 (.05) –.06 (.05)
—Never teen mom by Wave 3 a –.14** (.04) –.15** (.04)
Wave 2 age (years) .03** (.01)
Wave 3 age (years) –.01 (.01)
Constant .28** (.09) .92** (.15)
Design-based F test (df) 25.17** (3,128) 6.24** (3,128)
R-squared .03 .01
Source: National Longitudinal Study of Adolescent Health (1995)
Notes: a Pregnant after Wave 2 and Never teen mom differ significantly in post hoc tests.
b
Wave 3 was conducted 5 years after Wave 2, with Wave 3 ages ranging from 18 to 27.
Analyses account for sample design effects (weighting, stratification, and clustering).
* p < .05; ** p < .01; two-tailed tests. Standard errors in parentheses.

had already experienced teenage motherhood in Table 3 addressed this question using a bi-
when psychological distress was measured. variate logistic regression analysis that esti-
The second group had not, but was assumed to mated the association between distress scale
be otherwise similar to the first group with the scores at wave one and the likelihood of subse-
exception of being older. We controlled for age quently becoming a teenage mother sometime
in the regression models. Table 2 shows that in the following six years (pregnant at least 10
there were no significantly different wave two months after the month of the wave one inter-
or three distress levels between these groups, view and giving birth by wave three and before
suggesting that experiencing teenage mother- age 20). There was a very strong association
hood did not increase distress. between these variables: Each one unit increase
If the experience of teenage childbearing in the psychological distress scale measured at
was not related to distress, then why were wave one more than doubled a girl’s odds of
teenage mothers more distressed than their becoming a teenage mother by wave three (OR
childless peers? We suspect that underlying = 2.08, p < .01).
factors may have caused these girls to be dis- In sum, using natural comparison groups to
tressed before they became teenage mothers. establish causality more firmly than in past re-
Analyses reported in Table 2 addressed this search, our findings suggest that teenage moth-
question by comparing girls who became preg- ers’ higher distress scores were the result of
nant after wave two and who gave birth as distress prior to childbearing rather than the ex-
teenagers to those who were not teenage moth- perience of teenage motherhood itself. In the
ers by wave three. Wave two distress scores al- opposite causal direction, distress was posi-
low us to assess the former group’s mental tively related to subsequent teenage childbear-
health before they got pregnant. Post hoc tests ing; multivariate analyses reported below ex-
based on Table 2 show that girls who would plored potential causes of this association.
eventually become teenage mothers were sig-
nificantly more distressed at wave two than Question 3: Does Distress Lead to Teenage
their childless peers (p < .01). The difference Childbearing?
between these two groups’ distress levels re- Bivariate evidence suggested that psycho-
mained similar and significant (p < .01) at logical distress was associated with subsequent
wave three, after teen childbearing had oc- teenage childbearing. But was this relationship
curred for the former group. causal, or did underlying factors affecting both
This evidence suggests that teenage mothers distress and teenage motherhood explain the
may have been more distressed than their relationship between the two? Model 2 in Table
childless peers because they were already dis- 3 added controls to the distress measure to es-
tressed before becoming pregnant rather than timate the likelihood of teenage childbearing.
the experience of adolescent childbearing in- The psychological distress coefficient dropped
creasing distress. Does it follow that distress from .73 (OR = 2.08) in model 1 to .07 (OR =
among adolescents was associated with higher 1.07) in model 2 and lost statistical signifi-
odds of becoming a teenage mother? Model 1 cance; thus, background factors accounted for

Downloaded from hsb.sagepub.com at University of Strathclyde on March 20, 2015

320 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR
TABLE 3. Logistic Regression Coefficients Predicting Teen Childbearing Using Distress and
Controls (N = 6,737)
Wave 1 Variables Model 1 Model 2 Model 3
Distress scale (0–3) .73** (.09) .07 (.13)
—Moderate distressa .11 (.23)
—High distressa –.66 (.34)
Parents’ education (years) –.15** (.03) –.15** (.03)
Household poverty status (% FPL)b
—0–100 .60* (.26) .43 (.35)
—101–200 .65* (.26) .61* (.31)
—201–300 .40 (.23) .37 (.30)
—301–400 –.26 (.30) –.65 (.40)
—Missing income .62** (.24) .64** (.23)
Race-ethnicityc
—Non-Hispanic Black .33* (.17) .35* (.17)
—Hispanic .32* (.15) .34* (.15)
—American Indian .35 (.61) .39 (.57)
—Asian/Pacific Islander –.02 (.33) –.00 (.33)
—Other race .47 (.60) .42 (.61)
Grade point averaged
—1–3 .79** (.17) .78** (.17)
—3–3.49 .27 (.20) .27 (.20)
—Missing GPA 1.10** (.20) 1.10** (.21)
School attachment scale (1–5) –.06 (.06) –.06 (.06)
Age at Wave I (years) –.45** (.04) –.45** (.05)
—Family structuree
—2 parents (other types) .71** (.14) .71** (.14)
—Single mother .46** (.12) .44** (.13)
—Single father .36 (.37) .36 (.36)
—Other family structure .69** (.23) .68** (.23)
Church attendancef
—Never/no religion –.08 (.16) –.08 (.16)
—< Once/month .12 (.15) .11 (.15)
—≥ Once/month and < once/week .27 (.14) .28* (.14)
U.S.-born (1 = yes) .47 (.28) .47 (.29)
Satisfaction with parent relationship –.14* (.06) –.15* (.06)
Had vaginal intercourse (1 = yes) .94** (.13) .94** (.13)
In a romantic relationship (1 = yes) .35** (.12) .36** (.13)
Distress ⫻ poverty interactionsg
—Moderate ⫻ 0–100% FPL –.00 (.38)
—High ⫻ 0–100% FPL 1.44** (.53)
Constant –2.47** (.11) 5.26** (1.05) 5.42** (1.07)
F-test for model fit (df) 65.84** (1,130) 14.86** (28,103) 12.31** (37,91)
F-test compared to previous model .— 10.38** .—
Source: National Longitudinal Study of Adolescent Health (1995)
Notes: Reference categories: a Low distress; b > 400% FPL; c Non-Hispanic White ; d 3.5–4; e 2 biological parents;
f
Never/no Religion; g Interactions of distress with other poverty categories were not significant and are excluded here.
Analyses account for sample design effects (weighting, stratification, and clustering).
* p < .05; ** p < .01; two-tailed tests. Standard errors in parentheses.

the bivariate relationship between distress and Supplemental analyses (not shown) revealed
subsequent teenage motherhood. Significant that just four control variables in combination
wave one risk factors for becoming a teenage eliminated the significance of the distress co-
mother by wave three included low parental ed- efficient: parental education, grade point aver-
ucation; low income; being African American age, family structure, and having had sexual in-
or Hispanic; low grades; younger age; single- tercourse. At least one category of each of
mother, stepparent, or “other” family struc- these measures also significantly predicted dis-
tures compared to two biological parents; less tress at wave one. We conclude that the rela-
satisfying parent-teen relationships; having tionship between psychological distress and
had penile-vaginal intercourse by wave one; subsequent teenage motherhood was spurious
and being in a romantic relationship at wave in this analysis, based on the findings that these
one. four background factors eliminated the signif-

Downloaded from hsb.sagepub.com at University of Strathclyde on March 20, 2015

TEENAGE CHILDBEARING AND DISTRESS 321
icance of psychological distress, and that they None of the race-ethnicity interactions was
each significantly predicted distress. In other significant, but model 3 in Table 3 revealed a
words, underlying measures of socioeconomic large and significant interaction between high
status, academic achievement, family struc- levels of distress and having a household in-
ture, and sexual experience influenced both come at or below the poverty line (OR = 4.22,
teens’ psychological distress and their likeli- p < .01). It is interesting that the main effects
hood of becoming adolescent mothers, ac- of distress and poverty in combination with the
counting for what initially appeared to be a interaction showed that high distress did not
strong relationship between teenage mother- raise the odds of teen childbearing for wealthy
hood and distress. respondents, nor did poverty increase the like-
lihood of teen childbearing for respondents ex-
Question 4: Do Relationships between periencing low levels of distress. Rather, the
Distress and Teen Childbearing Vary across combination of poverty and psychological dis-
Groups? tress was predictive of teenage childbearing,
We found that, on average, the experience of and girls from poor families were the only
teenage childbearing did not increase girls’ group we tested for whom psychological dis-
psychological distress. Does this finding vary tress was associated with the likelihood of sub-
by race-ethnicity, age, or socioeconomic sta- sequently becoming an adolescent mother.
tus?6 To address this question, in supplemental To interpret this interaction, supplemental
analyses we first added interactions between analyses relied on predicted probabilities of be-
teenage childbearing indicators and each of coming a teenage mother for a “typical” hypo-
these factors to model 1 of Table 2. One mod- thetical respondent who had average values for
el included interactions of teenage childbear- all variables in model 3 of Table 3, except for
ing with African American and with Latina those being manipulated: distress level and
(reference category was white), another in- poverty level. A poor hypothetical respondent
cluded interactions with ages younger than 16 (wave one household income ≤ 100% of the
and ages 18 and older (reference category was federal poverty threshold, adjusted for house-
16 to 17 years old), and a third included inter- hold size) was compared to a high income one
actions with each of the poverty indicators (income > 400% of the poverty line) for each
from Table 1. None of the interactions between of the three categories of wave one psycholog-
being a teenage mother and these demograph- ical distress. At a low level of distress, high-in-
ic categories was significant, suggesting that come and poor respondents were both relative-
teenage childbearing did not increase psycho- ly unlikely to become teenage mothers (a 2%
logical distress on average for any of these so- chance for the poor respondent, compared to a
ciodemographic categories. 1% chance for the rich respondent). The prob-
Next, we investigated whether the relation- ability for both income categories at moderate
ship between psychological distress and teen levels of distress was 2 percent. At high dis-
childbearing was spurious for all groups. Here, tress levels, however, the high-income respon-
we excluded interactions of distress with age dent had a 1 percent chance of teenage child-
because of the Add Health study design, in bearing, while the poor respondent’s likelihood
which respondents’ staggered ages at wave one rose to 5 percent. For poor respondents, then,
led to widely varying lengths of time in which socioeconomic status, race-ethnicity, family
they could have become teenage mothers be- structure, sexual activity, and other factors did
tween wave one and wave three. This prohibit- not account for the relationship between dis-
ed attempts to disentangle actual age effects tress and subsequently becoming a teenage
from associations that were artifacts of the mother; high levels of psychological distress
study design. In order to allow for the possibil- still influenced their childbearing behaviors.
ity of nonlinear relationships, we used the
three-category measure of distress in our inter- DISCUSSION
action models rather than the continuous mea- Using two nationally representative, longitu-
sure. Models were identical to model 2 in Table dinal data sets, we found that teenage mothers
3, except for the categorical distress measure were more distressed than both childless ado-
and the addition of terms for interactions be- lescent peers and adult mothers. Contrary to
tween distress and sociodemographic vari- common assumptions, however, their increased
ables. psychological distress did not appear to be

Downloaded from hsb.sagepub.com at University of Strathclyde on March 20, 2015

322 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR
caused by experiencing teenage childbearing. assumption that becoming a teenage mother
Rather, teenage mothers’ distress levels were creates a high risk of distress, which is similar
already higher than their peers’ before they be- to the largely debunked supposition that ado-
came pregnant, and they remained more dis- lescent childbearing causes severely compro-
tressed than others after childbearing, and into mised educational outcomes, was not support-
early and middle adulthood. This is similar to ed in this study. Rather, many teenage mothers
Booth et al.’s (2008) finding that trajectories of experienced long-term psychological distress
depressive symptoms for those who experi- both before and after having a child, similar to
enced early childbearing did not widen or nar- (and potentially even partially caused by or
row compared to peers who had not formed causing) their typically worse educational per-
families. formance both before and after childbearing.
Just as we found that teenage motherhood Social disadvantage underlies both teenage
did not increase distress, our results also childbearing and distress, just as it has been
showed that distress did not raise the likelihood shown to underlie both teenage childbearing
of subsequent adolescent childbearing except and educational outcomes (Ribar 1994).
among teenage girls from poor families. In this Understanding the dynamics of the relation-
group, high distress levels markedly increased ship between teenage motherhood and distress
the probability of becoming a teenage mother. is important, and our study contributes to this
Among the other groups we tested, the rela- body of knowledge.
tionship between distress and subsequent These findings also speak to the broader lit-
teenage childbearing was spurious and caused eratures on selection and causation in mental
by underlying factors related to both, such as health and on the causes of social disparities in
socioeconomic background, school perfor- health outcomes. Empirical evidence on the
mance, family structure, and previous sexual salience of social disadvantage for understand-
experience. ing health outcomes is strong (Link and Phelan
The analyses reported here have several lim- 1995), and our study provided further support.
itations. First, the ECLS-B allowed us to ex- An array of factors including both socioeco-
amine levels of postpartum distress among for- nomic status and related phenomena (family
mer teenage mothers, but the upcoming wave structure, school performance, and sexual ac-
four of Add Health will provide information tivity) were important for understanding how
about the distress levels of teenage mothers be- social disadvantage is related to psychological
yond the postpartum period, a period that often distress and teenage childbearing. As re-
poses unique challenges to mothers’ mental searchers from both of these traditions would
health. Second, as noted above, data constraints anticipate, we found that underlying social dis-
prohibited us from extending our conclusions advantage increased psychological distress.
to teenage fathers, who represent another vul- Similarly, and as researchers of teenage child-
nerable population in terms of mental health bearing have documented in the past, social
(Heath, McKenry, and Leigh 1995) but who disadvantage made girls more likely to become
are currently understudied. Third, although the adolescent mothers. Our new finding was that
CES-D scale has been validated across a wide the combination of extreme psychological dis-
range of populations, there has been consider- tress and extreme socioeconomic disadvantage
able debate within the social science commu- was particularly predictive of teenage child-
nity about its and other measures’ utility for in- bearing. As Lorant et al. (2003) have suggest-
vestigating mental health as a predictor and ed, then, the relationship between socioeco-
outcome (see the June 2002 issue of JHSB for nomic disadvantage and mental health is com-
an extended discussion). Thus, other facets of plicated, and the two factors can work in tan-
mental health besides the CES-D would be dem.
useful for measuring the intersection of mental Perhaps surprisingly, our results did not sup-
health and social consequences for teen moth- port the prediction from life course theory that
ers. individuals who make a precocious transition
This study has implications for the adoles- to parenthood should suffer mental health con-
cent childbearing literature. Our research sug- sequences. Although the theoretical implica-
gests a parallel between the teen childbearing- tion was not borne out, our results are in line
distress relationship and the teen childbearing- with some earlier research that did not find
education relationship. The rarely-examined negative consequences of violating transition

Downloaded from hsb.sagepub.com at University of Strathclyde on March 20, 2015

TEENAGE CHILDBEARING AND DISTRESS 323
norms (Settersten 1998; Settersten 2003). We most teenage mothers are not at risk of in-
do not interpret our findings as identifying a creased distress due to childbearing. At the
weakness in the theory, especially because our same time, though, they typically start out
study did not capture other mental health out- more distressed than their peers before preg-
comes besides distress that may be affected by nancy and are more likely to have long-term
violating transition norms. Rather, other patterns of psychological distress. Because
processes that have emerged from the life maternal distress can compromise the out-
course paradigm besides the violation of tran- comes of both mother and child regardless of
sition norms may be more useful for under- its cause, distress in teenage mothers should be
standing our findings. taken seriously and addressed, and teenage
For example, two studies that have used the mothers should be considered at risk for de-
Add Health survey to examine the subjective pression problems.
aspect of aging may be relevant for under- All in all, the most compelling policy impli-
standing our findings. Foster, Hagan, and cation of our findings is that ameliorating so-
Brooks-Gunn (2008) showed that young peo- cial disadvantage may decrease both psycho-
ple who perceived themselves to have grown logical distress and early childbearing for ado-
up more quickly than their peers experienced lescent girls. This suggestion, that remedying
increased psychological distress in young the “upstream” socioeconomic root causes of
adulthood. Johnson and Mollborn (2009) health problems may be the most efficient way
found that early hardships (poverty, unsafe or to solve those problems, echoes the work of re-
violent environments, and family structure) in- searchers such as Link and Phelan (1995).
creased the likelihood of early childbearing Reducing socioeconomic disadvantage
and sped up subjective aging. The authors hy- through policies such as income supports
pothesized that one mechanism through which might prevent the complex process we identi-
this occurred was that young people who expe- fied of low SES increasing distress, and then
rienced hardship violated cultural norms con- compounding the effect of that distress on be-
ceptualizing childhood and adolescence as be- coming a teen mother. Future research should
ing innocent and free of responsibility. Perhaps explore whether such policy steps could ad-
to deal with this mismatch between what soci- dress both of these important public health is-
ety expected young people to experience and sues, mental health, and teenage childbearing,
what they actually experienced, they began to simultaneously.
perceive themselves as being older. This kind
of subjective aging process could also underlie NOTES
our study’s results, explaining why socioeco- 1. While we use the terms “depression” and
nomic disadvantage, family structure, and oth- “psychological distress” in the background
er related factors increased both psychological section, the mental health construct ana-
distress and teenage childbearing. In this ac- lyzed in this study measures only psycho-
count, it would not be the violation of cultural logical distress. Depression is a clinical di-
norms about the transition to parenthood, but agnosis with a specific cutoff, not a contin-
rather the violation of norms about childhood uum like our measure.
innocence and freedom from responsibility, 2. Because of ECLS-B confidentiality restric-
that is related to psychological distress. tions, all unweighted numbers were round-
Our findings suggest that many adolescent ed to the nearest 50.
girls who reported greater psychological dis- 3. Two items from the original scale were
tress were at increased risk of teenage child- dropped (“My sleep was restless” and “I
bearing. Distress was likely a symptom of oth- had crying spells”), and one new item was
er underlying problems rather than a direct added (“I felt that life was not worth liv-
cause of teenage motherhood, with the excep- ing”). In addition, two of the items on the
tion of poor, highly distressed girls, but it can scale were rephrased (Lehrer et al. 2006).
still be viewed as a “red flag” identifying tar- 4. Supplemental analyses showed that girls
gets for intervention. Future research should who experienced teenage pregnancy but not
examine whether treating symptoms of distress live birth had significantly higher wave one
in teenage girls can decrease rates of subse- distress levels than girls who had not be-
quent adolescent childbearing, as Kessler et al. come pregnant, but significantly lower than
(1997) have suggested. Second, it seems that girls who had a live teen birth. Omitting

Downloaded from hsb.sagepub.com at University of Strathclyde on March 20, 2015

324 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR
these respondents from analyses reported in Eshbaugh, Elaine M., Jacques Lempers, and Gayle
Tables 2 and 3 did not change the primary J. Luze. 2006. “Objective and Self-Perceived
findings. Resources as Predictors of Depression among
5. All responses were coded into approximate Urban and Non-Urban Adolescent Mothers.”
Journal of Youth and Adolescence 35:839–47.
years of education. Whenever available, we Falci, Christina and Jeylan Mortimer. 2007.
averaged the education levels of the parent “Teenage Parents and Depressed Mood in Adult-
and spouse/partner. If the spouse’s educa- hood: Gender, Selection, Stressors and
tion level was missing, the teenage respon- Resources.” Presented at the annual meeting of
dent’s report of it was substituted, and ab- the American Sociological Association, August
sent that the mother’s education stood in for 11, New York, NY.
both. If no parent completed the survey, Foster, Holly, John Hagan, and Jeanne Brooks-
adolescent respondents’ reports of both par- Gunn. 2008. “Growing up Fast: Stress Exposure
ents’ education levels were averaged. and Subjective ‘Weathering’ in Emerging Adult-
hood.” Journal of Health and Social Behavior
6. We would have liked to examine gender dif- 49:162–77.
ferences in the relationship between teenage Furstenberg, Frank F., Jr. 2003. “Teenage Childbear-
parenthood and psychological distress, but ing as a Public Issue and Private Concern.”
data limitations prevented it. Annual Review of Sociology 29:23–39.
Geronimus, Arline T. and Sanders Korenman. 1992.
REFERENCES “The Socioeconomic Consequences of Teen
Bearman, Peter S., Jo Jones, and J. Richard Udry. Childbearing Reconsidered.” Quarterly Journal
1997. “The National Longitudinal Study of Ado- of Economics 107:1187–1214.
Grogger, Jeff and Stephen Bronars. 1993. “The
lescent Health: Research Design.” Chapel Hill,
Socioeconomic Consequences of Teenage Child-
NC: Carolina Population Center.
bearing: Findings from a Natural Experiment.”
Belle, Deborah and Joanne Doucet. 2003. “Poverty,
Family Planning Perspectives 25:156–61, 174.
Inequality, and Discrimination as Sources of
Hallfors, Denise D., Martha W. Waller, Daniel
Depression among U.S. Women.” Psychology of
Bauer, Carol A. Ford, and Carolyn T. Halpern.
Women Quarterly 27:101–13.
2005. “Which Comes First in Adolescence—Sex
Booth, Alan, Elisa Rustenbach, and Susan McHale. and Drugs or Depression?” American Journal of
2008. “Early Family Transitions and Depressive Preventive Medicine 29:163–70.
Symptom Changes from Adolescence to Early Harris, Kathleen Mullan. 1999. “The Health Status
Adulthood.” Journal of Marriage and the Fami- and Risk Behaviors of Adolescents in Immigrant
ly 70:3–14. Families.” Pp. 286–347 in Children of Immi-
Brown, Joanna D., Elizabeth R. Woods, Matthew P. grants: Health, Adjustment, and Public Assis-
Buman, Sion Kim Harris, and Joanne E. Cox. tance, edited by D. J. Hernandez. Washington,
2007. “Depression and Social Support over Time DC: National Academy Press.
in Adolescent Mothers.” Journal of Adolescent Heath, D. Terri, Patrick C. McKenry, and Geoffery
Health 40:S44–S44. K. Leigh. 1995. “The Consequences of Adoles-
Cubbin, Catherine, John Santelli, Claire D. Brindis, cent Parenthood on Men’s Depression, Parental
and Paula Braveman. 2005. “Neighborhood Satisfaction, and Fertility in Adulthood.” Journal
Context and Sexual Behaviors among Adoles- of Social Service Research 20:127–48.
cents: Findings from the National Longitudinal Hoffman, Saul D. 1998. “Teenage Childbearing Is
Study of Adolescent Health.” Perspectives on Not So Bad after All .|.|.|Or Is It? A Review of
Sexual and Reproductive Health 37:125–34. the New Literature.” Family Planning Perspec-
Deal, Lisa W. and Victoria L. Holt. 1998. “Young tives 30:236–39, 243.
Maternal Age and Depressive Symptoms: Holub, Christina K., Trace S. Kershaw, Kathleen A.
Results from the 1988 National Maternal and Ethier, Jessica B. Lewis, Stephanie Milan, and
Infant Health Survey.” American Journal of Pub- Jeannette R. Ickovics. 2007. “Prenatal and Par-
lic Health 88:266–70. enting Stress on Adolescent Maternal Adjust-
Elder, Glen H., Jr. and Michael J. Shanahan. 2006. ment: Identifying a High-Risk Subgroup.”
“The Life Course and Human Development.” Maternal and Child Health Journal 11:153–59.
Pp. 665–715 in Handbook of Child Psychology. Hotz, V. Joseph, Susan Williams McElroy, and Seth
Vol. 1, Theoretical Models of Human Develop- G. Sanders. 2005. “Teenage Childbearing and Its
ment, edited by R. M. Lerner and W. Damon. Life Cycle Consequences: Exploiting a Natural
Hoboken, NJ: John Wiley & Sons. Experiment.” Journal of Human Resources
Eshbaugh, Elaine M. 2006. “Predictors of Depres- 40:683–715.
sive Symptomatology among Low-Income Ado- Howell, Elizabeth A., Pablo A. Mora, Carol R.
lescent Mothers.” Archives of Women’s Mental Horowitz, and Howard Leventhal. 2005. “Racial
Health 9:339–42. and Ethnic Differences in Factors Associated

Downloaded from hsb.sagepub.com at University of Strathclyde on March 20, 2015

TEENAGE CHILDBEARING AND DISTRESS 325
with Early Postpartum Depressive Symptoms.” Coie, Anne Maumary-Gremaud, Claire Hyman,
Obstetrics and Gynecology 105:1442–50. Robert Terry, and John Lochman. 1999. “Moth-
Hubbs-Tait, Laura, Joy D. Osofsky, Della M. Hann, erhood during the Teen Years: A Development
and Anne McDonald Culp. 1994. “Predicting Perspective on Risk Factors for Childbearing.”
Behavior Problems and Social Competence in Development and Psychopathology 11:85–100.
Children of Adolescent Mothers.” Family Rela- Mollborn, Stefanie. 2009. “American Adults’
tions 43:439–46. Norms about Nonmarital Pregnancy and Their
Hudson, Diane Brage, Susan M. Elek, and Christie Influence on Willingness to Provide Resources
Campbell-Grossman. 2000. “Depression, Self- to Parents.” Journal of Marriage and Family
Esteem, Loneliness and Social Support among 71:122–34.
Adolescent Mothers Participating in the New Mowbray, Carol, Daphna Oyserman, Deborah
Parents Project.” Adolescence 35:445–53. Bybee, and Peter MacFarlane. 2002. “Parenting
Johnson, Monica Kirkpatrick and Stefanie Moll- of Mothers with a Serious Mental Illness: Dif-
born. 2009. “Growing up Faster, Feeling Older: ferential Effects of Diagnosis, Clinical History,
Hardship in Childhood and Adolescence.” Social and Other Mental Health Variables.” Social Work
Psychology Quarterly 72:39–60. Research 26:225–40.
Kalil, Ariel and James Kunz. 2002. “Teenage Child- National Center for Education Statistics. 2006.
bearing, Marital Status, and Depressive Symp- “ECLS-B Longitudinal 9-Month-2-Year Restrict-
toms in Later Life.” Child Development ed-Use Data File and Electronic Codebook (CD-
73:1748–60. ROM).” NCES 2006–044.
Kaufmann, Rachel B., Alison M. Spitz, Lilo T. Neugarten, Bernice L., Joan W. Moore, and John C.
Strauss, Leo Morris, John S. Santelli, Lisa M. Lowe. 1965. “Age Norms, Age Constraints, and
Koonin, and James S. Marks. 1998. “The Adult Socialization.” American Journal of Soci-
Decline in U.S. Teen Pregnancy Rates, ology 70:710–17.
1990–1995.” Pediatrics 102:1141–47. Radloff, Lenore Sawyer. 1977. “The CES-D Scale:
Kessler, Ronald C., Patricia A. Berglund, Cindy L. A Self-Report Depression Scale for Research in
Foster, William B. Saunders, Paul E. Stang, and the General Population.” Applied Psychological
Ellen E. Walters. 1997. “Social Consequences of Measurement 1:385–401.
Psychiatric Disorders, II: Teenage Parenthood.” Reis, J. 1988. “Correlates of Depression According
American Journal of Psychiatry 154:1405–11. to Maternal Age.” Journal of Genetic Psycholo-
Lehrer, Jocelyn A., Lydia A. Shrier, Steven Gort- gy 149:535–45.
maker, and Stephen Buka. 2006. “Depressive Ribar, David C. 1994. “Teenage Fertility and High
Symptoms as a Longitudinal Predictor of Sexual School Completion.” Review of Economics and
Risk Behaviors among U.S. Middle and High Statistics 76:413–24.
School Students.” Pediatrics 118:189–200. Rindfuss, Ronald R., C. Gary Swicegood, and
Link, Bruce G. and Jo Phelan. 1995. “Social Condi- Rachel A. Rosenfeld. 1987. “Disorder in the Life
tions as Fundamental Causes of Disease.” Jour- Course: How Common and Does It Matter?”
nal of Health and Social Behavior 35:80–94. American Sociological Review 52:785–801.
Lorant, V., D. Deliege, W. Eaton, A. Robert, P. Schmidt, R. M., C. M. Weimann, V. I. Rickert, and
Phillippot, and M. Ansseau. 2003. “Socioeco- E. O. Smith. 2006. “Moderate to Severe Depres-
nomic Inequalities in Depression: A Meta- sive Symptoms among Adolescent Mothers Fol-
Analysis.” American Journal of Epidemiology lowed Four Years Postpartum.” Journal of Ado-
157:98–112. lescent Health 38:712–18.
Marini, Margaret Mooney. 1984. “Age and Settersten, Richard A., Jr. 1998. “A Time to Leave
Sequencing Norms in the Transition to Adult- Home and a Time Never to Return? Age Con-
hood.” Social Forces 63:229–44. straints on the Living Arrangements of Young
Mays, Vickie M. and Susan D. Cochran. 2001. Adults.” Social Forces 76:1373–1400.
“Mental Health Correlates of Perceived Discrim- ———. 2003. “Age Structuring and the Rhythm
ination among Lesbian, Gay and Bisexual Adults of the Life Course.” Pp. 81–98 in Handbook of
in the United States.” American Journal of Pub- the Life Course, edited by J. T. Mortimer and M.
lic Health 11:1869–76. J. Shanahan. New York: Kluwer Academic/
Meier, Ann M. 2007. “Adolescent First Sex and Plenum Publishers.
Subsequent Mental Health.” American Journal Shanahan, Michael J. 2000. “Pathways to Adulthood
of Sociology 112:1811–47. in Changing Societies: Variability and Mecha-
Milan, Stephanie, Jeannette R. Ickovics, Trace Ker- nisms in Life Course Perspective.” Annual
shaw, Jessica Lewis, Christina Meade, and Kath- Review of Sociology 26:667–92.
leen Ethier. 2004. “Prevalence, Course, and Pre- Shrier, Lydia A., Sion Kim Harris, Maya Sternberg,
dictors of Emotional Distress in Pregnant and and William R. Beardslee. 2001. “Associations
Parenting Adolescents.” Journal of Consulting of Depression, Self-Esteem and Substance Use
and Clinical Psychology 72:328–40. with Sexual Risk among Adolescents.” Preven-
Miller-Johnson, Shari, Donna-Marrie Winn, John tive Medicine 33:179–89.

Downloaded from hsb.sagepub.com at University of Strathclyde on March 20, 2015

326 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR
Sills, Marion R., Susan Shetterly, Stanley Xu, David Comparison of Young Women’s Psychosocial
Magid, and Allison Kempe. 2007. “Association Status Based on Age of Their First Childbirth.”
between Parental Depression and Children’s Family and Community Health 24:73–84.
Health Care Use.” Pediatrics 119:E829–E836. Williams, David R., Harold W. Neighbors, and
Smith, Marjorie. 2004. “Parental Mental Health: James S. Jackson. 2003. “Racial/Ethnic Dis-
Disruptions to Parenting and Outcomes for Chil- crimination and Health: Findings from Commu-
dren.” Child and Family Social Work 9:3–11. nity Studies.” American Journal of Public
Troutman, Beth R. and Carolyn E. Cutrona. 1990. Health 93:200–208.
“Nonpsychotic Postpartum Depression among Wooten, David B. 2006. “From Labeling Posses-
Adolescent Mothers.” Journal of Abnormal Psy- sions to Possessing Labels: Ridicule and Social-
chology 99:69–78. ization among Adolescents.” Journal of Con-
Turner, R. Jay, Ann M. Sorenson, and J. Blake Turn- sumer Research 33:188–98.
er. 2000. “Social Contingencies in Mental Yamaguchi, Kazuo and Denise Kandel. 1987. “Drug
Health: A Seven-Year Follow-up Study of Use and Other Determinants of Premarital Preg-
Teenage Mothers.” Journal of Marriage and the nancy and Its Outcome: A Dynamic Analysis of
Family 62:777–91. Competing Life Events.” Journal of Marriage
Vicary, Judith R. and Devon A. Corneal. 2001. “A and the Family 49:257–70.

Stefanie Mollborn is an assistant professor of sociology and faculty in the Health and Society Program of
the Institute of Behavioral Science at the University of Colorado at Boulder. Her research focuses on social
psychological approaches to understanding health over the life course. Current projects include an analysis
of the importance of resources for the early development of teenage parents’ children and an examination
of the antecedents and consequences of social norms about teen pregnancy.

Elizabeth Morningstar is a doctoral student in sociology at the University of Colorado at Boulder. Her re-
search focuses on gender, families, and health. She is currently researching quality of life and service needs
among leukemia and lymphoma survivors and plans to explore patients’ and families’ responses to anti-
retroviral treatments in sub-Saharan Africa.

Downloaded from hsb.sagepub.com at University of Strathclyde on March 20, 2015