Full text of DOI:10.1111/j.1741-3737.2000.00777.x

R. JAY
NN M. SORENSON University of Toronto
J. BLAKE URNER Columbia University
TURNER Florida International University
A
T
Social Contingencies in Mental Health: A Seven-Year
Follow-Up Study of Teenage Mothers
This paper reports analyses from a 7-year follow-
up investigation of women pregnant as teenagers
who had been studied during their pregnancy and
shortly following their child’s birth. The objective
of these analyses was to identify potentially mod-
ifiable factors that might influence or condition
psychological adaptation within this high-risk
population. Consistent with prior research, differ-
ences in social support and in personal resources
or attributes effectively predicted depressive
symptomatology, suggesting that such differences
constitute crucial mental health contingencies and
thereby represent promising intervention targets.
Contrary to prior research, differences in stress
exposure were found to be of substantial explan-
atory significance, with lifetime accumulation of
major, potentially traumatic events representing
the most significant element. These findings sug-
gest the need to develop a greater understanding
of socially or programatically modifiable deter-
minants of stress exposure and to take seriously
the prospect of developing interventions that re-
duce such exposure.
A variety of negative outcomes have been reli-
ably observed among teenage mothers, including
diminished employment success (Card & Wise,
1978; Klepinger, Lundberg, & Plotnick, 1998)
and increased risk of ending up as a single parent
(Kellam, Adams, Brown, & Ensminger, 1982)
and of enduring long-term poverty (Hoffman &
Foster, 1997; National Survey of Family Growth,
1982). At the same time, it is clear that the nature
and adequacy of teenage mothers’ social and
economic adaptations over time are highly vari-
able (Furstenburg, Brooks-Gunn, & Morgan,
1987). Furthermore, teen mothers vary substan-
tially in their psychological adjustment (Turner,
Grindstaff, & Phillips, 1990). Very little is
known, however, about factors that might influ-
ence this variation—information necessary for
the identification of appropriate intervention tar-
gets. The fact that the offspring of teenage moth-
ers tend to experience elevated levels of behavior
problems, school failure, and emotional impair-
ment, combined with the clear connection be-
tween maternal emotional well-being and risk for
these child outcomes (Kellam, Ensminger, &
Turner, 1977; Spivak & Weitzman, 1987), further
justifies attention to these issues.
Deuxieme Maison (DM) Building, Room 238, Florida In-
ternational University, 11200 SW 8th Street, Miami, FL
33199 (turnerja@fiu.edu).
One of the complexities in studying the mental
health of this population is its close association
with low socioeconomic status. Researchers who
have employed sister-pairs to control for social
Key Words: depression, mastery, self esteem, social sup-
port, stress, teenage pregnancy.
Journal of Marriage and the Family 62 (August 2000): 777–791
777

778
Journal of Marriage and the Family
and economic disadvantage have discovered ad-
ditional complexities. Specifically, it seems that an
array of adversities occurs with disproportionate
frequency in the lives of girls who become preg-
nant in their teens—both prior to and after the
pregnancy—and these associations hold across so-
cioeconomic statuses (Chilman, 1979; Stiffman,
Powell, Earls, & Robins, 1990). Subsequent stud-
ies have found that these associated adversities
contribute in large measure to the social and emo-
tional difficulties that tend to accompany teenage
motherhood (Geronimus & Korenman, 1993; Hoff-
man, Foster, & Furstenburg, 1993).
The fact that pregnancy and childrearing are
often just part of a constellation of adversities ex-
perienced by a subgroup of female teens raises the
question of whether teenage motherhood makes
an independent contribution to the social and
emotional difficulties that often characterize this
population. This study cannot address this ques-
tion. Rather, our focus is on the identification of
potentially modifiable factors that might influence
or condition psychological adaptation within a
sample of young women who gave birth as teen-
agers. Specifically, we consider the role and sig-
nificance of childhood social experience, along
with more proximal social and psychological fac-
tors, in predicting depressive symptomatology 7
years after the transition to early motherhood. We
discuss the results of our analyses in terms of their
implications for future research and for future ef-
forts at intervention.
These associations have been interpreted as pro-
viding clues to the etiology of mental health prob-
lems and have directed research efforts toward
identification of factors and mechanisms that
might be involved. Fundamental to most expla-
nations that have been offered is the assumption
that mental health risk varies with level of expo-
sure to social stressors. In this context, stressors
refer to situations or events of sufficient force to
challenge the capacities of those exposed to them
(Wheaton, 1994). The assumed relevance of social
stress, of course, has been well justified by data
establishing a compelling link between stressful
life events and mental health problems (Anesh-
ensel, 1992; Dohrenwend & Dohrenwend, 1974;
Monroe, 1992; Tausig, 1982; Thoits, 1983). More-
over, although it has been shown that mental
health status can influence subsequent experiences
of stressful events (Turner & Noh, 1988), it is
clear that a substantial portion of the causation
involved in this linkage goes from social stress to
mental health status (Thoits, 1983).
The Stress Process
Although there remains little doubt about a mean-
ingful link between stressful life events and psy-
chological distress, it is equally clear that these
relationships have been of rather modest magni-
tude, typically accounting for less than 10% of
oberved variation in psychological symptoms
(Rabkin & Struening, 1976). Moreover, only a mi-
nority of individuals who are exposed to even sub-
stantial levels of social stress exhibit significant
emotional distress.
These findings have led to two main theoretical
elaborations of the role that social stress plays in
mental health. One of these elaborations specifies
social and personal resources that modify the
emotional impact of stressors by conditioning in-
dividuals’ vulnerability to such stressors. These
stress process models (Billings & Moos, 1982;
Pearlin, Lieberman, Menaghan, & Mullan, 1981)
map out the pathways through which exposure to
stressors and the availability of personal and so-
cial resources impact (additively and interactively)
psychological well-being. Models that include in-
dicators of personal vulnerability to stress explain
substantially more variation in psychological dis-
tress than do stressful life events alone—a fact
that has led researchers to investigate the relative
importance of stress exposure differences and of
vulnerability differences in explaining adverse
mental health outcomes. This research suggests
We focus on depression as our outcome of in-
terest because it is well established that depression
represents a particularly salient problem for wom-
en in general and young women in particular (No-
len-Hoeksema, 1987). Depression is also salient
because of its potentially adverse effects on the
young women’s capacity to form and maintain so-
cial and intimate relationships, to perform effec-
tively in educational and occupational arenas, and
to provide the attention and nurturance required
by young children for optimal security and devel-
opment.
BACKGROUND
The persistent evidence for associations between
mental health and such factors as gender (Al-Issa,
1982; Nolen-Hoeksema, 1987), marital status
(Gove, 1972) and socioeconomic status (Dohren-
wend & Dohrenwend, 1969; Hollingshead & Red-
lich, 1958) has stimulated a substantial body of
both clinical and community-based research.

Social Contingencies in Mental Health
779
that differential exposure to life events is substan-
tially less important than differential vulnerability
to stress in determining the relationships between
mental health and social class, gender, and marital
status (Kessler, 1979; Kessler & Cleary, 1980;
Aneshensel, 1992; Thoits, 1987)—a conclusion
that has focused the attention of mental health re-
searchers on identifying and understanding the so-
cial and personal resources that condition vulner-
ability differences.
The second main theoretical elaboration, how-
ever, rejects the view that social stress differences
are of only minimal significance for understanding
variations in psychological well-being. Its propo-
nents argue that previous measures of stress ex-
posure substantially undersample what Wheaton
(1994) has called the ‘‘stress universe’’ and thus
they underestimate the significance of exposure
differences. They also hold that it is important to
recognize that the effects of unmeasured differ-
ences in stress exposure tend to be masked within
research findings as vulnerability effects (Turner,
Wheaton, & Lloyd, 1995).
This paper applies both of the described theo-
retical elaborations in assessing factors associated
with the mental health of young women who were
pregnant as teenagers. Although we certainly can-
not claim fully adequate measurements of stress
exposure, we consider a substantially broader ar-
ray of social stressors than has been typical. The
stress process model that we consider also assess-
es levels of mastery, interpersonal dependency,
and social support, along with stress exposure.
of mastery must be, at least substantially, a prod-
uct of one’s history of successes and failures in
social and environmental encounters, this sense
must also reflect at least gross differences in social
and instrumental effectiveness. Accordingly, in-
dividuals high in mastery might effectively avoid
or prevent some potentially stressful events and
circumstances. Second, assuming that an individ-
ual’s appraisal of an event depends importantly on
the perception of ability or inability to manage or
adapt to the situation, those high in mastery might
also experience a narrower range of events or cir-
cumstances as problematic. Third, a sense of per-
sonal agency might attenuate the emotional im-
pact of adverse events because of its implications
for the initiation and persistence of efforts to re-
solve problematic situations (Mirowsky & Ross,
1989; Ross & Mirowsky, 1989). Whatever mech-
anism or mechanisms might be involved, a greater
sense of mastery is associated with reduced risk
for psychological distress and depression, with or
without exposure to social stress (Pearlin &
Schooler, 1978; Pearlin et al., 1981; Turner &
Noh, 1983, 1988).
Interpersonal dependency. We define this con-
struct as the degree to which an individual relies
on the love and attention of others as a basis for
estimating and maintaining her sense of self-
worth. We include a measure of this construct in
the present study for two reasons. First, findings
from clinical studies have identified it as a char-
acteristic of people prone to depression (Chodoff,
1972; Hirschfeld, Klerman, Chodoff, Korchin, &
Barrett, 1976; Masserman, 1970). Second, some
evidence suggests that women are more likely
than men to experience such dependency. Gore
and Colten (1991) summarize evidence suggesting
that aspects of the socialization process result in
a dependency among women ‘‘. . . on the opinions
and evaluations of others in making their own
judgments of how they are doing, that is, in main-
taining self-image and self-esteem’’ (p. 150). The
central hypothesis associated with the construct of
interpersonal dependency is that individuals who
rely almost exclusively on the love and attention
of others in estimating and maintaining their sense
of self-worth are more vulnerable to depression
(Hirschfeld et al., 1976). This kind of dependency
is thought to result from developmental disrup-
tions in the process of establishing secure rela-
tionships. Such individuals may ‘‘. . . become
overly preoccupied with interpersonal security and
. . . experience problems in maintaining positive
Mastery. This construct has been addressed by
employing such labels as sense of powerlessness
(Seeman, 1959), locus of control (Lefcourt, 1976;
Rotter, 1966), personal control (Bandura, 1977,
1982; Gurin, Gurin, & Morrison, 1978), helpless-
ness (Seligman & Mair, 1967; Seligman, 1975),
and fatalism (Wheaton, 1983) as well as mastery
(Pearlin & Schooler, 1978). Despite differences in
emphasis, these concepts are highly similar, and
their meanings are reasonably captured by Pearlin
and Schooler’s (1978) definition of mastery as
‘‘. . . the extent to which one regards one’s life-
chances as being under one’s own control in con-
trast to being fatalistically ruled’’ (p. 5).
There are a number of bases on which to hy-
pothesize the mental health significance of mas-
tery or personal control. Variations in mastery
might affect the number and type of potentially
stressful events and circumstances that the indi-
vidual confronts. Because differences in the sense

780
Journal of Marriage and the Family
feelings about themselves without external sup-
port’’ (Barnett & Gotlib, 1988, p. 119).
2. Level of mastery and of social support
will relate inversely, and level of interper-
sonal dependency will relate directly, to
current depression levels. These factors will
also relate to changes in depression levels
over the previous 7 years. In the case of
mastery and some indices of social support,
both initial (during pregnancy) and current
measures are available, allowing assessment
of the comparitive importance of past and
current levels of mastery and social support.
Social support. Social support has been the sub-
ject of an immense body of research that clearly
establishes an important association with emotion-
al well-being (e.g., Cohen & Syme, 1985; Got-
tlieb, 1981, 1983; House, Umberson, & Landis,
1988; Sarason & Sarason, 1985; Turner, Frankel,
& Levine, 1983; Vaux, 1988). Moreover, evidence
drawn from laboratory animal studies, experimen-
tal human studies, and longitudinal field studies
collectively provide compelling evidence that a
significant portion of the causation involved in
this consistently observed connection goes from
social support to mental health status (House,
1981; 1987).
3. Personal resources and social support
will act to mediate or moderate the impact
of stress exposure on levels of depression.
Although there is general agreement in the lit-
erature that social support is a multidimensional
contruct that includes actual supportive transac-
tions along with the perception of being supported
by others, it was long ago noted that the bulk of
evidence pointing to the mental health signifi-
cance of social support has come from studies fo-
cusing upon perceived support (e.g., House, 1981;
Turner, 1992). In House’s terms, ‘‘social support
is likely to be effective only to the extent per-
ceived’’ (p. 27). More recent studies continue to
demonstrate that the most powerful correlations
between indices of various aspects of social sup-
port and measures of psychological distress are
found with measures of perceived social support
(Kessler, 1992). Wethington and Kessler (1986)
present specific evidence documenting ‘‘. . . not
only that perceptions of support availability are
more important than actual support transactions
but that the latter promote psychological adjust-
ment through the former, as much as by practical
resolutions of situational demands’’ (p. 85). Ac-
cordingly, in this paper we focus on the perception
of being supported by others, considering from
whom the perceived support is derived and in-
cluding measures taken at different times and with
differing time referents.
M
ETHOD
Sample Selection and Retention
The data for this paper derive from a longitudinal
prospective study of pregnant adolescents residing
in southwestern Ontario. Subjects were initially
recruited between 1983 and 1986, primarily from
the caseloads of family physicians and obstetri-
cian–gynecologists practicing in the area (84.5%),
supplemented by cases recruited through public
health nurses providing care to pregnant teens
(4.6%) and through newspaper ads (10.9%).
We conducted in-person interviews with re-
cruited teenagers at three different times: (a) as
soon as possible after medical confirmation of
pregnancy; (b) approximately four weeks follow-
ing the delivery of the baby; and (c) approxi-
mately 7 years later. Except when the participant
preferred an alternative site, we conducted the in-
terviews in the young woman’s residence. We
completed 284 prebirth interviews and 251 post-
birth interviews. Comparisons of cases lost with
the total initial sample, on an array of character-
istics, revealed only one significant difference.
Lost subjects tended to be slightly younger than
those for whom we obtained complete data. At the
Time 3 follow-up, 7 years after the postbirth in-
terviews, we interviewed 213 subjects, or 85% of
the 251 women for whom postbirth data had been
obtained. The mean age of participants at Time 1
was 17.6 years, with 22% being 16 or younger.
At Time 3 the median annual household income
for the sample was $23,000 (approximately
$17,000 U.S.), with 36.2% of participants being
married, 26.8% unmarried but cohabiting, 4.7%
Hypotheses. We hypothesize the following:
1. Level of stress exposure, including ex-
posure to traumatic events in the distant
past, will relate directly to levels of current
depression among women 7 years after their
transition to teenage motherhood. Stress ex-
posure will also predict changes in depres-
sion over the course of those 7 years.

Social Contingencies in Mental Health
781
still living with their family of orientation, and
32.4% remaining single and living on their own.
Of the 213 mothers, 162 retained custody of the
index child at the time of the 7-year follow-up
interview.
However, the twin facts that (a) much of the re-
search in the field, including highly influential
studies (Furstenberg, 1976; Furstenberg et al.,
1987), has focused exclusively or largely on Af-
rican-Americans, and (b) that the majority of teen-
age mothers are White, suggests that this sample
might fill an important scientific need. Another
strength is that we assembled the sample in the
context of the Canadian health-care system, which
provides free and genuinely equitable access to all
relevant services. The sample is thus unaffected
by the various selection biases, faced by most
studies, associated with health insurance or in-
come issues or related limitations on where and
from whom care could be recieved. We believe
that findings from this sample can be cautiously
generalized to a substantial proportion of the pop-
ulation of interest, with the understanding that
their applicability to minority groups is uncertain.
The analyses presented here are based on Time
1 and Time 3 observations, with the exception of
those that focus on the dependent variable of
changes in depressive symptomatology over time.
In these instances, change in depressive symptoms
is assessed in relation to scores obained at Time
2 (4 to 6 weeks following the birth).
Our inability to obtain information on young
women who chose not to participate in the study
limits our capacity to identify biases that might
have occurred at cohort formation and thus limits
our ability to estimate generalizability of findings.
Our research procedure involved monthly tele-
phone calls to the 120 family physicians and ob-
stetrician–gynecologists who cooperated in the
study, to determine whether any pregnant teen-
agers had appeared or had pending appointments.
The anonymous information provided was record-
ed in a log book as a basis for comparison with
interviewed subjects. From these data, we esti-
mate that participating physicians were involved
with some 447 pregnant teenagers during our
catchment period, yielding 240 participants. These
data suggest a rate of loss of about 46%. However,
we presume that a significant proportion of those
lost chose to terminate their pregnancy and hence
did not meet inclusion criteria. Ontario Ministry
of Health data reveal that approximately 50% of
all teenage pregnancies ended in abortion during
the period of our study recruitment, suggesting
that a substantial majority of those comprising the
46% lost might have terminated their pregnancy.
If the very conservative assumption is made that
only half of these young women chose abortion,
our loss rate would be about 30%. Data from the
Ontario Ministry of Health also allowed compar-
ison of our Time 1 sample with all teenagers giv-
ing birth in the same geographical area and over
the same period on the factors of infant birth-
weight, gestational age, infant sex, maternal age,
paternal age, and marital status of the mother.
With the exception of paternal age, we found no
substantial or statistically significant differences
between the study group and the total population
of parturient adolescent women. Although this is
reassuring, we cannot dismiss the possibility that
the original sample was biased with respect to oth-
er variables.
Measurement
Psychological adaptation. We assess psychologi-
cal adaptation in terms of depressive symptoma-
tology as measured by the Center for Epidemio-
logical Studies Depression Scale (CES-D). This
20-item instrument was designed to measure an
individual’s current level of depressive sympto-
matology, with emphasis on depressed mood.
Ample evidence suggests the reliability and valid-
ity of this widely used instrument (Devins &
Orme, 1985; Husaini, Neff, Harrington, Hughes,
& Stone, 1980; Radloff, 1977; Roberts & Vernon,
1983; Weissman, Sholomskas, Pottenger, Prusoff,
& Locke, 1977). In our data, the reliability of this
scale (Cronbach’s alpha) is .88 at Time 2 and .92
at Time 3.
Social stress. Our effort to achieve a more com-
prehensive estimate of variations in exposure to
social stress considers stressful life events, endur-
ing or chronic stressors, and major lifetime trau-
mas. We assessed stressful life events using a 31-
item checklist of negative events common to a
number of life event indices (Henderson, Byrne,
& Duncan-Jones, 1981; Holmes & Rahe, 1967;
Sarason, Johnson, & Siegel. 1978; Turner et al.,
Reflecting the population from which they
were drawn, study participants were largely of Eu-
ropean ancestry, with the exception of a very
small proportion from one of Canada’s native
groups. Because African-American and Hispanic
minorities are overrepresented among teenage
mothers within North America in general, the
present sample can hardly be fully representative.

782
Journal of Marriage and the Family
1995). The list includes such items as ‘‘Was there
a serious accident or injury?,’’ ‘‘Was there trouble
with the law?,’’ and ‘‘Was [a family member] fired
or layed off?’’ We asked respondents to indicate
which of the 31 events they had experienced per-
sonally over the preceding 12 months. We also
asked whether their spouse or partner had expe-
rienced any of 18 of these negative events, wheth-
er their children had experienced any of 11 of
these events, and, for 7 events, whether other rel-
atives or friends had been affected. In the analyses
presented here, we estimate exposure to eventful
stressors by a simple count of events that hap-
pened to the respondent or to someone close to
her.
We assessed enduring or chronic stressors us-
ing 36 items drawn from the inventory developed
by Wheaton (1991, 1994; see also Turner et al.,
1995). These items provide a subjective assess-
ment of life conditions and situations in such areas
as general or ambient stress, financial stress, work
stress, marital or relationship stress, and parenting
stress. Examples from the 10-item marital or re-
lationship stress section are: ‘‘Your partner doesn’t
understand you,’’ ‘‘Your partner expects too much
of you,’’ and ‘‘Your partner is always threatening
to leave or end the relationship.’’ Wheaton (1991)
has argued that the subjective component in these
stressors is an inherent part of what they are, and
he presented evidence that the observed associa-
tion between his measure of chronic stress and
mental health outcomes cannot be attributed to ei-
ther measurement or causal confounding.
life trauma and chronic stress inventories to the
questionnaire only at Time 3, and, because evi-
dence suggests that the effects of stressful life
events are limited to a year or less (Brown & Har-
ris, 1978; Murphy & Brown, 1980), effects over
a 7-year interval could not be expected and, in
fact, were not observed in the present data.
Personal resources. We assessed personal control
or mastery using a 7-item scale developed by
Pearlin and Schooler (1978). Respondents rate
each item on a 5-point scale ranging from strongly
agree to strongly disagree. In this study, the in-
ternal reliability of this measure as estimated by
Cronbach’s alpha is .71 in both the Time 1 and
Time 3 data. We administered a number of items
from the interpersonal dependency measure de-
veloped by Hirschfeld, Klerman, Gough, Barrett,
Korchin, & Chodoff (1977) at Time 3. These in-
cluded four items from the ‘‘emotional reliance’’
subscale and three items from the ‘‘assertion of
autonomy’’ subscale. Factor analysis confirmed
the distinctiveness of these two subscales as mea-
sured, and Cronbach’s alpha was .61 and .74 for
emotional reliance and assertion of autonomy, re-
spectively. These two subscales are inversely cor-
related (r ϭ Ϫ.08), and we hypothesize that high
levels of emotional reliance increase risk for de-
pression, whereas high levels of autonomy de-
crease the risk. These measures are available only
within the Time 3 data.
Social support. As noted above, available evi-
dence has suggested that perceived social support
is likely to be of substantial relevance for the out-
come assessed in this paper. The initial interview,
conducted during pregancy, included the Provi-
sions of Social Relations Scale (PSR) (Turner et
al., 1983). This measure was selected because of
its demonstrated reliability and construct validity
and because of evidence for a stable factor struc-
ture that effectively distinguishes between support
experienced from family and support experienced
from friends (Turner et al., 1983). At Time 3, we
measured family and friend support by separate 8-
item scales developed on the basis of the PSR
scale for use in the Toronto Mental Health Study
(Turner & Marino, 1994). The reliability coeffi-
cients were .96 and .97 for family support and
friend support, respectively. ‘‘Family’’ was left
undefined within both of these measures. Thus,
participants answered based on whatever ‘‘fami-
ly’’ meant to them.
We obtained data on major lifetime events or
traumas by means of an 18-item inventory that
asked whether each event had ever occurred and,
if so, the respondent’s age at the last occurrence.
This measure, which is a slightly shortened ver-
sion of Turner and colleagues’ checklist (1995),
addresses such experiences as parental death,
physical abuse by a parent, physical abuse by a
spouse or partner, and a life-threatening accident
or illness. Because this instrument and related is-
sues have been discussed in some detail elsewhere
(Turner & Lloyd, 1995), we will summarize just
two points that are pertinent for our purposes.
First, there is good evidence that most of the ma-
jor events considered, both singly and in combi-
nation, can have important long-term mental
health consequences. Second, respondents report
these potentially traumatic experiences with rea-
sonable reliability even when very distant in time.
In this paper we consider only measures of so-
cial stress obtained at Time 3. We added the major
During the course of pregnancy (Time 1), we

Social Contingencies in Mental Health
783
T
ABLE 1. REGRESSION OF
YMPTOMS (CESDT3) ON IMENSIONS OF
AND RIOR YMPTOMATOLOGY
C
URRENT
D
S
EPRESSIVE
the conventional wisdom that differences in ex-
posure to social stress are of only minimal ex-
planatory significance. The variance accounted for
(nearly 19%) represents about double the upper
range of previous reports that have considered
only stressful events (Rabkin & Struening, 1976).
Consistent with the findings of Turner and col-
leagues (1995), these results suggest that social
stress might represent a more important mental
health risk factor than has generally been as-
sumed. Recent events no longer contribute to the
prediction of depressive symptomatology once
both chronic stress and lifetime traumas are con-
sidered. The significance of these latter stress in-
dices remains clear when prior levels of depres-
sion are controlled (Model 4), indicating their
relevance for changes in depressive symptomatol-
ogy over time. These findings provide clear sup-
port for Hypothesis 1.
S
D
S
OCIAL TRESS
S
P
Model 1 Model 2 Model 3 Model 4
Predictor
␤
␤
␤
␤
Recent events
.29**** .21*** .08
.23*** .20***
.30****
—
Chronic stress
Lifetime traumas
CESDT2
.16**
.28****
.31****
.27
Adjusted R2
.08
.12
.19
Note: N ϭ 204.
**p Ͻ .05. ***p Ͻ .01. ****p Ͻ .001.
selected 9 of the 12 items from the ‘‘care dimen-
sion’’ of the Parental Bonding Instrument (Parker,
Tupling, & Brown, 1979) to estimate the extent of
parental support each adolescent percieved as hav-
ing been available during her developmental
years, as distinct from level of family support ex-
perienced in the context of the pregnancy. We ad-
ministered this measure separately with respect to
fathers and to mothers. Analyses confirmed a sin-
gle underlying dimension and identified 6 items
that demonstrated the property of factor invari-
ance when compared across scales describing re-
lationship with father and relationship with
mother. These 6 items were also sufficient to
achieve internal reliabilities between .75 and .94
when describing father or mother at both Time 1
and Time 3.
Personal resources. The relevance of mastery at
Time 1 and Time 3, and that of the two measured
subdimensions of interpersonal dependency (emo-
tional reliance and assertion of autonomy), are
considered in Table 2. Level of mastery assessed
during the course of pregancy shows a clear re-
lationship (Ϫ.26) with depressive symptomatolo-
gy measured more than 7 years later. Subsequent
models in this stepwise analysis illustrate that the
effects of Time 1 mastery are largely indirect
through Time 3 mastery and that mastery is an
even stronger predictor than is previous level of
depressive symptomatology. Moreover, the appar-
ent significance of mastery is little attenuated
when previous depression is controlled. With
Time 3 mastery controlled, the coefficient for
emotional reliance does not achieve statistical sig-
nificance by the usual criterion, whereas assertion
of autonomy shows a significant inverse associa-
tion with depressive symptoms independent of
mastery, emotional reliance, and previous symp-
tom level. Collectively, the personal resources
considered in these analyses account for a sub-
stantial 28.6% of observed variation in depressive
symptomatology.
RESULTS
Our initial analyses of risk and protective factors
involved the separate consideration of sets of mea-
sures comprising each domain of the stress pro-
cess model (stress exposure, personal resources,
and social resources). The goal was to assess the
individual predictive power of each domain and
to identify the minimum set of measures that
could effectively represent each domain within
combined analyses.
Stress exposure. Table 1 presents the result of
stepwise analyses in which we regressed Time 3
depressive symptomatology on the three measured
dimensions of social stress. It is noteworthy that
the standardized coefficient of .29, observed at the
first step matches closely relationships between re-
cent events and psychological distress that have
typically been reported in the literature (Thoits,
1983). However, when chronic stressors and life-
time traumas are also considered (Model 3), the
results are far from typical and clearly challenge
Social support. As noted earlier, our assessment
of social support involved measures of support
from family and from friends during the course of
the pregnancy and 7 years after the index birth,
as well as retrospective reports, at both of these
time points, of the warmth experienced from each
parent when the respondent was growing up. In
considering these latter data we were confronted

784
Journal of Marriage and the Family
TABLE 2. REGRESSION OF
CURRENT
D
EPRESSIVE
S
YMPTOMS (CES-DT3) ON
P
ERSONAL
R
ESOURCES AND PRIOR
SYMPTOMATOLOGY
Model 1
Model 2
Model 3
Model 4
Model 5
Model 6
Predictor
␤
␤
␤
␤
␤
␤
Time 1: mastery
Time 3: mastery
Time 3: emotional reliance
Time 3: assert. of autonomy
CESDT2
Ϫ.26****
Ϫ.09
Ϫ.47****
—
Ϫ.46****
.13*
—
—
Ϫ.38****
.08
Ϫ.18***
.30****
.37
—
Ϫ.40****
—
Ϫ.18***
.30****
.36
Ϫ.44****
.11*
Ϫ.17***
Adjusted R2
.06
.25
.26
.29
*p Ͻ .10. **p Ͻ .05. ***p Ͻ .01. ****p Ͻ .001.
with the fact that more than 10% of these young
women indicated that they had no father or father
substitute with whom to relate. Thus, the use of
continuous scores on the warmth dimension
would result in a significant loss of sample size
and of the capacity to evaluate the effects of hav-
ing no father within the analyses.
3, each of the three categories describing paternal
or maternal warmth, along with a category de-
scribing mother or father absence, is treated as a
dummy variable, scored 1 if the respondent is as-
signed to that category and 0 otherwise. As the
omitted category describes those with a ‘‘very
warm’’ relationship, our parameter estimates con-
trast this most desirable circumstance with lesser
levels of maternal or paternal warmth or with the
absence of the parental figure.
Models 1 and 2 evaluate the mental health rel-
evance of maternal and paternal warmth as re-
ported during the course of pregnancy. Those who
reported having experienced their mother as ‘‘not
warm’’ exhibited significantly higher levels of de-
pressive symptoms more than 7 years later. Al-
though this relationship is no longer observed
when Time 1 social support from family and
friends is considered (Model 3), level of parental
warmth from mother or from father, as remem-
bered at Time 3, makes a significant independent
contribution even when other measures of social
support from family and from friends are con-
trolled (Model 6).
It appears that a respondent’s relationship with
her mother or father that was recalled as being
emotionally cold is associated with greater mental
health risk than is the absence of that parent. With
all nontrimmed social support measures consid-
ered, more than 20% of the variability in depres-
sive symptoms is accounted for by this resource,
with clear contributions from friend support at
both Times 1 and 3 and from Time 3 accounts of
parental warmth during childhood. Morover, each
of these dimensions remains statistically signifi-
cant when Time 2 depression is controlled.
Tables 2 and 3 directly address Hypothesis 2,
confirming an inverse relationship between cur-
rent depression levels and both mastery and social
support. These relationships are also observed
with respect to changes in depressive symptoma-
We addressed this problem by employing a la-
tent class model (Clogg, 1981; McCutcheon,
1987) to develop categorical measures of parental
warmth. This approach conceptualizes the under-
lying contructs of paternal and maternal warmth
in terms of two or more latent categories or class-
es, rather than as a continuous latent scale. In all
four cases (at Times 1 and 3, with respect to both
mother and father), a three-class model was ade-
quate to account for the observed association
among five of the items measuring parental
warmth. In each case, the data were consistent
with the hypothesis that responses were generated
by a single latent variable describable by three
categories. Although the model does not rely on
the assumption of an ordered relationship among
the categories, inspection of the conditional prob-
abilities of a positive response to each individual
item within each of the three latent classes sug-
gests one category of respondents highly likely to
affirm a relationship with an affectionate parent
who ‘‘spoke warmly,’’ ‘‘enjoyed talking with
them,’’ and ‘‘could make them feel better’’; that
is, a parent not remembered as being emotionally
cold. At the other extreme, latent probabilities
suggest another category of respondents who were
quite unlikely to report these positive aspects of
the parental relationship and who were quite likely
to recall the parent as having been emotionally
cold toward them. Finally, we observed an inter-
mediate class of respondents whose experience
fell between these two extremes of parental
warmth.
In the regression analyses presented in Table

Social Contingencies in Mental Health
785
TABLE 3. REGRESSION OF CURRENT DEPRESSIVE SYMPTOMS (CESDT3) ON INDICES OF SOCIAL SUPPORT
Model 1 Model 2
Model 3
Model 4
Model 5
Model 6
Model 7
Predictor
␤
␤
␤
␤
␤
␤
␤
Time 1: mothera
Absentb
.10
.27***
.08
—
.22****
—
—
.10
—
—
—
—
—
—
—
—
—
—
—
—
—
Not warmb
Somewhat warmb
Time 1: fathera
Absentb
.03
.04
.08
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
Not warmb
Somewhat warmb
Time 1: PSR
Family
Friends
Ϫ.16**
Ϫ.20***
Ϫ.14*
Ϫ.19***
Ϫ.12*
Ϫ.20***
Ϫ.09
Ϫ.14**
—
Ϫ.13**
Time 3: mothera
Absentb
.08
.21***
.08
—
.16**
—
—
.14**
—
—
.16**
—
Not warmb
Somewhat warmb
Time 3: fathera
Absentb
.00
.25***
.02
—
.21***
—
—
.15**
—
Not warmb
Somewhat warmb
Time 3: relatives support
Ϫ.09
—
Time 3: friends’ support
Ϫ.17**
Ϫ.18***
CESDT2
.30****
.27
Adjusted R2
.05
.04
.11
.13
.17
.20
Note: N ϭ 204.
b
aA very warm relationship is the omitted category. 1; 0 otherwise.
*p Ͻ .10. **p Ͻ .05. ***p Ͻ .01. ****p Ͻ .001.
tology. Also apparent is a positive, though mar-
ginally significant, zero-order association between
emotional reliance and depression.
nificant in its effects on Time 3 depression, a find-
ing consistent with the expectation that important
earlier experiences are more likely to play a role
in shaping previous levels of depressive symp-
toms than in predicting subsequent changes in
such symptomatology. The remaining coefficients
are highly consistent when predicting changes in
depressive symptoms.
The unstandardized regression coefficients pre-
sented in Table 4 allow us to evaluate Hypothesis
3, that personal and social resources mediate the
linkage between stress exposure and depression.
The coefficient for lifetime traumas is reduced by
20% with the inclusion of social support measures
(Model 2) and by nearly 29% when the personal
resources of mastery and autonomy are also con-
sidered (Model 3). However, the primary relation-
ship is clearly direct rather than indirect, with a
strong association continuing to be observed after
all other factors are controlled. When both social
support and personal resource measures are con-
trolled, the coefficient for chronic stress declines
by 60% and is no longer statistically significant.
The stress process model. Each of the analyses so
far involves stepwise procedures in which mea-
sures of the particular stress process construct at
issue were added and in which nonsignificant
measures were trimmed at each successive step.
These results provided the basis for the analyses
reported in Table 4. We regressed Time 3 depres-
sive symptomatology on those measures of social
stress, social resources, and personal resources
that demonstrated significant independent associ-
ations within the separate analyses reported above.
When we consider all surviving measures si-
multaneously (Model 3), this refined stress pro-
cess model accounts for fully 42% of observed
differences in CES-D scores, with Time 3 mastery
and lifetime traumas representing the most pow-
erful predictors. Controlling Time 2 depressive
symptomatology (Models 4 and 5) renders child-
hood relationship with father statistically nonsig-

786
Journal of Marriage and the Family
TABLE 4. DEPRESSIVE
SYMPTOMS (CES-DT3) REGRESSED ON
SOCIAL
S
TRESS, SOCIAL
S
UPPORT AND
P
ERSONAL
RESOURCES
Model 1
Model 2
Model 3
Model 4
Model 5
Predictor
b
␤
b
␤
b
␤
b
␤
b
␤
Lifetime traumas
Chronic stress
1.22 .34****
.63 .22****
.97
.45
.27****
.16**
.87
.25
.24****
.09
.87
—
.24****
.90
—
.25****
T1 friends support
T3 mother not warm
T3 father not warm
T3 friends support
T3 mastery
Ϫ.27 Ϫ.15**
Ϫ.21 Ϫ.09
—
—
1.66
—
—
—
2.61
3.69
.10*
.17***
1.96
2.67
.07
.13**
.08
Ϫ.26 Ϫ.17***
Ϫ.17 Ϫ.14**
Ϫ.68 Ϫ.32**** Ϫ.69 Ϫ.32**** Ϫ.71 Ϫ.33****
Ϫ.25 Ϫ.16*** Ϫ.26 Ϫ.17***
T3 assert. of autonomy
CESDT2
.43 Ϫ.18****
.45 Ϫ.19**** Ϫ.46 Ϫ.19****
.21
.23****
.23
.25****
Adjusted R2
.19
.29
.42
.45
.45
Note: N ϭ 204.
*p Ͻ .10. **p Ͻ .05. ***p Ͻ .01. ****p Ͻ .001.
Thus, the effects of chronic stress are substantially
mediated through the social support and personal
resource dimensions of the stress process model.
A minor portion of the lifetime trauma–depression
relationship is also indirect through these re-
sources.
kinds of risks that might tend to characterize the
lives of women who become mothers as teen-
agers. This is so because they contrast rather
sharply with findings from a general community
study that empoyed essentially the same three
measures of social stress. Turner and colleagues
(1995) studied nearly 1,400 men and women, 18
to 55 years of age, randomly sampled from an
urban community a short distance from the loca-
tion of our sample. Although the risk of both de-
pressive symptoms and depressive disorder were
found to increase with number of life traumas, no
direct effects were observed for this variable when
chronic and eventful stressors were also consid-
ered. In the general community, the impact of ma-
jor traumas that occur over a lifetime appears to
be mediated through more contemporaneous stress
experiences.
It might be that the clear difference in the men-
tal health significance of life traumas across these
two samples can be accounted for by gender dif-
ferences, by age differences, or by associated dif-
ferences in temporal proximity to the experience
of these major childhood stressors. Because the
women in the present sample were 27 years old
or younger at the Time 3 interview, less time
would have elapsed on average since they expe-
rienced the reported traumas. It is also possible
that the greater significance of potentially trau-
matic events is associated with a greater frequency
of such events in the experience of that subgroup
of young women defined in terms of teenage
motherhood.
Results presented thus far confirm our hypoth-
eses with respect to the additive effects on de-
pression of stress exposure and of other dimen-
sions of the stress process model that we
considered. In order to address the possibility that
personal resources and social support moderate
(as opposed to mediate) the impact of stress ex-
posure on depression, we created multiplicative
interaction terms that combine each of the re-
source variables shown in Model 3 with each of
the social stress measures represented in that mod-
el. We tested these interactions one at a time in
the context of these main effects. None was sta-
tistically significant, which disconfirms the mod-
erating or buffering hypothesis. The stress process
model postulates that personal and social re-
sources condition stress vulnerability, and thus it
predicts stronger associations in the presence of
high stress. However, within our sample of young
women 7 years after their transition to mother-
hood, the importance of these resources for de-
pression appears independent of the level of stress
exposure and vice versa.
Life Traumas and Teenage Motherhood
The apparent importance of life traumas in our
results, and the fact that they represent the most
powerful factor among dimensions of social stress
(Table 1), are findings unique in the literature.
Moreover, they are suggestive with respect to the
To assess this latter possibility, we distin-
guished a subsample of 191 young women, also
between the ages of 21 and 27, within the com-
munity sample of Turner and colleagues (1995),

Social Contingencies in Mental Health
787
F
IGURE 1. TRAUMA
E
XPOSURE
LEVELS IN TWO YOUNG
the Figure, the teenage mothers were differentially
weighted to standardize the distribution of paren-
tal occupational prestige to that of the comparison
group. The distribution of life traumas in data
weighted to control on SES differences varied lit-
tle from those shown in the Figure. The proportion
of teenage mothers who reported fewer than three
such events increased only 1.3% (from 23.5% to
24.8%); those who reported three or four such
events decreased from 26.8% to 26.3%; and those
who reported five or more traumas decreased only
from 49.7% to 48.9%. The socioeconomic status
of family of origin, at least as estimated by our
index of parental occupational status, appears to
explain little of the differential risk of life traumas
experienced by the two groups—a conclusion sup-
ported by the fact that parents’ occupational level
is only very weakly correlated with number of
lifetime traumas reported (r ϭ .09 in the teen
pregnancy sample and .12 for the age-matched
comparison group).
Thus, the women who were pregnant as teen-
agers came disproportionately from lower socio-
economic backgrounds, with whatever implica-
tions that fact might have for health, well-being
and life chances. In addition, these women’s pre-
vious experience is characterized by significantly
elevated risk for multiple, potentially traumatic
adversities. Because our measures of cumulative
life traumas and SES are essentially independent,
their associated mental health risks are clearly ad-
ditive rather than redundant.
Our work suggests that exposure to social
stress represents an important mental health con-
tingency in general. The high risk of exposure to
traumatic events experienced by women who be-
came mothers as teenagers suggests the need for
special consideration of the role of social stress in
this group. Table 5 compares the responses to spe-
cific traumas of the teen pregnancy sample and
the age-matched community sample. The compar-
ison suggests a rather specific pattern of events
that contribute to these elevated levels of lifetime
trauma. During childhood and early adolescence,
the teenage mothers were no more likely than
their age-matched peers to have experienced a ma-
jor illness or accident, long-term parental unem-
ployment, or even major difficulty in school.
However, they were more than twice as likely to
have had parents who divorced and nearly four
times as likely to have experienced physical abuse
by a parent.
FEMALE
P
OPULATIONS
and we examined their cumulative exposure to the
17 trauma items common to both studies. The Fig-
ure 1 provides a representation of the dramatic
contrast between these two samples in their life-
time experience of major adversities. The women
who were pregnant as teenagers are two and a half
times more likely than their age-matched counter-
parts to have experienced 5 or more traumatic
events, and they are less than half as likely to have
experienced two or fewer such events. Neither
teenage pregnancy nor childbearing was included
within the trauma list, and exclusion of the two
items involving a respondent’s child produced lit-
tle change in the distributions shown, with the
number of teen mothers who had experienced five
or more life traumas declining by only about 3%,
to 46.5%. Thus, these striking differences are sub-
stantively important.
An obvious question that arises in relation to
these results is the extent to which observed ex-
posure differences are an artifact of differences in
socioeconomic status. Data available on the oc-
cupational prestige (Hollingshead, 1957) of the
main wage-earning parent, while respondents in
both studies were growing up, allows us to con-
sider this issue. As anticipated, socioeconomic sta-
tus (SES) of origin, thus indexed, is distributed
quite differently in these two populations. Women
who were pregnant as teenagers were more than
twice as likely as their age-matched counterparts
to have grown up in families whose main wage
earner worked at semi-skilled or unskilled jobs.
Correspondingly, they were less than one quarter
as likely to have had a parent who was a major
or minor professional.
To assess the role of SES in accounting for the
differential experiences of adversity displayed in
With respect to events that could have occurred
either when the respondent was a child or an

788
Journal of Marriage and the Family
TABLE 5. PROPORTION OF
POSITIVE
R
ESPONSES TO
M
AJOR
L
IFE
E
VENTS, WOMEN 21–28 YEARS OF
AMPLE
AGE. TEENAGE
MOTHERS AND
C
OMMUNITY S
Teenage Mothers
Community Sample
(n ϭ 213)
(n ϭ 191)
While a child or teenager:
Major illness or accident
Parents divorced
Repeated a year of school
Long-term parental unemployment
Physical abuse by parent
25.8
33.8
22.5
18.3
19.8
18.3
14.1****
15.7
11.5
5.2****
Lifetime:
Divorced or ended relationship with someone still loved
Parent died
Own child died or given up
Spouse or other loved one died
Witnessed violent event
46.0
19.2
23.0
30.3
33.8
9.9
42.4
18.3
2.5****
41.1**
23.7**
9.9
In a natural disaster
Disabling or life-threatening accident, injury or illness
Own child had life-threatening illness or accident
Spouse or partner unfaithful
Sexual abuse or assault
Physical abuse by spouse or partner
Immediate family member addicted to alcohol or drugs
23.0
19.9
36.8
39.6
39.0
36.0
12.1***
1.6****
23.2***
15.5****
12.6****
14.8****
**p Ͻ .05. ***p Ͻ .01. ****p Ͻ .001.
adult, the two samples are nearly indistinguishable
on the experiences of divorcing or ending a rela-
tionship with someone they were still in love with,
losing a parent in death, and going through a nat-
ural disaster. In the realm of other close interper-
sonal and familial relationships, however, clear
differences are apparent. Teenage mothers were
significantly more likely to have discovered that
a spouse or partner was unfaithful, more than
twice as likely to have been sexually abused or
assaulted, more than three times as likely to have
been physically abused by a spouse or partner, and
more than three times as likely to have had an
immediate family member addicted to alcohol or
drugs.
personal resources of mastery and assertion of au-
tonomy represent potential targets for interven-
tions aimed at fostering emotional well-being and
preventing depression among these young women.
These factors tend to be conceptualized as coping
resources and to be seen as defining individual
differences in vulnerability to stress.
The importance of vulnerability differences for
understanding variations in mental health risk has
been emphasized by reports that differential ex-
posure to stressful life events is substantially less
important than differential vulnerability to stress
for explaining social status differences in mental
health (Kessler, 1979; Kessler & Cleary, 1980;
Thoits, 1987). These findings, along with the per-
ception that social stress is only weakly associated
with mental health, have focused the attention of
researchers on the task of understanding individ-
ual differences in responsiveness to stress and the
attention of interventionists on the goal of de-
creasing such responsiveness through the en-
hancement of coping resources.
DISCUSSION
The results of this study support a number of con-
clusions that are suggestive with respect to both
future research and the focus of future interven-
tion efforts. At the most general level, stress ex-
posure and social and personal resources as mea-
sured here appear to effectively specify crucial
mental health contingencies within this important
high-risk population. Social stress and social and
personal resources accounted for fully 42% of ob-
served differences in depression.
Although the findings presented here provide
additional support for the promise of such efforts,
there are grounds for questioning the nearly ex-
clusive preoccupation with vulnerability differ-
ences that appears to characterize current thinking.
We argued earlier that research that considers only
recent life events has encouraged the perception
that social stress is of only minimal mental health
As the final equation of Table 4 reveals, the
social resource of support from friends and the

Social Contingencies in Mental Health
789
significance. In counterpoint to this perception, we
contend that our Hypothesis 1, that level of social
stress is an important determinant of mental health
status, has never been effectively tested because
variations in exposure to social stress have never
been adequately estimated.
onstrate their importance. We believe, however,
that it might be possible to identify social, struc-
tural, familial, or developmental factors that influ-
ence levels of exposure to stressful events and cir-
cumstances. Research is needed that aims at
developing a greater understanding of socially or
programatically modifiable determinants of stress
exposure. Such an understanding is necessary in
order to take seriously the prospect of developing
interventions that reduce stress exposure as well
as interventions that ameliorate the consequences
of such exposure.
Finally, findings based on a simple count of
major experienced events or traumas suggest a po-
tentially important point. The kind of cumulative
adversity we have observed among young women
in our study signifies elevated risk for depressive
symptoms later in life and has elsewhere been
shown to constitute a marker for elevated risk for
lifetime psychiatric and substance disorders (Tur-
ner & Lloyd, 1995). In addition, high levels of
early adversity appear to represent a marker for
increased risk for various subsequent adversities
such as teenage pregnancy. Thus, assessment of
trauma exposure might represent an additional ef-
fective strategy for identifying those most in need
of intervention.
With a more comprehensive estimate of stress
exposure than has been typically available, this
study, and the findings of Turner and colleagues’
(1995) community study, have demonstated that
social stress accounts for substantially more vari-
ability in depressive symptoms than previous re-
search has suggested. Moreover, these results are
likely to represent a conservative estimate of the
role and significance of social stress. This is so
because, whatever advance might be involved in
this more comprehensive measurement effort, it
seems clear that a fully adequate estimate of vari-
ations in stress exposure remains to be achieved.
In our view, the conclusion is warranted that
exposure to social stress represents a crucial men-
tal health contingency. Thus, there is a need to
develop an understanding of the factors that influ-
ence or condition exposure to stress and to take
seriously the prospect of developing interventions
to reduce such exposure. It seems clear that some
stressors are adventitious or fateful in nature. The
fact that very low correlations were observed be-
tween SES and exposure to potentially traumatic
events (r ϭ .09 in the teen mothers sample and
.12 for the age-matched community comparison
group), suggests that many of the traumatic ex-
periences we have assessed might be of this na-
ture. However, it is equally clear that stressful cir-
cumstances and events, as distinct from major life
traumas, are differentially distributed across social
statuses (Turner et al., 1995), suggesting that such
circumstances and events are socially conditioned
and potentially amenable to social or programatic
modification.
To this point, the efforts of prevention scien-
tists, whether the outcome of interest is depressive
symptoms, psychiatric disorders, substance abuse,
delinquency or criminality, or violence, have con-
centrated on the goal of modifying individual so-
cial, motivational, and behavioral characteristics.
This largely exclusive focus has been encouraged
by the assumption, which seems pervasive in the
behavioral sciences, that the conditions that influ-
ence stress exposure are immutable. It is our view
that this assumption should be rejected. We are
not suggesting that efforts aimed at improving
personal effectiveness and social competence
should be abandoned. Indeed, our findings dem-
NOTE
This is a revision of a paper presented at the American
Sociological Association Annual Meetings, New York,
August, 1996. This study was supported by a research
grant from the National Health Research and Develop-
ment Program (NHRDP) of Health Canada and by a
National Health Scientist Award to R. Jay Turner.
REFERENCES
Al-Issa, I. (1982). Gender and adult psychopathology.
Gender and psychopathology. New York: Academic
Press.
Aneshensel, C. S. (1992). Social stress: Theory and re-
search. Annual Review of Sociology, 18, 15–38.
Barnett, P. A., & Gotlib, I. H. (1988). Psychosocial func-
tioning and depression: distinguishing among ante-
cedents, concomitants, and consequences. Psycholog-
ical Bulletin, 104, 97–126.
Bandura, A. (1977). Self-efficacy: Toward a unifying
theory of behavioral change. Psychological Review,
84, 191–215.
Bandura, A. (1982). Self-efficacy mechanism in human
agency. American Psycholgist, 37, 122–147.
Billings, A. G., & Moos, R. H. (1982). Stressful life
events and symptoms: A longitudinal model. Health
Psychology, 1, 99–117.
Brown, G. W., & Harris. T. (1978). Social origins of
depression: A study of psychiatric disorder in women.
New York: Free Press.
Card, J. J., & Wise, L. L. (1978). Teenage mothers and

790
Journal of Marriage and the Family
teenage fathers: The impact of early childbearing on
the parents’ personal and professional lives. Family
Planning Perspectives, 10, 199–205.
Chilman, C. S. (1979). Teenage pregnancy: A research
review. Social Work, 24, 492–498.
relates of nonmarital childbearing among adult wom-
en. Family Planning Perspectives, 29, 137–140.
Hoffman, S. D., Foster, E. M., & Furstenberg, F. F.
(1993). Reevaluating the costs of teenage childbear-
ing. Demography, 30, 1–14.
Chodoff, P. (1972). The core problem in depression: In-
terpersonal aspects. In J. H. Messerman (Ed.), Sci-
ence and psychoanalysis (pp. 56–65). New York:
Grune & Stratton.
Clogg, C. C. (1981). New developments in latent struc-
ture analysis. In D. M. Jackson & E. F. Borgatta
(Eds.), Factor analysis and measurement (pp. 215–
246). Beverly Hills, CA: Sage.
Hollingshead, A. B. (1957). Two factor index of social
position. New Haven, CT: Author.
Hollingshead, A. B., & Redlich, F. C. (1958). Social
class and mental illness: A community study. New
York: Wiley.
Holmes, T. H., & Rahe, R. H. (1967). The social read-
justment rating scale. Journal of Psychosomatic Re-
search, 11, 213–218.
Cohen, S., & Syme, S. L. (Eds.). (1985). Social support
and health. San Francisco, CA: Academic Press.
Devins, G. M., & Orme, C. M. (1985). Centre for Ep-
idemiological Studies Depression Scales. In D. J.
Keyser & R. C. Sweetland (Eds.), Test critiques (pp.
144–160). Kansas City, MO: Test Corporation of
America.
Dohrenwend, B. P., & Dohrenwend, B. S. (1969). Social
status and psychological disorder: A causal inquiry.
New York: Wiley.
Dohrenwend, B. S., & Dohrenwend, B. P. (1974). Over-
view and prospects for research on stressful life
events. In B. S. Dohrenwend & B. P. Dohrenwend
(Eds.), Stressful life events: Their nature and events
(pp. 313–331). New York: Wiley.
House, J. S. (1981). Work stress and social support.
Reading, MA: Addison-Wesley.
House, J. S. (1987). Social support and social structure.
Sociological Forum, 2, 135–146.
House, J. S., Umberson, D., & Landis, K. (1988). Struc-
tures and processes of social support. Annual Review
of Sociology, 14, 293–318.
Husaini, B., Neff, J. A., Harrington, S. B., Hughes, M.
D., & Stone, R. H. (1980). Depression in rural com-
munities: Validating the CES-D scale. Journal of
Community Psychology, 8, 20–27.
Kellam, S. G., Adams, R., Brown, C. H., & Ensminger,
M. (1982). The long-term evolution of the family
structure of teenage and older mothers. Journal of
Marriage and the Family, 44, 539–554.
Furstenburg, F. F. (1976). The social consequences of
teenage childbearing. New York: Free Press.
Furstenburg, F. F., Lincoln, R., & Menken, J. (Eds.).
(1981). Teenage sexuality, pregnancy, and childbear-
ing. Philadelphia: University of Pennsylvania Press.
Furstenburg, F. F., Brooks-Gunn, J., & Morgan, S. P.
(1987). Adolescent mothers and their children in later
life. Family Planning Prospectives, 19, 142–151.
Geronimus, A., & Korenman, S. (1993). The socioeco-
nomic costs of teenage childbearing: Evidence and
interpretation. Demography, 30, 281–290.
Gore, S., & Colten, M. E. (1991). Gender, stress, and
distress. In J. Eckenroke (Ed.), The social context of
coping (pp. 139–163). New York: Plenum Press.
Gottlieb, B. H. (1981). Social networks and social sup-
port in community mental health. In B. H. Gottlieb
(Ed.), Social networks and social support (pp. 11–
42). Beverly Hills, CA: Sage.
Gottlieb, B. H. (1983). Social networks and social sup-
port. Beverly Hills, CA: Sage.
Gove, W. (1972). The relationship between sex roles,
mental illness and marital status. Social Forces, 51,
34–44.
Gurin, P., Gurin, G., & Morrison, B. M. (1978). Personal
and ideological aspects of internal and external con-
trol. Social Psychology Quarterly, 41, 175–196.
Henderson, S., Byrne, D. G., & Duncan-Jones, P.
(1981). Neurosis and the social environment. New
York: Academic Press.
Hirschfeld, Robert M. A., G. L. Klerman, P. Chodoff,
S. Korchin, and J. Barrett. (1976). Dependency-self-
esteem-clinical depression. Journal of the American
Academy of Psychoanalysis, 4, 373–388.
Kellam, S. G., Ensminger, M., & Turner, R. J. (1977).
Family structure and the mental health of children:
Concurrent and longitudinal community-wide studies.
Archives of General Psychiatry, 34, 1012–1022.
Kessler, R. C. (1979). A strategy for studying differ-
ential vulnerability to the psychological consequences
of stress. Journal of Health and Social Behavior, 20,
100–108.
Kessler, R. C. (1992). Perceived support and adjustment
to stress: Methodological considerations. In H. O. F.
Veiel & U. Baumann (Eds.), The meaning and mea-
surement of social support (pp. 259–271). New York:
Hemisphere.
Kessler, R. C., & Cleary, P. D. (1980). Social class and
psychological distress. American Sociological Re-
view, 45, 463–478.
Klepinger, D., Lundberg, S., & Plotnick, R. D. (1998).
Teen childbearing and human capital: Does timing
matter? (Working Paper No. 98–2). Seattle, WA: Uni-
versity of Washington, Seattle Population Research
Center.
Lefcourt, H. M. (1976). Locus of control: Current
trends on theory and research. New York: Erlbaum.
Masserman, J. H. (1970). Preface: An historical review
of the psychodynamic theories of affect. In J. H. Mas-
serman (Ed.), Science and psychodynamics (pp. viii-
xviii). New York: Grune & Stratton.
McCutcheon, A. L. (1987). Latent class analysis (Sage
University Paper Series on Quantitative Applications
in the Social Sciences, Series No. 07–064). Beverly
Hills, CA: Sage.
Mirowsky, J., & Ross, C. E. (1989). Psychiatric diag-
nosis as reified measurement. Journal of Health and
Social Behavior, 30, 11–25.
Monroe, S. M. (1992). Life events assessment: Current
practices, emerging trends. Clinical Psychology Re-
view, 2, 435–453.
Hirschfeld, R. M. A., Klerman, G. L., Gough, H. G.,
Barrett, J., Korchin, S., & Chodoff, P. (1977). A mea-
sure of interpersonal dependency. Journal of Person-
ality Assessment, 41, 610–618.
Hoffman, S. D., & Foster, E. M. (1997). Economic cor-
Murphy, E., & Brown, G. W. (1980). Life events, psy-

Social Contingencies in Mental Health
791
chiatric disturbance and physical illness. British Jour-
nal of Psychiatry, 136, 326–338.
National Survey of Family Growth. (1982). Working
women and childbearing. Washington, DC: U.S. Bu-
reau of the Census.
Nolen-Hoeksema, S. (1987). Sex differences in unipolar
depression: Evidence and theory. Psychological Bul-
letin, 2, 259–282.
Parker, G., Tupling, H., & Brown, L. B. (1979). A pa-
rental bonding instrument. British Journal of Medical
Psychology, 52, 1–10.
synthesis of the literature. In H. Kaplan (Ed.), Psy-
chosocial stress: Trends in theory and research (pp.
33–103). New York: Academic Press.
Thoits, P. A. (1987). Gender and marital status differ-
ences in control and distress: Common stress versus
unique stress explanations. Journal of Health and So-
cial Behavior, 28, 7–22.
Turner, R. J. (1992). Measuring social support: Issues
of concept and method. In H. O. F. Veiel & U. Bau-
mann (Eds.), The meaning and measurement of social
support. New York: Hemisphere.
Pearlin, L. I., Lieberman, M. A., Menaghan, E. G., &
Mullan, J. T. (1981). The stress process. Journal of
Health and Social Behavior, 22, 337–356.
Pearlin, L. I., & Schooler, C. (1978). The structure of
coping. Journal of Health and Social Behavior, 19,
2–21.
Rabkin, J. G., & Struening, E. I. (1976). Life events,
stress, and illness. Science, 194, 1013–1020.
Radloff, L. (1977). The CES-D scale: A self report de-
pression scale for research in the general population.
Applied Psychological Measurement, 1, 385–401.
Roberts, R. E., & Vernon, S. W. (1983). The Center for
Epidemiologic Studies Depression Scale: Its use in a
community sample. American Journal of Psychiatry,
140, 41–46.
Turner, R. J., Frankel, B. G., & Levine, D. M. (1983).
Social support: Conceptualization, measurement and
implications for mental health. In J. Greenley (Ed.),
Community and mental health (Vol. 3). Greenwich,
CT: JAI Press.
Turner, R. J., Grindstaff, C. F., & Phillips, N. (1990).
Social support and outcome in teenage pregnancy.
Journal of Health and Social Behavior, 31, 43–57.
Turner R. J., & Lloyd, D. A. (1995). Lifetime traumas
and mental health: The significance of cumulative ad-
versity. Journal of Health and Social Behavior, 36,
360–376.
Turner, R. J., & Marino, F. (1994). Social support and
social structure: A descriptive epidemiology. Journal
of Health and Social Behavior, 35, 193–212.
Turner, R. J., & Noh, S. (1983). Class and psychological
vulnerability among women: The significance of so-
cial support and personal control. Journal of Health
and Social Behavior, 24, 2–15.
Ross, C. E., & Mirowsky, J. (1989). Explaining the so-
cial patterns of depression: Control and problem solv-
ing—or support and talking? Journal of Health and
Social Behavior, 30, 206–219.
Rotter, J. B. (1966). Generalized expectancies for inter-
nal vs. external control of reinforcement. Psycholog-
ical Monographs, 80, 1–28.
Sarason, I. G., Johnson, J. H., & Siegel, J. M. (1978).
Assessing the impact of life changes: Development
of the life experiences survey. Journal of Consulting
and Clinical Psychology, 46, 932–946.
Sarason, I. G., & Sarason, B. R. (Eds). (1985). Social
support: Theory, research, and applications. Boston,
MA: Martinus Nijhoff.
Seeman, M. (1959). On the meaning of alienation.
American Sociological Review, 24, 783–791.
Seligman, M. E. P. (1975). Helplessness: On depression,
development and death. San Francisco: Freeman.
Seligman, M. E. P., & Maier, S. F. (1967). Failure to
escape traumatic shock. Journal of Experimental Psy-
chology, 74, 1–9.
Turner, R. J., & Noh, S. (1988). Physical disability and
depression: A longitudinal analysis. Journal of
Health and Social Behavior, 29, 23–27.
Turner, R. J., Wheaton, B., & Lloyd, D. A. (1995). The
epidemiology of social stress. American Sociological
Review, 60, 104–124.
Vaux, A. (1988). Social support: Theory, research, and
intervention. New York: Praeger.
Weissman, M. M., Sholomskas, D., Pottenger, M., Pru-
soff, B. A., & Locke, B. A. (1977). Assessing de-
pressive symptoms in five psychiatric populations: A
validation study. American Journal of Epidemiology,
106, 203–214.
Wethington, E., & Kessler, R. C. (1986). Perceived sup-
port, received support, and adjustment to stressful life
events. Journal of Health and Social Behavior, 27,
78–89.
Spivak, H., & Weitzman, M. (1987). Social barriers
faced by adolescent parents and their children. Jour-
nal of the American Medical Association, 258, 1500–
1504.
Stiffman, A. R., Powell, J., Earls, F., & Robins, L. N.
(1990). Pregnancies, childbearing, and mental health
problems in adolescents. Youth and Society, 21, 483–
495.
Tausig, M. (1982). Measuring life events. Journal of
Health and Social Behavior, 23, 52–64.
Thoits, P. A. (1983). Dimensions of life events that in-
fluence psychological distress: An evaluation and
Wheaton, B. (1983). Stress, personal coping resources,
and psychiatric symptoms: An investigation of inter-
active models. Journal of Health and Social Behav-
ior, 24, 208–29.
Wheaton, B. (1991, August). The specifications of
chronic stress: Models and measurement. Paper pre-
sented at the Annual Meeting of the Society for the
Study of Social Problems, Cincinnati, OH.
Wheaton, B. (1994). Sampling the stress universe. In W.
R. Avison & I. H. Gotlib (Eds.) Stress and mental
health: Contemporary issues and prospects for the
future. New York: Plenum Press.