httpsdoi.org10.11770956797619835147Psychological Scie

https://doi.org/10.1177/0956797619835147
Psychological Science
2019, Vol. 30(5) 798 –803
© The Author(s) 2019
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/0956797619835147
www.psychologicalscience.org/PS
ASSOCIATION FOR
PSYCHOLOGICAL SCIENCEShort Report
Research has consistently shown that life satisfaction is
associated with longevity (for a review, see Diener &
Chan, 2011). For example, meta-analyses of long-term
prospective studies have shown that higher life satisfac-
tion predicts lower risk of mortality over decades (Chida
& Steptoe, 2008). Although this literature has demon-
strated an intrapersonal effect of life satisfaction (i.e.,
an effect of an individual’s life satisfaction on that indi -
vidual’s mortality), it is less clear whether life satisfac-
tion has interpersonal effects as well. In particular, does
an individual’s life satisfaction affect the mortality risk
of his or her spouse?
Epidemiological studies have demonstrated the impor-
tance of contextual characteristics (e.g., neighborhood
characteristics; Bosma, Dike van de Mheen, Borsboom,
& Mackenbach, 2001) for individuals’ longevity. Adopting
the interpersonal perspective (Zayas, Shoda, & Ayduk,

2002), I propose that the characteristics (e.g., life satisfac-
tion) of the people who are close to an individual can
also make up that person’s context and, potentially,
affect his or her life outcomes. For example, life satis-
faction has been associated with healthy behaviors such
as physical exercise (Kim, Kubzansky, Soo, & Boehm,
2017). Given that spouses tend to affect each other’s
lifestyle ( Jackson, Steptoe, & Wardle, 2015), having a
happy spouse might increase one’s likelihood of engag-
ing in healthy behaviors. In addition, happiness has been
associated with helping behavior (O’Malley & Andrews,
1983). Hence, having a happy partner might be related
to experiencing support from that partner and, conse-
quently, might improve one’s health and longevity.
Indeed, a recent study found that spousal life satisfac-
tion was associated with individuals’ self-rated health
(Chopik & O’Brien, 2017), although such interpersonal
effects were not detected for doctor-diagnosed chronic
conditions (Chopik & O’Brien, 2017) or for inflammation
markers (Uchino et al., 2018). None of the existing studies
have explored whether spousal life satisfaction predicts
individuals’ mortality. The present research examined this
question using panel data of approximately 4,400 elderly
couples in the United States. In addition, a set of
835147PSSXXX10.1177/0956797619835147StavrovaLife
Satisfaction and Mortality
research-article2019
Corresponding Author:
Olga Stavrova, Department of Social Psychology, Tilburg
University,
Warandelaan 2, 5000 LE, Tilburg, The Netherlands
E-mail: [email protected]

Having a Happy Spouse Is Associated
With Lowered Risk of Mortality
Olga Stavrova
Department of Social Psychology, Tilburg University
Abstract
Studies have shown that individuals’ choice of a life partner
predicts their life outcomes, from their relationship
satisfaction to their career success. The present study examined
whether the reach of one’s spouse extends even further,
to the ultimate life outcome: mortality. A dyadic survival
analysis using a representative sample of elderly couples
(N = 4,374) followed for up to 8 years showed that a 1-
standard-deviation-higher level of spousal life satisfaction was
associated with a 13% lower mortality risk. This effect was
robust to controlling for couples’ socioeconomic situation
(e.g., household income), both partners’ sociodemographic
characteristics, and baseline health. Exploratory mediation
analyses pointed toward partner and actor physical activity as
sequential mediators. These findings suggest that life
satisfaction has not only intrapersonal but also interpersonal
associations with longevity and contribute to the fields of
epidemiology, positive psychology, and relationship research.
Keywords
life satisfaction, mortality, dyadic analyses, couples, open
materials
Received 8/2/18; Revision accepted 12/30/18
http://www.psychologicalscience.org/ps
mailto:[email protected]
https://sagepub.com/journals-permissions
http://crossmark.crossref.org/dialog/?doi=10.1177%2F09567976
19835147&domain=pdf&date_stamp=2019-03-21

Life Satisfaction and Mortality 799
exploratory mediation analyses tested the role of partner
support as well as partner and actor physical activity as
potential mechanisms for such an association.
Finally, it is possible that the level of spousal life
satisfaction per se matters much less than the extent to
which it is similar to individuals’ own life satisfaction.
A growing body of research has underscored the level
of congruence between partners’ dispositional charac-
teristics as an important factor for their relationship and
life outcomes (Dyrenforth, Kashy, Donnellan, & Lucas,
2010). Therefore, in an additional set of analyses, I
explored whether the level of actor-partner similarity
in life satisfaction was associated with actor mortality.
Method
Participants
The data for this study came from the Health and Retire-
ment Study (HRS; http://hrsonline.isr.umich.edu/), a
nationally representative panel study of American
adults ages 50 and older and their spouses. It is spon-
sored by the National Institute on Aging (Grant No. NIA
U01AG009740) and is conducted by the University of
Michigan. HRS is particularly well suited for the present
investigation because it collects data from both spouses.
Starting in 2006, the study has included a measure of
life satisfaction, as part of a self-report questionnaire
that participants are asked to complete on their own
and return by mail. For one half of the sample, life sat-
isfaction was first measured in 2006, and for the other

half, it was first measured in 2008. These data were
combined into a baseline assessment. I selected partici -
pants who had a spouse or a live-in partner at baseline
(95.7% of the participants who had a live-in partner
were officially married).1 After I removed cases with
missing values on key variables (actor life satisfaction,
partner life satisfaction, survival time), the final sample
consisted of 8,748 individuals (mean age at baseline =
67.17, SD = 9.75; 50.0% male). This sample size was large
enough for even small effects to be detected with 80%
power (at α = .05). Of the 4,374 couples, 99.5% were
heterosexual. Data from participants who remained alive
throughout the observation period (n = 6,643) or were
lost to follow-up (n = 656) were censored.2
The data and materials for HRS can be accessed at its
Website (http://hrsonline.isr.umich.edu/). The computer
code for the analyses reported here can be accessed at
the Open Science Framework (https://osf.io/geq9x/).
Measures
Life satisfaction. Life satisfaction was measured with
the Satisfaction With Life Scale (Diener, Emmons, Larsen,
& Griffin, 1985). This scale includes five items (e.g., “I am
satisfied with my life”). Because a 6-point response scale
was used in 2006 and a 7-point response scale was used
in 2008 (both scales ranged from strongly disagree to
strongly agree), I rescaled the responses to range from 1
to 10.3 The scale had good reliability (2006 subsample:
α = .89; 2008 subsample: α = .88). The analyses included
both partner and actor life satisfaction.
Mortality. The HRS data set included information on
participants’ vital status (1 = deceased, 0 = alive) through

December 2014. This information came from the National
Death Index (Centers for Disease Control and Prevention,
2017), the spouse’s report, or both. Survival time was
computed in months, starting from the month of the
baseline interview and ending with death or censoring
(in December 2014).
Additional variables. Perceived partner support was
measured by participants’ ratings of the extent to which
their partners provided them with social support (seven
items; e.g., “How much can you rely on [your partner] if
you have a serious problem?” “How much can you open
up to [your partner] if you need to talk about your wor-
ries?” “How much does [your partner] let you down when
you are counting on him/her?”; all items are provided in
the Supplemental Material available online). Responses
were given on a 4-point scale (1 = a lot, 4 = not at all)
and were recoded such that higher values reflected stron-
ger support. Each person’s recoded responses were then
averaged (2006 subsample: α = .82; 2008 subsample: α =
.84).
Actor and partner physical activity were assessed
with two questions. Both partners indicated how often
they engaged in vigorous activities (e.g., jogging,
cycling, digging with a spade or shovel) and moderately
energetic activities (e.g., gardening, cleaning the car,
walking at a moderate pace, dancing). Responses to
both questions were given on a 4-point scale (1 = more
than once a week, 2 = once a week, 3 = one to three
times a month, 4 = hardly ever or never) and were
recoded such that higher values reflected higher fre-
quency. The frequencies of vigorous and moderately
energetic activity were related to each other (r = .36,
p < .001), so I combined the responses to these two
questions to form an indicator of physical activity.

To make sure that any observed association between
partner life satisfaction and actor mortality was not
driven by an overlap with sociodemographic charac-
teristics or baseline health (e.g., one spouse’s health
problems might negatively affect both spouses’ life sat-
isfaction and mortality), I included a range of control
variables in the analyses. Specifically, I controlled for
actor and partner self-rated health (1 = poor, 5 = excel-
lent), as well as morbidity, measured with the number of
doctor-diagnosed chronic conditions (hypertension,
http://hrsonline.isr.umich.edu/
https://osf.io/geq9x/
800 Stavrova
diabetes, cancer, lung disease, coronary heart disease,
stroke, arthritis, incontinence, psychiatric problems;
although this list is not comprehensive, it covers major
causes of death). Further control variables included
actor gender (1 = male, 0 = female), actor and partner
age at baseline, actor and partner ethnicity (1 =
Caucasian, 0 = other), actor and partner education (1 =
less than high school, 2 = general education diploma,
3 = high school diploma, 4 = some college, 5 = college
and above), and baseline year (1 = 2008, 0 = 2006). Given
that the household financial situation is likely to affect
both partners’ life satisfaction and longevity, the analyses
also included baseline household income (total annual
household income in dollars, log transformed). To
account for partner mortality, the analyses included a
variable indicating whether the partner died during the
observation period (1 = deceased, 0 = alive).

Results
Means and standard deviations of the variables, as well
as their zero-order correlations, are provided in Table
S1 in the Supplemental Material. During the observation
period, 16.6% (n = 1,449) of the sample died. The sur-
vival time ranged from 2 to 104 months (8.67 years)
and averaged 50.5 months (4.21 years). An examination
of differences between survivors and decedents
revealed that the latter were older, t(8746) = 33.57, p <
.001; were more likely to be male, χ2(1, N = 8,748) =
191.90, p < .001; were less educated, t(8745) = 11.33,
p < .001; and were less wealthy, t(2703) = 14.43, p <
.001. They also were more likely to have chronic dis-
eases, t(1955) = 20.66, p < .001; were less likely to
engage in physical activity, t(8591) = 17.84, p < .001;
and reported poorer self-rated health, t(1959) = 23.51,
p < .001, and lower life satisfaction, t(1976) = 6.55, p <
.001. Similarly, decedents’ spouses, compared with sur-
vivors’ spouses, were older, t(8746) = 24.53, p < .001;
were less educated, t(2031) = 9.70, p < .001; reported
more chronic conditions, t(8745) = 11.00, p < .001;
reported a lower level of physical activity, t(8591) = 10.91,
p < .001; and had poorer self-rated health, t(8741) = 8.13,
p < .001. They also reported lower relationship satisfac-
tion, t(1922) = 7.97, p < .001, and lower life satisfaction,
t(1986) = 5.09, p < .001. Finally, decedents’ spouses were
more likely than survivors’ spouses to die within the
observation period, χ2(1, N = 8,748) = 202.61, p < .001.
To determine whether partner life satisfaction pre-
dicted actor mortality, I used multilevel (dyadic) sur-
vival analysis. Specifically, because time was measured
on a continuous scale (in months), I used the Cox
proportional hazards model. Given the clustered time-

to-event data (individuals were clustered within dyads),
I used an extension of the Cox model that accounts for
correlated observations by implementing robust sand-
wich variance estimators. The analyses were conducted
with the survival package (Therneau, 2015) in R. Note
that using a frailty model with penalized likelihood
estimation produced the same results (see Table S4 in
the Supplemental Material).
Time to event was measured in months, from the
baseline measurement of life satisfaction until death or
censoring. I additionally checked for robustness of the
results by conducting analyses using participants’ age
as a time scale. These analyses provided the same
results and are reported in the Supplemental Material
(Table S4). All continuous variables were standardized
before analysis, so the coefficients can be interpreted
in terms of standard deviations.
The full estimation results are presented in Table S2 in
the Supplemental Material. Model 1 showed that greater
partner life satisfaction at baseline was associated with
lower actor mortality risk. Specifically, a 1-standard-
deviation-higher level of spousal life satisfaction was asso-
ciated with a 13% lower risk of dying within the following
8 years (hazard ratio, or HR = 0.87, 95% confidence i nter-
val, or CI = [0.83, 0.91], p < .001). Figure 1 plots the
cumulative hazard of death during the observation period,
separately for individuals with a happy spouse (life satis -
faction above the median) and individuals with an
unhappy spouse (life satisfaction below the median). The
figure shows that as time went by, the mortality risk of
individuals with a happy spouse rose more slowly than
the mortality risk of individuals with an unhappy spouse.

To make sure that this effect was not just a result of
confounding with participants’ own life satisfaction, I
added actor life satisfaction at baseline in Model 2 (see
Table S2 in the Supplemental Material). The results
showed that both greater actor life satisfaction at base-
line (HR = 0.86, 95% CI = [0.82, 0.91], p < .001) and
greater partner life satisfaction at baseline(HR = 0.92,
95% CI = [0.87, 0.97], p = .001) were associated with
lower mortality risk.
Model 3 (see Table S2 in the Supplemental Material)
showed that these effects were robust to controlling for
major sociodemographic variables: actor gender, actor
and partner age, actor and partner ethnicity, actor and
partner education level, household income, baseline
year, and couple type (same-sex vs. heterosexual4). A
1-standard deviation-higher level of actor life satisfac-
tion was associated with an 18% lower mortality risk
(HR = 0.82, 95% CI = [0.78, 0.86], p < .001), and a
1-standard deviation-higher level of partner life satisfac-
tion was associated with a 10% lower mortality risk
(HR = 0.90, 95% CI = [0.85, 0.95], p < .001).
In Model 4, I added actor and partner health indica-
tors (self-rated health and morbidity) and partner mor-
tality (whether the partner died during the observation
period). The effect of actor life satisfaction on actor
mortality was rendered nonsignificant (HR = 0.96, 95%
CI = [0.90, 1.02], p = .155). In contrast, the effect of
partner life satisfaction remained (HR = 0.92, 95% CI =
[0.87, 0.97], p = .005).

Similarity effect
To explore whether the level of actor-partner similarity
in life satisfaction was associated with actor mortality,
I used a dyadic polynomial regression analysis, the
state-of-the-art approach to testing similarity effects
(Weidmann, Schönbrodt, Ledermann, & Grob, 2017).
Actor mortality was regressed on actor and partner life
satisfaction (xa and xp), their interaction term (xaxp),
and the quadratic terms (xa
2 and xp
2). The quadratic
and interaction terms were not significant (ps > .57).
The only terms with significant effects were the linear
terms of actor life satisfaction (HR = 0.85, 95% CI = [0.79,
0.92], p < .001) and partner life satisfaction (HR = 0.93,
95% CI = [0.86, 0.996], p = .039). Hence, I concluded
that the data do not provide evidence for a similarity
effect. Overall, these results suggest that having a part-
ner who is more satisfied with life is associated with
lower mortality regardless of one’s own level of life
satisfaction.
Exploratory mediation analyses
The variables available in the data set allowed me to
explore two potential mediation processes. First, I
hypothesized that individuals with a happier partner
experience more partner support, and that greater per-
ceived partner support is associated with lower mortal-
ity. However, an examination of the zero-order
associations among partner life satisfaction, perceived

partner support, and actor mortality revealed that this
mediation path is unlikely: Although having a happier
partner was indeed associated with greater perceived
partner support (r = .27, 95% CI = [.25, .29], p < .001),
perceived partner support was not related to actor mor -
tality (HR = 0.99, 95% CI = [0.94, 1.04], p = .69).
Second, I explored the role of partner and actor
physical activity as sequential mediators. Specifically,
on the basis of previous research (Kim et al., 2017), I
hypothesized that greater partner life satisfaction is
associated with increased partner physical activity,
which in turn is associated with greater actor physical
activity ( Jackson et al., 2015) and, consequently, lower
actor mortality. A look at the zero-order associations
showed that, indeed, partner life satisfaction was posi -
tively associated with partner physical activity (r = .17,
95% CI = [.15, .19], p < .001), partner and actor physical
activity were positively related to each other (r = .24,
95% CI = [.22, .26], p < .001), and actor physical activity
negatively predicted actor mortality (HR = 0.75, 95%
CI = [0.71, 0.79], p < .001).
Therefore, I proceeded to test for sequential media-
tion using multilevel structural equation modeling.
The model (see Fig. S1 in the Supplemental Material
included a set of multilevel (participants nested within
couples) regression equations, in which partner life
satisfaction predicted partner physical activity (path
a, multilevel linear regression), partner physical activ-
ity predicted actor physical activity (path d, multilevel
linear regression), and actor physical activity pre-
dicted actor mortality (path b, multilevel Cox regres-
sion). The indirect effect was computed by multiplying
the a, d, and b paths, and its significance was tested
using the delta method. The model included random

intercepts for actor and partner physical activity and
actor mortality and used clustered robust standard
errors. The analyses were conducted with Stata/MP
Version 14.2.
The results showed that partner life satisfaction was
positively associated with partner physical activity (b =
0.08, 95% CI = [0.07, 0.09], p < .001), which in turn was
positively associated with actor physical activity (b =
0.23, 95% CI = [0.20, 0.26], p < .001), which was nega-
tively associated with actor mortality (HR = 0.64, 95%
CI = [0.61, 0.68], p < .001). The coefficient for the
indirect effect was significant, b = −0.008, 95% CI =
[−0.01, −0.006], p < .001, which provided support to the
sequential mediation. The indirect effect was robust to
adding the control variables as predictors of both the
mediators and the dependent variable (see Table S5 in
the Supplemental Material).
+
+
0.00
0.05
0.10
0.15
0.20
0.25
0 25 50 75 100

Survival Time (Months Since Baseline)
C
um
ul
at
iv
e
H
az
ar
d
Partner Life Satisfaction Below Median
Partner Life Satisfaction Above Median
Fig. 1. Cumulative hazard of death (including 95% confidence
bands) during the observation period. Results are shown
separately
for individuals whose spouses reported life satisfaction below
the
median at baseline and those whose spouses reported life
satisfaction
above the median at baseline.
802 Stavrova
Exploratory moderation analyses
I explored whether the effect of partner life satisfaction

on actor mortality depended on various actor and part-
ner characteristics: gender, age, ethnicity, education,
income, health indicators, physical activity, perceived
partner support, and partner mortality. I ran 16 models
testing the interactions between partner life satisfaction
and these variables (by adding the respective interaction
terms, one at a time, to Model 4; see Table S2 in the
Supplemental Material). The only significant interaction
was between partner life satisfaction and partner mortal -
ity (HR = 1.15, 95% CI = [1.02, 1.31], p = .027). Partner
life satisfaction was negatively associated with actor
mortality only when the partner remained alive through
the end of the observation period (partner alive: HR =
0.90, 95% CI = [0.83. 0.96], p = .003; partner deceased:
HR = 1.03, 95% CI = [0.93, 1.15], p = .553). Yet the
exploratory nature of these analyses and the multiple
testing do not allow strong conclusions to be drawn.
Discussion
Previous research has shown that individuals’ career
success and relationship and life satisfaction are pre-
dicted by their spouses’ dispositional characteristics
(Dyrenforth et al., 2010; Solomon & Jackson, 2014). The
present research suggests that spouses’ reach might
extend even further. A dyadic survival analysis using
the data from 4,374 couples showed that having a
spouse who was more satisfied with life was associated
with reduced mortality.
What explains this interpersonal effect of life satis-
faction? Exploratory mediation analyses established
partner and actor physical activity as sequential media-
tors. One partner’s life satisfaction was associated with
his or her increased physical activity, which in turn was
related to increased physical activity in the other part-

ner, which predicted that partner’s mortality. Yet, given
the correlational nature of these data, these results
should be interpreted with caution.
It is noteworthy that the effect of spousal life satis-
faction was comparable in size to the effects of other
well-established predictors of mortality, such as educa-
tion and income (in the present study, HRs = 0.90 for
partner life satisfaction, 0.93 for household income, and
0.91 for actor education). In fact, spousal life satisfac-
tion predicted mortality as strongly as (and even more
robustly than) an individual’s own life satisfaction and
as strongly as basic personality traits, such as neuroti -
cism and extraversion, predicted mortality in previous
work ( Jokela et al., 2013).
Although most existing research on predictors of
mortality has focused nearly exclusively on individuals’
own characteristics, the present analyses revealed that
the characteristics of a person who is close to an
individual, such as a spouse, might be an equally
important determinant of that individual’s mortality.
Continuing this line of research, future studies might
explore whether the interpersonal effect of life satisfac-
tion on mortality is restricted to (marital) dyads or
whether it extends to larger social networks.
To conclude, happiness is a desirable trait in a
romantic partner, and marriage to a happy person is
more likely to last than is marriage to an unhappy per-
son (Lucas, 2005). The present study showed that hav-
ing a happier spouse is associated not only with a
longer marriage but also with a longer life.
Action Editor

James K. McNulty served as action editor for this article.
Author Contributions
O. Stavrova is the sole author of this article and is responsible
for its content.
ORCID iD
Olga Stavrova https://orcid.org/0000-0002-6079-4151
Acknowledgments
I would like to thank Anthony M. Evans for his statistical
advice and general support.
Declaration of Conflicting Interests
The author(s) declared that there were no conflicts of interest
with respect to the authorship or the publication of this
article.
Supplemental Material
Additional supporting information can be found at http://
journals.sagepub.com/doi/suppl/10.1177/0956797619835147
Open Practices
All analysis code for this study has been made publicly avail -
able via the Open Science Framework and can be accessed
at https://osf.io/geq9x/. The data are available through the
Health and Retirement Study’s Web site (http://hrsonline.isr
.umich.edu/). The design and analysis plans for this study were
not preregistered. The complete Open Practices Disclosure for

this article can be found at http://journals.sagepub.com/doi/
suppl/10.1177/0956797619835147. This article has received the
badge for Open Materials. More information about the Open
Practices badges can be found at http://www.psychological
science.org/publications/badges.
Notes
1. Additional analyses using only married couples produced
identical results (see Table S4 in the Supplemental Material).
https://orcid.org/0000-0002-6079-4151
http://journals.sagepub.com/doi/suppl/10.1177/09567976198351
47
47
https://osf.io/geq9x/
47
47
http://www.psychologicalscience.org/publications/badges
http://www.psychologicalscience.org/publications/badges
2. These two groups of censored observations did not dif-
fer from each other on any variable included in the analyses,
except for the number of chronic conditions: Participants who
dropped out reported fewer chronic conditions (M = 1.87, SD =
1.40) than did participants who stayed in the panel (M = 2.07,
SD = 1.46), t(7296) = 3.46, p = .001.
3. As a robustness check, I used standardization to normalize

the data instead (i.e., I standardized the values within the two
subsamples). The analyses using the standardized scale pro-
duced the same results as the analyses presented in the main
text (see Table S3 in the Supplemental Material).
4. Being part of a same-sex couple positively predicted mortal-
ity (HR = 2.72, p = .018; there were 9 gay and 11 lesbian
couples
in the sample). The interaction between couple type (gay vs.
lesbian) and actor gender was not significant (HR = 0.38, p =
.260).
References
Bosma, H., Dike van de Mheen, H., Borsboom, G. J. J. M.,
& Mackenbach, J. P. (2001). Neighborhood socioeco-
nomic status and all-cause mortality. American Journal
of Epidemiology, 153, 363–371.
Centers for Disease Control and Prevention. (2017). National
Death Index [Data file]. Retrieved from http://www.cdc
.gov/nchs/ndi.htm
Chida, Y., & Steptoe, A. (2008). Positive psychological well -
being and mortality: A quantitative review of prospec-
tive observational studies. Psychosomatic Medicine, 70,
741–756.
Chopik, W. J., & O’Brien, E. (2017). Happy you, healthy me?
Having a happy partner is independently associated with
better health in oneself. Health Psychology, 36, 21–30.
Diener, E., & Chan, M. Y. (2011). Happy people live longer:
Subjective well-being contributes to health and longev-
ity. Applied Psychology: Health and Well-Being, 3, 1–43.
Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985).

The Satisfaction With Life Scale. Journal of Personality
Assessment, 49, 71–75.
Dyrenforth, P. S., Kashy, D. A., Donnellan, M. B., & Lucas,
R. E. (2010). Predicting relationship and life satisfaction
from personality in nationally representative samples
from three countries: The relative importance of actor,
partner, and similarity effects. Journal of Personality and
Social Psychology, 99, 690–702.
Jackson, S. E., Steptoe, A., & Wardle, J. (2015). The influ-
ence of partner’s behavior on health behavior change:
The English Longitudinal Study of Ageing. JAMA Internal
Medicine, 175, 385–392.
Jokela, M., Batty, G. D., Nyberg, S. T., Virtanen, M., Nabi, H.,
Singh-Manoux, A., & Kivimäki, M. (2013). Personality and
all-cause mortality: Individual-participant meta-analysis
of 3,947 deaths in 76,150 adults. American Journal of
Epidemiology, 178, 667–675.
Kim, E. S., Kubzansky, L. D., Soo, J., & Boehm, J. K. (2017).
Maintaining healthy behavior: A prospective study of
psychological well-being and physical activity. Annals
of Behavioral Medicine, 51, 337–347.
Lucas, R. (2005). Time does not heal all wounds: A lon-
gitudinal study of reaction and adaptation to divorce.
Psychological Science, 16, 945–950.
O’Malley, M. N., & Andrews, L. (1983). The effect of mood
and incentives on helping: Are there some things money
can’t buy? Motivation and Emotion, 7, 179–189.
Solomon, B. C., & Jackson, J. J. (2014). The long reach of

one’s spouse: Spouses’ personality influences occupa-
tional success. Psychological Science, 25, 2189–2198.
Therneau, T. M. (2015). A package for survival analysis in S
(Version 2.38) [Computer software]. Retrieved from
https://CRAN.R-project.org/package=survival
Uchino, B. N., Kent, R. G., Cronan, S., Smith, T. W., Diener,
E.,
Joel, S., & Bosch, J. (2018). Life satisfaction and inflam-
mation in couples: An actor–partner analysis. Journal of
Behavioral Medicine, 41, 22–30.
Weidmann, R., Schönbrodt, F. D., Ledermann, T., & Grob, A.
(2017). Concurrent and longitudinal dyadic polynomial
regression analyses of Big Five traits and relationship
satisfaction: Does similarity matter? Journal of Research
in Personality, 70, 6–15.
Zayas, V., Shoda, Y., & Ayduk, O. N. (2002). Personality in
context: An interpersonal systems perspective. Journal of
Personality, 70, 851–900.
http://www.cdc.gov/nchs/ndi.htm
http://www.cdc.gov/nchs/ndi.htm
https://CRAN.R-project.org/package=survival
Does College-Based Relationship Education Decrease
Extradyadic
Involvement in Relationships?
Scott R. Braithwaite
Brigham Young University

Nathaniel M. Lambert, Frank D. Fincham,
and Kay Pasley
Florida State University
We used latent growth curve modeling to examine the
effectiveness of a relationship
education intervention (Relationship U, or RU) on rates of
extradyadic involvement in a
sample of 380 college students in committed romantic
relationships. RU is designed to be
integrated into existing college courses; it educates students
about partner selection, making
healthy relationship transitions, communica tion skills, and the
potentially negative conse-
quences of cheating in romantic relationships and how to
prevent its occurrence. Participants
who received the intervention reported trajectories of less
extradyadic involvement over time
relative to control participants. Being female was not associated
with less extradyadic
involvement at baseline, but it did predict less extradyadic
involvement over time across both
intervention and control conditions. Implications for
dissemination of relationship education
are discussed.
Keywords: relationship education, infidelity, dissemination,
emerging adulthood
Relationship education is the provision of information
designed to help couples and individuals experience suc-
cessful, stable romantic relationships. It has progressed to
the point at which at least eight preventive interventions
have been developed and shown to be efficacious (Braith-
waite & Fincham, 2009; Jakubowski et al., 2004). However,

in a pattern that mirrors the wider field of clinical psychol -
ogy, these empirically supported interventions are not typ-
ically used in everyday practice (Barlow, Levitt, & Bufka,
1999). Thus, the challenge facing relationship educators is
not only to develop more efficacious interventions, but also
to further refine them and find methods of dissemination
that will increase the likelihood that empirically supported
treatments will be used, especially in target populations.
One target population identified by relationship education
researchers is college students (see Fincham, Stanley, &
Rhoades, in press). This population is important because
many individuals experience emerging adulthood in the
context of college (e.g., 57% of young adults between 25
and 29 have attended some college; Stoops, 2004). During
this period, many health-relevant habits are formed, and the
individual and contextual changes that occur throughout
college push to the forefront a host of risky behaviors that
can increase risk for immediate and future problems
(Braithwaite, Delevi, & Fincham, 2010). For example,
infidelity— defined as “a secret sexual, romantic, or emo-
tional involvement that violates the commitment to an
exclusive relationship” (Glass, 2002, p. 489)— occurs in
as many as 65%–75% of college students (Shackelford,
LeBlanc, & Drass, 2000; Wiederman & Hurd, 1999) and
is associated with adverse mental and physical health
(Hall & Fincham, 2009). When infidelity entails unpro-
tected sex, there is direct risk of exposure to sexually
transmitted diseases and indirect risk of exposing the
faithful partner to HIV and other sexually transmitted
diseases. In short, this population is made up of individ-
uals engaging in risky behaviors, developing relational
habits, and forming relationships that often culminate in
marriage (Blossfeld & Timm, 2003).

To increase the reach of relationship education, research-
ers have begun to examine the impact of interventions when
they move from the laboratory to the real world (Markman
et al., 2004). In contrast to efficacy trials in which strict
experimental control is paramount, effectiveness trials ex-
amine treatment outcomes under the more practical condi-
tions of everyday implementation (Flay, 1986). The present
research is one such trial and examines the effectiveness of
relationship education delivered as part of a college course.
Based on the Within My Reach curriculum (Pearson, Stan-
ley, & Kline, 2005), RU educates students about risk and
protective factors for relationship dysfunction and provides
tools to diminish the influence of risk factors and enhance
protective factors—it is designed to be applicable to stu-
dents regardless of their current romantic relationship sta-
tus. Over the 13 weeks of an academic semester, RU em-
Scott R. Braithwaite, Department of Psychology, Brigham
Young University; Nathaniel M. Lambert, Frank D. Fincham,
and
Kay Pasley, Department of Family and Child Sciences, Florida
State University.
This research was supported by a grant from the U.S. Depart-
ment of Health and Human Services.
Correspondence concerning this article should be address to
Scott R.
Braithwaite, Brigham Young University Department of
Psychology 1070
SWKT Provo, UT 84602-5543. E-mail: [email protected]
Journal of Family Psychology © 2010 American Psychological
Association
2010, Vol. 24, No. 6, 740 –745 0893-3200/10/$12.00 DOI:
10.1037/a0021759

740
T
hi
s
do
cu
m
en
t i
s
co
py
ri
gh
te
d
by
th
e
A
m
er
ic
an

P
sy
ch
ol
og
ic
al
A
ss
oc
ia
tio
n
or
o
ne
o
f i
ts
a
lli
ed
p
ub

lis
he
rs
.
T
hi
s
ar
tic
le
is
in
te
nd
ed
s
ol
el
y
fo
r t
he
p
er

so
na
l u
se
o
f t
he
in
di
vi
du
al
u
se
r a
nd
is
n
ot
to
b
e
di
ss

em
in
at
ed
b
ro
ad
ly
.
phasizes the implications of “sliding” into relationship
transitions versus making conscious decisions, how to cre-
ate safe relationships, how to communicate effectively, the
potentially negative consequences of extradyadic involve-
ment, and how to create healthy boundaries to prevent its
occurrence. RU was designed to fit seamlessly into existing
college courses with the idea that such integration might be
a feasible way of disseminating this program on a larger
scale in colleges across the United States and abroad.
Notwithstanding its public health relevance, little re-
search has focused directly on interventions for infidelity in
emerging adulthood despite the pivotal nature of this devel -
opmental period when individuals are particularly focused
on romantic relationships. In this study, we therefore exam-
ine rates of change in extradyadic involvement over time in
response to the RU intervention. We hypothesized that
individuals who received RU would engage in less extra-
dyadic involvement. Because extradyadic involvement has
been shown to differ across genders in married samples

(Atkins, Baucom, & Jacobson, 2001), we also accounted for
the impact of biological sex on such behavior.
Method
Participants and Procedure
Data from this study come from a larger study that
examined students in a three-credit university-wide course
(Introduction to Families Across the Lifespan) that met
liberal studies requirements in social science; thus, students
could potentially represent all programs of study available
at the university. At the beginning of the semester, all
students in all sections were invited to participate in “a
study examining the effect of this class on your relation-
ships.” The study was one of multiple options for students
to obtain course credit. Participants were selected for inclu-
sion in the study if they reported being in an exclusive
heterosexual romantic relationship; those in nonexclusive
relationships were excluded. To determine this, we first
asked, “Are you currently in a romantic relationship?” If
students indicated that they were, we asked, “Which state-
ment best describes your relationship?” followed by the
options “dating nonexclusively,” “dating exclusively,” “en-
gaged,” “married,” or “other.” Only those who indicated
that they were dating exclusively, engaged, or married were
included (only 23 students indicated they were engaged, and
only 8 indicated they were married).
The larger sample consisted of 770 (175 men, 593 women;
data were missing for two participants) students (64% White,
16% African American, 10% Hispanic, 10% other) with an
average age of 19.74 years. The subsample in this study
consisted of 380 (69 men, 310 women; one with missing data)
students (64% White, 16% African American, 10% Hispanic,
10% other) with an average age of 19.96. We examined

differences between those in the treatment and control condi-
tions on the basis of relevant demographic characteristics and
found no difference for age, t(377) � 1.02, p � .05; relation-
ship length, t(364) � �1.08, p � .05; or parental divorce, F(1,
377) � 1.13, p � .05. The modal relationship duration for the
sample was 1–2 years (6%, less than 2 months; 12%, 3– 4
months; 7%, 5– 6 months; 13%, 7–12 months; 32%, 1–2 years;
and 26%, 2 years or more). Institutional review board approval
was obtained before any data collection.
After providing informed consent, participants completed
a battery of questionnaires at the beginning of the semester
and twice more at 6-week intervals. Assignment to condi-
tion was not random because students were free to sign up
for any available course section; however, students were
unaware of condition. Specific sections were designated as
treatment or control conditions before the semester began.
RU. Students (n � 312) in this condition had one class
each week (50 min/week for 13 weeks) in which they met in
small groups (20 –30) and received the intervention. These
breakout sessions (but not the lecture sessions) were led by
graduate student and postdoctoral instructors (naı̈ ve to study
hypotheses) who had received a minimum of 24 hr of
training in curriculum delivery. These breakout sessions
were not extra classes; rather, they took the place of one of
the existing class sessions each week. A brief description of
the content of the 13 RU sessions can be seen in Table 1.
Control condition. Owing to pressure from our funding
agency to offer the intervention as widely as possible, the ratio
of participants in the intervention condition to participants in
the control condition was approximately 5:1. Students (n � 67)
in the control condition received instruction that was identical
to that in the treatment condition except that they did not

receive RU content in one of their weekly classes. The class
content was based on a widely used introductory text (La-
manna & Riedmann, 2009) that provides an overview of the-
ory and research on marriage and families. Learning this kind
of information (e.g., information about mate selection, com-
munication in close relationships, “hooking up” and “friends
with benefits”) may serve to promote healthier relationship
choices, but the class as usual did not have an applied, skill-
based focus as did the RU breakout sessions.
Measurement
At each time point, participants in committed romantic
relationships were asked to report whether they had engaged
in several sexual or romantic behaviors with someone other
than their romantic partner in the previous 6 weeks. These
activities were kissing, sexual intimacy without intercourse
(two items; one specifically used the phrase “sexually inti -
mate without intercourse,” and the second assessed caress-
ing and hugging) and sexual intercourse (each coded 1 �
yes, 0 � no).
The sex of participants was coded as 0 � male and 1 �
female. Condition was coded as 0 � control and 1 � RU. At
baseline, 11% of participants admitted to extradyadic sexual
intercourse, 13% admitted to extradyadic sexual behavior
(nonintercourse), 44% admitted to extradyadic caressing
and hugging, and 22% admitted to extradyadic kissing.
Results
The data were analyzed using latent growth curve (LGC)
modeling in Mplus 5.2. Because we used latent variables
741RELATIONSHIP EDUCATION

T
hi
s
do
cu
m
en
t i
s
co
py
ri
gh
te
d
by
th
e
A
m
er
ic
an
P

sy
ch
ol
og
ic
al
A
ss
oc
ia
tio
n
or
o
ne
o
f i
ts
a
lli
ed
p
ub
lis

he
rs
.
T
hi
s
ar
tic
le
is
in
te
nd
ed
s
ol
el
y
fo
r t
he
p
er
so

na
l u
se
o
f t
he
in
di
vi
du
al
u
se
r a
nd
is
n
ot
to
b
e
di
ss
em

in
at
ed
b
ro
ad
ly
.
consisting of our binary indicators of extradyadic involve-
ment, we first tested the fit of the measurement models of
the latent variables. Each of the models (for Times 1–3)
provided a good fit, with the indicators loading well onto the
latent variables. Estimation of this type of LGC model
requires measurement invariance of the factors at each time
point; thus, the loadings from the items that made up the
latent factors were constrained to be equal across each of
the three time points. When these constraints were imposed,
Heywood cases arose as a result of correlations between
items that approached 1. As such, we summed the items to
form a composite scale of extradyadic involvement and fit a
latent growth curve to this scale using a Poisson model for
count data. For the purpose of providing evidence of psy-
chometric validity, the measurement model can be seen in
Figure 1. In the growth curve model, the intercepts for the
three manifest variables were fixed to unity, and each of the
paths from the slope latent variable to the three manifest
variables was constrained to be equal to the number of
weeks from the baseline assessment (0, 6, and 12 weeks).
Slope and intercept were regressed on condition and sex. The
specified latent growth model can be seen in Figure 2 and

descriptive statistics for the two groups across time can be seen
in Table 2. Specifying the model as described provides infor-
mation about the impact of RU on patterns of change or
“growth” in extradyadic involvement over time. Because we
could not randomly assign students to condition, and in
any case, interventions that are completed by groups are
susceptible to clustering effects that can lead to Type I
errors (Baldwin, Murray, & Shadish, 2005), we ac-
counted for the effect of clustering by using the TYPE �
COMPLEX with CLUSTER � ID functions in Mplus.
Condition and intercept were significantly associated, sug-
gesting that those in the RU condition reported more extrady-
adic involvement at baseline (� � .17, p � .05); however, the
association between intercept and slope was not significant,
suggesting that scores at baseline were not associated with
rates of change over time. Condition predicted slope, such that
those who received the RU intervention reported larger de-
creases in extradyadic involvement over time relative to the
control group (� � �.47, p � .05). Sex was also associated
with slope (� � �.44, p � .05), such that being female was
associated with larger decreases in extradyadic involvement
over time irrespective of condition. To explore this finding
further, we conducted an exploratory analysis in which we
included a Sex � Treatment interaction; the interaction term
did not significantly predict the slope of extradyadic involve -
ment, indicating that both condition and sex contributed reli -
able main effects that cannot be accounted for by an interaction
of these two variables.
Table 1
Description of Relationship U Intervention
Class Objective

1 To introduce students to the course, foster commitment from
students to put forth effort, and seriously engage in the class
and create an educational environment conducive to
implementing the intervention.
2 To help raise students’ awareness about their relationship
beliefs, clarify how setting relationship goals can lead to
healthier relationship decisions, and illustrate the connection
between family of origin experiences and relationship
expectations.
3 To introduce students to the differences between sex, gender,
and sexual orientation and how expectations surrounding
these constructs can influence romantic relationships.
4 To raise students’ awareness about their own personality traits
and provide information about ways they can identify
positive and negative traits within themselves and within
potential partners and illustrate how awareness of personality
and traits can lead to better mate selection and relationship
decision making.
5 To raise awareness of students’ expectations regarding
relationships, help students clarify which expectations they
possess
that are realistic or unrealistic, and promote safety in
relationships by identifying that expectations cannot be
appropriately addressed in unsafe relationships.
6 To reinforce the ideas of smart love (being thoughtful,
purposeful, and judicious in negotiating relationship transitions
such
as initiation of a sexual relationship), demonstrate how active
decision making in the present can affect the future of
relationships, introduce the concept of “sliding versus
deciding,” and help students identify barriers that prevent

healthy
relationships.
7 To integrate the material from the last two sessions (“Being
smart in reaching relationship goals” and “sliding versus
deciding”) and explore ideas about commitment (negotiation,
compromise, sacrifice) and what makes a marriage choice
healthy.
8 To help students identify the differences between good and
bad communication, briefly familiarize students with the
speaker–listener technique, and identify situations in which it
might be usefully applied.
9 To continue to explore the speaker–listener technique as a
guide for good communication, focusing on active listening;
provide students with additional practice of active listening
through role-plays and coaching; and introduce the concept
of time-out as a strategy for enhancing communication through
deescalation.
10 To assist students in continuing to practice and use the
speaker–listener technique, review and practice using time-outs,
and
introduce and assist students in practicing XYZ statements.
11 To continue to explore the use of XYZ statements, briefly
explore ideas about problem solving in relationships, provide an
overview of the elements of good communication, and assist
students with practicing the speaker–listener technique.
12 To help students identify sources of social support in their
lives, have students recognize conditions that suggest the need
to end a relationship, help students learn how to break up
effectively, and increase students’ sense of self-worth in
relationships by emphasizing the importance of safety and

respect for healthy relationships.
13 To help students reflect on their goals in relationships and
help students gain closure on the course.
742 BRAITHWAITE, LAMBERT, FINCHAM, AND PASLEY
T
hi
s
do
cu
m
en
t i
s
co
py
ri
gh
te
d
by
th
e
A
m

er
ic
an
P
sy
ch
ol
og
ic
al
A
ss
oc
ia
tio
n
or
o
ne
o
f i
ts
a
lli

ed
p
ub
lis
he
rs
.
T
hi
s
ar
tic
le
is
in
te
nd
ed
s
ol
el
y
fo
r t

he
p
er
so
na
l u
se
o
f t
he
in
di
vi
du
al
u
se
r a
nd
is
n
ot
to
b

e
di
ss
em
in
at
ed
b
ro
ad
ly
.
Regarding the clinical significance of these findings, we
examined the occurrence or absence of each of the behaviors
and found that RU produced a 58% reduction in the occurrence
of extradyadic sexual intercourse; the reduction in the control
group was 33%. Being sexually intimate without intercourse
was reduced by 50% among RU participants, whereas it in-
creased by 50% among control participants. Caressing and
hugging decreased by 37% for RU participants, but increased
by 8% for control participants. Extradyadic kissing reduced by
52% among RU participants, whereas it remained the same
among participants in the control condition.
Alternate Models

To rule out the possibility that the changes were being
driven by socially desirable responding, we analyzed an
alternate model in which social desirability (scores from the
Marlowe–Crowne Social Desirability Scale; Reynolds,
1982) was added as an exogenous predictor of the slope and
the intercept of extradyadic involvement. Social desirability
predicted the intercept but not the slope of extradyadic
involvement and did not attenuate the associations between
treatment and slope, indicating that individual desire to look
good influenced initial reports of how much extradyadic
involvement the participant was engaging in, but not the
change in extradyadic involvement over time.
Discussion
Our results suggest that RU reduces the overall frequency
of extradyadic involvement among college students over the
course of an academic semester. We found that individuals
in the RU group reported more extradyadic involvement at
baseline, but examination of the association between base-
line scores and individual slopes showed that initial rates of
extradyadic involvement were not reliably associated with
changes in extradyadic involvement over time, whereas
condition did reliably predict trajectories over the 12 weeks
of the study. In addition, we found that being female did not
predict less extradyadic involvement at baseline, but it did
predict less extradyadic involvement over time across both
conditions.Figure 2. Latent growth curve model. T1 � Time 1.
Figure 1. Measurement model for infidelity variables at each
time point. �2(12) � 2,736.21, p �
.01, comparative fit index � .99, Tucker–Lewis Index �.99,
root-mean square error of approxi-
mation � .06. T1 � Time 1.

in
at
ed
b
ro
ad
ly
.
Why would being female predict changes in extradyadic
involvement over time across both treatment and control
conditions? It is possible that our education-as-usual control
condition was more potent than we had anticipated. Recall
that the control condition was identical to the treatment
condition in terms of course content with the exception of
the weekly breakout sessions. Perhaps the information
taught in the introductory family science class was sufficient
to change female students’ behavior (but not male students’)
without the addition of the RU component, which would
suggest that RU may be more beneficial for men and that
non–skills-based relationship education may be sufficient to
change women’s rates of extradyadic involvement, but fur-
ther research is needed to examine this question more spe-
cifically. A selection bias for an introductory family science
class may also be possible—these students (in particular, the
women for whom there was a significant main effect) may
be more interested in improving their relationships and thus
more likely to respond not only to our intervention, but also
to the class as usual. If this is true, however, it also suggests
that the RU intervention had a higher hurdle to clear to

demonstrate incremental decreases in extradyadic involve-
ment than if we had compared it with another class that
taught unrelated content (e.g., an introductory biology
course). This question would be interesting to explore in
future research.
Implications for Dissemination
This study shows that relationship education can be ef-
fectively delivered as part of a college course. Even with al l
of the practical limitations imposed by the structure of a
college course (e.g., the need to meet university curriculum
requirements), RU can be seamlessly integrated into exist-
ing college courses, especially those in family science or
psychology departments. This fact is important to establish
because one of the obstacles to implementing relationship
education in colleges is that it is not feasible or desirable to
start new courses that are designed to serve only as a vehicle
for relationship education and that have little relevance to
the other goals of universities.
The applied nature of this curriculum may seem to differ
somewhat from the usual content of college coursework, but
it is consistent with the stated goals of most universities. For
example, the university at which this intervention took place
lists as its primary mission “the development of new gen-
erations of citizen leaders, based on the concepts inscribed
in our seal: Vires, Artes, Mores—Strength, Skill and Char-
acter.” Thus, to develop not only theoretical knowledge but
practical knowledge— or skill— of how to develop and
maintain healthy relationships across the life span is surely
within the scope of a university education. Moreover, many
universities seek to discourage alcohol abuse and promote
physical and mental health; not only does promoting healthy
relationships complement these kinds of effort, but it may

also prove a means to improve outcomes across the multiple
domains university administrations seek to improve (Braith-
waite & Fincham, 2009). The college years are a pivotal
opportunity for intervention because risky behaviors predic-
tive of extradyadic involvement such as heavy drinking are
prevalent on college campuses (Graham, Fincham, & Lam-
bert, 2009; Slutske, 2005). Also, receiving college credit
provides a useful incentive to increase student participation.
As such, colleges and universities may become effective
platforms to administer empirically validated relationship
education without requiring much in the way of start-up
costs.
Although this research was conducted at a large univer-
sity, future research will ideally examine RU when it is
delivered at educational institutions that have higher pro-
portions of populations that have not traditionally received
relationship education, such as community colleges or col-
leges in rural areas. It may also be possible to adapt this
curriculum to high school populations, which would extend
the potential reach of relationship education to an even
greater degree and provide for even earlier entry of empir-
ically supported relationship education. This research is the
first step in showing that such a dissemination model is
viable.
Limitations and Future Directions
Inevitably, our study suffers from a number of limitations.
First, because this was an effectiveness study, we did not have
the experimental control of an efficacy study; specifically,
participants were not randomly assigned to condition. As such,
these encouraging findings need to be viewed as preliminary.
We have no reason to believe that the sections students choose
to sign up for, however, would have an influence on their
extradyadic experiences. Moreover, this limitation also repre-

sents a strength because it allows for an examination of the
impact of RU under the conditions in which it is likely to be
Table 2
Means Across Three Time Points
Scale
Relationship U Control
Hedges’ gBaseline 6 Weeks 13 Weeks Baseline 6 Weeks 13
Weeks
Intercourse 0.12 (0.32) 0.07 (0.26) 0.05 (0.22) 0.06 (0.24) 0.08
(0.27) 0.04 (0.20) .05
Sexual intimacy 0.14 (0.34) 0.09 (0.29) 0.07 (0.26) 0.04 (0.21)
0.08 (0.27) 0.06 (0.24) .04
Caressing and hugging 0.46 (0.50) 0.37 (0.48) 0.29 (0.46) 0.36
(0.48) 0.38 (0.49) 0.39 (0.49) �.22
Kissing 0.25 (0.43) 0.18 (0.38) 0.12 (0.33) 0.10 (0.31) 0.13
(0.34) 0.10 (0.31) .06
Composite 0.96 (1.32) 0.71 (1.15) 0.54 (0.99) 0.57 (0.99) 0.66
(1.13) 0.59 (0.98) �.06
Note. Values in parentheses are standard deviations.
744 BRAITHWAITE, LAMBERT, FINCHAM, AND PASLEY
T
hi
s
do
cu
m

en
t i
s
co
py
ri
gh
te
d
by
th
e
A
m
er
ic
an
P
sy
ch
ol
og
ic
al

A
ss
oc
ia
tio
n
or
o
ne
o
f i
ts
a
lli
ed
p
ub
lis
he
rs
.
T
hi
s

ar
tic
le
is
in
te
nd
ed
s
ol
el
y
fo
r t
he
p
er
so
na
l u
se
o
f t
he

in
di
vi
du
al
u
se
r a
nd
is
n
ot
to
b
e
di
ss
em
in
at
ed
b
ro
ad

ly
.
used. Also, it is time for relationship education to move beyond
efficacy studies to examine the impact of these interventions
under the more practical conditions of impleme ntation. Sec-
ond, it would be ideal if we could track students for longer
periods of time to see whether these gains are maintained. We
are currently conducting research to address this limitation; it is
likely that the gains will be maintained given the fact that that
the interventions RU builds on have demonstrated robust and
lasting effects for newlywed couples (e.g., Laurenceau, Stan-
ley, Olmos-Gallo, Baucom, & Markman, 2004). Finally, future
research is needed to examine the impact of this kind of
intervention on individuals who do not pursue higher educa-
tion; as previously mentioned, RU could easily be adapted for
late adolescents and young adults for delivery through com-
munity organizations or high school courses.
In conclusion, this study provides evidence that under
real-world conditions, RU constitutes an effective preven-
tive intervention for extradyadic involvement. Because cou-
ples therapists have indicated that infidelity is the third most
difficult problem to treat (Whisman, Dixon, & Johnson,
1997), it addresses early a behavior that, if left unaddressed,
is difficult to later treat. Finally, it also represents an im-
portant step forward in the dissemination of empirically
supported relationship education. As we continue to work to
find novel ways of extending the reach of relationship
education using existing structures such as colleges and
other educational institutions, we will move closer to the
goal of getting these interventions to the individuals who
will benefit from them the most.

References
Atkins, D. C., Jacobson, N. S., & Baucom, D. H. (2001). Under-
standing infidelity: Correlates in a national random sample.
Jour-
nal of Family Psychology, 15, 735–749.
Baldwin, S. A., Murray, D. M., & Shadish, W. R. (2005).
Empir-
ically supported treatments or Type I errors? Problems with the
analysis of data from group-administered treatments. Journal of
Consulting and Clinical Psychology, 73, 924 –935.
Barlow, D. H., Levitt, J. T., & Bufka, L. F. (1999). The
dissemi-
nation of empirically supported treatments: A view to the
future.
Behaviour Research and Therapy, 37, 147–162.
Blossfeld, H. P., & Timm, A. (2003). Who marries whom? Edu-
cational systems as marriage markets in modern societies.
Springer: Netherlands.
Braithwaite, S. R., Delevi, R., & Fincham, F. D. (2010).
Romantic
relationships and the physical and mental health of college stu-
dents. Personal Relationships, 17(1), 1–12.
Braithwaite, S. R., & Fincham, F. D. (2009). A randomized clin-
ical trial of a computer based preventive intervention: Replica-
tion and extension of ePREP. Journal of Family Psychology, 23,
32–38.
Fincham, F. D., Stanley, S. M., & Rhoades, G. (in press). Rela-
tionship education in emerging adulthood: Problems and pros-

pects. In F. D. Fincham & M. Cui (Eds.), Romantic
relationships
in emerging adulthood. Cambridge: Cambridge University
Press.
Flay, B. R. (1986). Efficacy and effectiveness trials (and other
phases of research) in the development of health promotion
programs. Preventive Medicine, 15, 451– 474.
Glass, S. P. (2002). Couple therapy after the trauma of
infidelity. In
A. S. Gurman & N. S. Jacobson (Eds.), Clinical handbook of
couple
therapy (3rd ed., pp. 488 –507). New York, NY: Guilford Press.
Graham, S. M., Fincham, F. D., & Lambert, N. M. (2009).
Getting
drunk and sleeping around: Problem drinking and its association
with infidelity. Manuscript submitted for publication.
Hall, J. H., & Fincham, F. D. (2009). Psychological distress:
Precursor or consequence of dating infidelity? Personality and
Social Psychology Bulletin, 35, 143–159.
Jakubowski, S. F., Milne, E. P., Brunner, H., & Miller, R. B.
(2004).
A review of empirically supported marital enrichment
programs�.
Family Relations, 53(5), 528 –536. doi:10.1111/j.0197-
6664.2004
.00062.x
Lamanna, M. A., & Riedmann, A. (2009). Marriages and
families:
Making choices in a diverse society (10th ed.). Belmont, CA:

Wadsworth.
Laurenceau, J. P., Stanley, S. M., Olmos-Gallo, A., Baucom, B.,
&
Markman, H. J. (2004). Community-based prevention of marital
dysfunction: Multilevel modeling of a randomized effectiveness
study. Journal of Consulting and Clinical Psychology, 72, 933–
942.
Markman, H. J., Whitton, S. W., Kline, G. H., Stanley, S. M.,
Thompson, H., Peters, M. S., . . . Cordova, A. (2004). Use of an
empirically based marriage education program by religious or -
ganizations: Results of a dissemination trial. Family Relations,
53, 504 –512.
Muthén, L. K., & Muthén, B. O. (2007). Mplus User’s Guide
(5th
ed.). Los Angeles: Muthén & Muthén.
Pearson, M., Stanley, S., & Kline, G. (2005). Within My Reach:
Instructor manual. Greenwood Village, CO: PREP for Individ-
uals, Inc.
Reynolds, W. M. (1982). Development of reliable and valid
short
forms of the Marlowe–Crowne Social Desirability Scale.
Journal
of Clinical Psychology, 38, 119 –125.
Shackelford, T. K., LeBlanc, G. J., & Drass, E. (2000).
Emotional
reactions to infidelity. Cognition & Emotion, 14(5), 643– 659.
Slutske, W. (2005). Alcohol use disorders among US college
students and their non-college-attending peers. Archives of
Gen-

eral Psychiatry, 62, 321–327.
Stoops, N. (2004). Educational attainment in the United States:
2003 (Current Population Reports, P20 –550). Washington, DC:
U.S. Census Bureau.
Whisman, M. A., Dixon, A. E., & Johnson, B. (1997).
Therapists’
perspectives of couple problems and treatment issues in couple
therapy. Journal of Family Psychology, 11, 361–366.
Wiederman, M., & Hurd, C. (1999). Extradyadic involvement
during dating. Journal of Social and Personal Relationships, 16,
265–274.
Received November 11, 2009
Revision received September 2, 2010
Accepted September 9, 2010 �
T
hi
s
do
cu
m
en
t i
s
co

py
ri
gh
te
d
by
th
e
A
m
er
ic
an
P
sy
ch
ol
og
ic
al
A
ss
oc
ia

tio
n
or
o
ne
o
f i
ts
a
lli
ed
p
ub
lis
he
rs
.
T
hi
s
ar
tic
le
is

in
te
nd
ed
s
ol
el
y
fo
r t
he
p
er
so
na
l u
se
o
f t
he
in
di
vi
du

al
u
se
r a
nd
is
n
ot
to
b
e
di
ss
em
in
at
ed
b
ro
ad
ly
.

https://doi.org/10.1177/0956797618776942
Psychological Science
2018, Vol. 29(9) 1540 –1547
© The Author(s) 2018
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/0956797618776942
www.psychologicalscience.org/PS
ASSOCIATION FOR
PSYCHOLOGICAL SCIENCEShort Report
Two core values guide human behavior: communion
(i.e., promotion of other people) and agency (i.e., self-
promotion; Bakan, 1966). Pursuing communion is espe-
cially predictive of psychological health (Le, Impett,
Kogan, Webster, & Cheng, 2013). Yet men value com-
munion relatively less than do women (Donnelly &
Twenge, 2017), a difference mirrored by men’s lower
family orientation and participation in communal careers
(Croft, Schmader, & Block, 2015; Diekman, Steinberg,
Brown, Belanger, & Clark, 2017). Because men’s lower
communal engagement might negatively affect them-
selves as well as their families (Croft et al., 2015), it is
important to understand the early development of gen-
der differences in communal values and future role
expectations. In the current study, we examined whether
young boys explicitly devalue communion (and perhaps
accentuate agency) in ways that relate to lower antici-
pated prioritization of family over career.
By the age of 6 years, boys expect to prioritize career
over family (Croft, Schmader, Block, & Baron, 2014).

Cognitive-development theory suggests that these
gender differences in anticipated roles might arise
because children conform to gendered behavioral scripts
that align with their gender identity (e.g., Kohlberg,
1966; Martin & Ruble, 2009). However, theoretically,
gender identification (i.e., as feminine or masculine)
and gender expression (i.e., exhibiting stereotypically
feminine or masculine preferences; American Psycho-
logical Association Task Force on Gender Identity and
Gender Variance, 2009) are related to, but distinct from,
more fundamental communal and agentic values
(Spence & Buckner, 2000). On the basis of goal-congruity
theory (Diekman et al., 2017), we expected that the
internalization of values, more than gender identifica-
tion or expression, would predict children’s expectations
of their future roles.
776942PSSXXX10.1177/0956797618776942Block et al.Early
Gender Differences in Core Values
research-article2018
Corresponding Author:
Katharina Block, Department of Psychology, The University of
British
Columbia, 2136 West Mall, V6T 1Z4, Vancouver, British
Columbia,
Canada
E-mail: [email protected]
Early Gender Differences in Core
Values Predict Anticipated Family
Versus Career Orientation
Katharina Block, Antonya Marie Gonzalez, Toni Schmader,
and Andrew Scott Baron
Department of Psychology, The University of British Columbia

Abstract
Communion and agency are often described as core human
values. In adults, these values predict gendered role
preferences. Yet little work has examined the extent to which
young boys and girls explicitly endorse communal and
agentic values and whether early gender differences in values
predict boys’ and girls’ different role expectations. In a
sample of 411 children between the ages of 6 and 14 years, we
found consistent gender differences in endorsement
of communal and agentic values. Across this age range, boys
endorsed communal values less and agentic values more
than did girls. Moreover, gender differences in values partially
accounted for boys’ relatively lower family versus career
orientation, predicting their orientation over and above gender
identification and parent reports of children’s gender
expression. These findings suggest that gender differences in
core values emerge surprisingly early in development
and predict children’s expectations well before they make
decisions about adopting adult roles in their own families.
Keywords
childhood development, personal values, sex differences, gender
roles, open data, open materials
Received 8/24/17; Revision accepted 4/10/18
http://www.psychologicalscience.org/ps
mailto:[email protected]
https://us.sagepub.com/en-us/journals-permissions
http://crossmark.crossref.org/dialog/?doi=10.1177%2F09567976
18776942&domain=pdf&date_stamp=2018-06-22
Early Gender Differences in Core Values 1541

According to goal-congruity theory, people seek out
roles that afford their values. Adults self-segregate into
different occupations not only to conform to gendered
expectations but also because of the different values
that men and women have internalized (Diekman et al.,
2017). If these values are internalized early, they could
shape how children imagine their future. Whereas no
research has directly examined children’s endorsement
of core values, children’s preferences provide indirect
evidence. Girls, more than boys, want to connect rather
than compete in friendships (Ojanen, Grönroos, &
Salmivalli, 2005), and girls value altruism rather than
status in careers (Weisgram, Bigler, & Liben, 2010).
The current research examined gender differences
in core values and their relationship to family versus
career orientation in a large sample of children. We
hypothesized that boys endorse communal values less
than do girls (Hypothesis 1a). We had no predictions
for agentic values, given mixed evidence in adults
(Croson & Gneezy, 2009; Diekman et al., 2017). We also
hypothesized that boys (relative to girls) anticipate less
future family orientation than future career orientation
(Hypothesis 1b). To the degree that values are distinct
from gender identity and expression but important for
role preferences, they should explain unique variance
in children’s family versus career orientation. We thus
hypothesized that gender differences in communal (and
perhaps agentic) values mediate boys’ relatively lower
family versus career orientation, over and above explicit
and implicit gender identification (Hypothesis 2a) and
gender expression (Hypothesis 2b). Additionally, we
explored the developmental trajectory of these relation-
ships from childhood and early adolescence.
Method

Participants and procedure
Our final sample consisted of 411 children (216 boys,
195 girls) between the ages of 6 and 14 years (M = 9.84
years, SD = 2.23), who were recruited from a commu-
nity science center. We excluded 41 participants because
of experimenter error (n = 12), incomplete data (n =
9), or technical issues (n = 20). We aimed to recruit at
least 25 usable participants per age (in years) and gen-
der (target n = 400), given the feasibility of recruiting
children from the science center. Participants were pre-
dominantly Caucasian (54.5%) or East Asian (17.8%)
but also identified as South Asian (6.6%), Aboriginal/
Canadian First Nation (4.6%), Middle Eastern (1.5%),
mixed race (10.5%), Black (0.5%), or outside of the race
categories provided (2.7%). Each participant was tested
individually in a soundproof room by one of four
research assistants (one man, three women; research
assistant gender did not moderate the results; see the
Supplemental Material available online), who read all
instructions aloud to participants. Participants com-
pleted measures of implicit and explicit gender identi-
fication and their own values. They also answered
open-ended questions about their occupational inter-
ests and made ratings of their family versus career ori -
entation. The open-ended responses to the question,
“What do you want to be when you grow up?” (missing
data for 85 children) were coded for gender stereotypical -
ity and goal affordances. Because these coded variables
were unrelated to other variables in the study (for details,
see the Supplemental Material), they are not discussed
further. In addition, for 392 children, a parent completed
other demographics and rated his or her child’s gender
expression as masculine or feminine. A full list of mea-

sures can be found in the Supplemental Material.
Measures
Communal and agentic values. For the purposes of
this study, we developed a child-friendly scale of com-
munal and agentic value orientation by adapting items
used in adult samples (Diekman et al., 2017; Trapnell &
Paulhus, 2012). We first explained to children that “some
things are important to some people but not at all impor-
tant to other people. I want you to tell me how important
these things are to YOU!” Children were then asked to
rate the personal importance of four communal values
(“How important do YOU think it is to always help oth-
ers, even if it takes effort?” “How important do YOU think
it is to do things together with others?” “How important
do YOU think it is to be kind to others?” “How important
do YOU think it is to think about others’ feelings?” α =
.65) and three agentic values (“How important do YOU
think it is to be the one who gets to make decisions?”
“How important do YOU think it is to win?” “How impor-
tant do YOU think it is to be good at things?” α = .68) on
a 5-point scale ranging from 1 (not very important) to 5
(super important). A fourth agentic value, “doing things
all by yourself,” was not highly related to other items
(item-total correlation < .25) and was thus excluded from
the measure. Submitting the remaining seven items to an
exploratory maximum likelihood factor analysis with direct
oblimin rotation revealed that communal and agentic items
loaded on two distinct primary factors with minimal cross-
loading (see the analyses in the Supplemental Material).
Children’s reports of communal and agentic values were
weakly but significantly negatively correlated (r = −.18, p <
.001). An example item can be seen in the Appendix.
Family versus career orientation. To assess children’s

expected family versus career orientation for adulthood,
we had children rate two items taken from Croft et al.
1542 Block et al.
(2014). For each of the two items, children saw two indi -
viduals (matched to participant gender and similar to
each other in the physical appearance of skin color, hair-
cut, hair color, and facial features). These individuals
were described as childhood friends who grew up to
have different priorities as adults. For each of the two
pairs, one target was described as family oriented and
one was described as career oriented. After being read
each description, participants were asked, “Someday you
will also be all grown up. When you are grown up, who
do you think you will be more like?” Participants indi -
cated to whom they thought they would be more similar
on a 5-point scale from 1 (a lot similar to [name of career-
oriented exemplar]) to 5 (a lot similar to [name of family-
oriented exemplar]), in which options were represented
as dots of different sizes. Responses to these two i tems
(r = .34, p < .001) were averaged to represent an index of
children’s family versus career orientation. See the Sup-
plemental Material for a depiction of the measure.
Implicit gender identification. The extent to which
children implicitly identified as female versus male was
measured with a child-friendly Implicit Association Test
(IAT; Baron & Banaji, 2006). The IAT has been used and
validated as a measure of implicit gender identity in chil -
dren as young as 6 years old (e.g., Cvencek, Meltzoff, &
Greenwald, 2011). The IAT assesses the strength of asso-
ciations between concepts by measuring participants’
reaction times to categorize word and picture stimuli into

congruent versus incongruent categories (e.g., self + girl
vs. self + boy). Initially, participants completed separate
practice blocks to familiarize themselves with the identity
stimuli delivered auditorily (self = “I,” “me,” “my,” “myself”
vs. other = “they,” “them,” “their,” “themselves”; Dunham,
Baron, & Banaji, 2007) and gender stimuli delivered visu-
ally (pictures of boys and girls; see the Supplemental
Material for all stimuli used). Each practice block consisted
of 12 trials in which participants had to decide (using one
of two large response buttons) whether a word they heard
or picture they saw belonged to the category shown on
the left or right.
After finishing two practice blocks, participants com-
pleted two critical test blocks (40 trials each, in coun-
terbalanced order) requiring them to categorize stimuli
from both self and gender categories simultaneously.
In one block, category pairings congruent with a female
identity, that is, self and girl (vs. other and boy), were
mapped onto the same response button. In the other
block, category pairings congruent with a male identity,
that is, self and boy (vs. other and girl), were mapped
onto the same response button. Following scripts from
Baron and Banaji (2006), we computed d scores to
represent strength of implicit female (vs. male) identi -
fication. This scoring allowed gender identification to
positively correlate with other variables that were also
coded so that higher numbers equaled more feminine
or female. In analyses using this measure, we followed
recent general recommendations (Nosek, Bar-Anan,
Sriram, Axt, & Greenwald, 2014) to exclude participants
with more than 10% of responses faster than 300 ms or
with more than 30% errors on this task (19 participants,
4.6% of total sample).

Explicit gender identification. Four questions were
designed to assess the extent to which children explicitly
identified with other females versus males. For each of
these four questions, participants were presented with
clip-art depictions of one boy and one girl, matched to
each other in ethnicity (e.g., “Sarah and David”), and
then prompted, “I want you to tell me who you are more
like.” Participants made their ratings on a 5-point scale
that asked, “Are you . . . a lot more like X (1), a little more
like X (2), in the middle between X and Y (3), a little more
like Y (4), or a lot more like Y (5)?” In two of the four
pairs, the female character was presented on the right
(corresponding to a rating of 5), and in the other two, the
male character was presented on the right. To approxi-
mate ethnic representation in the sample community, we
assigned two boy-girl pairs to appear Caucasian, one pair
to appear East Asian, and one pair to appear medium
dark-skinned (meant to be interpretable as South Asian or
Latino). Ethnicity was thus never confounded with target
gender, and a composite score of all four items was reli -
able (α = .89).
However, to avoid contaminating scores with chil-
dren’s ethnic identity, we followed the recommendation
of an anonymous reviewer and operationalized gender
identification as the responses made only to the items
that matched the participant’s own parent-reported eth-
nicity (i.e., a Caucasian participant’s score is the average
of two Caucasian items, an East Asian participant’s score
is the rating of the East Asian item, a Black/Hispanic/
South Asian participant’s score is the score on the
medium dark-skinned item, and the full four-item com-
posite was used for children of mixed or nonreported
ethnicity). This ethnicity-matched coding of explicit
female identification correlated highly with the four-
item composite (r = .94, p < .001), and results are the

same with either measure (see the Supplemental Mate-
rial for these analyses and all stimuli). Scores were
coded so that higher numbers represented greater iden-
tification with females (vs. males).
Parent-reported gender expression. To assess the
extent to which children currently exhibited feminine ver-
sus masculine gender expressions, we asked parents to
complete a 12-item measure (Johnson et al., 2004) assessing
the frequency with which their child showed a number of
gendered behaviors and preferences (e.g., “He/She plays
with girl-type dolls, such as Barbie”). Parents also com-
pleted two face-valid items rating how (a) feminine and
(b) masculine they perceived their child to be compared
with other children of the same age. Because these latter
two items were highly related to the 12-item question-
naire (item-total correlations r > .74), we standardized all
14 items and averaged the standardized responses into
an overall index of parent-reported gender expression.
Higher scores on this measure indicated more feminine
versus masculine gender expression. See Table 1 for
means and correlations for all key variables.
Results
Gender and age differences
We first examined gender differences in our focal vari -
ables (communal values, agentic values, and family vs.
career orientation; Hypotheses 1a and 1b). To control
for age variation within our sample and explore the

possible developmental trajectory of boys’ and girls’
value endorsement, we also included age as a moderator
in these analyses. To examine gender and age effects,
we entered children’s gender (male = 0, female = 1) and
age (standardized) as predictors in Step 1 and their
interaction in Step 2 of a hierarchical linear regression
model for each outcome.
Value orientation. On average, older children endorsed
communal values less than did younger children, β =
−0.14, SE = 0.03, t(406) = −2.85, p = .005, whereas age did
not predict agentic values, β = 0.01, SE = 0.05, t(406) =
0.25, p = .803. Over and above these effects of age, and
as predicted, boys endorsed communal values signifi-
cantly less than did girls, β = 0.12, SE = 0.05, t(406) = 2.43,
p = .015. Boys also endorsed agentic values significantly
more than did girls, β = −0.14, SE = 0.10, t(406) = −2.93,
p = .004. There were no significant interactions between
gender and age in predicting values, βs < 0.10, ps > .16,
suggesting that the observed gender differences in values
varied little within the age range of our sample. Thus,
results suggest that gender differences in children’s core
values emerge by the age of 6 years and parallel gender
differences in valued careers in children (Weisgram et al.,
2010). In addition, values reported by children resemble
patterns of values in adults, with communal value
endorsement being generally high but higher for women
than for men (e.g., Diekman et al., 2017).
Family versus career orientation. We next examined
gender differences in children’s role orientation. In line
with Hypothesis 1b and replicating the results of Croft
et al. (2014), boys (compared with girls) expected to be
less family oriented, β = 0.18, SE = 0.09, t(394) = 3.53,
p < .001. Neither age, β = −0.08, SE = 0.05, t(394) = −1.61,

p = .108, nor the age-by-gender interaction, β = −0.04,
SE = 0.09, t(393) = −0.58, p = .564, predicted children’s
family versus career orientation. These results suggest
that by the time children are 6 years old, we can clearly
observe gender differences in values as well as family
versus career orientation.
Do gender differences in values
explain expected family versus career
orientation?
Having established an early gender difference in core
values, we next tested these gender differences as
mediators of gender differences in seeing one’s future
as more family oriented rather than career oriented. To
test this, we conducted mediational analyses with the
PROCESS macro (Hayes, 2013; 10,000 resamples for
bootstrapped confidence intervals, or CIs), entering
gender as a predictor and communal and agentic values
Table 1. Bivariate Correlations and Descriptive Statistics for
All Variables
Variable Boys (M) Girls (M)
Correlations
1 2 3 4 5 6 7
1. Age 9.98a (2.29) 9.69a (2.17) — −.15* −.06 −.12 −.37* .08
.12
†
2. Communal values (1–5) 4.38a (0.59) 4.52b (0.48) −.13
† — −.03 .15* .19* .03 .13†

3. Agentic values (1–5) 2.77a (1.00) 2.48b (0.99) .08 −.27* —
−.09 .07 −.01 .12
4. Family orientation (1–5) 3.05a (0.94) 3.38b (0.87) −.05 .18*
−.21* — .03 −.03 −.10
5. Gender expression (z score) −0.63a (0.33) 0.67b (0.38) .06
−.08 −.09 .12
† — .001 .12
6. Implicit female identification −0.22a (0.39) 0.27b (0.38) −.11
.08 .05 −.05 −.04 — −.07
7. Explicit female identification 2.05a (0.85) 4.01b (0.67) −.19*
−.001 −.01 .04 .21* .03 —
Note: Correlations for boys are presented below the diagonal,
and correlations for girls are presented above the diagonal.
Standard deviations
for means are given in parentheses. Within a row, means with
different subscripts are significantly different.
*p < .05. †p ≤ .10.
1544 Block et al.
as simultaneous mediators, predicting children’s family
versus career orientation (all standardized; see Fig. 1).
As hypothesized, expecting a family-oriented rather
than a career-oriented future was predicted by both
higher communal values, β = 0.14, SE = 0.05, t(395) =
2.81, p = .005, and lower agentic values, β = −0.13,
SE = 0.05, t(395) = −2.56, p = .011, even after we con-
trolled for gender. In addition, significant indirect
effects were consistent with boys’ lower communal val-
ues, indirect effect = 0.02, 95% CI = [0.003, 0.04], p <
.05, and higher agentic values, indirect effect = 0.02,
95% CI = [0.003, 0.05], p < .05, accounting, at least in

part, for their relatively low family versus career orien-
tation, compared with girls’. These indirect relationships
suggest that gender differences in both communal and
agentic values partly account for gender differences in
family versus career orientation. These results held
when we controlled for both participant age and
research assistant gender, and paths were not moder-
ated by age. For more information about these as well
as mediation analyses on younger versus older children,
refer to the Supplemental Material.
Next, we examined whether our results could be
accounted for simply by the extent to which children
implicitly or explicitly identified as female (vs. male)
or the extent to which children currently displayed
feminine versus masculine gender expression. Either
might suggest that children’s tendency toward identify-
ing as female or male or outwardly expressing feminine
or masculine behaviors and preferences, and not their
endorsement of communal values (constructs that are
distinct; Spence & Buckner, 2000), better predicts future
role expectations.
First, to better understand these variables in our data
set, we tested for gender and age effects on all variables
using linear regression analyses with children’s gender
(male = 0, female = 1) and age (standardized) as pre-
dictors in Step 1 and their interaction in Step 2 for each
outcome. As we expected on the basis of past research
(Cvencek et al., 2011), results showed that girls implic-
itly identified more as girls than did boys, β = 0.54,
SE = 0.04, t(387) = 12.47, p < .001, suggesting that
implicit gender identification corresponded sensibly to
children’s binary gender identity. Neither age, β = −0.02,
SE = 0.02, t(387) = −0.34, p = .733, nor the age-by-

gender interaction, β = 0.11, SE = 0.04, t(386) = 1.87,
p = .062, significantly predicted implicit gender identi -
fication.
Similarly, results showed a large gender difference
in explicit female identification: Girls explicitly identi-
fied more strongly with females than did boys, β = 0.78,
SE = 0.08, t(406) = 25.47, p < .001. Although we observed
no main effect of age, β = −0.04, SE = 0.04, t(406) =
−1.33, p = .184, the main effect of gender was qualified
by a significant age-by-gender interaction, β = 0.13,
SE = 0.06, t(405) = 3.23, p = .001. Decomposing this
interaction suggested that the tendency to explicitly
identify with girls more than with boys did not signifi-
cantly increase with age for girls, β = 0.07, SE = 0.06,
t(405) = 1.49, p = .137, but significantly decreased with
age among boys, β = −0.13, SE = 0.05, t(405) = −3.16,
p = .002.
We next examined how parents’ reports of their chil-
dren’s gender expression differed between boys and
girls. Results suggest that our sample showed substan-
tial gender differences in gender expression, with boys’
gender expression being rated as less feminine (more
masculine) by their parents than girls’ gender expres-
sion, β = 0.87, SE = 0.04, t(379) = 35.79, p < .001. Nota-
bly, gender and age interacted significantly when
Gender
Communal Values
Family Orientation
Agentic Values βb2 = –0.13*

βb1 = 0.14*βa1 = 0.11*
βa2 = –0.15*
βunmediated = 0.18*
βmediated = 0.14*
a1 × b1 = 0.02, 95% CI = [0.003, 0.04]
a2 × b2 = 0.02, 95% CI = [0.003, 0.05]
Fig. 1. Model showing how communal and agentic values
mediate the relationship between gender
(boys = 0, girls = 1) and family (vs. career) orientation, as
mediated by communal and agentic values.
Analyses included 399 children; additional children were
excluded for missing data. Asterisks indicate
significant paths (p < .05). CI = confidence interval.
predicting parent-rated gender expression, β = −0.18,
SE = 0.04, t(378) = −5.57, p < .001. Decomposing this
interaction suggested that, for girls, β = −0.23, SE = 0.03,
t(378) = −6.53, p < .001, but not for boys, β = 0.04,
SE = 0.02, t(378) = 1.13, p = .260, increasing age was
associated with less feminine (more masculine) gender
expression, suggesting that girls express fewer uniquely
feminine behaviors as they age, whereas boys tend to
express relatively more masculine behaviors at all ages
examined.
Do values predict family orientation

over and above identification and
expression?
Given the significant gender differences in gender
identification and parents’ reports of children’s gender
expression, it is possible that the relationship between
values and family versus career orientation is simply
a reflection of children’s explicit or implicit gender
identity (Hypothesis 2a) or gender expression
(Hypothesis 2b). To rule out these alternative expla-
nations, we repeated the above mediational analyses
2 times to test (a) the extent to which children implic-
itly and explicitly identified as female versus male
and (b) the extent to which children showed a femi-
nine versus masculine gender expression, as control
variables (z-scored) in the relationship between val-
ues and orientation (b path). These analyses were
done separately for gender identification and gender
expression to preserve degrees of freedom, because
some children were excluded on the basis of IAT error
rates, and a different subsample of children was
excluded because their parents did not complete their
questionnaire.
Results of analyses controlling for children’s implicit
and explicit gender identification (Hypothesis 2a)
revealed neither explicit identification as female versus
male, β = −0.08, SE = 0.08, t(374) = −1.02, p = .307, nor
implicit identification as female versus male, β = −0.04,
SE = 0.06, t(374) = −0.63, p = .527, significantly predict-
ing children’s family versus career orientation (over and
above dichotomous gender and values). Importantly,
both communal values, β = 0.13, SE = 0.05, t(374) =
2.45, p = .015, and agentic values, β = −0.11, SE = 0.05,
t(374) = −2.24, p = .025, remained significant predictors
of family versus career orientation when we controlled

for these two gender-identification variables. Moreover,
the indirect effects of child gender on family versus
career orientation through communal values, indirect
effect = 0.01, SE = 0.01, 95% CI = [0.002, 0.04], and
agentic values, indirect effect = 0.02, SE = 0.01, 95% CI =
[0.001, 0.05], also remained significant, although small.
Finally, results of parallel analyses that controlled
for gender expression (Hypothesis 2b) revealed that
this measure did not significantly predict children’s
family versus career orientation, β = 0.14, SE = 0.11,
t(367) = 1.30, p = .196. Importantly, both communal
values, β = 0.12, SE = 0.05, t(367) = 2.25, p = .025,
and agentic values, β = −0.12, SE = 0.05, t(367) =
−2.30, p = .022, remained significant predictors of
future family versus career orientation when gender
expression was added into the model. Moreover, the
indirect effects of gender on future family versus
career orientation through communal values, indirect
effect = 0.01, SE = 0.01, 95% CI = [0.001, 0.04], and
agentic values, indirect effect = 0.02, SE = 0.01, 95%
CI = [0.001, 0.05], also remained significant. Thus, we
found no evidence that either children’s current iden-
tification as female (vs. male) or their current gender
expression (as rated by their parents) was a better
explanation for their expected future selves than were
their own values.
General Discussion
In adults, communal and agentic values are important
predictors of gender roles. Little research has investi -
gated these values in children. Our results indicate that,
by the age of 6 years, boys show lower communal and
higher agentic values than do girls. Over and above
children’s gender identification and expression, valuing

communion less and agency more predicted boys’ rela-
tively lower family versus career orientation. Although
effect sizes were modest, effects could have cumula-
tively meaningful consequences for children’s aspira-
tions over development. Correlational designs preclude
causal inference. However, this evidence is consistent
with our hypothesis that values relate to children’s ori -
entation well before children confront the realities of
balancing family and career.
There was no evidence that effects were moder-
ated by age. Further research is needed to (a) under-
stand how, and how early, communal and agentic
values are internalized; (b) assess their causal impact
with prospective or experimental designs; and (c)
identify effects on actual behavioral outcomes. More-
over, studies should examine effects alongside non-
binary conceptions of gender identity (Martin,
Andrews, England, Zosuls, & Ruble, 2017). Societal
gender disparities in childcare and housework are
pressing issues (Croft et al., 2015). Our data under-
score the importance of understanding the early
development of core values and how they might set
the stage for gendered expectations about career and
family for adulthood.
1546 Block et al.
Appendix
Not Very ImportantSUPER Important
How important do YOU think it is to be kind to others?

Fig. A1. One of the communal-values questions and the
associated rating scale, as shown
to participants.
Action Editor
Erika E. Forbes served as action editor for this article.
Author Contributions
All the authors contributed to the development of the ques-
tions and study design. K. Block supervised data collection,
analyzed the data, and prepared the manuscript under the
supervision of A. S. Baron and T. Schmader. All the authors
provided critical feedback and edits to drafts of the manu-
script and approved the final version for submission.
Acknowledgments
We thank the members of the Living Lab for their hard work
collecting data and the members of Science World for their
partnership, which enables us to conduct research in their
community.
Declaration of Conflicting Interests
The author(s) declared that there were no conflicts of interest
with respect to the authorship or the publication of this
article.
Funding
This research was supported by Social Sciences and Humani -
ties Research Council grants to A. S. Baron (Grant No. 435-
2013-0286) and T. Schmader (Grant No. 895-2016-2011 and
Grant No. 895-2017-1025).

Supplemental Material
Additional supporting information can be found at http://
journals.sagepub.com/doi/suppl/10.1177/09567976 18776942
Open Practices
All data and materials have been made publicly available via
the Open Science Framework and can be accessed at https://
osf.io/fc8yt/. The complete Open Practices Disclosure for this
article can be found at http://journals.sagepub.com/doi/
suppl/10.1177/0956797618776942. This article has received
the badges for Open Data and Open Materials. More informa-
tion about the Open Practices badges can be found at http://
www.psychologicalscience.org/publications/badges.
References
American Psychological Association Task Force on Gender
Identity and Gender Variance. (2009). Report of the APA
Task Force on gender identity and gender variance.
Washington, DC: Author. Retrieved from http://www.apa
.org/pi/lgbt/resources/policy/gender-identity-report.pdf
Bakan, D. (1966). The duality of human existence: An essay
on psychology and religion. Chicago, IL: Rand McNally.
Baron, A. S., & Banaji, M. R. (2006). The development of
implicit attitudes: Evidence of race evaluations from ages
6 and 10 and adulthood. Psychological Science, 17, 53–58.
doi:10.1111/j.1467-9280.2005.01664.x
Croft, A., Schmader, T., & Block, K. (2015). An underex-

amined inequality: Cultural and psychological barriers
to men’s engagement with communal roles. Personality
and Social Psychology Review, 19, 343–370. doi:10
.1177/1088868314564789
Croft, A., Schmader, T., Block, K., & Baron, A. S. (2014).
The second shift reflected in the second generation: Do
parents’ gender roles at home predict children’s aspi -
rations? Psychological Science, 25, 1418–1428. doi:10
.1177/0956797614533968
Croson, R., & Gneezy, U. (2009). Gender differences in pref-
erences. Journal of Economic Literature, 47, 448–474.
doi:10.1257/jel.47.2.448
Cvencek, D., Meltzoff, A. N., & Greenwald, A. G. (2011).
Math-gender stereotypes in elementary school children.
Child Development, 82, 766–779. doi:10.1111/j.1467-8624
.2010.01529.x
Diekman, A. B., Steinberg, M., Brown, E. R., Belanger, A. L.,
& Clark,
E. K. (2017). A goal congruity model of role entry, engage-
ment, and exit: Understanding communal goal processes
in STEM gender gaps. Personality and Social Psychology
Review, 21, 142–175. doi:10.1177/1088868316642141
42
42
https://osf.io/fc8yt/
https://osf.io/fc8yt/
42

httpsdoi.org10.11770956797619835147Psychological Scie

httpsdoi.org10.11770956797619835147Psychological Scie

Recommended

Recommended

More Related Content

Similar to httpsdoi.org10.11770956797619835147Psychological Scie

Similar to httpsdoi.org10.11770956797619835147Psychological Scie (16)

More from PazSilviapm

More from PazSilviapm (20)

Recently uploaded

Recently uploaded (20)

httpsdoi.org10.11770956797619835147Psychological Scie