The association between discontinuing hormonal
contraceptives and wives’marital satisfaction
depends on husbands’facial attractiveness
V. Michelle Russell
, James K. McNulty
, Levi R. Baker
, and Andrea L. Meltzer
Department of Psychology, Florida State University, Tallahassee, FL 32306; and
Department of Psychology, Southern Methodist University, Dallas, TX 75275
Edited by Steven J. C. Gaulin, University of California, Santa Barbara, CA, and accepted by the Editorial Board October 16, 2014 (received for review August 1, 2014)
How are hormonal contraceptives (HCs) related to marital well-
being? Some work suggests HCs suppress biological processes
associated with women’s preferences for partner qualities reflective
of genetic fitness, qualities that may be summarized by facial attrac-
tiveness. Given that realizing such interpersonal preferences posi-
tively predicts relationship satisfaction, any changes in women’s
preferences associated with changes in their HC use may interact
with partner facial attractiveness to predict women’s relationship
satisfaction. We tested this possibility using two longitudinal studies
of 118 newlywed couples. Trained observers objectively rated hus-
bands’facial attractiveness in both studies. In study 1, wives reported
their marital satisfaction every6mofor4yandthenreportedthe
history of their HC use for their relationship. In study 2, wives re-
ported whether they were using HCs when they met their husbands
and then their marital satisfaction and HC use every 4 mo for up to
three waves. In both studies, and in an analysis that combined the
data from both studies, wives who were using HCs when they
formed their relationship with their husband were less satisfied with
their marriage when they discontinued HCs if their husband had
a relatively less attractive face, but more satisfied if their husband
had a relatively more attractive face. Beginning HCs demonstrated no
consistent associations with marital satisfaction. Incongruency be-
tween HC use at relationship formation and current HC use was
negatively associated with sexual satisfaction, regardless of hus-
bands’facial attractiveness. These findings suggest that HC use
may have unintended implications for women’s close relationships.
The majority of women living in industrialized nations use
hormonal contraceptives (HCs) at some point during their
lives (1, 2). Adding to a robust literature that already documents
important intrapersonal implications of HCs (3), a growing litera-
ture suggests that HCs may also have an important interpersonal
implication—they may affect women’s evaluations of their romantic
According to biological perspectives on human mating, women
evolvedtobeattractedtopartnerqualities reflective of genetic
fitness, qualities that may be summarized by facial attractiveness (4–
7). However, there is some evidence that HC use weakens the
hormonal processes that partially account for these preferences (8)
(for exceptions, see refs. 9 and 10). Specifically, a few studies de-
monstrate that women using HCs show weaker cyclical shifts in
preferences for cues of genetic fitness than do nonusers (11–14).
Further, a few additional studies demonstrate that women using
HCs demonstrate weaker overall preferences for cues of genetic
fitness than do nonusers (15, 16). In one study, for example, women
showed a weaker preference for facial masculinity when using HCs
versus not (16).
What then are the implications of HCs for women’slong-term
relationships? Some women use HCs before entering a committed
relationship and thus choose a long-term partner while using HCs.
However, at some point during the relationship, women must
discontinue using HCs to conceive, which the majority of women
eventually do (17). Other women, in contrast, may choose a part-
ner while not using HCs and at some point during the relationship
may begin using HCs as they become sexually active. It is possible
that any changes in preferences for partner genetic fitness associ-
ated with changes in HC use may have implications for women’s
relationship satisfaction. According to interdependence theory (18,
19) and supportive research (20), having a partner who meets one’s
interpersonal preferences is positively associated with overall re-
lationship satisfaction. Thus, the changes in women’spreferences
for cues of partner genetic fitness that may accompany corre-
sponding changes in their HC use may interact with actual cues of
their partners’genetic fitness, such as overall facial attractiveness,
to predict women’s relationship satisfaction. Specifically, women
who form a relationship when using HCs and later discontinue
using HCs may begin to prioritize cues of partner genetic fitness to
a greater extent and thus experience increases in satisfaction if
their partner’s face contains numerous cues of genetic fitness (i.e.,
is relatively attractive), but decreases in satisfaction if their part-
ner’s face contains fewer cues of genetic fitness (i.e., is relatively
less attractive). In contrast, women who form a relationship when
not using HCs and later begin using HCs may begin to prioritize
cues of partner genetic fitness to a lesser extent and thus experi-
ence decreases in satisfaction if their partner’s face contains nu-
merous cues of genetic fitness (given that such cues may become
less important to them), but subsequent increases in satisfaction if
their partner’s face contains fewer cues of genetic fitness.
We are aware of two published studies that have examined the
implications of HCs for women’s established relationships.
Roberts et al. (21) reported that women who used HCs when
Hormonal contraceptives (HCs) are believed to suppress
biological processes associated with women’s preferences for
cues of partner genetic fitness, cues that may be summarized by
men’s facial attractiveness. Two longitudinal studies of marriage
demonstrate that wives who used HCs at relationship formation
became less satisfied when they discontinued HCs if their hus-
band had a relatively less attractive face, but more satisfied if
their husband had a relatively more attractive face. Incongruency
between HC use at relationship formation and current HC use
was negatively associated with sexual satisfaction, regardless of
husbands’facial attractiveness. Practically, these findings suggest
that discontinuing HCs may have critical unintended effects on
women’s relationships. Theoretically, they indicate that evolved
mating processes have implications for established relationships.
Author contributions: V.M.R. and J.K.M. conceptualized the idea; L.R.B. and A.L.M. helped
conceptualize the idea; V.M.R. and J .K.M. designed res earch; V.M.R. and J. K.M. per-
formed research; V.M.R., J.K.M., and L. R.B. analyzed data; an d V.M.R., J.K.M., L .R.B.,
and A.L.M. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission. S.J.C.G. is a guest editor invited by the Editorial Board.
To whom correspondence may be addressed. Email: firstname.lastname@example.org or mcnulty@
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
www.pnas.org/cgi/doi/10.1073/pnas.1414784111 PNAS Early Edition
they chose their partner and then bore children with those
partners (and thus must have discontinued using HCs) were less
satisfied with the sexual aspects of their relationships, but more
satisfied with their partners’financial provision. Likewise, Roberts
et al. (22) reported that women who discontinued using HCs
during an ongoing relationship reported lower levels of sexual
satisfaction and no changes in satisfaction with their partners’
financial provision and intelligence. However, neither of these
studies examined the role of cues of partner genetic fitness. As
noted, any shifts in women’s preferences for partner genetic
fitness that coincide with changes in HC use should interact
with such cues to predict women’s relationship satisfaction.
We used data obtained from two longitudinal studies to examine
whether the association between wives’HC use and their marital
satisfaction depended on (i)whetherwiveswereusingHCswhen
theyenteredintotheirrelationship with theirhusbandand(ii )their
husband’s facial attractiveness. Based on the possibility that HC dis-
continuation leads wives to more strongly prioritize partner facial
attractiveness, we expected HC discontinuation to interact with hus-
bands’facial attractiveness to predict wives’marital satisfaction, such
that HC discontinuation would be negatively associated with marital
satisfaction among wives married to husbands with relatively less at-
tractive faces but positively associated with marital satisfaction among
wives married to husbands with relatively more attractive faces. We
also examined whether beginning the use of HCs was positively as-
sociated with marital satisfaction among wives married to husbands
with relatively less attractive faces but negatively associated with
satisfaction among wives married to husbands with relatively more
attractive faces. Finally, we attempted to replicate the association
between changing HC use and wives’sexual satisfaction (21, 22).
Materials and Methods
Participants. The participants in study 1 were 48 couples who participated in
a broader longitudinal study of 135 newlywed couples. These 48 couples were
those for whom the wife provided HC data during the fourth and final year of
the study. The wives who did not provide the HC data either had discontinued
the study (n=11), had divorced or separated (n=14), or did not respond
to the inquiry (n=62). No wives refused to provide HC data. The couples who
provided th e data necessary to be included in these analyses did not dif fer
from those who did not provide such data on any of the variables examined
here (P>0.45). The participants in study 2 were 70 couples participating in
a broader longitudinal study of 79 newlywed couples. Nine couples were ex-
cluded because the wives had experienced menopause. See SI Materials and
Methods for details regarding recruitment and sample characteristics.
Procedure. At baseline, couples in both studies were either mailed a packet of
surveys to complete at home and bring with them to a laboratory session or
emailed a link to Qualtrics.com, where they completed surveys online before
their laboratory session. These surveys included a consent form approved by
the University of Tennessee Institutional Review Board and the Florida State
University Human Subjects Committee, measures of marital and sexual sat-
isfaction, other measures beyond the scope of the current analyses, and
a letter instructing couples to complete their questionnaires independently
of one another. In study 2, wives also reported whether they were pregnant
and, if not, whether they were currently using HCs and whether the couple
was trying to get pregnant. During the laboratory session, each member of
the couple was photographed and couples completed other tasks beyond
the scope of the current analyses. Couples in study 1 were paid US$80 and
couples in study 2 were paid US$100 for completing this baseline phase.
Subsequent to the baseline session, couples in study 1 were contacted ap-
proximately every 6 to 8 mo for ∼4 y and mailed the same marital and sexual
satisfaction measures, as well as other measures beyond the scope of the
current analyses. Couples in study 2 were contacted every 4 mo up to three
times and emailed a link to the same marital and sexual satisfaction measures,
the same measure of HC use and pregnancy-related issues, and other measures
beyond the scope of the current analyses. Couples in study 1 were paid US$50
and couples in study 2 were paid US$25 for completing these phases.
During the fourth and final year of study 1 (when the current hypotheses
were developed), wives were contacted via phone or email and asked to
provide information regarding their HC use over the entire course of their
relationship. Wiveswho responded wereprovided a link to Surveymonkey.com,
where they were asked to provide consent to answer additional questions
regarding their HC use. Participating wives were mailed US$25.
Wives’HC use at relationship formation. We asked wives in both studies whether
they were using HCs when they began their relationship with their husband
and dummy-coded their responses (0 =no, 1 =yes).
Wives’HC use during their relationship. In study 1, we asked wives to retrospec-
tively list the types of HCs they used since they began their relationship and the
start and end dates for the periods during which they used each type of HC.
Although prior research has demonstrated that women are relatively accurate
when reporting their HC history (23), we took two steps to increase the accuracy
of their reports. First, we encouraged wives to consult their physician and/or
medical records if they were unsure of the type used or the period during which
they used it. Second, we verified that the type (and brand, if provided) of birth
control listed was in fact hormonal. We used this information to create a
dummy code indicating whether or not wives had used HCs during each phase
of data collection (0 =no, 1 =yes). [Progesterone-only HC formulations may
affect mate preferences differently than HC formulations containing estrogen
(24). No women reported using progesterone-only formulations when they
began their relationships with their husbands and one woman reported using
a progesterone-only formulation during the final two assessments of the study.
Notably, subsequent analyses indicated the three-way interaction remained
significant when excluding the two assessments during which that wife used
a progesterone-only HC; t(219) =−2.26, P=0.03.]
In study 2, we asked wives at each assessment whether they were using
HCs. We then formed a dummy code indicating whether they were using HCs
at the time they completed the assessment (0 =no, 1 =yes).
Facial attractiveness. According to some perspectives, facial attractiveness sum-
marizes overall genetic fitness (4–7). Given that our predictions regard overall
genetic fitness, rather than any specific cue of genetic fitness (e.g., masculinity,
symmetry, MHC heterogeneity), a group of trained research assistants (n=5in
study 1; n=4 in study 2) rated the facial attractiveness of each spouse from the
photographs that were taken at baseline, using a scale ranging from 1 to 10,
where higher ratings indicated more physically attractive faces. In both studies,
participants stood in front of a blank, white wall and were told to maintain
a comfortable expression while their photograph was taken. The photographs
used in study 1 depicted the spouses’faces from the shoulders up, and all
judges were instructed to rate the facial attractiveness only. The photographs
used in study 2 were similar, except they were additionally cropped so that only
the spouses’faces were shown. To further ensure that ratings of husbands’
facial attractiveness were not confounded with ratings of their body attrac-
tiveness, which may be less indicative of genetic fitness, judges also rated
separate photographs that depicted the bodies of husbands in both studies,
and we controlled for these ratings in supplemental analyses. Coders rated
each spouse independently. The reliability of our coders was adequate [in
study 1, interclass correlation (ICC) =0.78 for husbands and ICC =0.92 for
wives; in study 2, ICC =0.85 for husbands and ICC =0.88 for wives]. We used
the mean attractiveness ratings across judges as our measure of each spouse’s
Marital satisfaction. Although prior work examining the role of HCsfor women’s
satisfaction with nonsexual aspects of their relationship (21, 22) has examined
their satisfaction with specific aspects of the relationship (e.g., partners’fi-
nancial provision, intelligence), we used a global measure of marital satisfac-
tion to capture the extent to which wives were generally satisfied with their
marriages because we believed such a global evaluation would be most likely
to capture the extent to which wives were susceptible to the interactive effects
of their shifting preferences for cues of genetic fitness and their husbands’
facial attractiveness. Indeed, relationship scholars (25, 26) have pointed out
that examining the effects of specific processes (e.g., behavior, cognition,
preferences) on evaluations of a relationship requires assessing global evalu-
ations of the marriage. We assessed marital satisfaction in both studies with
the Quality Marriage Index (27), a measure that requires spouses to report
their agreement with six general statements regarding their marital satisfac-
tion (e.g., “My relationship with my partner makes me happy”). Five items use
a seven-point scale and one item uses a 10-point scale. All items were summed
for each participant. Internal consistency was acceptable (in study 1, α>0.85
for husbands and 0.88 for wives at each assessment; in study 2, α>0.95 for
husbands and 0.92 for wives at each assessment).
Sexual satisfaction. In study 1, sexual satisfaction was assessed with the Index of
Sexual Satisfaction (28), a measure that requires spouses to report their
agreement with 25 statements regarding their satisfaction with the sexual
relationship with their spouse using a seven-point scale (sample items in-
clude, “Sex is fun for my partner and me”). All items were summed for each
participant. Internal consistency was acceptable (α>0.94 for husbands and
wives at each assessment). In study 2, sexual satisfaction was assessed with
one item (i.e., “How satisfied are you with the quality of the sex you have
had with your spouse over the past 4 mo?”) using a seven-point scale.
www.pnas.org/cgi/doi/10.1073/pnas.1414784111 Russell et al.
Sixteen (33%) of the 48 wives in study 1 and 37 (53%) of the 70
wives in study 2 reported using HCs at relationship formation. Ten
(63%) of these wives in study 1 and 17 (46%) of these wives in study
2 reported discontinuing HCs at some point during their relation-
ship. Ratings of husbands’facial attractiveness were normally dis-
tributed in both studies (in study 1, skewness =0.12, kurtosis =
−0.42; in study 2, skewness =0.43, kurtosis =0.20). Growth curve
analyses indicated that these wives experienced declines in their
satisfaction over time [in study 1, b=−0.17, SE =0.05, t(234) =
−3.23, P<0.01; in study 2, b=−0.21, SE =0.06, t(156) =−3.40,
P<0.01]. Although partial correlations between wives’HC use at
relationship formation and husbands’facial attractiveness, con-
trolling for wives’facial attractiveness, did not reach significance in
study 1 (r=−0.15, P=0.31) or study 2 (r=−0.09, P=0.47), this
correlation was marginally significant in an analysis that combined
thedatafrombothstudies(r=−0.16, P=0.09), providing some
evidence that wives who used HCs at relationship formation chose
less attractive husbands. Husbands’facial attractiveness was not
significantly associated with whether or not wives discontinued HCs
[in study 1, b=−0.09, SE =0.13, t(44) =−0.71, P=0.48; in study 2,
b=0.09, SE =0.09, t(66) =0.88, P=0.32]. Descriptive statistics
and bivariate correlations for each study appear in Table S1.
To ease interpretation of parameter estimates, all of the fol-
lowing primary analyses were conducted using standardized ver-
sions of all continuous independent variables and grand-centered
values of all HC variables. We first tested the main effect of wives’
HC use for their marital and sexual satisfaction, without accounting
for whether they were using HCs at relationship formation or their
husbands’attractiveness. Using multilevel modeling, we regressed
wives’reports of satisfaction at each assessment onto the dummy
code indicating whether wives used HCs at each assessment. To
control the influence of relationship-specific factors and changes in
each type of satisfaction over time, we controlled for month of
assessment and husbands’corresponding levels of satisfaction by
estimating the following two-level model using the HLM 7 com-
puter program (Scientific Software International) (Eq. 1):
Ytiðwives’satisfactionÞ=b0i +b1i ðmonth of assessmentÞ
+b3iðHC statusÞ+e0ti +r0i :
We estimated Eq. 1six times: once for each type of satisfaction
(marital and sexual) in each study and once for each type of
satisfaction combining the data across studies, but controlling
for idiosyncratic differences between studies using a dummy
code. In the combined analyses, wives’sexual satisfaction was
standardized before combining the data due to the different
measures used in the two studies.
Results are presented in the section of Table 1 labeled as
model A. HC use was unassociated with either type of satisfac-
tion on average across all analyses except one: Among wives in
study 1, HC use was negatively associated with wives’marital
satisfaction, indicating that wives in study 1 were less satisfied
with their marriages on average when they used HCs. This as-
sociation did not reach significance in the other five analyses.
Next, we tested whether the associations between wives’HC
use and satisfaction estimated in the previous analyses were
moderated by whether they were using HCs at relationship for-
mation—i.e., whether beginning or discontinuing HCs was as-
sociated with wives’marital or sexual satisfaction. To do this, we
repeated the previous analyses except we also entered the
dummy code indicating whether wives were using HCs at re-
lationship formation to account for variance in the intercept and
current HC status slope estimates in the second level of the
model to create the current HC status ×HC status at relation-
ship formation interaction with the following equation (Eq. 2):
=b0i +b1iðmonth of assessmentÞ
+b3iðcurrent HC statusÞ
+b4iðHC status at relationship formationÞ
+b5iðcurrent HC status
×HC status at relationship formationÞ+e0ti +r0i :
In this model, the current HC status ×HC status at relationship
formation interaction tests the effect of HC congruency (i.e., begin-
ning or discontinuing HC use) on satisfaction. These tests are very
similar to the tests conducted by Roberts et al. (22), with the ex-
ception that our measure of relationship satisfaction was intention-
ally more global.
Results are presented in the section of Table 1 labeled as
model B. The current HC status ×HC status at relationship
formation interaction was not associated with wives’marital sat-
isfaction, indicating that HC congruency was unrelated with
general relationship satisfaction, on average. Nevertheless, as can
be seen in Table 1, the current HC status ×HC status at re-
lationship formation interaction was positively associated with
wives’sexual satisfaction in all three analyses, indicating that HC
congruency was positively associated with wives’sexual satisfac-
tion in both studies and the analysis that combined the two
studies. Replicating prior research (21, 22), wives who were using
HCs at relationship formation became less sexually satisfied when
they discontinued HCs in all three analyses [in study 1, t(188) =
−2.23, P=0.03; in study 2, t(82) =−2.07, P=0.04; in combined, t
(274) =−2.89, P<0.01]. Further, wives who were not using HCs
at relationship formation also became less sexually satisfied when
they began using HCs in study 1 and the combined analysis, [in
study 1, t(188) =−2.13, P=0.03; in combined, t(274) =−1.99,
P<0.05]. Although this effect did not quite reach significance
in study 2 [t(82) =−1.06, not significant (NS)], it also did not
differ statistically across the two studies [t(272) =0.00, NS].
Finally, we tested our primary prediction that the association
between HC congruency and wives’marital satisfaction depends on
husbands’facial attractiveness. To do this, we repeated the previous
analyses except this time we additionally added husbands’facial
attractiveness, and the HC status at relationship formation ×hus-
bands’facial attractiveness interaction to account for variance in the
intercept and current HC status slope estimates in the second level
of the model to create the crucial current HC status ×HC status at
relationship formation ×husbands’facial attractiveness interaction
and all lower level interactions with the following model (Eq. 3):
=b0i +b1iðmonth of assessmentÞ
+b2iðhusbands’satisfactionÞ+b3i ðcurrent HC statusÞ
+b4iðHC status at relationship formationÞ
+b6iðcurrent HC status
×HC status at relationship formationÞ
+b7iðcurrent HC status
+b8iðHC status at relationship formation
+b9iðcurrent HC status
×HC status at relationship formation
×husbands’facial attractivenessÞ+e0ti +r0i:
In this model, the current HC status ×HC status at relationship
formation ×husbands’facial attractiveness interaction tests
Russell et al. PNAS Early Edition
whether the association between HC congruency and satisfaction
depends on husbands’facial attractiveness.
Results are presented in the section of Table 1 labeled as
model C. The current HC status ×HC status at relationship
formation ×husbands’facial attractiveness interaction was not
associated with wives’sexual satisfaction in any of the analyses.
Nevertheless, the current HC status ×HC status at relationship
formation ×husbands’facial attractiveness interaction was
negatively associated with wives’marital satisfaction in all three
analyses. Notably, this interaction remained significant when
controlling for wives’facial attractiveness and husbands’body
attractiveness in study 1 [t(176) =−2.34, P=0.02] and the
combined analysis [t(261) =−2.41, P=0.02], and wives’facial
attractiveness, husbands’body attractiveness, whether wives were
pregnant, and whether the couple was trying to get pregnant in
study 2 [t(75) =−3.41, P<0.01].
Tests of the simple two-way interactions among wives not
using HCs when they formed their relationships revealed that
husbands’facial attractiveness did not moderate the effects of
HC use in any of the three analyses, although it was marginally
significant in study 2 [in study 1, t(176) =1.17, NS; in study 2,
t(79) =1.98, P=0.05; in combined, t(261) =0.58, NS]. Further,
the pattern of main effects of HC use among wives not using HCs
at relationship formation was inconsistent across studies 1 and 2
[in study 1, t(176) =−3.11, P<0.01; in study 2, t(79) =3.01, P<
0.01] and nonsignificant in the analysis that combined the data
from the two studies [t(261) =−0.53, NS]. In other words, be-
ginning HC use did not have consistent effects on wives’marital
satisfaction and the effects that did emerge did not reliably de-
pend on husbands’attractiveness.
Tests of the simple two-way interactions among wives who were
using HCs at relationship formation, in contrast, revealed that
husbands’facial attractiveness did moderate the effects of HC use
in all three analyses [in study 1, t(176) =−2.01, P<0.05; in study
2, t(79) =−2.56, P=0.01; in combined, t(261) =−2.45, P=0.02].
These significant interactions are plotted in Fig. 1. Given that
husbands’facial attractiveness was a continuous variable, we
followed the recommendation of Preacher et al. (29) to use the
Johnson–Neyman method to identify one-tailed regions of sig-
nificance of the simple effects of HC use among these wives—i.e.,
the exact levels of husbands’attractiveness at which HC discon-
tinuation demonstrated significant associations with wives’mari-
tal satisfaction. Consistent with predictions, discontinuing HCs
was positively associated with marital satisfaction among wives
Table 1. Associations between wives’marital satisfaction and initial contraceptive status, current contraceptive status, husbands’
facial attractiveness, and their interactions
Marital satisfaction Sexual satisfaction
Study 1 Study 2 Combined Study 1 Study 2 Combined
brbrbr b rbrbr
Intercept 40.23 —40.08 —39.95 —139.36 —5.82 —−0.02 —
Study ————−1.49 0.17 — ———−0.01 0.01
Month of assessment −1.11** 0.24 −1.05** 0.33 −1.22** 0.24 −0.64 0.04 −0.11 0.19 −0.05 0.05
Husbands’satisfaction 1.44* 0.17 3.91** 0.45 2.57** 0.24 10.71** 0.45 0.29** 0.29 0.36** 0.35
CHCS −1.57* 0.17 1.48 0.17 −0.37 0.04 −1.97 0.06 0.16 0.09 0.01 0.01
Intercept 40.23 —40.28 —39.95 —139.20 —5.77 —−0.03 —
Study ————−1.72* 0.19 — ———−0.04 0.03
Month of assessment −1.10** 0.24 −1.03** 0.32 −1.22** 0.24 −0.65 0.04 −0.11 0.19 −0.05 0.06
Husbands’satisfaction 1.44* 0.17 3.87** 0.47 2.55** 0.24 10.35** 0.44 0.31** 0.32 0.37** 0.36
CHCS −1.44* 0.15 1.39 0.17 −0.46 0.05 0.26 0.01 0.11 0.06 0.08 0.06
HCRF 0.54 0.07 0.64 0.08 1.00 0.12 −2.85 0.10 0.21 0.11 0.06 0.04
CHCS ×HCRF 1.19 0.07 −3.62 0.21 −0.16 0.01 15.08** 0.21 0.83* 0.23 0.64** 0.20
Intercept 40.27 —40.29 —40.00 —138.63 —5.84 —−0.03 —
Study ————−2.00* 0.21 — ———−0.06 0.04
Month of assessment −1.09** 0.23 −0.96** 0.30 −1.23** 0.23 −0.57 0.04 −0.12 0.20 −0.05 0.07
Husbands’satisfaction 1.41* 0.17 4.25** 0.52 2.60** 0.25 10.65** 0.45 0.31** 0.34 0.36** 0.36
CHCS −1.71** 0.19 1.63 0.20 −0.59 0.06 0.59 0.02 0.21 0.13 0.07 0.04
HCRF 0.59 0.07 1.20 0.15 1.21 0.14 −3.58 0.13 0.17 0.10 0.08 0.05
CHCS ×HCRF 1.04 0.07 −4.52* 0.27 −0.36 0.02 17.07** 0.27 0.47 0.15 0.60** 0.18
HFA −0.40 0.09 −0.50 0.14 −0.42 0.10 0.31 0.02 −0.05 0.07 −0.03 0.04
CHCS ×HFA −0.32 0.04 −0.67 0.09 −0.87 0.08 2.60 0.08 −0.43* 0.28 −0.13 0.08
HCRF ×HFA −0.00 0.00 1.65 0.23 0.10 0.01 −5.95 0.19 0.50* 0.30 0.00 0.00
CHCS ×HCRF ×HFA −3.64* 0.17 −5.46** 0.34 −2.96* 0.14 3.79 0.04 −0.24 0.08 −0.29 0.09
CHCS, current HC status; HCRF, HC status at relationship formation; HFA, husbands’facial attractiveness; r, effect size r.
Study 1, df =179 for marital satisfaction and df =189 for sexual satisfaction; study 2, df =82 for all other effects in marital satisfaction model and df =83 for
all other effects in sexual satisfaction model; combined, df =116 for study, df =264 for all other effects in marital satisfaction model, and df =275 for all other
effects in sexual satisfaction model.
Study 1, df =46 for HCRF, df =178 for all other effects in marital satisfaction model, and df =188 for all other effects in sexual satisfaction model; study 2, df =
68 for HCRF, df =81 for all other effects in marital satisfaction model, and df =82 for all other effects in sexual satisfaction model; combined, df =115 for study
and HCRF, df =263 for all other effects in marital satisfaction model, and df =274 for all other effects in sexual satisfaction model.
Study 1, df =44 for HFA, HCRF, and HFA ×HCRF, df =176 for all other effects in marital satisfaction model, and df =186 for all other effects in sexual
satisfaction model; study 2, df =66 for HFA, HCRF, and HFA ×HCRF, df =79 for all other effects in marital satisfaction model and df =80 for all other effects in
sexual satisfaction model; combined, df =113 for study, HFA, HCRF, and HFA ×HCRF, df =261 for all other effects in marital satisfaction model, and df =272
for all other effects in sexual satisfaction model. Wives’sexual satisfaction was standardized in the combined analyses.
www.pnas.org/cgi/doi/10.1073/pnas.1414784111 Russell et al.
with husbands who had faces relatively more attractive than the
mean (>0.46 SDs in study 1, >0.53 SDs in study 2, and >0.25 SDs
in the combined analysis), but negatively associated with marital
satisfaction among wives with husbands who had faces relatively
less attractive than the mean (<1.80 SDs in study 1, <1.16 SDs in
study 2, and <1.34 SDs in the combined analysis).
Although our primary focus regarded the moderating role of
husbands’attractiveness, we also examined the simple associa-
tions between husbands’attractiveness and wives’marital satis-
faction for wives who did versus did not discontinue HCs.
Consistent with expectations, husbands’facial attractiveness
trended toward being positively associated with marital satis-
faction among wives who discontinued HCs [in study 1, t(44) =
1.73, P=0.09; in study 2, t(66) =1.97, P=0.05; in combined,
t(113) =1.69, P=0.09]. In contrast, husbands’facial attrac-
tiveness was unassociated with marital satisfaction among wives
who continued using HCs in studies 1 and 2 [in study 1, t(44) =
−0.93, P=0.36; in study 2, t(66) =1.61, P=0.11], although it
was marginally negatively associated with marital satisfaction
among such wives in the combined analysis [t(113) =−1.731,
A growing body of research demonstrates that evolved processes
have implications for established relationships (30–36). In one set
of studies, naturally cycling women who described their partners as
more physically desirable experienced increased satisfaction with
those partners while fertile, whereas such women who described
their partners as less physically desirable experienced decreased
satisfaction with those partners while fertile (36).
However, HCs are believed to suppress the biological processes
that account for these effects and thus may have unintended con-
sequences for relationships (8). Providing some initial evidence for
such effects, Roberts et al. (21, 22) demonstrated that women who
began or discontinued HCs during their relationships were less
satisfied with sexual aspects of their relationships. The current re-
search provided a valuable independent replication of that research
by revealing that beginning or discontinuing HCs was negatively
associated with women’s sexual satisfaction in two longitudinal
studies of marriage. However, these two studies also extended this
prior work by revealing that the association between discontinuing
HCs and women’s marital satisfaction depended on their husbands’
facial attractiveness; whereas discontinuing HCs was associated
with greater marital satisfaction among wives with relatively more
attractive husbands, it was associated with lower satisfaction among
wives with relatively less attractive husbands. Likewise, husbands’
attractiveness was positively associated with satisfaction among
wives who discontinued using HCs. Partner attractiveness was un-
associated with wives’satisfaction among wives who continued using
HCs in each study, but negatively associated with wives’satisfaction
in the analysis that combined the data from the two studies. Hus-
bands’facial attractiveness did not moderate the effects of wives’
HC use on sexual satisfaction.
The fact that husbands’facial attractiveness moderated the
association between HC use and wives’marital but not sexual
satisfaction suggests that HC congruency may be associated with
marital and sexual satisfaction for different reasons. As others
have argued (30, 37, 38), global relationship satisfaction may
serve as a barometer of the extent to which the relationship
meets evolved preferences and needs. If so, the hormonal fluc-
tuations that coincide with ovulation may be an important cir-
cuitry through which this link emerges. Given that HCs are
thought to disrupt this circuitry, it makes sense that they would
weaken the extent to which relationship satisfaction is sensitive
to any cues of partner genetic fitness, including facial attrac-
tiveness. Discontinuing HCs may interact with husbands’facial
attractiveness because it recalibrates women’s relationship sat-
isfaction to be more sensitive to such cues. Sexual satisfaction as
measured here (28), in contrast, captures the extent to which
partners derive physical pleasure from their sexual activities. In
this light, any changes in sexual satisfaction should be due to
either changes in preferences for or attitudes toward sexual activi-
ties, or changes in sexual activities themselves. Perhaps, then, the
association between changes in HC use and decreases in sexual
satisfaction is due to hormonal shifts associated with changes
in sexual preferences, attitudes, and/or behaviors that are inde-
pendent of cues of partner genetic fitness. Future research may
benefit from addressing these possibilities.
Notably, we did not find consistent evidence that beginning
HCs since relationship formation was associated with wives’
marital satisfaction. Interestingly, prior research (22) also failed
to document consistent effects of beginning HCs for women’s
satisfaction with their partner’s financial provision/intelligence.
As others have argued (16), HCs may have their effects through
partner choice; women who make such choices while not using
HCs may be immune from later effects of HC use on their
general satisfaction. Indeed, prior research indicates that women
who used HCs at relationship formation tend to choose partners
with less masculine faces (16) and there was some weak evidence
that the women using HCs at relationship formation in our
studies chose less attractive husbands. Future research may
benefit by examining whether the null effects of beginning HCs
on marital satisfaction observed here and in prior research are
moderated by other factors, such as specific cues of partner ge-
netic fitness or factors not linked to genetic fitness.
Our confidence in these findings is enhanced by several
strengths. First, the key findings emerged (i) across two in-
dependent longitudinal studies, (ii) in an analysis that combined
the data across both studies, and (iii) controlling for numerous
potential confounds. Second, both studies relied partly on
within-person changes in HCs, helping to rule out the influence
Fig. 1. Husbands’facial attractiveness moderating the association between
wives’HC discontinuation and their marital satisfaction in study 1 (A), study
2(B), and the combined analyses (C).
Russell et al. PNAS Early Edition
of any between-person differences associated with HC discon-
tinuation. Third, the predicted effect remained significant con-
trolling for wives’reports of whether couples were attempting to
become pregnant in study 2, which helps rule out the possibility
that it was wives’motivations to become pregnant, rather than
hormonal changes associated with discontinuing HCs, that
interacted with husbands’attractiveness to account for wives’
satisfaction. Finally, the analyses provided evidence of a theo-
retical moderator of the influence of HCs on relationship satis-
faction—partner facial attractiveness—strengthening the extent
to which these results provide evidence that the effects emerged
due to the influence of HCs on evolved preferences for cues of
partner genetic fitness.
Nevertheless, several factors limit the interpretations and
generalizability of these results until they can be replicated and
extended. First, although the within-person design of these
studies helps rule out the influence of between-person differ-
ences, and although we controlled for important within-person
confounds, it remains possible that other confounding factors not
controlled in our analyses account for the interactive effects of
HC discontinuation. Only experimental research can definitively
rule out such alternative explanations. Second, the two samples
were relatively homogenous and relied on relatively modest
sample sizes. Any generalizations should be made with caution
until the findings can be replicated with a larger and more rep-
resentative sample. Third, given that previous research suggests
that estrogen may partially account for women’s mate prefer-
ences (39), and given that HCs vary in the amount of estrogen
they contain, it is possible that the current effects may be
stronger among women using HCs that contain lower amounts of
estrogen and weaker, or even nonexistent, among women using
HCs that contain higher amounts of estrogen. Of course, given
that HCs contain synthetic rather than endogenous hormones, it
is also possible that their effects may differ from the effects as-
sociated with endogenous hormones. Future research may ben-
efit from examining this possibility. Likewise, future research
may benefit from exploring whether the effects observed here
were driven by the peaks in endogenous estrogen experienced by
women once they discontinue HCs. Fourth, although some
perspectives posit that facial attractiveness is a powerful pre-
dictor of genetic fitness (4–7), the current study did not examine
the role of specific indicators of genetic fitness, such as sym-
metry, dominance, or partners’MHC dissimilarity. Future re-
search may benefit by attempting to conceptually replicate these
results using specific measures of these constructs to determine if
one or multiple indicators have similar or different implications.
Finally, it is worth highlighting the potential practical impli-
cations of this research. Marital satisfaction is strongly associated
with mental (40) and physical health (41); occupational out-
comes (42); life satisfaction (43); and a host of physical, mental,
and social outcomes for children (44). The fact that wives’HC
use was linked to their marital satisfaction suggests that HCs may
have far-reaching implications, both beneficial and harmful.
ACKNOWLEDGMENTS. This work was supported by a National Science
Foundation Graduate Research Fellowship under Grant DGEI246794 (to
V.M.R.) and the National Institute of Child Health and Development Grant
RHD058314 (to J.K.M.).
1. Daniels K, Mosher WD, Jones J (2013) Contraceptive methods women have ever used:
United States, 1982-2010. Natl Health Stat Rep 62(62):1–15.
2. Darroch JE (2013) Trends in contraceptive use. Contraception 87(3):259–263.
3. Burrows LJ, Basha M, Goldstein AT (2012) The effects of hormonal contraceptives on
female sexuality: A review. J Sex Med 9(9):2213–2223.
4. Fink B, Penton-Voak I (2002) Evolutionary psychology of facial attractiveness. Curr Dir
Psychol Sci 11(5):154–158.
5. Grammer K, Thornhill R (1994) Human (Homo sapiens) facial attractiveness and sexual
selection: The role of symmetry and averageness. J Comp Psychol 108(3):233–242.
6. Johnston VS, Hagel R, Franklin M, Fink B, Grammer K (2001) Male facial attractiveness:
Evidence for hormone mediated adaptive design. Evol Hum Behav 21(4):251–267.
7. Rhodes G (2006) The evolutionary psychology of facial beauty. Annu Rev Psychol 57:
8. Alvergne A, Lummaa V (2010) Does the contraceptive pill alter mate choice in hu-
mans? Trends Ecol Evol 25(3):171–179.
9. Little AC, Jones BC, Burriss RP (2007) Preferences for masculinity in male bodies
change across the menstrual cycle. Horm Behav 51(5):633–639.
10. Roberts SC, Gosling LM, Carter V, Petrie M (2008) MHC-correlated odour preferences
in humans and the use of oral contraceptives. Proc Biol Sci 275(1652):2715–2722.
11. Penton-Voak IS, Perrett DI (2000) Female preferences for male faces changes cycli-
cally: Further evidence. Evol Hum Behav 21(1):39–48.
12. Penton-Voak IS, et al. (1999) Menstrual cycle alters face preference. Nature 399(6738):
13. Wedekind C, Seebeck T, Bettens F, Paepke AJ (1995) MHC-dependent mate prefer-
ences in humans. Proc Biol Sci 260(1359):245–249.
14. Vukovic J, et al. (2010) Self-rated attractiveness predicts individual differences in
women’s preferences for masculine men’s voices. Pers Individ Dif 45(6):451–456.
15. Feinberg DR, DeBruine LM, Jones BC, Little AC (2008) Correlated preferences for
men’s facial and vocal masculinity. Evol Hum Behav 29(4):233–241.
16. Little AC, Burriss RP, Petrie M, Jones BC, Roberts SC (2013) Oral contraceptive use in
women changes preferences for male facial masculinity and is associated with partner
facial masculinity. Psychoneuroendocrinology 38(9):1777–1785.
17. Monte LM, Ellis RR (2014) Fertility of Women in the United States: June 2012, Current
Population Reports (US Census Bureau, Washington, DC), P20-575.
18. Kelley HH, Thibaut JW (1978) Interpersonal Relations: A Theory of Interdependence
(Wiley, New York).
19. Thibaut JW, Kelley HH (1959) The Social Psychology of Groups (Wiley, New York).
20. Fletcher GJO, Simpson JA, Thomas G (2000) Ideals, perceptions, and evaluations in
early relationship development. J Pers Soc Psychol 79(6):933–940.
21. Roberts SC, et al. (2012) Relationship satisfaction and outcome in women who meet
their partner while using oral contraception. Proc Biol Sci 279(1732):1430–1436.
22. Roberts SC, et al. (2014) Partner choice, relationship satisfaction, and oral contra-
ception: The congruency hypothesis. Psychol Sci 25(7):1497–1503.
23. Coulter A, Vessey M, McPherson K, Crossley B (1986) The ability of women to recall
their oral contraceptive histories. Contraception 33(2):127–137.
24. Jones BC, et al. (2005) Commitment to relationships and preferences for femininity
and apparent health in faces are strongest on days of the menstrual cycle when
progesterone level is high. Horm Behav 48(3):283–290.
25. Fincham FD, Bradbury TN (1987) The assessment of marital quality: A reevaluation.
J Marriage Fam 49(4):797–809.
26. Sprecher S, Metts S, Burleson B, Hatfield E, Thompson A (1995) Domains of expressive
interaction in intimate relationships: Associations with satisfaction and commitment.
Fam Relat 44(2):203–210.
27. Norton R (1983) Measuring marital quality: A critical look at the dependent variable.
J Marriage Fam 45(1):141–151.
28. Hudson WW, Harrison DF, Crosscup PC (1981) A short-form scale to measure sexual
discord in dyadic relationships. J Sex Res 17(2):157–174.
29. Preacher KJ, Curran PJ, Bauer DJ (2006) Computational tools for probing interaction
effects in multiple linear regression, multilevel modeling, and latent curve analysis.
J Educ Behav Stat 31(4):437–448.
30. Meltzer AL, McNulty JK, Jackson GL, Karney BR (2014) Sex differences in the im-
plications of partner physical attractiveness for the trajectory of marital satisfaction.
J Pers Soc Psychol 106(3):418–428.
31. Haselton MG, Gangestad SW (2006) Conditional expression of women’s desires and
men’s mate guarding across the ovulatory cycle. Horm Behav 49(4):509–518.
32. Buss DM, Shackelford TK (1997) Susceptibility to infidelity in the first year of mar-
riage. J Res Pers 31(2):193–221.
33. Buss DM, Shackelford TK (1997) From vigilance to violence: Mate retention tactics in
married couples. J Pers Soc Psychol 72(2):346–361.
34. Gangestad SW, Thornhill R, Garver-Apgar CE (2005) Women’s sexual interests across
the ovulatory cycle depend on primary partner developmental instability. Proc Biol Sci
35. Garver-Apgar CE, Gangestad SW, Thornhill R, Miller RD, Olp JJ (2006) Major histo-
compatibility complex alleles, sexual responsivity, and unfaithfulness in romantic
couples. Psychol Sci 17(10):830–835.
36. Larson CM, Haselton MG, Gildersleeve KA, Pillsworth EG (2013) Changes in women’s feelings
about their romantic relationships across the ovulatory cycle. Horm Behav 63(1):128–135.
37. Meltzer AL, McNulty JK, Jackson GL, Karney BR (2014) Men still value physical at-
tractiveness in a long-term mate more than women: Rejoinder to Eastwick, Neff,
Finkel, Luchies, and Hunt (2014). J Pers Soc Psychol 106(3):435–440.
38. Shackelford TK, Buss DM (1997) Marital satisfaction in evolutionary psychological
perspective. Satisfaction in Close Relationships, eds Sternberg RH, Jojjat M (Guilford
Press, New York), pp 7–25.
39. Roney JR, Simmons ZL (2008) Women’s estradiol predicts preference for facial cues of
men’s testosterone. Horm Behav 53(1):14–19.
40. Whisman MA (2001) The association between depression and marital dissatisfaction.
Marital and Family Processes in Depression: A Scientific Foundation for Clinical Practice,
ed Beach SRH (American Psychological Association, Washington, DC), pp 3–24.
41. Kiecolt-Glaser JK, Newton TL (2001) Marriage and health: His and hers. Psychol Bull
42. Leiter MP, Durup MJ (1996) Work, home, and in-between: A longitudinal study of
spillover. J Appl Behav Sci 32(1):29–47.
43. Glenn ND, Weaver CN (1981) The contribution of marital happiness to global hap-
piness. J Marriage Fam 43(1):161–168.
44. Amato PR (2000) The consequences of divorce for adults and children. J Marriage Fam
www.pnas.org/cgi/doi/10.1073/pnas.1414784111 Russell et al.
Russell et al. 10.1073/pnas.1414784111
SI Materials and Methods
At baseline in both studies, couples were recruited from com-
munities in and around the study locations (eastern Tennessee for
study 1 and northwestern Florida for study 2). Couples in both
studies were recruited using advertisements in community news-
papers and bridal shops, and letters sent to couples who had ap-
plied for marriage licenses in the area; couples in study 2 were
additionally recruited using Facebook advertisements. Responding
couples were screened via telephone interview to determine eli-
gibility according to the following criteria important to the broader
aims of the study: (i) the couple had been married fewer than 3
mo, (ii) each partner was at least 18 y of age, and (iii )eachpartner
spoke English. Additionally, couples in study 1 were required to
(i) be in their first marriage, (ii) have completed at least 10 y of
education (to ensure comprehension of the questionnaires), and
(iii) not already have children and involve wives who were not
older than 35 y (to allow a similar probability of transitioning to
first parenthood for all couples).
At baseline, wives in study 1 were 24.25 y old (SD =2.87) and
had completed 16.34 y of education (SD =1.96); wives in study 2
were 28.90 y old (SD =5.94) and had completed 16.81 y of
education (SD =2.42). The majority of wives were Caucasian
(96% in study 1; 73% in study 2). At baseline, husbands in study
1 were 25.60 y old (SD =3.57) and had completed 16.54 y of
education (SD =2.24); husbands in study 2 were 31.03 y old
(SD =7.16) and had completed 16.07 y of education (SD =
2.62). The majority of husbands also were Caucasian (94% in
study 1; 73% in study 2).
Table S1. Descriptive statistics and bivariate correlations among variables
Independent variables 1 2 3 4 5 6 M SD
1. HC status at relationship formation ——————33%
2. Wives’marital satisfaction at baseline 0.00 —————42.31 3.48
3. Husbands’marital satisfaction at baseline −0.04 0.67** ————42.67 3.06
4. Wives’sexual satisfaction at baseline −0.15 0.46** 0.42** ———146.71 21.92
5. Husbands’sexual satisfaction at baseline −0.14 0.28
0.58** 0.64** ——148.26 23.17
6. Wives’facial attractiveness −0.05 −0.11 −0.04 0.06 −0.15 —4.73 1.31
7. Husbands’facial attractiveness −0.15 −0.04 0.04 −0.06 −0.21 0.52** 4.83 1.01
1. HC status at relationship formation ——————53%
2. Wives’marital satisfaction at baseline 0.07 —————42.13 4.04
3. Husbands’marital satisfaction at baseline −0.06 0.55** ————42.00 5.11
4. Wives’sexual satisfaction at baseline 0.10 0.11 0.00 ———6.07 0.97
5. Husbands’sexual satisfaction at baseline −0.17 0.08 0.04 0.23
—— 5.70 1.55
6. Wives’facial attractiveness 0.20 0.03 −0.11 −0.09 0.02 —4.44 1.33
7. Husbands’facial attractiveness 0.01 0.01 0.05 −0.02 −0.05 0.44** 4.31 1.00
P<0.10; *P<0.05; **P<0.01.
Russell et al. www.pnas.org/cgi/content/short/1414784111 1of1