ARTICLE IN PRESS

Social Science & Medicine 61 (2005) 2177–2189

www.elsevier.com/locate/socscimed

Longevity following the experience of parental divorce

Leslie R. Martina,, Howard S. Friedmanb, Kathleen M. Clarkb, Joan S. Tuckerc
a
Department of Psychology, La Sierra University, 4500 Riverwalk Parkway, Riverside, CA 92515-8247, USA
b
Department of Psychology, University of California, Riverside, CA 92521-0426, USA
c
RAND Corporation, 1776 Main Street, Santa Monica, CA 90407-2138, USA

Available online 3 June 2005

Abstract

An archival prospective design was used to study mediating and moderating variables for the association between
parental divorce and increased mortality risk, using a sub-group (n ¼ 1183) of individuals from the US Terman Life
Cycle Study covering the period 1921–2000. In childhood, both socioeconomic status (SES) and family psychosocial
environment were related to parental divorce but did little to explain its effects. The higher mortality risk associated
with experiencing parental divorce was ameliorated among individuals (especially men) who achieved a sense of
personal satisfaction by mid-life. Behaviorally, smoking was the strongest mediator of the divorce-mortality link. This
study extends previous work on the long-term effects of parental divorce and reveals some reasons why the stress of
parental divorce in childhood need not necessarily lead to negative later-life outcomes.
r 2005 Elsevier Ltd. All rights reserved.

Keywords: Parental divorce; Mortality risk; Longevity; USA

Introduction Ingersoll-Dayton, 2004) are all predictive of poor

Unstable families and disruptive home environments
can be damaging to children, both at the time they
occur, and years into the future. Repetti, Taylor, and
Seeman (2002) summarize a large body of work
indicating that children in ‘‘risky families’’, those
characterized by conflict, aggression and lack of
nurturance, are vulnerable to a host of physical and
mental health problems. Regarding physical health,
researchers have found that abuse in childhood (Walker
et al., 1999); family conflict and aggression (Lundberg,
1993; Mechanic & Hansell, 1989; Montgomery, Bartley,
& Wilkinson, 1997); and neglect (Wickrama, Lorenz, &
Conger, 1997; Gottman & Katz, 1989; Russek &
Schwartz, 1997; Shaw, Kraus, Chattes, Connell, &
Corresponding author. Tel./fax: +1 951 785 2454.
E-mail address: Lmartin@LaSierra.edu (L.R. Martin).
outcomes.
The negative effects on psychological and physical
health associated with the particular stress of the divorce
of one’s parents during childhood are well documented
(Amato, 2001; Amato & Keith, 1991a, b; Cherlin et al.,
1991; Emery, 1999; Frustenberg & Teitler, 1994;
Hetherington, Bridges, & Insabella, 1998; Tucker et
al., 1997; Wallerstein, 1991). Our own work using
prospective archival data found a striking effect of
parental divorce on mortality risk across decades of life:
on average, children from divorced families died 4 years
earlier compared to their peers from non-divorced
households (Schwartz et al., 1995). That study also
found that parental divorce was the primary early life
social predictor of life-span mortality risk and appeared
independent of childhood personality. The present paper
follows up on that finding.
It is not the case, however, that all children of parental
divorce suffer the same increases in risk. Many

0277-9536/$ - see front matter r 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.socscimed.2005.04.027
 ARTICLE IN PRESS
2178 L.R. Martin et al. / Social Science & Medicine 61 (2005) 2177–2189
participants who experienced parental divorce in the
Schwartz et al. (1995) study did not die earlier than
average, suggesting that these individuals mitigated their
risk and did not embark on a path to increased
vulnerability. In much the same way that Repetti et al.
(2002) propose that a combination of environmental risk
factors and their physiological correlates will sometimes
lead to deleterious health outcomes, a combination of
mediating and/or moderating factors may allow certain
individuals to withstand or even flourish in the face of a
traumatic event such as parental divorce. At the
psychophysiological level, there is variability in the
degree to which exposure to stress creates dysregulation
in responses of the sympathetic-adrenomedullary sys-
tem, the hypothalamic-pituitary-adrenocortical axis,
and the serotonergic system (Repetti et al., 2002;
Kaufman et al., 1998; Koob, Sanna, & Bloom, 1998;
Rosen & Schulkin, 1998). Similarly, at the socio-
behavioral level, there is variability in the extent to
which early stresses lead to health-impairing behaviors,
whether they be detrimental coping strategies (such as
substance abuse) or lowered levels of achievement
leading to further stressors (e.g., economic difficulties
after dropping out of school) (Frustenberg & Teitler,
1994; Repetti et al., 2002; Tucker et al., 1997).
The present study thus addresses the question of
which life pathways lead to health and well-being versus
psychological maladjustment and premature mortality,
in the face of parental divorce. On the one hand, a
diathesis-stress model would predict that parental
divorce, in combination with other risk factors, may
initiate or prime a developmental process whose
eventual outcome is premature death. Conversely, the
stress of parental divorce combined with salutary
variables may produce a strengthening experience (Park,
1998), consequently reducing the health risks. For
example, coping with parental divorce may provide
children with opportunities for the development of
skills, mastery, and personal relationships that promote
thriving (e.g., Carver, 1998; Ickovics & Park, 1998). A
number of such mediating and moderating variables are
suggested by past research, including family variables,
socioeconomic variables, personal and behavioral vari-
ables, and social variables.
Family environment
Interparental conflict and poor parenting skills con-
sistently appear as negative correlates of psychological
well-being for children from broken homes (Emery &
Forehand, 1994; Shaw & Emery, 1987; Shaw, Emery, &
Tuer, 1993). Block, Block, and Gjerde (1988) reported
that prior to the actual divorce, fathers in their study
tended to disengage from their relationships with their
children and mothers often acted resentful toward their
children.
After divorce, changes in contact with the non-
custodial parent represent a significant challenge for
most children (Emery & Forehand, 1994), but a strong,
positive relationship with one competent parent seems to
buffer children from a negative or absent relationship
with the other parent (Amato, 2001; Deater-Deckard &
Dunn, 1999; Emery & Forehand, 1994; Fauber, Fore-
hand, Thomas, & Wierson, 1990; Forehand, Thomas,
Wierson, Brody, & Fauber, 1990; Hetherington et al.,
1998; Rutter, 1987; Zill, Morrison, & Coiro, 1993).
Further, the stress of parental divorce seems to be
experienced differently by boys and girls, and correlates
of their successful adaptation may also differ. For
example, boys who adapt well often come from homes
that combine structure and rules with emotional
expressiveness, while the most important factor for girls
is a household with reliable support from a caring,
competent adult, especially a female caregiver who
emphasizes risk-taking and independence (Hetherington
& Kelly, 2002; Werner, 1995).
Socioeconomic status (SES)
Divorce generally results in a decline in the family’s
standard of living (Emery & Forehand, 1994), especially
for women and children (Bianchi, Subaiya, & Kahn,
1997), presenting another stressor. Some research
suggests that income level or loss explains only a small
portion of adjustment difficulty for children (Amato &
Keith, 1991b; Shaw & Emery, 1987; Weitoft, Hjern,
Haglund, & Rosen, 2003) but others argue that these
changes are important in explaining differences between
children from divorced vs. intact families (Brown &
Moran, 1997; McLanahan, 1999; McLanahan & Sande-
fur, 1994) and that particularly low economic status may
result in poorer nutrition, lower educational quality, and
increased chronic stress (Amato, 2001; Hetherington,
1993).
The relation of parental divorce to SES may extend
into adulthood, as well, through economically relevant
patterns that begin early in life. For example, children
from divorced families are more likely to choose non-
marital cohabitation and earlier sexual relations than
are children from intact families and are more likely to
drop out of school and achieve a lower level of
education (Frustenberg & Teitler, 1994; Tucker et al.,
1997) which may impact their earning power later on.
Personal characteristics and behaviors
Individual characteristics such as a child’s sex,
temperament, personality, and personal achievement
have been studied in order to determine their harmful or
protective effects when coupled with a trauma such as
parental divorce (Amato, 2001; Clarke-Stewart &
 ARTICLE IN PRESS
L.R. Martin et al. / Social Science & Medicine 61 (2005) 2177–2189
Hayward, 1996; Hetherington et al., 1998; Hetherington
& Clingempeel, 1992; Rutter, 1987).
Children with difficult temperaments are less adap-
table to change and more likely to be recipients of
parental criticism, displaced anger, and anxiety than are
children with easy temperaments (Hetherington, Stan-
ley-Hagan, & Anderson, 1989; see also Kulka &
Weingarten, 1979), whereas children with easy tempera-
ments, normal to high self-esteem, and intelligence are at
an advantage. Some have suggested that divorce may
actually enhance the adaptive skills of children with
these characteristics, increasing their ability to cope with
subsequent challenges (Amato, 2001; Hetherington et
al., 1989). Beyond childhood characteristics, the genetic
factors that maintain stability in individual differences,
such as temperament (Plomin, 1986; Plomin & Nessel-
roade, 1990), foster consistency in the ways that people
engage their environments. Life transitions have been
proposed as important mechanisms for shaping indivi-
dual differences (Healy, 1989) and a high magnitude
event such as parental divorce, interacting with existing
temperament, would likely strengthen certain early
personality inclinations, raising the possibility that
personality may moderate the relationship between
parental divorce and mortality risk.
Life satisfaction, achievement, and marital stability
Upon reaching adulthood, one’s social relations and
personal perceptions are associated with one’s own
earlier experience of parental divorce. For example,
consistent with other research, participants in Terman’s
Life-Cycle study who experienced parental divorce were
themselves more likely to divorce than were those whose
childhood families remained intact (Tucker et al., 1997),
thus decreasing any protective buffer marriage may
provide (Hemstroem, 1996). Furthermore, participants
who divorced by midlife had a significantly higher
mortality risk than those who did not divorce (Tucker,
Friedman, Wingard, & Schwartz, 1996). Individuals
who experience traumatic childhood events have also
been reported to have lower life satisfaction, at least
partly because they tend to pursue less education and
therefore be less likely to achieve their goals (Royse,
Rompf, & Dhooper, 1993).
The present study utilizes a rich set of lifetime data
(now nearly 8 decades) to examine how a key set of
variables might explain an individual’s propensity for
long-term successful adaptation to a highly stressful
event—the divorce of one’s parents during childhood.
This study significantly expands our previous work by
drawing on more complete death information (through
the year 2000), including adult personality traits as
potential moderating factors, and assessing the plausi-
bility of a modified diathesis-stress model in which
factors that may not be particularly harmful or
2179
protective in the absence of a major stressor become
important predictors of long-term mortality risk in its
presence.
Method
Participants
The data in this study come primarily from the
Terman Life-Cycle Study, begun in 1921 by Lewis
Terman, and supplemented by information (especially
death certificates) gathered by Friedman and colleagues
(e.g., Friedman et al., 1995). This is the only data set we
know of that has detailed psychosocial and longevity
information across more than 7 decades. The original
sample consisted of 856 boys and 672 girls who were
recruited from California schools after being identified
by their teachers as gifted. All participants recruited had
a Stanford-Binet IQ score of at least 135. The average
year of birth for participants was 1910 and their mean
age at the onset of the study was 11.8 years for boys and
11.1 years for girls. The sample is homogeneous in terms
of ethnicity, consisting mostly of individuals from White
(99%) middle class families (Terman & Oden, 1959).
Participants (and sometimes parents, teachers, spouses,
or interviewers) completed questionnaires every 5–10
years, beginning in 1921, on personality, family life,
education, career, interests, and accomplishments.
In the present study, the following exclusions were
made: (1) 155 participants (10%) were excluded because
they were not of school age when data collection began
(that is, they were born prior to 1904 or after 1915); (2)
190 participants (12%) died, dropped out of the study,
were lost to follow-up prior to 1950 (the year during
which the most recent adult variables used in this study
were collected), or were missing information on all of the
variables from 1940 and 1950. This is a remarkably
small loss rate across many decades, and participants
who were lost from the Terman study are not known to
differ in any systematic way from those who remained in
the study (Friedman et al., 1993; Sears, 1984). The final
sample for the present study consisted of 1183 partici-
pants who were analyzed over the period of 1950
through 2000, or until the date of their death (if prior to
2000).1 Of these 1183 individuals, 160 experienced
1
This paper is one of a series developed from our multi-year,
multi-disciplinary project on psychosocial predictors of health
and longevity (supported by NIA; AG08825) using data partly
derived from Terman’s Life-Cycle Study archives (1921–1991),
and partly collected by us as a follow-up to Terman’s study. All
relevant findings are included in each manuscript to the extent
feasible and prior publications from our project are cited when
appropriate; care should be taken not to include overlapping
findings in meta-analyses or other reviews. Note also that
 ARTICLE IN PRESS
2180 L.R. Martin et al. / Social Science & Medicine 61 (2005) 2177–2189
parental divorce prior to age 21 (calculated from
parental reports) and 64% of the sample had died by
2000 (ascertained, except for 67 cases, through national
death indexes and collection of death certificates).
Validity of archival measures
Research using archival data involves constructing
new measures based on assessment tools chosen by
others; in some cases, the old measures are not well
validated and may be psychometrically questionable
(e.g., Martin & Friedman, 2000; Strein, 1993; Tomlin-
son-Keasey, 1993). To assess the validity of archival
scales (created by averaging standardized scores from
the original questionnaires), a comparison of these scales
with contemporary measures is advisable, and we
conducted such a study for childhood (1922) and early
adult (1940) personality scales within this data set
(Martin & Friedman, 2000). For the current study,
additional validity testing was performed. This validity
analysis entailed administering selected archival ques-
tions, along with contemporary measures of the same
constructs, to a group of 73 new participants and
assessing the congruence between the measures. Validity
coefficients are reported below as appropriate.
Family environment
A positive family attributes scale included information
from 1940 regarding participants’ childhood attachment
to each parent and ratings from the 1950 questionnaire
regarding the degree of admiration felt for each parent
and the degree of affection, understanding, and help-
fulness each parent had provided the participants during
childhood. Cronbach’s alpha for this scale was .79 and it
was highly associated (r ¼ :70, po:001) with the
Cohesion sub-scale of Moos and Moos’ (1974) Family
Environment Scale.
A negative family attributes scale included data from
1940 regarding childhood conflict the participants had
with each parent; ratings made in 1950 regarding
parental resistance to the participants’ independence,
and related feelings of rebellion and rejection; and
ratings of the amount of familial friction during the
(footnote continued)
sample sizes change from paper to paper, as old data are
refined, new data are gathered, or time periods change. This
paper is a follow-up to two other studies: Schwartz et al. (1995)
on parental divorce during childhood and its relation to life
span mortality risk; and Tucker et al. (1997) which examined
social ties, SES, psychological well being, and health behaviors
in relation to parental divorce and mortality risk from 1950 to
1991. This study assesses additional potentially explanatory
variables (including moderators) and includes new mortality
data through the year 2000.
participants’ childhoods. Cronbach’s alpha for this scale
was .78, and it was correlated .64 (po:001) with the
Conflict sub-scale of Moos and Moos’ (1974) Family
Environment Scale. Family environment, as defined
here, is a perceived (subjective) construct.
Childhood socioeconomic status
Childhood SES evaluated the children’s families based
on father’s occupational status (a 7-point scale ranging
from ‘‘professional’’ to ‘‘unskilled blue collar’’) and both
parents’ educational levels as of 1922 (highest grade
completed, up to 6+ years of postgraduate education).
These three items were standardized then averaged;
Cronbach’s alpha for this scale was .79.2
Adulthood education level
Education level was indicated by cumulative level of
education as of 1950 (highest grade completed, up to 10
years of postgraduate education).3 The average educa-
tion level for this sample was completion of the fourth
year of college (standard deviation of two educational
years).
Personal characteristics and health-relevant behaviors
To better understand individual differences that might
play a role in successful adaptation to parental divorce,
measures of adulthood Conscientiousness, Agreeable-
ness, Extraversion, and Neuroticism were derived and
validated from the 1940 and 1950 assessments (Martin &
Friedman, 2000; Martin, Friedman, & Schwartz, sub-
mitted for publication).
In 1950 alcohol use was assessed and participants were
classified into one of three categories: abstainers (never
take a drink or only on rare occasions), light to
moderate drinkers (never or seldom intoxicated), and
heavier drinkers.4 Smoking information was gathered by
Friedman and colleagues in 1991–1992 (Friedman et al.,
2
Data on income were not available at this assessment.
3
Although information on income was available at this
assessment, a notation appears in Terman’s original codebook
warning of its unreliability as it failed to distinguish between
‘‘no income’’ and other responses such as ‘‘sufficient’’ and also
left out income that was reported as room and board or any
other non-standard compensation.
4
Although moderate alcohol use is sometimes found to be
protective, we found no such effect in this sample, but did find a
threshold effect with no differences between abstainers and
light-moderate drinkers but increased mortality associated with
heavier drinking (Martin, Friedman, Tucker, Schwartz, Criqui,
Wingard, & Tomlinson-Keasey, 1995). Approximately 50% of
both men and women reported light-moderate drinking; fewer
women were heavier drinkers (by half), and fewer men reported
as abstainers (also by half).
 ARTICLE IN PRESS
L.R. Martin et al. / Social Science & Medicine 61 (2005) 2177–2189 2181

1995). If participants were dead or could not be located, Table 1

this information was requested from a close relative (if Correlations of potential mediating and moderating variables
known). Not all participants or families could be with parental divorcea, by sex, in the Terman Cohort
contacted and so the sample size for analyses involving Total Male Female
smoking is smaller and is potentially biased (since those
who smoked heavily and died young were less likely to Childhood (N’s)
be located or to have identifiable close family members, Socioeconomic status (SES)
thus tending to underestimate the mediating effect of (1078/598/480) .09** .10* .07
smoking). Pack-years, a standard indicator of the Positive family attributes
amount of smoking, were calculated as: (total number (965/534/431) .24*** .25*** .23***
Negative family attributes
of years smoked  average number of cigarettes per
(1078/598/530) .04 .03 .07
day smoked during those years)/20 (20 cigarettes per
pack) and ranged from 0 (never smoked) to 180. Body Adulthood (N’s)
mass index (BMI) was calculated from participant Marital stability
weight and height reports in 1940 using Quetelet’s (1039/594/445) .07* .08* .04
formula: BMI ¼ weight in kg/height in m2. Marital satisfaction
(789/449/340) .00 .04 .05
Life satis/achievement
Life satisfaction, achievement, and marital stability
(973/542/431) .07* .10* .05
Neuroticism
Four 1940 variables were used to create the marital (1043/583/460) .03 .03 .05
satisfaction scale. Participants were asked if they ever Extraversion
contemplated divorce, if they ever regretted their (1043/583/460) .05+ .09* .01
marriage, and two questions regarding the happiness Agreeableness
of their marriage (a ¼ .58). Marital stability in 1950 was (1043/583/460) .08** .09* .05
classified by us from least to most stable as follows Conscientiousness
(1 ¼ currently divorced; 2 ¼ currently married but pre- (1043/583/460) .05 .06 .06
Body mass index (BMI)
viously divorced; 3 ¼ consistently married as of 1950).
(1038/572/466) .01 .05 .08+
Because studies suggest that the ability to recognize
Education level
one’s own strengths and accomplishments is adaptive (1091/615/476) .14*** .15*** .14**
(e.g., Werner & Smith, 1992), two measures of subjective Alcohol use
accomplishment were examined. In 1950, participants (1138/643/495) .03 .01 .05
were asked to indicate how well they had lived up to Smoking
their intellectual abilities and how they felt about their (686/355/331) .07+ .03 .11+
present vocation. These items comprised the life
satisfaction/achievement scale (a ¼ .59), which was Note: Correlations are point biserial.
þ
po:10,  po:05,  po:01,  po:001.
correlated (r ¼ :50, po:001) with Diener, Emmons, a
(0,1; no parental divorce ¼ 0).
Larsen, & Griffin’s (1985) Satisfaction with Life Scale.

Analyses
Correlational associations between theory-based po-
tential mediating and moderating variables and parental
divorce were first assessed (Table 1) and then these
variables were tested for relationships with mortality
risk (Table 2) using Cox’s proportional hazards regres-
sions. Hypothesized mediators demonstrating associa-
tions to both parental divorce and mortality risk were
then tested in survival analyses by entering them
simultaneously with the main predictor (parental di-
vorce) into the proportional hazards models (Tables 3
and 4); results were cross-checked using Gompertz
survival equations. The resulting relative hazards (rhs)
estimate the risks associated with a 1-point change in the
predictor variable (controlling for age and sex) and are
the strongest and most appropriate test of differences
between groups in longevity (survival). Mediation was
also tested more formally using Sobel’s test of indirect
effects (Preacher & Hayes, 2004; Sobel, 1982). Last,
variables that were differentially related to the parental
divorce versus no parental divorce groups were further
examined by creating interaction terms that were then
used to test the modified diathesis-stress hypothesis
(Tables 5 and 6).
Results
Parental divorce, mortality risk, and potentially
explanatory variables
The previous study examining mortality risk
and parental divorce in the Terman sample concluded
 ARTICLE IN PRESS
2182 L.R. Martin et al. / Social Science & Medicine 61 (2005) 2177–2189

Table 2
Associations of potential mediating and moderating variables with mortality risk, by sex, in the Terman Cohort

Total Male Female

(b) (rh) (b) (rh) (b) (rh)

Childhood (N’s)
Socioeconomic status (SES)
(1078/598/480) .05 .95 .02 .97 .06 .95
Positive family attributes
(965/534/431) .04 .96 .05 .95 .05 .96
Negative family attributes
(1078/598/530) .01 1.01 .04 1.04 .03 .97

Adulthood (N’s)
Marital stability
(1039/594/445) .27 .76*** .30 .74*** .25 .78**
Marital satisfaction
(789/449/340) .09 .91 .14 .87** .03 1.03
Life satis/achievement
(973/542/431) .21 .81** .23 .79** .16 .85*
Neuroticism
(1043/583/460) .02 .97 .03 .97 .01 .99
Extraversion
(1043/583/460) .01 .99 .02 1.02 .06 .94
Agreeableness
(1043/583/460) .06 .95 .07 .94 .04 .96
Conscientiousness
(1043/583/460) .18 .84** .15 .86** .21 .81**
Body mass index (BMI)
(1038/572/466) .03 1.02 .01 1.00 .03 1.02
Education level
(1091/615/476) .12 .89* .14 .87* .08 .93
Alcohol use
(1138/643/495) .14 1.15* .16 1.17* .08 1.08+
Smoking
(686/355/331) .35 1.42*** .28 1.32*** .37 1.45***
þ   
Note: po:10, po:05, po:01, po:001.
Beta coefficients and relative hazards are interquartile except when the variable is categorical. Relative hazards above 1.0 indicate an
increase in risk; those below 1.0 indicate protection.

that participants who experienced parental divorce
prior to 21 years of age had a significantly higher
mortality risk as of the year 1991 than those
whose parents remained married (n ¼ 1285, rh ¼ 1.44,
po:01) (Schwartz et al., 1995) and that these risks
were independent of childhood personality. This
relative hazard indicated that the experience of
parental divorce was associated with an appro-
ximately 44% increase in mortality risk across the
lifespan.
In the present 10-year follow-up study, death data
were gathered for the 1990s, and the association between
experiencing parental divorce in childhood and a
subsequently increased mortality risk was confirmed
for the period of 1950 through 2000 (n ¼ 1183,
rh ¼ 1.50, po:001). Patterns of association between
parental divorce status and potential mediating and
moderating variables were next examined; these correla-
tions appear in Table 1.
Some correlations were strong and consistent for both
males and females. Parental divorce was associated with
fewer positive family attributes and lower adult education
level for both men and women. Others demonstrated
significant or marginally significant associations with
parental divorce in the full sample, but failed to reach
statistical significance in both the female and male sub-
samples. Other variables did not demonstrate significant
associations with parental divorce status in the full
sample or for either gender group.
Each variable was also tested for relevance to
mortality risk, and the resulting relative hazards
appear in Table 2. SES, the only hypothesized mediator
 ARTICLE IN PRESS
L.R. Martin et al. / Social Science & Medicine 61 (2005) 2177–2189 2183

Table 3 Table 4
Survival analyses: examinations of variables potentially mediat- Survival analyses: examinations by sex of variables potentially
ing the parental divorce to mortality risk association mediating the parental divorce to mortality risk association

Models b rh p Models b rh p

Sex .257 .77 o.001 Male

Parental divorce (n ¼ 1183) .403 1.50 o.001 Parental divorce .375 1.46 o.01
Education level (n ¼ 615) .261 .77 o.001
Sex .215 .81 o.01
Parental divorce .319 1.38 o.01 Parental divorce .376 1.46 o.01
Marital stability (n ¼ 1039)a .231 .79 o.01 Marital stability (n ¼ 594) .245 .78 o.01
Sex .224 .80 o.01 Parental divorce .237 1.27 o.06
Parental divorce .307 1.36 o.01 Life satis/achievement (n ¼ 542)a .294 .75 o.03
Life satis/achievement (n ¼ 973)a .065 .94 ns
Female
Sex .318 .73 o.001 Parental divorce .351 1.42 o.05
Parental divorce .334 1.40 o.01 Smoking (n ¼ 331)a .369 1.45 o.01
Education level (n ¼ 1091) .189 .83 o.001
Note: The other initially hypothesized mediators failed to
Sex .212 .81 o.04 demonstrate the necessary associations either with parental
Parental divorce .260 1.30 o.10 divorce or with mortality risk. Beta coefficients and relative
Smoking (n ¼ 686)a .302 1.35 o.001 hazards are interquartile except when the variable is categorical.
Combined models Relative hazards above 1.0 indicate an increase in risk; those
Sex .210 .81 o.04 below 1.0 indicate protection.
a
Parental divorce .211 1.23 ns Sobel test of indirect effects significant, po:05.
Education level .209 .81 o.03
Marital stability .250 .78 o.03
Life satis/achievement .085 .92 ns
Smoking ðn ¼ 665) .305 1.36 o.001

Sex .212 .81 o.01

Table 5
Parental divorce .289 1.34 o.02
Survival analyses: examinations of parental divorce interactions
Education level .195 .82 o.02
as predictors of mortality risk across the life span
Marital stability .232 .79 o.01
Life satis/achievement (n ¼ 942) .105 .90 ns Models b rh p
Note: The remaining initially hypothesized mediators failed to Sex .249 .78 o.001
demonstrate the necessary associations either with parental PD  childhood SES (n ¼ 1078) .313 1.37 o.01
divorce or with mortality risk. Beta coefficients and relative
hazards are interquartile except when the variable is categorical. Sex .223 .80 o.01
Relative hazards above 1.0 indicate an increase in risk; those PD  Pos. family attributes (n ¼ 965) .068 1.07 ns
below 1.0 indicate protection. Being female is protective (i.e.,
Sex .232 .79 o.01
females live longer).
a PD  marital stability (n ¼ 1039) .075 .93 ns
Sobel test of indirect effects significant, po:05.
Sex .223 .80 o.01
PD  life satis/achievement (n ¼ 973) .333 .72 o.02
Sex .220 .80 o.01
from childhood, was not significantly related to mortal-
PD  adult extraversion (n ¼ 1043) .011 1.01 ns
ity risk, but several of the proposed mediators from
adulthood were predictive of mortality risk for both men Sex .222 .80 o.01
and women (marital stability, life satisfaction/achieve- PD  adult agreeableness (n ¼ 1043) .008 1.01 ns
ment, alcohol use, and smoking) and, as with the Sex .270 .76 o.001
correlations, other variables demonstrated significant PD  education level (n ¼ 1091) .014 .99 ns
associations with mortality risk in only one gender
group. Sex .220 .80 o.03
PD  smoking (n ¼ 686) .246 1.28 o.001
Our next step was to conduct survival analyses,
controlling for likely mediating variables (e.g., those Note: PD ¼ parental divorce. Beta coefficients and relative
that were linked to both parental divorce and mortality hazards are interquartile except when the variable is categorical.
risk). Table 3 summarizes these analyses, showing that Relative hazards above 1.0 indicate an increase in risk; those
the association between parental divorce and mortality below 1.0 indicate protection.
 ARTICLE IN PRESS
2184 L.R. Martin et al. / Social Science & Medicine 61 (2005) 2177–2189
Table 6
Examinations by sex of parental divorce interactions as
predictors of mortality risk across the life span
Models b rh
Males
PD  childhood SES (n ¼ 598) .014 1.01
PD  Pos. family attributes (n ¼ 534) .110 1.12
PD  marital stability (n ¼ 594) .036 1.04
PD  life satis/achievement (n ¼ 542) .214 .81
PD  adult extraversion (n ¼ 583) .021 1.02
PD  adult agreeableness (n ¼ 583) .047 1.05
PD  education level (n ¼ 615) .082 .92
Females
PD  Pos. family attributes (n ¼ 431) .096 1.10
PD  body mass index (n ¼ 466) .092 1.10
PD  education level (n ¼ 476) .058 1.06
PD  smoking (n ¼ 331) .334 1.40
Note: PD ¼ parental divorce. Relative hazards above 1.0
indicate an increase in risk; those below 1.0 indicate protection.
risk remained strong and significant when marital
stability, and life satisfaction/achievement were con-
trolled, although the Sobel’s test indicates that each of
these is significantly mediating the parental divorce-
mortality association (pso:05). Education level did not
significantly mediate the relationship (p4:30) and
although smoking reduced the effect to only marginal
significance this seems most likely due to the substantial
decrease in sample size than to a particularly strong
mediational effect. The final two models in Table 3
include simultaneously each of the predictors that
diminished (even minimally) the parental divorce to
mortality risk association in the previous survival
analyses (the last model is identical to the prior except
that it leaves out smoking since many participants were
missing data on this variable). When these variables are
included simultaneously, the parental divorce to mor-
tality association decreases to non-significance (but not
to unity) in the subsample with smoking data.
Because not all associations were the same for males
and females, the survival analyses were also conducted
separately by sex. These analyses are presented in Table
4. For men, parental divorce remained a strong and
significant predictor of mortality risk when controlling
education level and marital stability and neither was a
significant mediator (although both approached signifi-
cance at po:12). The association was reduced substan-
tially, however, when life satisfaction/achievement was
included (becoming only marginally significant), indicat-
ing that differences on these dimensions help to explain
a significant (po:05) portion of the difference in
mortality risk seen for men from divorced vs. non-
divorced families. For women, including smoking in the
survival analysis model did not render the parental
divorce-mortality risk relationship non-significant5
although smoking was a significant (po:05) mediator.
p
The diathesis-stress hypothesis
ns
o.09 Our final step was to test the key hypothesis that some
ns variables might be relatively unimportant in terms of
o.05 predicting mortality for the sample as a whole, but for
ns those who experienced a significant stressor (parental
ns divorce) these variables might become more important,
ns
either placing participants at increased risk or indicating
an opportunity to thrive. Interactions between parental
ns divorce and all variables that were significantly related to
ns it were therefore used as predictors in survival analyses.
ns These results are listed in Tables 5 (whole sample) and 6
o.001 (by sex). For the sample as a whole, the interactions of
parental divorce with childhood SES, life satisfaction/
achievement and smoking were all significant predictors of
mortality risk. Thus, these variables carry more weight in
terms of their relation to mortality risk for those who are
already at risk (the parental divorce sample) than they do
for those from non-divorced households. In other words,
individuals who had experienced parental divorce were
more ‘‘protected’’ by positive effects associated with
higher SES in childhood, better later life satisfaction and
achievement, and not smoking than were those whose
parents had not divorced.
For males, the interaction between parental divorce
and life satisfaction/achievement was significant, indicat-
ing greater risk from low satisfaction/achievement for
men who had experienced parental divorce. The inter-
action between positive family attributes and parental
divorce also approached significance for males, with
boys whose seemingly functional families ended in
divorce experiencing slightly higher risk than those with
less positive familial qualities. Interactions between
parental divorce and the remaining variables were all
non-significant. For women, the only variable that
interacted significantly with parental divorce was smok-
ing, and this association was robust, indicating that
despite the fact that smoking is a significant risk factor
for earlier mortality, it was an even greater risk factor
for women whose parents had divorced.
5
This finding differs from that of our earlier study (Tucker et
al., 1997) where the relative hazard for women with smoking
data was initially smaller (rh ¼ 1.23) and decreased to 1.00
when smoking behavior was controlled. This is likely due to
differences in exclusionary criteria resulting in samples that are
somewhat different in both size and composition, our location
of more death certificates for women, and the 10-year extension
of follow-up in the present study which allowed the life-span
effects of smoking to more fully reveal themselves. Addition-
ally, the 1997 study collapsed pack-years into a less sensitive
four-category variable.
 ARTICLE IN PRESS
L.R. Martin et al. / Social Science & Medicine 61 (2005) 2177–2189
Discussion
Why is it that some individuals have the ability to
withstand a significant stressor over the long term—to
face one of life’s perils and move on to healthy
outcomes? How is it that some children whose parents
divorced grew to old age, whereas others found
themselves at significantly increased mortality risk for
decades to come? This study helps explain avenues to
longevity (as well as likely dead-ends) by examining lives
through time.
We initially identified a number of variables as likely
mediators or moderators of the parental divorce-
mortality risk association based on theory, but not all
of these hypothesized mediators were correlated with
parental divorce status, and only a few indicated an
explanatory pathway by changing the association
between parental divorce and mortality risk when they
were included in the proportional hazards regression
equations. Similarly, not all of the proposed moderators
were supported by the data. Overall, however, several
key findings did emerge.
Beginning in childhood, we failed to find evidence that
positive aspects of family were helpful to children of
divorce, at least in terms of their later mortality risk.
Instead, we found hints that boys with more positive
family attributes were at slightly higher risk than those
with fewer. We cannot ascertain from the available data
whether or not these family attributes refer mostly to the
pre-divorce or post-divorce family or elements of each,
but the findings are consistent with the conclusions of
other researchers (e.g., Amato, Loomis, & Booth, 1995;
Wheaton, 1990) who have found that dismantling a
seemingly functional family may be more traumatic and
have more significant long-term negative effects than the
dissolution of a clearly troubled family. (Also, the
sample is not large enough to ascertain any effects of age
at parental divorce.)
SES in childhood was important as individuals
from divorced homes and with lower childhood
SES lived significantly shorter lives than those
who had higher childhood SES. This effect was
stronger for females. Childhood SES was partially
determined by the mother’s education level and,
after divorce, educated mothers may have been more
likely to gain employment, thus maintaining higher SES
and providing a healthy role model for active coping
with a difficult situation. This kind of example,
especially for daughters, may have increased the like-
lihood of coping with parental divorce as well as later
life stressors (which was an important component of
successful adjustment in Werner’s Kauai study, 1992).
Level of adult education was associated with parental
divorce and with mortality risk, in the expected
directions, but had only a modest role in explain-
ing risk. Perhaps this highly intelligent sample was
2185
well-educated enough that even those with fewer adult
resources, relative to the group, had their needs
sufficiently met. The possibility of such a threshold
effect is deserving of further research.
In terms of personal characteristics and behaviors, we
found that none of the adult personality variables
significantly moderated the relationship between par-
ental divorce and mortality risk. This is consistent with
our previous findings that the risks associated with
parental divorce were independent of childhood person-
ality factors. Of the mediators, smoking behavior was
the strongest (when controlled, substantially decreasing
the risk hazard associated with parental divorce from 1.5
to 1.3).6 Smoking is consistently our most robust
predictor of early mortality and it is worth noting that
even in this reduced sample (n ¼ 686) and alongside this
powerful behavioral predictor, the experience of par-
ental divorce maintains a substantial effect size and
marginal statistical significance.
We also found, through a significant interaction term,
that the risks associated with smoking were accentuated
for those whose parents had divorced, perhaps serving
also as markers for other high-risk health behaviors in
this group, or perhaps because of differing physiological
reactions to the behavior itself. It is also interesting that
even though the sample of individuals with smoking
data is smaller, the proportion of the parental divorce
group for whom we have no smoking data is greater
than that for the sample as a whole; this hints that
smoking itself may have been a particularly likely
behavior in this group, resulting in earlier deaths and
thus making their smoking data more difficult to obtain.
It might be expected that life satisfaction at middle-
age would be predictive of subsequently lowered
mortality risk for all. We found that it makes a
difference only for those from divorced homes; those
who got their lives back on track by mid-life showed no
elevated risk, while those who failed to do so were at
substantially higher risk. For those who had not
experienced parental divorce, satisfaction at mid-life
was unrelated to mortality risk (these group associations
are illustrated in Fig. 1). Males, in particular, who were
resistant to the negative effects of parental divorce were
more satisfied with their vocations and accomplishments
by mid-life (approximately 40 years of age). Perhaps this
satisfaction with life reflects that they had learned
effective coping strategies earlier in life.
Key life patterns and characteristics that predicted
increased life spans of those whose parents divorced did
not significantly predict longevity for the participants
6
Note that the sub-sample of individuals for whom smoking
data are available is only about half the size of the sample as a
whole, and because the smoking data were collected in 1991
they are biased toward those who lived longer, regardless of
their smoking habits.

2186 L.R. Martin et al. / Social Science & Medicine 61 (2005) 2177–2189
0.70
0.60
0.50
Probability of Death
0.40
0.30
0.20
0.10
0.00
40
Fig. 1. Probability of dying at a given age, by life satisfaction/achievement and parental divorce status.
whose parents did not divorce. This interactive process is
consistent with a modified diathesis-stress model in
which some strengths emerge only when the individual is
presented with a stressor. The present study supports the
idea that emotional recovery can occur, strengths and
achievements can emerge, and that early trauma does
not necessarily continue to exert its effects throughout
life.
As in any research that precludes random assignment
to conditions, a limitation of this study is that we cannot
eliminate the possibility that the associations are
explainable by underlying third variables. Some have
suggested that biological factors may provide important
explanatory mechanisms for the link between parental
divorce and subsequent problems—that is, children
from divorced families may experience later difficulties
due to their inherited characteristics; and parenting
problems are due to deficient characteristics of the
parent, as opposed to the actual experience of parental
divorce. The possibility of a constitutional or tempera-
mental contribution is supported by studies that show a
higher concordance for divorce and greater similarity on
low impulse control and antisocial behavior in mono-
zygotic than dizygotic twins (Jockin, McGue, & Lykken,
1996; McGue & Lykken, 1992). We do know, however,
that childhood personality (which certainly has a genetic
ARTICLE IN PRESS
High Life Satisfaction/Achievement
Low Life Satisfaction/Achievement
Parental Divorce
No Parental Divorce
50 60 70 80 90
Age
component) is independent of the experience of parental
divorce in this sample (Schwartz et al., 1995). So,
although the mechanisms for the observed associations
may include underlying factors which we are unable to
assess, the provocative questions this raises serve to
highlight the critical need for ongoing research in this
area.
Few longitudinal studies of psychosocial variables and
health are based on a true random sample of the
population for a variety of reasons including focus, cost,
numbers, availability, and time frame. The important
question of generality that therefore arises should not be
whether any results are fully generalizable (since they
usually are not); rather, the focus should be on the likely
limits on generality and the purposes for which the
results are best employed. This data-set is especially
valuable in helping to avoid Type II errors; that is, it
helps uncover important relationships that might not be
recognizable in cross-sectional studies or in studies of
shorter duration. Further, although the sample’s homo-
geneity on the dimensions of intelligence, social class,
and ethnicity present limitations on generality, there is
also an advantage in that these people had the ability to
understand medical advice, had routine health care, and
did not face stresses associated with racial discrimina-
tion. This allows a clearer focus on the effects of certain
 ARTICLE IN PRESS
L.R. Martin et al. / Social Science & Medicine 61 (2005) 2177–2189
psychosocial variables than would be possible in studies
on more heterogeneous samples. Analyses by Sears
(1984) show that attrition has not significantly changed
the makeup of the group with respect to a variety of
demographic characteristics, including age, income, and
education, and we too find no differential attrition; for
example, survival analyses do not predict who drops
out. Other research confirms that participants in the
Terman sample are similar to other bright, middle-class
cohorts and might most profitably be regarded as a
sample of an important and productive segment of
20th century White US society, but not representative
of the full population (Sears, 1984; Subotnik, Karp, &
Morgan, 1989).
Current societal attitudes and norms regarding
divorce differ from those prevalent when the Terman
cohort experienced their parents’ divorces. Thus, the
impacts of parental divorce on life outcomes, including
health-related outcomes, may differ as well. The precise
salutary behavioral patterns of biopsychosocial recovery
may also differ in other cultures. Nevertheless, the
significant stress felt by parents and children as a result
of the dissolution of the family still prevails in most
cases, and questions of how to best adjust following
divorce remain important social issues.
References
Amato, P. R. (2001). The consequences of divorce for adults
and children. In R. M. Milardo (Ed.), Understanding
families into the new millennium: A decade in review (pp.
488–506). Minneapolis, MN: National Council on Family
Relations.
Amato, P. R., & Keith, B. (1991a). Parental divorce and adult
well-being: A meta-analysis. Journal of Marriage and the
Family, 53, 43–58.
Amato, P. R., & Keith, B. (1991b). Parental divorce and the
well-being of children: A meta-analysis. Psychological
Bulletin, 110, 26–46.
Amato, P. R., Loomis, L. S., & Booth, A. (1995). Parental
divorce, marital conflict, and offspring well-being during
early adulthood. Social Forces, 73, 895–915.
Bianchi, S. M., Subaiya, L., & Kahn, J. (1997). Economic well-
being of husbands and wives after marital disruption. Paper
presented at the annual meeting of the Population Associa-
tion of America, Washington, DC.
Block, J., Block, J. H., & Gjerde, P. F. (1988). Parental
functioning and the home environment in families of
divorce: Prospective and concurrent analyses. Journal of
American Academy of Child and Adolescent Psychiatry, 27,
207–213.
Brown, G., & Moran, P. (1997). Single mothers, poverty, and
depression. Psychological Medicine, 27, 21–33.
Carver, C. S. (1998). Resilience and thriving: Issues, models, &
linkages. Journal of Social Issues, 54, 245–266.
Cherlin, A. J., Frustenberg, F. F., Jr., Chase-Lansdale, P. L.,
Kiernan, K. E., Robins, P. K., Morrison, D. R., & Teitler, J.
2187
O. (1991). Longitudinal studies of effects of divorce on
children in Great Britain and the United States. Science,
252, 1386–1389.
Clarke-Stewart, K. A., & Hayward, C. (1996). Advantages of
father custody and contract for the psychological well-being
of school-age children. Journal of Applied Developmental
Psychology, 17, 239–270.
Deater-Deckard, K., & Dunn, J. (1999). Multiple risks and
adjustment in young children growing up in different family
settings. In E. M. Hetherington (Ed.), Coping with divorce,
single parenting and remarriage: A risk and resiliency
perspective (pp. 47–64). Mahwah, NJ: Erlbaum.
Diener, E., Emmons, R., Larsen, J., & Griffin’s, S. (1985). The
satisfaction with life scale. Journal of Personality Assess-
ment, 49, 71–75.
Emery, R. E. (1999). Marriage, divorce, and children’s adjust-
ment (2nd ed.). Thousand Oaks, CA: Sage.
Emery, R. E., & Forehand, R. (1994). Parental divorce and
children’s well-being: A focus on resilience. In R. J.
Haggerty, et al. (Eds.), Stress, risk, and resilience in children
and adolescents (pp. 100–146). Cambridge, MA: Harvard
University Press.
Fauber, R., Forehand, R., Thomas, A. M., & Wierson, M.
(1990). A mediational model of the impact of marital
conflict on adolescent adjustment in intact and divorced
families: The role of disrupted parenting. Child Develop-
ment, 61, 1112–1123.
Forehand, R., Thomas, A. M., Wierson, M., Brody, G., &
Fauber, R. (1990). Role of maternal functioning and
parenting skills in adolescent functioning following parental
divorce. Journal of Abnormal Psychology, 99, 278–283.
Friedman, H. S., Tucker, J. S., Schwartz, J. E., Tomlinson-
Keasey, C., Martin, L. R., Wingard, D. L., & Criqui, M. H.
(1995). Psychosocial and behavioral predictors of longevity:
The aging and death of the ‘‘Termites’’. American Psychol-
ogist, 50, 69–78.
Friedman, H. S., Tucker, J. S., Tomlinson-Keasey, C.,
Schwartz, J. E., Wingard, D. L., & Criqui, M. H. (1993).
Does childhood personality predict longevity? Journal of
Personality and Social Psychology, 65, 176–185.
Frustenberg, F. F., Jr., & Teitler, J. O. (1994). Reconsidering
the effects of marital disruption. Journal of Family Issues,
15, 173–190.
Gottman, J. M., & Katz, L. F. (1989). Effects of marital discord
on young children’s peer interaction and health. Develop-
mental Psychology, 25, 373–381.
Healy, J. M., Jr. (1989). Emotional adaptation to life
transitions: Early impact on integrative cognitive processes.
In D. M. Buss, & N. Cantor (Eds.), Personality psychology:
Recent trends and emerging directions (pp. 115–127). New
York: Springer-Verlag.
Hemstroem, O. (1996). Is marriage dissolution linked to
differences in mortality risks for men and women? Journal
of Marriage and the Family, 58, 366–378.
Hetherington, E. M. (1993). An overview of the Virginia
longitudinal study of divorce and remarriage with a focus
on early adolescence. Journal of Family Psychology, 7,
39–65.
Hetherington, E. M., Bridges, M., & Insabella, G. M. (1998).
What matters? What does not? Five perspectives on the
 ARTICLE IN PRESS
2188 L.R. Martin et al. / Social Science & Medicine 61 (2005) 2177–2189
association between marital transitions and children’s
adjustment. American Psychologist, 53, 167–184.
Hetherington, E. M., & Clingempeel, G. W. (1992). Coping
with marital transitions: A family systems perspective.
Monographs of the Society for Research in Child Develop-
ment, 57, 1–206.
Hetherington, E. M., & Kelly, J. (2002). For better or for worse:
Divorce reconsidered. NY: Norton.
Hetherington, E. M., Stanley-Hagan, M., & Anderson, E. R.
(1989). Marital transitions. American Psychologist, 44,
303–312.
Ickovics, J. R., & Park, C. L. (1998). Paradigm shift: Why a
focus on health is important. Journal of Social Issues, 54,
237–244.
Jockin, V., McGue, M., & Lykken, D. T. (1996). Personality
and divorce: A genetic analysis. Journal of Personality and
Social Psychology, 71, 288–299.
Kaufman, J., Birmaher, B., Perel, J., Stull, S., Brent, D.,
Trubnick, L., Al-Shabbout, M., & Ryan, N. D. (1998).
Serotonergic functioning in depressed abused children:
Clinical and familial correlates. Biological Psychiatry, 44,
973–981.
Koob, G. F., Sanna, P. P., & Bloom, F. E. (1998). Neuroscience
of addiction. Neuron, 21, 467–476.
Kulka, R. A., & Weingarten, H. (1979). The long-term effects
of parental divorce in childhood on adult adjustment.
Journal of Social Issues, 35, 50–78.
Lundberg, O. (1993). The impact of childhood living conditions
on illness and mortality in adulthood. Social Science &
Medicine, 36, 1047–1052.
Martin, L. R., & Friedman, H. S. (2000). Comparing
personality scales across time: An illustrative study of
validity and consistency in life-span archival data. Journal of
Personality, 68, 85–110.
Martin, L. R., Friedman, H. S., & Schwartz, J. E. Personality
and mortality risk across the lifespan. Health Psychology,
submitted for publication.
Martin, L. R., Friedman, H. S., Tucker, J. S., Schwartz, J. E.,
Criqui, M. H., Wingard, D. L., & Tomlinson-Keasey, C.
(1995). An archival prospective study of mental health and
longevity. Health Psychology, 14, 381–387.
McGue, M., & Lykken, D. T. (1992). Genetic influence on risk
of divorce. Psychological Science, 6, 368–373.
McLanahan, S. (1999). Father absence and the welfare of
children. In E. M. Hetherington (Ed.), Coping with divorce,
single parenting and remarriage: A risk and resiliency
perspective (pp. 117–146). Mahwah, NJ: Erlbaum.
McLanahan, S., & Sandefur, G. (1994). Growing up with a single
parent: What hurts, what helps. Cambridge, MA: Harvard
University Press.
Mechanic, D., & Hansell, S. (1989). Divorce, family conflict,
and adolescents’ well-being. Journal of Health and Social
Behavior, 30, 105–116.
Montgomery, S. M., Bartley, M. J., & Wilkinson, R. G. (1997).
Family conflict and slow growth. Archives of Disease in
Childhood, 77, 326–330.
Moos, R. H., & Moos, B. S. (1974). Manual for the family
environment scale. Palo Alto, CA: Consulting Psychologists
Press.
Park, C. L. (1998). Stress-related growth and thriving through
coping: The roles of personality and cognitive processes.
Journal of Social Issues, 54, 267–277.
Plomin, R. (1986). Development, genetics, and psychology.
Hillsdale, NJ: Erlbaum.
Plomin, R., & Nesselroade, J. R. (1990). Behavioral genetics
and personality change. Journal of Personality, 58, 191–220.
Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS
procedures for estimating indirect effects in simple media-
tion models. Behavior Research Methods, Instruments, and
Computers, 36, 717–731.
Repetti, R. L., Taylor, S. E., & Seeman, T. E. (2002). Risky
families: Family social environment and the mental and
physical health of offspring. Psychological Bulletin, 128,
330–366.
Rosen, J. G., & Schulkin, J. (1998). From normal fear to
pathological anxiety. Psychological Review, 105, 325–360.
Royse, D., Rompf, B. L., & Dhooper, S. S. (1993). Childhood
trauma and adult life satisfaction. The Journal of Applied
Social Sciences, 17, 179–189.
Russek, L. G., & Schwartz, G. E. (1997). Feelings of parental
caring can predict health status in mid-life: A 35-year
follow-up of the Harvard Mastery of Stress study. Journal
of Behavioral Medicine, 20, 1–13.
Rutter, M. (1987). Psychosocial resilience and protective
mechanisms. American Journal of Orthopsychiatry, 57,
316–331.
Schwartz, J. E., Friedman, H. S., Tucker, J. S., Tomlinson-
Keasey, C., Wingard, D. L., & Criqui, M. H. (1995).
Childhood sociodemographic and psychosocial factors as
predictors of mortality across the life-span. American
Journal of Public Health, 85, 1237–1245.
Sears, R. R. (1984). The Terman gifted children study. In S.
Mednick, M. Harway, & K. Finello (Eds.), Handbook of
longitudinal research, Vol. 1 (pp. 398–414). NY: Praeger.
Shaw, B. A., Kraus, N., Chattes, L. M., Connell, C. M., &
Ingersoll-Dayton, B. (2004). Emotional support from
parents early in life, aging, and health. Psychology and
Aging, 19, 4–12.
Shaw, D. S., & Emery, R. E. (1987). Parental conflict and other
correlates of the adjustment of school-age children whose
parents have separated. Journal of Abnormal Child Psychol-
ogy, 15, 269–281.
Shaw, D. S., Emery, R. E., & Tuer, M. D. (1993). Parental
functioning and children’s adjustment in families of divorce:
A prospective study. Journal of Abnormal Child Psychology,
21, 119–134.
Sobel, M. E. (1982). Asymptotic confidence intervals for
indirect effects in structural equation models. In S. Leinhart
(Ed.), Sociological methodology (pp. 290–312). San Francis-
co: Jossey-Bass.
Strein, W. (1993). Structural validity of an academic self-
concept scale: An illustrative case study of large-scale
archival data set use. Measurement and Evaluation in
Counseling and Development, 26, 125–136.
Subotnik, R. F., Karp, D. E., & Morgan, E. R. (1989). High IQ
children at mid-life: An investigation into the general-
izability of Terman’s genetic studies of genius. Roeper
Review, 11, 139–144.
 ARTICLE IN PRESS
L.R. Martin et al. / Social Science & Medicine 61 (2005) 2177–2189
Terman, L. M., & Oden, M. H. (1959). Genetic studies of genius,
V: The gifted group at mid-life. Stanford, CA: Stanford
University Press.
Tomlinson-Keasey, C. (1993). Opportunities and challenges
posed by archival data sets. In D. C. Funder, R. D. Parke,
C. Tomlinson-Keasey, & K. Widaman (Eds.), Studying lives
through time: Personality and development (pp. 65–92).
Washington, DC: APA.
Tucker, J. S., Friedman, H. S., Schwartz, J. E., Criqui, M. H.,
Tomlinson-Keasey, C., Wingard, D. L., & Martin, L. R.
(1997). Parental divorce: Effects on individual behavior and
longevity. Journal of Personality and Social Psychology, 73,
381–391.
Tucker, J. S., Friedman, H. S., Wingard, D. L., & Schwartz, J.
E. (1996). Marital history at mid-life as a predictor of
longevity: Alternative explanations to the protective effect
of marriage. Health Psychology, 15, 94–101.
Walker, E. A., Gelfand, A., Katon, W. J., Koss, M. P.,
Won Korff, M., Bernstein, D., & Russo, J. (1999). Adult
health status of women with histories of childhood abuse
and neglect. The American Journal of Medicine, 107,
332–339.
2189
Wallerstein, J. S. (1991). The long-term effects of divorce
on children: A review. Journal of the American
Academy of Child and Adolescents Psychiatry, 30,
349–360.
Weitoft, G. R., Hjern, A., Haglund, B., & Rosen, M. (2003).
Mortality, severe morbidity, and injury in children living
with single parents in Sweden: A population-based study.
Lancet, 361, 289–295.
Werner, E. E. (1995). Resilience in development. Current
Directions, 4, 81–85.
Werner, E. E., & Smith, R. S. (1992). Overcoming the odds.
Ithaca, NY: Cornell University Press.
Wheaton, B. (1990). Life transitions, role histories, and mental
health. American Sociological Review, 55, 209–223.
Wickrama, K. A. S., Lorenz, F. O., & Conger, R. D. (1997).
Parental support and adolescent physical health status: A
latent growth-curve analysis. Journal of Health and Social
Behavior, 38, 149–163.
Zill, N., Morrison, D. R., & Coiro, M. J. (1993). Long-term
effects of parental divorce on parent-child relationships,
adjustment, and achievement in young adulthood. Journal
of Family Psychology, 7, 91–103.