Oxytocin receptor gene (OXTR) is related

to psychological resources
Shimon Saphire-Bernstein
a
, Baldwin M. Way
b
, Heejung S. Kim
c
, David K. Sherman
c
, and Shelley E. Taylor
a,1
a
Department of Psychology, University of California, Los Angeles, CA 90095;
b
Department of Psychology, Ohio State University, Columbus, OH 43210;
and
c
Department of Psychology, University of California, Santa Barbara, CA 93106
Contributed by Shelley E. Taylor, August 10, 2011 (sent for review May 19, 2011)
Psychological resourcesoptimism, mastery, andself-esteembuffer
the deleterious effects of stress and are predictors of neurophysiolog-
ical and psychological health-related outcomes. These resources have
been shown to be highly heritable, yet the genetic basis for this her-
itability remains unknown. Here, we report a link between the
oxytocin receptor (OXTR) SNP rs53576 and psychological resources,
such that carriers of the A allele have lower levels of optimism,
mastery, and self-esteem, relative to G/G homozygotes. OXTR was
also associated with depressive symptomatology. Mediation analysis
indicates that the effects of OXTR on depressive symptoms may be
largelymediatedbythe inuence of OXTRonpsychological resources.
P
sychological resources refer to individual differences that are
directly predictive of physical and psychological health (13).
The most well-studied of these resources are optimism, mastery,
and self-esteem (46). Optimism refers to the extent to which
people hold favorable expectations about the future (4, 7); as a
dispositional variable, it reects positive expectations across a
broad array of outcomes. Mastery involves the belief that one
can determine ones own behavior, inuence ones environment,
and bring about desired outcomes; it also has a strong disposi-
tional component (8). Self-esteem is a dispositional concept that
refers to a persons overall evaluation of self-worth (9). Previous
research has established that these three resources are closely
interrelated (4, 6, 10) and, both independently and as a cluster,
they are known to buffer the effects of stressful life events and
experimentally manipulated stressors on physiological stress
responses (for reviews, see refs. 2 and 5). Moreover, considerable
research demonstrates that susceptibility to depression and other
forms of psychological distress is lower among individuals high in
optimism (4, 11, 12), mastery (13, 14), and self-esteem (1517).
The model guiding the present research attributes the origins
of psychological resources to developmental and genetic factors
(2, 5). Aspects of the early environment that affect the devel-
opment of these resources include family socioeconomic status
(15, 18, 19), childhood adversities (20), and parental practices
(18, 20). Psychological resources may continue to be inuenced
by life experiences in adolescence, young adulthood, and beyond
(18), but less research is available on this question. Although we
acknowledge the importance of these environmental factors in
the developmental origins of psychological resources, the present
research is primarily concerned with the second primary source:
human genetics.
Twin studies have shown that a large proportion of the variance
in psychological resources is heritable (2125). The extant research
suggests a narrow range for the heritability of optimism, with in-
dependent reports ranging from 20% (21) to 36% (24). Estimates
for the heritability of self-esteem range more widely, from a low of
29% (23) to a high of 73% (22). Research on the behavioral ge-
netics of mastery has not, to our knowledge, been conducted.
Despite evidence for the heritability of psychological resources, the
genetic bases of this heritability have yet to be elucidated.
The oxytocin system appears to play a role in socioemotional
functioning and positive emotion (2629). Located on chromo-
some 3p25.3, OXTR codes for the oxytocin receptor, the receptor
by which the neurohormone oxytocin exerts a range of effects
throughout the body and the brain (30; for review, see ref. 31).
Several studies report associations of the SNP rs53576, located in
intron 3 of OXTR, with stress-related and psychological traits, with
the majority suggesting that carriers of the A allele (i.e., A/G and
A/A genotypes) have an increased sensitivity to stress, reduced
social skills, and more negative mental health outcomes relative to
individuals with two copies of the G allele (3238; but see ref. 39).
For example, Lucht et al. (35) found that adult A/A homozygotes
had lower self-reported positive affect and higher negative affect
and loneliness. Another study found that G/G homozygotes out-
performed carriers of the A allele of rs53576 on the ability to
detect emotion from pictures of human faces showing only the
eyes (37). Other research has found that among mothers of re-
cently born infants, OXTR rs53576 A-allele carriers demonstrated
lower maternal sensitivity relative to G-allele carriers (32). A
subsequent study of reproductive-age women without children
revealed greater heart-rate response to baby cries in GG women
relative to A-allele carriers (36), which was interpreted as in-
dicating greater sensitivity to the babys needs and emotional state
among GGs. Taken together, these ndings suggest that OXTR
rs53576 may play a role in human social behaviors and psycho-
logical resources (see also refs. 2629).
Although these are intriguing and consistent ndings, it is
difcult to assess their meaning without a functional account of
how variation at the rs53576 locus inuences behavior via the
circuits and networks of the brain. Although not complete, a
picture has begun to emerge regarding the consequences of
variation at this SNP for the structure and function of brain areas
involved in a wide range of cognitive and social-cognitive pro-
cesses. For example, Tost et al. (38) found that A-allele carriers
had reduced volume of the hypothalamus and increased struc-
tural (correlated structure sizes) and functional (coactivation in
response to faces picturing emotions) connectivity of the hypo-
thalamus to both the amygdala and the dorsal anterior cingulate
cortex (dACC). These brain regions have been tied previously to
stress responses (10, 40, 41), which have also been associated
with rs53576 (37) and vulnerability to social distress (dACC)
(4042). As such, these ndings provide suggestive evidence for
a biological stress-related phenotype associated with OXTR
rs53576. The link between OXTR and the dACC is especially
interesting in light of ndings that self-esteem attenuates dACC
activation in response to social rejection (42). The connection to
the amygdala is consistent with previous research showing that
intranasal administration of oxytocin modulates activation of the
amygdala to emotional and neutral faces (4246). Petrovic et al.
(46) also found an effect of intranasal oxytocin on dACC acti-
vation. Thus, a growing literature supports a role for OXTR in
the structure and function of the amygdala, hypothalamus, and
Author contributions: S.S.-B., B.M.W., H.S.K., D.K.S., and S.E.T. designed research; B.M.W.
and S.E.T. performed research; S.S.-B. and B.M.W. analyzed data; and S.S.-B. and S.E.T.
wrote the paper.
The authors declare no conict of interest.
1
To whom correspondence should be addressed. E-mail: taylors@psych.ucla.edu.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1113137108/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1113137108 PNAS Early Edition | 1 of 5
P
S
Y
C
H
O
L
O
G
I
C
A
L
A
N
D
C
O
G
N
I
T
I
V
E
S
C
I
E
N
C
E
S
dACC, structures of known relevance to psychological resources,
and their inuence on stress processes (10, 4042).
On the basis of this literature, we hypothesized that variation in
OXTR at rs53576 may be associated with psychological resources,
such that self-esteem, optimism, and mastery would be higher
among G/G homozygotes relative to carriers of the A allele. Ad-
ditionally, because the A allele may confer risk for lower psy-
chological resources, which are protective against stress and
negative self-views, we predicted that OXTR A allele carriers
would have higher levels of depressive symptomatology, which
may potentially be mediated by (lack of) psychological resources.
Results
The distribution of genotypes for OXTR was 108 G/Gs (33.1%),
153 A/Gs (46.9%), and 65 A/As (19.9%). This distribution does
not deviate from the Hardy-Weinberg equilibrium,
2
(1) = 0.65,
P = 0.419.
To test the relation between OXTR and psychological re-
sources and depressive symptomatology, we computed the scale
means for the three resource measures and a sum-score for the
depression measure (the Beck Depression Inventory-IA, BDI-
IA) (47). In addition, we created factor scores for psychological
resources and depressive symptomatology on the basis of an
exploratory factor analysis (see Methods, SI Text, and Table S1
for more details). Descriptive statistics of the measures are
presented in Table 1, along with the correlations between the
different scales and factors.
Tests of the hypothesized association between OXTR and psy-
chological resources are presented in Table 2. The results showthat
A-allele carriers (M = 3.53, SD = 0.85) were lower than G allele
homozygotes (M = 3.74, SD = 0.79) in optimism [t(324) = 2.17,
P = 0.031, d = 0.26, R
2
= 0.016], mastery [M = 3.06, SD= 0.47 vs.
M = 3.17, SD = 0.44), t(324) = 2.03, P = 0.043, d = 0.24,
R
2
= 0.014], and self-esteem [M = 3.20, SD = 0.54 vs. M = 3.38,
SD= 0.46), t(324) = 2.91, P = 0.004, d = 0.34, R
2
= 0.029]. OXTR
was also signicantly related to the resources factor, such that A-
allele carriers had lower levels of psychological resources relative
to G-allele homozygotes: t(324) = 3.07, P = 0.002, d = 0.36,
R
2
= 0.032. OXTR rs53576 appears to account for between 4.4%
(0.016/0.360) and 8% (0.016/0.200) of the genetic variance in op-
timism. Likewise, OXTR may account for between 4% (0.029/
0.730) and 10% (0.029/0.290) of the genetic variance in self-
esteem. A-allele carriers also had signicantly greater levels of
depressive symptomatology for the measure of depression [for
BDI-IA sum, t(277.6) = 2.56, P = 0.011, d = 0.27, R
2
= 0.018; for
depressive symptomatology factor, t(266.6) = 2.51, P = 0.013,
d = 0.27, R
2
= 0.018].Variances for the BDI-IA sum and factor
score were signicantly unequal between G/Ghomozygotes and A-
allele carriers according to Levenes test for equality of variances.
The t-statistics, df, and P values provided are corrected for this
violation of the equality of variances assumption.
The allelic distribution differed between the Asian and non-
Asian participants [
2
(2) = 51.6, P < 0.0001], with lower prev-
alence of G-allele homozygotes among Asians and conversely
lower prevalence of A-allele homozygotes among non-Asians
(Table S2). This difference is consistent with past reports in the
literature (33, 34, 48). In addition, levels of psychological
resources were lower and levels of depressive symptomatology
were higher in the Asian participants (Table S3). The overall
pattern of relations between OXTR genotypes and psychological
resources, however, was the same for Asians and non-Asians (SI
Text and Table S4). Additional analyses for ethnicity and anal-
yses for sex appear in the SI Text and Table S5.
Mediation Analysis. The effect of OXTR on depressive symptom-
atology may be indirectly mediated by psychological resources,
which are known to be protective against depression (4, 6, 11
17). We tested this possibility using mediation analysis (49). The
hypothesized mediation model is presented in Fig. 1.
Simultaneous regression of the depressive symptomatology
factor on OXTR and the psychological resources factor suggested
that resources may fully mediate the effect of OXTR on distress.
Specically, the effect of OXTR on depressive symptomatology
dropped from = 0.127 (P = 0.021) without resources in the
model to = 0.050 (P = 0.319) with resources in the model, and
the effect of resources on depressive symptomatology was highly
signicant, = 0.462 (P < 0.001). The OXTR A allele carriers
had lower levels of psychological resources, and the lower level
of resources was in turn associated with higher levels of de-
pression. These results are summarized in Fig. 1. The indirect
effect of OXTR on distress as mediated by resources was sig-
nicant (Sobels Z = 2.91, P = 0.004). A model-based bootstrap
(49) using 1,000 samples from the original data revealed a mean
indirect effect of 0.0835 (SE = 0.0297), with a 95% condence
interval of 0.0295, 0.1442. Because the value 0 lies outside this
interval, we can reject the null hypothesis of no indirect effect.
As such, the ndings are supportive of the hypothesis that OXTR
affects depressive symptomatology by means of its inuence on
psychological resources.
An alternative model testing whether the relation of OXTR to
psychological resources is mediated by depressive symptomatol-
ogy suggested partial mediation, as the effect of the OXTRAallele
on psychological resources remained signicant ( = 0.110,
P = 0.026; from = 0.168, P = 0.002) when depressive symp-
tomatology was added to the model, and yet the indirect effect
remained signicant (Sobels Z = 2.26, P = 0.024). The OXTR
A-allele carriers had higher levels of depressive symptomatology,
which was in turn associated with lower levels of psychological
resources. A model-based bootstrap using 1,000 samples from
the original data revealed a mean indirect effect of 0.0610
(SE = 0.0244), with a 95% condence interval of 0.0137,
0.1081. The fact that the value 0 lies outside this interval allows
us to reject the null hypothesis of no indirect effect. Thus, al-
Table 1. Descriptive statistics for measures of psychological resources and depression
Correlations
Scale No. of Items Range Mean SD 1 2 3 4 5
Self-esteem 10 0.89 1.54.0 3.26 0.52 1.00
Mastery 7 0.76 1.74.0 3.09 0.46 0.50 1.00
Optimism 6 0.82 1.25.0 3.60 0.84 0.54 0.50 1.00
BDI-IA, sum 20* 0.84 0.042.0 6.06 5.60 0.54 0.43 0.47 1.00
Psychological resources factor 21

0.91 3.31.9 0.00 1.00 0.90 0.73 0.80 0.58 1.00

BDI-IA, factor 14

0.79 1.06.6 0.00 1.00 0.43 -0.39 0.39 0.96 0.47

*The BDI-IA contains 21 items; however, one item (suicidality) was dropped at the request of the Institutional Review Board.

On the basis of the exploratory factor analysis (Table S1) (48), 21 items were used to compute the resources factor.

On the basis of the exploratory factor analysis (Table S1) (48), 14 items were used to compute the BDI-IA factor.
2 of 5 | www.pnas.org/cgi/doi/10.1073/pnas.1113137108 Saphire-Bernstein et al.
though the indirect effect is smaller than in the previous analysis,
there is evidence for partial mediation of the effect of OXTR on
psychological resources by depressive symptomatology. However,
because the direct effect of OXTR on resources is still signicant
when depressive symptomatology is in the model, the mediation is
modest, whereas the results from the previous model suggest
full mediation of the effect of OXTR on symptomatology via
psychological resources.
Discussion
Psychological resources of optimism, mastery, and self-esteem
have been found to be signicant predictors of effective stress
management, neurophysiological responses to stress, and physical
and psychological health-related outcomes in previous research
(2, 3, 5, 7). Researchers have long known that these psychological
resources have genetic bases (2125), but investigations have not
previously identied which genes may be implicated. The present
results suggest that OXTR is one gene that is linked to these
resources. Carriers of the A allele of the OXTR SNP rs53576 were
less optimistic, felt less personal mastery, and had lower levels of
self-esteem. In addition, carriers of the A allele had higher levels
of depressive symptomatology. Mediation analysis suggested that
the effect of OXTR on depression may be mediated by psycho-
logical resources.
Previous research on OXTR and on the oxytocin system more
generally has underscored connections between the oxytocin
system and social afliation and bonding (2629), and links be-
tween oxytocin and positive emotional experiences have been
explored primarily within a social context (cf. 26). To our
knowledge, the present ndings are unique in linking OXTR
directly to individual psychological resources, suggesting an ex-
panded role for the oxytocin system in the management of stress
and distress that does not depend on social contact (see also refs.
33 and 37). Nonetheless, psychological resources are likely to
facilitate social bonding, and this may be one route whereby the
oxytocin system promotes social contact.
The association of variation in intron 3 of the OXTR with
psychological resources and depressive symptomatology is con-
sistent with an emerging body of evidence associating variation in
this region with similar psychological phenotypes (35, 50), psy-
chological responses in experimental studies (33, 34, 37), and
neural reactivity to emotional stimuli (38). Although the precise
molecular mechanisms responsible for these associations are
unclear, intron 3 is emerging as an important region in the
regulation of OXTR expression. For example, variation in intron
3 has been associated with OXTR expression in both peripheral
blood cells and the amygdala (51), a brain region exhibiting
volumetric differences associated with OXTR intron 3 variation
(52, 53). Further support for intron 3 being involved in the
regulation of OXTR expression comes from epigenetic studies.
Differential methylation of a CpG island within intron 3 is as-
sociated with relative levels of OXTR expression between pe-
ripheral blood cells and myometrial cells (54). Such effects may
be relevant for OXTR expression in the brain, as Gregory et al.
(55) found that methylation patterns within the OXTR were
similar in peripheral blood cells and the temporal cortex. The
methylation pattern within the OXTR promoter, but not intron 3,
was associated with OXTR expression in this cortical area. The
potential interacting inuences of polymorphic variation and
epigenetic modications on OXTR expression will be an in-
teresting area for future research.
The present study also highlights the need for additional re-
search on the genetic bases of positive well-being and psycho-
logical ourishing. Ample evidence demonstrates the heritability
of psychological resources and related constructs, yet few studies
have explored the molecular genetic basis of this heritability. The
results presented here represent a step toward understanding the
genetics of positive well-being, but many unanswered questions
remain. Consequently, this will likely be a source of productive
research in the future.
Limitations to the present research should be noted. The cross-
sectional design of the study precludes denitive conclusions re-
garding the causal relations among the OXTR polymorphism,
psychological resources, and depression. In particular, the medi-
ation analyses can only disconrm hypothetical causal relations,
not conrm their presence. Accordingly, one cannot completely
reject the alternative model in which the effect of OXTR on psy-
chological resources is mediated by depressive symptomatology.
Nonetheless, only partial mediation was indicated, and some
longitudinal research supports a causal relation between lower
levels of psychological resources and subsequent increases in the
incidence of depression (12, 16, 17).
In conclusion, scientists have long known that psychological
resources protect against the psychological and physiological
ravages of stress. Research has also made clear that there are
genetic bases to these resources. The present study identies
Fig. 1. Mediation model of OXTR, psychological resources and depression. For
the path from OXTR to depressive symptomatology, the upper -weight indi-
cates the effect without resources in the model and the lower -weight indi-
cates the effect withresources inthe model. *P <0.05, **P <0.01, ***P <0.001.
Table 2. Means for psychosocial resources by OXTR genotype
Scale G/G (n = 108) G/A/A/A (n = 218) Signicance
Self-esteem 3.38 (0.46) 3.20 (0.54) P = 0.004
Mastery 3.17 (0.44) 3.06 (0.47) P = 0.043
Optimism 3.74 (0.79) 3.53 (0.85) P = 0.031
BDI-IA, sum* 5.05 (4.45) 6.56 (6.04) P = 0.011
Psychological resources, factor score 0.238 (0.91) 0.118 (1.02) P = 0.002
BDI-IA, factor score* 0.181 (0.83) 0.089 (1.07) P = 0.013
Cell values indicate group means. SDs are in parentheses.
*For these variables, the equality of variances assumption was violated according to Levenes test. Corrected
P values are provided.
Saphire-Bernstein et al. PNAS Early Edition | 3 of 5
P
S
Y
C
H
O
L
O
G
I
C
A
L
A
N
D
C
O
G
N
I
T
I
V
E
S
C
I
E
N
C
E
S
OXTR as one genetic contributor and also provides evidence that
psychological resources may mediate the relation between OXTR
and depressive symptomatology.
Methods
Sample. Participants were 344 students and employees of a large Western
university recruited in two cohorts for participation in a study of stress and
coping (further details on sample characteristics can be found in refs. 10 and
56 for cohorts 1 and 2, respectively). Data were collected between 2004
and 2007. The sample was genotyped for OXTR for the purpose of this study,
and no previous publications have reported on this data. Participants were
recruited via posted yers advertising compensation of $125; the high level
of compensation was because of participants completing several subsequent
tasks that were not relevant to the present investigation. Recruitment and
study procedures were approved by the University of California Los Angeles
Institutional Review Board.
After excluding participants who were missing information for OXTR
genotype (n = 18), the nal sample consisted of 326 individuals. Ages for the
nal sample ranged from 18 to 36, with a mean of 21.3. There were 199
female participants (61.0%) and 127 male participants (39.0%). Of the par-
ticipants, 117 (35.9% of the sample) were of Asian ancestry, 87 were Euro-
peans or European-Americans (26.7%), 49 were Hispanic (15.0%), 9 were
African-American (2.8%), 34 reported their ethnicity as other (10.4%)
(usually Indian, Pakistani, or Middle Eastern ancestry), and 30 were
mixed (9.2%).
Missing Data. To retain all successfully genotyped participants, missing
responses to the psychological resources measures and the BDI-IA were im-
puted using the multiple imputation procedure, as implemented in AMELIAII
(57). This procedure is preferred to more traditional approaches to impu-
tation, such as mean replacement (cf. ref. 58). All statistics reported in the
article and SI Text were computed using the imputed data.
Measures. Optimism was measured using the Life Orientation Test (LOT-R, 4),
a six-item measure of dispositional optimism. Personal control was measured
by the Pearlin Mastery scale, a seven-item measure of perceived control over
ones circumstances and environment (8). Self-esteem was measured using
the Rosenberg Self-Esteem Scale (9), a widely used 10-item measure of self-
regard and perceived self-worth. Depressive symptomatology was measured
with the BDI-IA (47), a 21-item inventory of depressive symptoms experi-
enced in the past week. All of these measures are widely used and well-
validated standardized instruments. Details regarding the internal reli-
abilities of the scales in the present sample are presented in Table 1, along
with scale ranges, means, and SDs. Means were computed for the optimism,
mastery and self-esteem scales, and the BDI-IA was summed to facilitate
comparison with past studies using the BDI.
Factor Analysis and Factor Scores. Exploratory factor analysis was conducted
on the items from the optimism, mastery, self-esteem, and BDI-IA ques-
tionnaires. Two factors were specied for extraction using principal axis
factoring with direct oblimin rotation. This analysis yielded two factors
corresponding to psychological resources and depressive symptomatology
which were signicantly negatively correlated (r = 0.497, P < 0.001) and
combined to explain 28.6% of the variance in the 43 items.
Mediation Analysis. Sobels Z was calculated using a Web-based interactive
calculation tool available at http://quantpsy.org/sobel/sobel.htm. Unbiased
estimates of the indirect effects and their 95% condence intervals were
obtained via bootstrap analysis using the SPSS macro provided by Preacher
and Hayes (49).
Genotyping. In cohort 1, DNA was obtained using the Orasure oral specimen
collection device (Orasure Technologies Inc.). Samples were immediately
placed on ice in a cooler and transferred within the next few minutes to
a freezer. The samples were stored at 20 C for 12 to 18 mo before being
extracted using the Puregene DNA purication kit (Gentra Systems, Inc.). All
samples were whole-genome amplied using a GenomiPhi V2 DNA Ampli-
cation Kit (GE Healthcare) according to the manufacturers instructions. For
cohort 2, the DNA was collected from saliva with Oragene kits (DNA Gen-
otek) and extracted according to the manufacturers recommendations. The
OXTR rs53576 SNP was identied using a 59 nuclease assay to discriminate
between the two alleles (Taqman SNP Genotyping Assay C_3290335_10;
Applied Biosystems Inc.). Polymerase chain reactions were performed using
5-L reaction volumes in 384-well plates with 5 ng of DNA. The standard
protocol provided with the kit was followed. End point reads of uorescence
levels were obtained with an ABI 7900HT Sequence Detection System.
ACKNOWLEDGMENTS. The authors thank Kate Haltom and Ian Boggero for
assistance in preparing the DNA samples for analysis, and the members of
the Social Neuroscience Laboratory Group at the University of California at
Los Angeles for comments on an earlier draft. This research was supported
by the National Science Foundation (BCS-0729532) and by the National
Institute on Aging (AG030309). S.S.-B. was supported by a National Science
Foundation Graduate Research Fellowship while working on this project.
1. Aspinwall LG, Taylor SE (1992) Modeling cognitive adaptation: A longitudinal in-
vestigation of the impact of individual differences and coping on college adjustment
and performance. J Pers Soc Psychol 63:989e1003.
2. Taylor SE (2010) Mechanisms linking early life stress to adult health outcomes.
Proc Natl Acad Sci USA 107:8507e8512.
3. Taylor SE, Lerner JS, Sherman DK, Sage RM, McDowell NK (2003) Are self-enhancing
cognitions associated with healthy or unhealthy biological proles? J Pers Soc Psychol
85:605e615.
4. Scheier MF, Carver CS, Bridges MW (1994) Distinguishing optimism from neuroticism
(and trait anxiety, self-mastery, and self-esteem): A reevaluation of the Life Orien-
tation Test. J Pers Soc Psychol 67:1063e1078.
5. Taylor SE, Broffman JI (2011) Psychosocial resources: Functions, origins, and links to
mental and physical health. Adv Exp Soc Psychol 44:1e57.
6. Taylor SE, Lerner JS, Sherman DK, Sage RM, McDowell NK (2003) Portrait of the self-
enhancer: Well adjusted and well liked or maladjusted and friendless? J Pers Soc
Psychol 84(1):165e176.
7. Carver CS, Scheier MF, Segerstrom SC (2010) Optimism. Clin Psychol Rev 30:879e889.
8. Pearlin LI, Schooler C (1978) The structure of coping. J Health Soc Behav 19(1):2e21.
9. Rosenberg M (1965) Society and the Adolescent Self-image (Princeton University
Press, Princeton, NJ).
10. Taylor SE, et al. (2008) Neural bases of moderation of cortisol stress responses by
psychosocial resources. J Pers Soc Psychol 95:197e211.
11. Andersson G (1996) The benets of optimism: A meta-analytic review of the life
orientation test. Pers Individ Dif 21:719e725.
12. Vickers KS, Vogeltanz ND (2000) Dispositional optimism as a predictor of depressive
symptoms over time. Pers Individ Dif 28:259e272.
13. Ben-Zur H (2002) Coping, affect and aging: The roles of mastery and self-esteem. Pers
Individ Dif 32:357e372.
14. Taylor SE, Helgeson VS, Reed GM, Skokan LA (1991) Self-generated feelings of control
and adjustment to physical illness. J Soc Issues 47(4):91e109.
15. Finkelstein DM, Kubzansky LD, Capitman J, Goodman E (2007) Socioeconomic dif-
ferences in adolescent stress: The role of psychological resources. J Adolesc Health 40
(2):127e134.
16. Orth U, Robins RW, Roberts BW (2008) Low self-esteem prospectively predicts de-
pression in adolescence and young adulthood. J Pers Soc Psychol 95:695e708.
17. Orth U, Robins RW, Trzesniewski KH, Maes J, Schmitt M (2009) Low self-esteem is
a risk factor for depressive symptoms from young adulthood to old age. J Abnorm
Psychol 118:472e478.
18. Ek E, Remes J, Sovio U (2004) Social and developmental predictors of optimism from
infancy to early adulthood. Soc Indic Res 69:219e242.
19. Taylor SE, Seeman TE (1999) Psychosocial resources and the SES-health relationship.
Ann N Y Acad Sci 896:210e225.
20. Korkeila K, et al. (2004) Childhood adversities, parent-child relationships and dispo-
sitional optimism in adulthood. Soc Psychiatry Psychiatr Epidemiol 39:286e292.
21. Alessandri G, et al. (2010) Much more than model tting? Evidence for the heritability
of method effect associated with positively worded items of the Life Orientation Test
Revised. Struct Equ Modeling 17:642e653.
22. Caprara GV, et al. (2009) Human optimal functioning: The genetics of positive ori-
entation towards self, life, and the future. Behav Genet 39:277e284.
23. Kendler KS, Gardner CO, Prescott CA (1998) A population-based twin study of self-
esteem and gender. Psychol Med 28:1403e1409.
24. Mosing MA, Zietsch BP, Shekar SN, Wright MJ, Martin NG (2009) Genetic and envi-
ronmental inuences on optimism and its relationship to mental and self-rated
health: A study of aging twins. Behav Genet 39:597e604.
25. Plomin R, et al. (1992) Optimism, pessimism and mental health: A twin/adoption
analysis. Pers Individ Dif 13:921e930.
26. Campbell A (2010) Oxytocin and human social behavior. Pers Soc Psychol Rev 14:
281e295.
27. Heinrichs M, von Dawans B, Domes G (2009) Oxytocin, vasopressin, and human social
behavior. Front Neuroendocrinol 30:548e557.
28. Ishak WW, Kahloon M, Fakhry H (2011) Oxytocin role in enhancing well-being: A
literature review. J Affect Disord 130(1-2):1e9.
29. Uvnas-Moberg K, Arn I, Magnusson D (2005) The psychobiology of emotion: The role
of the oxytocinergic system. Int J Behav Med 12(2):59e65.
30. Inoue T, et al. (1994) Structural organization of the human oxytocin receptor gene.
J Biol Chem 269:32451e32456.
4 of 5 | www.pnas.org/cgi/doi/10.1073/pnas.1113137108 Saphire-Bernstein et al.
31. Gimpl G, Fahrenholz F (2001) The oxytocin receptor system: structure, function, and
regulation. Physiol Rev 81:629e683.
32. Bakermans-Kranenburg MJ, van Ijzendoorn MH (2008) Oxytocin receptor (OXTR) and
serotonin transporter (5-HTT) genes associated with observed parenting. Soc Cogn
Affect Neurosci 3(2):128e134.
33. Kim HS, et al. (2010) Culture, distress, and oxytocin receptor polymorphism (OXTR)
interact to inuence emotional support seeking. Proc Natl Acad Sci USA 107:
15717e15721.
34. Kim HS, et al. (2011) Gene-culture interaction: Oxytocin receptor polymorphism
(OXTR) and emotion regulation. Soc Psychol Pers Sci, 10.1177/1948550611405854.
35. Lucht MJ, et al. (2009) Associations between the oxytocin receptor gene (OXTR) and
affect, loneliness and intelligence in normal subjects. Prog Neuropsychopharmacol
Biol Psychiatry 33:860e866.
36. Riem MME, Pieper S, Out D, Bakermans-Kranenburg MJ, van Ijzendoorn MH (2011)
Oxytocin receptor gene and depressive symptoms associated with physiological re-
activity to infant crying. Soc Cogn Affect Neurosci 6:294e300.
37. Rodrigues SM, Saslow LR, Garcia N, John OP, Keltner D (2009) Oxytocin receptor
genetic variation relates to empathy and stress reactivity in humans. Proc Natl Acad
Sci USA 106:21437e21441.
38. Tost H, et al. (2010) A common allele in the oxytocin receptor gene (OXTR) impacts
prosocial temperament and human hypothalamic-limbic structure and function. Proc
Natl Acad Sci USA 107:13936e13941.
39. Costa B, et al. (2009) Oxytocin receptor polymorphisms and adult attachment style in
patients with depression. Psychoneuroendocrinology 34:1506e1514.
40. Eisenberger NI, Lieberman MD, Williams KD (2003) Does rejection hurt? An fMRI
study of social exclusion. Science 302:290e292.
41. Eisenberger NI, Taylor SE, Gable SL, Hilmert CJ, Lieberman MD (2007) Neural path-
ways link social support to attenuated neuroendocrine stress responses. Neuroimage
35:1601e1612.
42. Onoda K, et al. (2010) Does low self-esteem enhance social pain? The relationship
between trait self-esteem and anterior cingulate cortex activation induced by os-
tracism. Soc Cogn Affect Neurosci 5:385e391.
43. Baumgartner T, Heinrichs M, Vonlanthen A, Fischbacher U, Fehr E (2008) Oxytocin
shapes the neural circuitry of trust and trust adaptation in humans. Neuron 58:
639e650.
44. Domes G, et al. (2007) Oxytocin attenuates amygdala responses to emotional faces
regardless of valence. Biol Psychiatry 62:1187e1190.
45. Kirsch P, et al. (2005) Oxytocin modulates neural circuitry for social cognition and fear
in humans. J Neurosci 25:11489e11493.
46. Petrovic P, Kalisch R, Singer T, Dolan RJ (2008) Oxytocin attenuates affective evalu-
ations of conditioned faces and amygdala activity. J Neurosci 28:6607e6615.
47. Beck AT, Rush AJ, Shaw BF, Emery G (1979) Cognitive Therapy of Depression (Guilford,
New York).
48. Sasaki JY, Kim HS, Xu J (2011) Religion and well-being: The moderating role of
culture and the oxytocin receptor (OXTR) gene. J Cross-Cultural Psychol, 10.1177/
0022022111412526, in press.
49. Preacher KJ, Hayes AF (2004) SPSS and SAS procedures for estimating indirect effects
in simple mediation models. Behav Res Methods Instrum Comput 36:717e731.
50. Campbell DB, et al. (2011) Association of oxytocin receptor (OXTR) gene variants with
multiple phenotype domains of autism spectrum disorder. J Neurodev Disord 3:
101e112.
51. Tansey KE, et al. (2010) Oxytocin receptor (OXTR) does not play a major role in the
aetiology of autism: Genetic and molecular studies. Neurosci Lett 474(3):163e167.
52. Furman DJ, Chen MC, Gotlib IH (2011) Variant in oxytocin receptor gene is associated
with amygdala volume. Psychoneuroendocrinology 36:891e897.
53. Inoue H, et al. (2010) Association between the oxytocin receptor gene and amygdalar
volume in healthy adults. Biol Psychiatry 68:1066e1072.
54. Mizumoto Y, Kimura T, Ivell R (1997) A genomic element within the third intron of
the human oxytocin receptor gene may be involved in transcriptional suppression.
Mol Cell Endocrinol 135(2):129e138.
55. Gregory SG, et al. (2009) Genomic and epigenetic evidence for oxytocin receptor
deciency in autism. BMC Med 7:62.
56. Way BM, Taylor SE (2010) The serotonin transporter promoter polymorphism is as-
sociated with cortisol response to psychosocial stress. Biol Psychiatry 67:487e492.
57. Honaker J, King G, Blackwell M (2011) AMELIA II: A program for missing data (Ver 1.5-
2). http://gking.harvard.edu/amelia. Accessed on June 27, 2011.
58. Schafer JL, Graham JW (2002) Missing data: Our view of the state of the art. Psychol
Methods 7(2):147e177.
Saphire-Bernstein et al. PNAS Early Edition | 5 of 5
P
S
Y
C
H
O
L
O
G
I
C
A
L
A
N
D
C
O
G
N
I
T
I
V
E
S
C
I
E
N
C
E
S
Supporting Information
Saphire-Bernstein et al. 10.1073/pnas.1113137108
SI Text
Additional Details on Construction of Factor Scores. Exploratory
factor analysis of the combined items from the Beck Depression
Inventory-IA (BDI-IA) and from the three measures of psy-
chological resources revealed a reasonably clear two-factor so-
lution (Table S1). Four items from the BDI-IA loaded more
strongly (negatively) on the resources factor than on the symp-
toms factor. These items (BDI-IA #s 2, 3, 7, and 8) each assess
the absence of self-esteem or positive self-feeling, and so the
cross-loadings on resources are not surprising. In addition, one
item from the mastery scale (MAS #7) had a stronger loading on
the depressive symptomatology factor (0.31) than it did on the
resources measure (0.30). In addition, three items (MAS #2 and
BDI-IA #s 14 and 19) failed to achieved loadings greater than
0.20 on either factor (see Table S1 for the full pattern matrix of
factor loadings).
After the elimination of these eight items, a satisfactory simple
structure solution was produced by extracting two factors, which
combined to account for 29.2% of the total item variance. All
items attained a loading greater than 0.30 on the intended factor
and none displayed loadings greater than 0.30 on the other factor.
The correlation between the two factors in this second exploratory
factor analysis was 0.433. Factor scores were estimated by av-
eraging the items loading on each factor in the second analysis,
after rst standardizing each item to assure the equivalence of
item scaling across measures. Thus, 21 items were used to
compute the psychological resources factor, and 14 items were
used to compute the depressive symptomatology factor (see
Table S1 for more details). Unit-weighting was used to compute
factor scores, as this method is believed to avoid overtting to
the sample at hand and therefore to yield factor scores that are
more replicable across samples (1, 2). The factor scores for
psychological resources and depressive symptomatology were
standardized to enable the direct comparison of the boot-
strapped estimates of the indirect effects.
Ethnicity Analyses. Molecular genetic ndings regarding human
behavior may not generalize across different ethnic groups (35).
For example, signicant moderation by race (African-American
relative to Caucasian) has been reported for an association be-
tween the serotonin transporter promoter length polymorphism
(5HTTLPR) and levels of 5-hydroxyindoleacetic acid (5-HIAA)
in cerebrospinal uid (3). Previous studies of OXTR rs53576 in
mixed-ethnicity samples (Asians and Caucasians) have reported
signicantly greater prevalence of A alleles relative to G alleles
in individuals of Asian ancestry relative to Caucasians (68). As
our sample was characterized by a relatively high degree of
ethnic heterogeneity (as noted in the main text), we explored the
extent to which the effects uncovered in the present study were
generally consistent across the largest ethnic subgroups. In the
analyses that follow, we rst investigated whether the effects of
OXTR were similar for individuals in the largest ethnic subgroup,
Asians and Asian-Americans (n = 117), and for individuals in
the remainder of the sample (n = 209). Follow-up analyses were
then conducted on a subset of the sample consisting exclusively
of individuals from the three largest ethnic subgroups: Asian,
Caucasian (n = 87), and Hispanic (n = 49). Together, these
three ethnicities accounted for 77.6% of the study participants
(combined n = 253).
A series of multiple regression analyses were conducted to
examine whether the effects of OXTR A-allele carrier status on
psychological resources and depressive symptomatology were
moderated by or confounded with ethnicity. Initial models in-
cluded a term for the interaction of OXTR and ethnicity, but no
signicant interaction effects were detected in any of the anal-
yses of ethnicity and OXTR, including additional analyses de-
scribed below. The absence of signicant interaction effects
indicates that ethnicity did not moderate the effect of OXTR on
any of the dependent variables included in the study, and this
possibility is not considered or discussed further.
We next investigated whether the effects of OXTR are con-
founded with ethnicity by regressing the psychological resources
and depressive symptomatology factor scores on OXTR and eth-
nicity simultaneously, absent the interaction term. These analyses
demonstrated that the variance in psychological resources ac-
counted for by OXTRA-allele carrier status partially overlaps with
the variance explained by Asian ethnicity, as the effect of OXTR
on psychological resources was reduced and no longer signicant
( = 0.094, P = 0.097; from = 0.168, P = 0.002 without
ethnicity in the model), whereas the effect of Asian ethnicity re-
mained signicant ( =0.229, P < 0.001). These results indicate
that lower resources were reported by Asians relative to non-
Asians and by A-allele carriers relative to G-allele homozygotes,
although the latter effect was not signicant. Follow-up analyses
restricted to participants of Asian, Caucasian, and Hispanic an-
cestry (n = 253) found that the effect of OXTR on resources in
this subsample remained signicant ( =0.142, P = 0.029) when
controlling for both Asian ( = 0.232, P = 0.001) and Hispanic
ethnicity ( = 0.029, P = 0.665). Caucasian ethnicity served as
the comparison group for this analysis. As with the preceding
analysis, the results indicated lower resources among A-allele
carriers relative to G-allele homozygotes and among Asians (but
not Hispanics) relative to Caucasians. Note that the zero-order
effect of OXTR A-allele carrier status was somewhat larger in the
subsample of Asians, Hispanics, and Caucasians ( =0.222, P <
0.001) than it was in the full sample ( = 0.168), a difference
that may have contributed to the greater robustness of the OXTR
effect in this subsample.
With respect to depressive symptomatology, the effect of Asian
ethnicity was not signicant ( = 0.050, P = 0.391) when entered
simultaneously with OXTR A-allele carrier status, the effect of
which fell just shy of signicance ( = 0.111, P = 0.057). This
situation was somewhat reversed in the subsample of the three
largest ethnicities, as the effect of carrying the OXTR A allele
was not signicant ( = 0.108, P = 0.107), but the effect of Asian
ethnicity ( = 0.111, P = 0.050) was right at the threshold of
nominal signicance, and the effect of Hispanic ethnicity was
signicant ( = 0.172, P = 0.014). These results indicate higher
levels of depressive symptomatology among both Asians and
Hispanics, relative to Caucasians, and a nonsignicant trend for
more depressive symptomatology in OXTR A-allele carriers.
One factor that may account in part for the shared variance
between the A allele of rs53576 and Asian ethnicity is the greater
prevalence of the A allele in Asian individuals, relative to indi-
viduals in all other ethnic groups (Table S2). Of the 117 Asians
and Asian-Americans in our sample, only 15 (12.8%) had the G/G
genotype, whereas 102 (87.2%) were A-allele carriers (of the
carriers, 58 were heterozygotes and 44 were A-allele homo-
zygotes) (Table S2). In contrast, A-allele homozygotes were
a relative rarity among the non-Asian participants in our study,
numbering only 21 in total, and they were even rarer (n = 3 of 87)
among Caucasian participants (see Table S2 for complete in-
formation on the distributions of genotype by ethnicity). It may
be the case that OXTR rs53576 variation in Caucasians is best
Saphire-Bernstein et al. www.pnas.org/cgi/content/short/1113137108 1 of 4
studied using a dominant model, whereas the consequences of
variation at this locus for Asian individuals may be better cap-
tured by an additive model, in which each additional A allele
contributes added risk for negative outcomes. In support of this
notion, we found that when the number of A alleles (i.e., 0, 1, or
2, coded as 1, 0, and 1, respectively) was used to predict psy-
chological resources instead of the A-allele carrier contrast in
the full sample, the effect of OXTR genotype survived the ad-
dition of Asian ethnicity to the model (for OXTR, = 0.115,
P = 0.050; for Asian ethnicity, = 0.214, P < 0.001; model
adjusted R
2
= 0.073). Likewise, the effect of OXTR on psycho-
logical resources in the reduced sample of Asian, Caucasian,
and Hispanic participants was more robust using the additive
model: OXTR genotype was signicantly related to psychological
resources ( = 0.189, P = 0.005) even when controlling
for both Asian ( = 0.197, P = 0.009) and Hispanic ethnicity
( = 0.011, P = 0.871; model adjusted R
2
= 0.092).
Separate analyses of the effects of OXTR on each of the de-
pendent variables are provided for the three largest ethnic
groups (Asians, Caucasians, and Hispanics) and for non-Asians
as a whole in Table S4. Although none of the effects are sig-
nicant when the sample is stratied by ethnicity because of the
reduction in sample size, comparison of the effect sizes obtained
in each ethnic group reveals that the effects of OXTR on the
individual resource measures and the psychological resources
factor are consistently positive and moderately comparable
across Asians, Caucasians, and Hispanics, although some varia-
tion in the size of the effect is present. The effect of OXTR on
depressive symptomatology was also present in Asians, non-
Asians overall, and in Hispanics, but was less evident for Cau-
casians (Table S4).
Finally, the mediation model held for both Asians and non-
Asians when examined separately by ethnicity, but the effects of
OXTR were once again not signicant because of the reduction
in sample size. Correlations between the psychological resources
and depressive symptomatology factor scores were signicant at
P < 0.001 in each subsample: for Asians (r = 0.36), for Cau-
casians (r = 0.47), Hispanics (r = 0.55), and non-Asians
overall (r = 0.53). Bootstrapped estimates of the indirect effect
were: for Asians [0.0648, 95% condence interval (CI): 0.0137,
0.1456], for Hispanics (0.0741, 95% CI: 0.0395, 0.2027), and for
Caucasians (0.0304, 95% CI: 0.0532, 0.1147).
Our supplemental analyses provide support for the hypothesis
that the effect of OXTR on psychological resources documented
in the present study (see main text) is roughly similar for Asians,
Caucasians, and Hispanics. Nevertheless, future research should
conrm that the rs53576 SNP is a marker for functionally similar
genetic effects in individuals from different ethnicities.
Sex Analyses. The distribution of OXTR alleles did not differ sig-
nicantly between males and females [
2
(2) = 3.50, P = 0.173],
and there were no signicant sex differences in self-esteem, op-
timism, depressive symptomatology, and the psychological re-
sources and depressive symptomatology factors (see Table S5 for
more details). We also tested for interactions between OXTR and
sex in their effects on each of these measures, and none were
signicant. A signicant difference did emerge on the mastery
scale, with men scoring higher on average (M = 3.17, SD = 0.43)
than women: (M = 3.05, SD = 0.48), t(324) = 2.37, P = 0.018,
d = 0.27, R
2
= 0.018. Simultaneous regression analysis revealed
this sex effect to be largely independent from the effect of OXTR,
as signicant effects were observed for both OXTR A-allele
carriers ( =0.116, P = 0.035) and sex ( =0.134, P = 0.015),
indicating lower levels of mastery in women and OXTR A-allele
carriers. The two variables combined to explain 2.4% of the
variance in self-reported mastery. As with the other dependent
variables, the interaction between OXTR and sex was not signi-
cant in predicting mastery (P = 0.745). Similar results were ob-
tained when the effect of OXTR genotype was assessed using the
additive model (for OXTR genotype, = 0.137, P = 0.013; for
sex, = 0.127, P = 0.021; model adjusted R
2
= 0.030). Fur-
thermore, although the effect of sex on psychological resources
did not reach the threshold for nominal signicance when entered
on its own ( = 0.102, P = 0.066), the P value for the sex effect
( = 0.107, P = 0.051) was only slightly greater than 0.05 when
entered simultaneously with the OXTR A-allele carrier contrast
(the effect of OXTR was not appreciably different, = 0.171,
P = 0.002), indicating lower levels of psychological resources
among females and among A-allele carriers, relative to men and to
GG homozygotes, respectively. Finally, correlations between the
psychological resources and depressive symptomatology factor
scores were signicant at P < 0.001: for men, r = 0.44; and
women, r = 0.49.
1. Floyd FJ, Widaman KF (1995) Factor analysis in the development and renement of
clinical assessment instruments. Psychol Assess 7:286e299.
2. Russell DW (2002) In search of underlying dimensions: The use (and abuse) of factor
analysis in Personality and Social Psychology Bulletin. Pers Soc Psychol Bull 28:
1629e1646.
3. Williams RB, et al. (2003) Serotonin-related gene polymorphisms and central nervous
system serotonin function. Neuropsychopharmacology 28:533e541.
4. Propper C, Willoughby M, Halpern CT, Carbone MA, Cox M (2007) Parenting quality,
DRD4, and the prediction of externalizing and internalizing behaviors in early
childhood. Dev Psychobiol 49:619e632.
5. Widom CS, Brzustowicz LM (2006) MAOA and the cycle of violence: Childhood abuse
and neglect, MAOA genotype, and risk for violent and antisocial behavior. Biol
Psychiatry 60:684e689.
6. KimHS, et al. (2010) Culture, distress, andoxytocinreceptor polymorphism(OXTR) interact
to inuence emotional support seeking. Proc Natl Acad Sci USA 107:15717e15721.
7. Kim HS, et al. (2011) Gene-culture interaction: Oxytocin receptor polymorphism (OXTR)
and emotion regulation. Soc Psychol Pers Sci, 10.1177/1948550611405854.
8. Sasaki JY, Kim HS, Xu J (2011) Religion and well-being: The moderating role of
culture and the oxytocin receptor (OXTR) gene. J Cross-Cultural Psychol, 10.1177/
0022022111412526, in press.
Saphire-Bernstein et al. www.pnas.org/cgi/content/short/1113137108 2 of 4
Table S1. Factor loadings from exploratory factor analysis
Factor loading Factor loading
Item I II Item I II
SE6 0.80 0.07 MAS3 0.39 0.10
SE5 0.75 0.15 OPT1 0.39 0.01
SE7 0.68 0.08 OPT3 0.37 0.23
SE10 0.67 0.07 MAS2 0.24 0.00
SE2 0.66 0.12 BDI-IA18 0.01 0.54
SE1 0.65 0.06 BDI-IA15 0.05 0.53
SE3 0.63 0.05 BDI-IA17 0.01 0.53
SE9 0.62 0.02 BDI-IA13 0.06 0.51
SE4 0.59 0.09 BDI-IA10 0.02 0.44
OPT4 0.58 0.01 BDI-IA4 0.23 0.43
SE8 0.55 0.04 BDI-IA1 0.23 0.42
OPT10 0.54 0.11 BDI-IA16 0.02 0.42
OPT9 0.50 0.13 BDI-IA5 0.19 0.38
BDI-IA7 0.49 0.32 BDI-IA11 0.02 0.37
OPT7 0.49 0.15 BDI-IA21 0.10 0.37
MAS6 0.48 0.20 BDI-IA12 0.11 0.36
MAS5 0.43 0.12 BDI-IA6 0.13 0.32
BDI-IA8 0.43 0.29 MAS7 0.30 0.31
MAS1 0.42 0.01 BDI-IA20 0.07 0.31
MAS4 0.41 0.06 BDI-IA14 0.21 0.27
BDI-IA2 0.40 0.29 BDI-IA19 0.07 0.23
BDI-IA3 0.40 0.31
Loadings with an absolute value greater than or equal to 0.20 are dis-
played in boldface text. Items excluded from the computation of the factor
scores are italicized.
Table S2. Distributions of OXTR genotypes by ethnicity
Ethnic grouping G/G A/G A/A
White (n = 87) 42 (47.7%) 42 (48.8%) 3 (3.5%)
Asian (n = 117) 15 (12.8%) 58 (49.6%) 44 (37.6%)
Hispanic (n = 49) 21 (42.9%) 17 (34.7%) 11 (22.4%)
Other (n = 73) 30 (41.9%) 36 (48.6%) 7 (9.5%)
Table S3. Scale means by ethnic group
Scale
Non-Asian
(n = 209)
Asian
(n = 117) White n = 87)
Hispanic
(n = 49)
t test Asian vs. Other
(df = 324)
F-test three ethnicities
(df = 2, 250)
Self-esteem 3.36 (0.49) 3.09 (0.52) 3.38 (0.44) 3.28 (0.57) 4.70 P < 0.001 8.62 P < 0.001
Mastery 3.16 (0.46) 2.98 (0.43) 3.15 (0.43) 3.15 (0.57) 3.55 P < 0.001 4.35 P = 0.014
Optimism 3.70 (0.83) 3.41 (0.82) 3.73 (0.80) 3.77 (0.83) 3.08 P = 0.002 5.29 P = 0.006
BDI sum 5.55 (5.37) 6.98 (5.91) 4.62 (4.50) 6.76 (5.67) 2.18 P = 0.031 5.15 P = 0.006
Resources factor 0.19 (0.96) 0.35 (0.99) 0.22 (0.87) 0.13 (1.05) 4.83 P < 0.001 9.99 P < 0.001
BDI factor 0.06 (0.98) 0.11 (1.03) 0.25 (0.80) 0.19 (1.06) 1.55 P = 0.122 4.68 P = 0.010
Cell values indicate group means. SDs are in parentheses.
Saphire-Bernstein et al. www.pnas.org/cgi/content/short/1113137108 3 of 4
Table S4. Means of major variables for G/Gs and A carriers by ethnicity and sex
Scale G/G G/A/A/A Effect sizes
Self-esteem
Asian (n = 117) 3.17 (0.51) 3.07 (0.52) d = 0.20, R
2
= 0.009
Non-Asian (n = 209) 3.41 (0.44) 3.32 (0.52) d = 0.19, R
2
= 0.009
Caucasian (n = 87) 3.40 (0.39) 3.35 (0.48) d = 0.13, R
2
= 0.004
Hispanic (n = 49) 3.40 (0.52) 3.19 (0.60) d = 0.38, R
2
= 0.033
Male (n = 127) 3.41 (0.47) 3.26 (0.50) d = 0.29, R
2
= 0.020
Female (n = 199) 3.36 (0.45) 3.16 (0.55) d = 0.38, R
2
= 0.035
Mastery
Asian (n = 117) 3.04 (0.46) 2.97 (0.43) d = 0.17, R
2
= 0.007
Non-Asian (n = 209) 3.19 (0.44) 3.14 (0.48) d = 0.11, R
2
= 0.003
Caucasian (n = 87) 3.18 (0.43) 3.11 (0.43) d = 0.16, R
2
= 0.007
Hispanic (n = 49) 3.19 (0.51) 3.11 (0.62) d = 0.14, R
2
= 0.005
Male (n = 127) 3.23 (0.44) 3.14 (0.42) d = 0.22, R
2
= 0.011
Female (n = 199) 3.13 (0.44) 3.00 (0.49) d = 0.27, R
2
= 0.018
Optimism
Asian (n = 117) 3.74 (0.80) 3.36 (0.81) d = 0.48, R
2
= 0.053
Non-Asian (n = 209) 3.74 (0.79) 3.68 (0.87) d = 0.08, R
2
= 0.002
Caucasian (n = 87) 3.86 (0.75) 3.60 (0.83) d = 0.33, R
2
= 0.027
Hispanic (n = 49) 3.95 (0.66) 3.64 (0.94) d = 0.39, R
2
= 0.035
Male (n = 127) 3.78 (0.84) 3.60 (0.86) d = 0.22, R
2
= 0.012
Female (n = 199) 3.72 (0.76) 3.48 (0.85) d = 0.28, R
2
= 0.020
BDIIA sum
Asian (n = 117) 5.13 (4.19) 7.25 (6.09) d = 0.36, R
2
= 0.031
Non-Asian (n = 209) 5.02 (4.52) 5.96 (5.95) d = 0.17, R
2
= 0.007
Caucasian (n = 87) 4.69 (5.00) 4.53 (4.06) d = 0.04, R
2
= 0.000
Hispanic (n = 49) 5.19 (4.57) 7.93 (6.19) d = 0.50, R
2
= 0.058
Male (n = 127) 4.10 (3.74) 6.66 (5.74) d = 0.49, R
2
= 0.056
Female (n = 199) 5.59 (4.77) 6.50 (6.25) d = 0.16, R
2
= 0.006
Resources, factor
Asian (n = 117) 0.03 (1.00) 0.38 (1.00) d = 0.38, R
2
= 0.034
Non-Asian (n = 209) 0.27 (0.83) 0.12 (0.94) d = 0.17, R
2
= 0.007
Caucasian (n = 87) 0.30 (0.77) 0.16 (0.85) d = 0.23, R
2
= 0.013
Hispanic (n = 49) 0.30 (0.97) 0.07 (0.96) d = 0.42, R
2
= 0.042
Male (n = 127) 0.30 (0.89) 0.01 (0.95) d = 0.31, R
2
= 0.023
Female (n = 199) 0.19 (0.84) 0.20 (1.02) d = 0.41, R
2
= 0.039
BDI-IA factor
Asian (n = 117) 0.11 (0.71) 0.15 (1.01) d = 0.33, R
2
= 0.027
Non-Asian (n = 209) 0.16 (0.72) 0.01 (0.96) d = 0.22, R
2
= 0.011
Caucasian (n = 87) 0.20 (0.85) 0.20 (0.70) d = 0.03, R
2
= 0.000
Hispanic (n = 49) 0.09 (0.85) 0.39 (1.17) d = 0.46, R
2
= 0.049
Male (n = 127) 0.32 (0.60) 0.13 (0.96) d = 0.57, R
2
= 0.074
Female (n = 199) 0.05 (0.77) 0.04 (1.01) d = 0.12, R
2
= 0.004
Cell values indicate group means. SDs are in parentheses.
Table S5. Means for study variables by sex
Scale Male (n = 127) Female (n = 199) t test (df = 324) Effect sizes
Self-esteem 3.31 (0.49) 3.23 (0.53) 1.34 P = 0.180 d = 0.15 R
2
= 0.006
Mastery 3.17 (0.43) 3.05 (0.48) 2.37 P = 0.018 d = 0.27 R
2
= 0.018
Optimism 3.65 (0.85) 3.56 (0.83) 0.97 P = 0.334 d = 0.11 R
2
= 0.003
BDI-IA sum 5.86 (5.31) 6.19 (5.79) 0.52 P = 0.602 d = 0.06 R
2
= 0.001
Resources factor score 0.13 (0.94) 0.08 (1.03) 1.84 P = 0.066 d = 0.21 R
2
= 0.011
BDI-IA factor score 0.02 (0.93) 0.02 (1.05) 0.34 P = 0.734 d = 0.04 R
2
= 0.000
Cell values indicate group means. SDs are in parentheses.
Saphire-Bernstein et al. www.pnas.org/cgi/content/short/1113137108 4 of 4