Psychological Bulletin
Do Women’s Mate Preferences Change Across the
Ovulatory Cycle? A Meta-Analytic Review
Kelly Gildersleeve, Martie G. Haselton, and Melissa R. Fales
Online First Publication, February 24, 2014. http://dx.doi.org/10.1037/a0035438
CITATION
Gildersleeve, K., Haselton, M. G., & Fales, M. R. (2014, February 24). Do Women’s Mate
Preferences Change Across the Ovulatory Cycle? A Meta-Analytic Review. Psychological
Bulletin. Advance online publication. http://dx.doi.org/10.1037/a0035438
Psychological Bulletin
2014, Vol. 140, No. 2, 000
© 2014 American Psychological Association
0033-2909/14/$12.00 DOI: 10.1037/a0035438
Do Women’s Mate Preferences Change Across the Ovulatory Cycle?
A Meta-Analytic Review
Kelly Gildersleeve, Martie G. Haselton, and Melissa R. Fales
University of California, Los Angeles
Scientific interest in whether women experience changes across the ovulatory cycle in mating-related
motivations, preferences, cognitions, and behaviors has surged in the past 2 decades. A prominent
hypothesis in this area, the ovulatory shift hypothesis, posits that women experience elevated immediate
sexual attraction on high- relative to low-fertility days of the cycle to men with characteristics that
reflected genetic quality ancestrally. Dozens of published studies have aimed to test this hypothesis, with
some reporting null effects. We conducted a meta-analysis to quantitatively evaluate support for the
pattern of cycle shifts predicted by the ovulatory shift hypothesis in a total sample of 134 effects from
38 published and 12 unpublished studies. Consistent with the hypothesis, analyses revealed robust cycle
shifts that were specific to women’s preferences for hypothesized cues of (ancestral) genetic quality (96
effects in 50 studies). Cycle shifts were present when women evaluated men’s “short-term” attractiveness
and absent when women evaluated men’s “long-term” attractiveness. More focused analyses identified
specific characteristics for which cycle shifts were or were not robust and revealed areas in need of more
research. Finally, we used several methods to assess potential bias due to an underrepresentation of small
effects in the meta-analysis sample or to “researcher degrees of freedom” in definitions of high- and
low-fertility cycle phases. Neither type of bias appeared to account for the observed cycle shifts. The
existence of robust relationship context-dependent cycle shifts in women’s mate preferences has implications for understanding the role of evolved psychological mechanisms and the ovulatory cycle in
women’s attractions and social behavior.
Keywords: mate preferences, ovulation, menstrual cycle, evolution, masculinity
Supplemental materials: http://dx.doi.org/10.1037/a0035438.supp
The question of whether women experience systematic changes
across the ovulatory cycle in mating-related motivations, preferences, cognitions, and behaviors has become a target of increasing
empirical, theoretical, and popular attention over the past 2 decades. In particular, research examining ovulation-related “cycle
shifts” in women’s mate preferences has reached landmark status
in the evolutionary social sciences. Dozens of published studies
have found evidence for cycle shifts in women’s mate preferences,
and several lines of work have documented related effects (e.g.,
cycle shifts in women’s mating motivations, attraction to current
relationship partners and other men, relationship satisfaction, and
partner jealousy; reviewed by Gangestad & Thornhill, 2008; see
also Larson, Haselton, Gildersleeve, & Pillsworth, 2013). Scientists and laypeople alike have increasingly cited these findings as
evidence of the footprints of evolution in modern human sexuality
and as revealing a potentially important, yet often overlooked, role
of the ovulatory cycle in attraction, sexual behavior, and relationship dynamics.
However, there are ongoing debates as to whether current findings provide compelling evidence for ovulation-related cycle shifts
in women’s mate preferences. Several recently published nonreplications have cast doubt on the robustness of these cycle shifts
Kelly Gildersleeve, Department of Psychology and Center for Behavior,
Evolution, and Culture, University of California, Los Angeles; Martie G.
Haselton, Departments of Psychology and Communication Studies, Institute for Society and Genetics, and Center for Behavior, Evolution, and
Culture, University of California, Los Angeles; Melissa R. Fales, Department of Psychology and Center for Behavior, Evolution, and Culture,
University of California, Los Angeles.
We gratefully acknowledge the University of California, Los Angeles
(UCLA), for awarding Kelly Gildersleeve the Graduate Research Mentorship and Dissertation Year Fellowship to support her throughout her
completion of this project and for awarding Martie G. Haselton an Academic Senate Grant to support this work. We also thank the National
Science Foundation Graduate Research Fellowship Program for providing
Melissa R. Fales with fellowship support while working on this project.
For this work, Gildersleeve received the Harold H. Kelley Award for
best student research paper in social psychology at UCLA and the New
Investigator Award (2013) from the Human Behavior and Evolution
Society. We enthusiastically thank the many researchers who generously shared data with us or conducted additional analyses to make it
possible for us to include their studies in this meta-analysis. We also
thank Steve Gangestad, Britt Ahlstrom, Jennifer Hahn-Holbrook, Christina Larson, Aaron Lukaszewski, Damian Murray, David Pinsof, and
Shimon Sapphire-Bernstein for their constructive feedback on earlier
versions of this article.
Correspondence concerning this article should be addressed to Kelly
Gildersleeve, Department of Psychology, University of California, 5437B
Franz Hall, Los Angeles, CA 90095. E-mail: kellygildersleeve@gmail.com
1
2
GILDERSLEEVE, HASELTON, AND FALES
(e.g., Koehler, Rhodes, & Simmons, 2002), and some researchers
have questioned whether the abundance of positive findings in the
published literature reflects publication bias or other sources of
bias.
Given the important implications of the existence of ovulationrelated cycle shifts in women’s mate preferences for scientific and
popular understandings of human sexuality, a rigorous evaluation
of the extant empirical literature is clearly needed. However,
published cycle shift studies have used a wide variety of methods
and have examined preferences for a wide variety of characteristics in men. Furthermore, many cycle shift studies remain unpublished, possibly due to barriers to publishing null effects. Thus,
even an exceptionally thorough narrative review of the published
literature would be inadequate to compel firm conclusions about
the existence and robustness of cycle shifts in women’s mate
preferences.
To address these issues, we conducted a meta-analysis on a large
sample of 134 effects from 38 published and 12 unpublished
studies. The goals of this meta-analysis were to use quantitative
methods to assess the magnitude and robustness of predicted cycle
shifts across the published and unpublished literatures, identify
specific preferences for which cycle shifts are or are not robust and
identify areas still in need of more research, and assess and adjust
for bias that could have contributed to the observed pattern of
cycle shifts.
Theoretical Background
For nearly all female mammals, the brief high-fertility window
that precedes and includes the day of ovulation is the only time
when sex can result in conception. Research on mating patterns in
nonhuman mammals suggests that females of many mammalian
species are more selective or differently selective at high fertility
compared with low fertility, possibly reflecting adaptive cycle
shifts in their underlying mate preferences (e.g., for evidence in
orangutans, chimpanzees, capuchins, and vervet monkeys, see
Knott, Thompson, & Stumpf, 2007; Pieta, 2008; Stumpf &
Boesch, 2005; for an early review, see Keddy-Hector, 1992). For
example, one study found that female chimpanzees in the sexually
active phase of their ovulatory cycle were more likely to mate
repeatedly with high-ranking males on days of this phase when
their fertility was maximally high than on days when their fertility
was still relatively low. In contrast, the rate at which females
mated repeatedly with low-ranking males did not increase with
their fertility (Matsumoto-Oda, 1999).
The Ovulatory Shift Hypothesis
Observations such as these raise the question of whether women
might also experience ovulation-related cycle shifts in their mate
preferences. The ovulatory shift hypothesis, first discussed by
Gangestad and Thornhill (1998) and later named as such in a
review by Gangestad, Thornhill, and Garver-Apgar (2005b), proposes that women experience a nuanced pattern of relationship
context-dependent cycle shifts in their preferences for certain
characteristics in men. Specifically, the ovulatory shift hypothesis
makes three key predictions that dozens of studies have aimed to
test (reviewed in DeBruine et al., 2010; Gangestad & Thornhill,
2008; Thornhill & Gangestad, 2008).
Prediction 1
The first prediction of the ovulatory shift hypothesis is that
women are more sexually attracted to characteristics in men that
reflected relatively high genetic quality in ancestral males1—for
example, the presence of genes with beneficial effects, absence of
genes with harmful effects, or a low overall number of mutated
genes— on high-fertility days of the ovulatory cycle as compared
with low-fertility days of the cycle. This cycle shift in women’s
preference for cues of (ancestral) genetic quality is proposed to
reflect psychological mechanisms that initially evolved because
they increased ancestral females’ likelihood of passing on certain
genetic benefits to their offspring, thereby increasing their own
reproductive success (roughly, their number of surviving descendants).
Cycling reproductive hormones, which underlie changes in female fertility across the ovulatory cycle, could potentially exert a
wide range of effects on female sexual motivations and attractions.
According to the ovulatory shift hypothesis, ancestral females who
experienced a shift in their attractions across the ovulatory cycle
such that they experienced greater sexual attraction to males exhibiting cues of relatively high genetic quality at high fertility than
at low fertility would have been more likely to have conceptive sex
with such males and produce offspring who were also relatively
high in genetic quality. Consequently, these females would have
had higher reproductive success, on average, than females whose
attractions did not shift across the cycle in this way. Also, importantly, their descendants would have been more likely to possess
any heritable aspects of the psychological mechanisms that produced the cycle shift in their mate preferences, making female
descendants more likely to experience this cycle shift themselves.
As long as conditions remained relatively stable, a cycle shift in
preferences for males displaying cues of genetic quality would
thereby have become increasingly common in females over evolutionary time.
Importantly, the ovulatory shift hypothesis predicts that the
proposed cycle shift in women’s attraction to men with characteristics that reflected genetic quality ancestrally will be present
specifically when women evaluate men’s immediate desirability as
sex partners. Only if ancestral females’ heightened preferences at
high fertility for males displaying cues of genetic quality at least
occasionally translated into higher rates of sex with such males
during the fleeting high-fertility window would the posited cycle
shift have been associated with higher reproductive success on
average. Thus, it follows that the predicted cycle shift will be
present specifically in the context of evaluating prospective partners for a short-term sexual affair or other types of relationships in
which ancestral females’ preferences would have been relatively
likely to influence their immediate sexual behavior.
1
We use the terms genetic quality and reproductive success as they are
used in the field of biology. These terms do not imply that, because of their
genetic constitution, some individuals are (or were ancestrally) superior to
others in any way not outlined above. In addition, the ovulatory shift
hypothesis makes no predictions regarding cycle shifts in women’s preferences for female partners. Accordingly, most studies in this literature
limit samples to women who identify as heterosexual, and our discussion
of this literature is likewise limited to this group of women.
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
Prediction 2
The second prediction of the ovulatory shift hypothesis is that
the proposed cycle shift in women’s attraction to characteristics
that reflected genetic quality in ancestral males will be absent or
only weakly present when they evaluate men’s desirability as a
social partner in the long run. If ancestral females’ heightened
preferences at high fertility for males displaying cues of genetic
quality did not translate into higher rates of sex with such males
during the high-fertility window or translated instead into higher
rates of nonsexual behaviors with such males during the highfertility window (e.g., courtship behaviors that might lead to the
formation of a long-term pair bond), the posited cycle shift would
not have been associated with higher reproductive success on
average. Thus, it follows that the predicted cycle shift in women’s
preferences for cues of ancestral genetic quality will be absent or
only weakly present in the context of evaluating prospective partners for a long-term relationship (e.g., marriage) or other types of
relationships in which ancestral females’ preferences would have
been relatively unlikely to influence their immediate sexual behavior.
Many studies aiming to test the ovulatory shift hypothesis have
asked women to evaluate men as potential partners for a “shortterm relationship” or a “long-term relationship.” To the extent that
these terms imply a sexual affair and a long-term social partnership
(such as marriage), respectively, it follows from Predictions 1 and
2 that the cycle shift in women’s attraction to cues of genetic
quality will be present and relatively pronounced in the former
context but absent or only weakly present in the latter. Notably,
however, the ovulatory shift hypothesis does not predict that the
magnitude of the cycle shift will depend on how long women
expect a relationship to last per se but rather on whether they
expect the relationship to involve having sex in the immediate
future.
In addition, many studies in this literature have asked women to
evaluate men’s attractiveness, physical attractiveness, sexual attractiveness, or sexiness or to evaluate the importance or desirability of a specific characteristic in a prospective partner without
specifying any particular relationship context. The majority of
these studies have assessed ratings of attractiveness, physical attractiveness, sexual attractiveness, or sexiness, whereas ratings of
importance or desirability are very rare. Given previous research
showing that women value physical attractiveness more when
evaluating short-term sex partners than when evaluating long-term
relationship partners (e.g., Li & Kenrick, 2006; Regan, 1998), it
follows from the ovulatory shift hypothesis that women in these
unspecified-context studies will generally exhibit a pattern of cycle
shifts more similar to the pattern observed in a short-term context
than to the pattern observed in a long-term context.
Predictions 1 and 2 highlight an implicit claim of the ovulatory
shift hypothesis—namely, that certain potentially observable phenotypes in men constituted reliable “cues” to genetic quality in
ancestral males. This claim rests on the following logic: Differences between ancestral males in heritable genetic factors likely
contributed to differences between males in immune function,
vulnerability to environmental stressors, ability to compete with
other males to attract mates, and other qualities that affected their
reproductive success. Some of these genetic differences between
males likely also contributed directly or indirectly (e.g., via effects
3
on health) to detectable differences between males in physical
appearance, body scents, vocal properties, and other phenotypes.
For example, symmetry and masculinity are widely thought to
have served as indicators of genetic quality in ancestral males
(discussed in more detail below). In turn, selection could have
acted on females to be sensitive to this phenotypic variation in
males and, possibly, experience enhanced attraction to indicators
of genetic quality under certain conditions.
Prediction 3
The third prediction of the ovulatory shift hypothesis is that,
regardless of relationship context, women are not more sexually
attracted to characteristics in men that reflected relatively high
suitability as a long-term social partner and coparent in ancestral
males on high-fertility days of the ovulatory cycle as compared
with low-fertility days of the cycle. The ovulatory shift hypothesis
posits that females could have reproductively benefited by mating
with such males regardless of their current fertility (and in a
variety of relationship contexts). For example, regardless of their
fertility when they initiated the relationship, ancestral females who
entered into long-term pair bonds with males who were cooperative, caring, and highly investing partners and coparents would
plausibly have had higher reproductive success, on average, than
females who entered into long-term pair bonds with males who
were uncooperative, negligent, or in other ways less suitable as a
long-term partner and coparent.
Given the hypothesized reproductive benefits of mating with
males relatively high in genetic quality, the ovulatory shift hypothesis raises the question of why females did not evolve to prefer
males exhibiting cues of genetic quality at all times in the ovulatory cycle. One possible answer to this question is that cycle shifts
in mate preferences initially evolved in an ancestral species (predating humans) that did not engage in high rates of pair bonding.
In that context, females whose preferences shifted across the cycle
in such a way that they were more likely to have sex with males
displaying cues of genetic quality at high fertility but more likely
to have sex with males offering nongenetic reproductive benefits
(e.g., material investment or protection) in the remainder of the
cycle might have had greater reproductive success, on average,
than females whose preferences did not shift across the cycle in
this way. In humans, for whom rates of pair bonding are high,
these cycle shifts could simply be vestigial, reflecting remnants of
psychological adaptations that now have a negligible impact on
women’s reproductive success or have a negative impact on women’s reproductive success but have not yet been fully removed by
selection (Gangestad & Garver-Apgar, 2013).
The dual mating hypothesis (Pillsworth & Haselton, 2006b)
presents another possible answer to the question of why females
did not evolve to prefer males with characteristics associated with
relatively high genetic quality throughout the cycle. Like the
ovulatory shift hypothesis, the dual mating hypothesis does not
stipulate whether cycle shifts in mate preferences initially evolved
in humans or in an ancestral species. However, unlike the ovulatory shift hypothesis (which is agnostic on this point), the dual
mating hypothesis proposes that cycle shifts in mate preferences
were associated with greater reproductive success among ancestral
women and therefore are not merely vestigial.
4
GILDERSLEEVE, HASELTON, AND FALES
According to the dual mating hypothesis, ancestral women
would generally have maximized reproductive benefits by forming
long-term pair bonds with men who were both high in genetic
quality and highly suitable as a long-term social partner and
coparent. However, these characteristics were distributed across
the population of men, and therefore, not all women could have
formed long-term pair bonds with men who were high in both
types of characteristics. The dual mating hypothesis proposes that
women who formed long-term pair bonds with men who were
relatively high in suitability as long-term partners but relatively
low in genetic quality would have had higher reproductive success,
on average, than women who formed long-term pair bonds with
men who were relatively high in genetic quality but relatively low
in suitability as long-term partners. This claim rests on the notion
that high-quality biparental care and investment were critical for
children’s survival in ancestral environments (Geary, 2000; but see
Sear & Mace, 2008). This claim is further reinforced by the notion
that ancestral men who were relatively high in genetic quality
might have been relatively less suitable and less available as
long-term mates. Briefly, if men displaying cues of genetic quality
were generally relatively desirable as sex partners, they might have
tended to pursue short-term sexual relationships instead of pair
bonds or outside of established pair bonds (thus diverting resources away from their long-term mate and children; see Gangestad & Simpson, 2000).
Following this line of reasoning, the dual mating hypothesis
proposes that, among women who formed long-term pair bonds
with men who were relatively high in suitability as long-term
partners but relatively low in genetic quality, women who maintained their primary pair bond but also occasionally engaged in
extra-pair sex with men of high genetic quality at high fertility
(and when their sexual infidelity was unlikely to be discovered)
would have had greater reproductive success, on average, than
women who did not pursue this “dual mating” strategy. Evidence
from nonhuman species in which females sometimes pursue this
reproductive strategy suggests that behavioral adaptations that
facilitate dual mating could have evolved even if rates of extra-pair
sex were quite low (e.g., as low as 1%–5% in some bird species;
see Thornhill & Gangestad, 2008).
Although many writings in this literature have suggested that
cycle shifts in women’s mate preferences reflect a long evolutionary history of dual mating in humans, the ovulatory shift hypothesis does not require that ancestral women engaged in extra-pair
sex. For example, cycle shifts could be vestigial, as noted above.
Alternatively, it is possible that cycle shifts have been maintained
by selection in humans because they were historically associated
with certain reproductive benefits in the context of sexually monogamous pair bonds, although this idea is not well developed in
the current literature. In sum, if women experience the posited
ovulation-related cycle shifts in their mate preferences, many
interesting questions remain about the precise evolutionary pathways giving rise to them.
Cues of Genetic Quality in Ancestral Males
Research on cycle shifts in mate preferences has focused primarily on symmetry and masculinity as candidates for potentially
observable characteristics that are likely to have been reliably
associated with genetic quality in ancestral males.2 Here we briefly
summarize the rationales typically given in support of claims that
symmetry and masculinity were cues of genetic quality in ancestral
males.
Symmetry
In biology, developmental stability is defined as “the ability of
an organism to withstand genetic and environmental disturbances
encountered during development so as to produce a predetermined
optimum phenotype” (Clarke, 1993, p. 15). Developmental stability is thought to reflect genetic quality as defined earlier (see, e.g.,
Thornhill & Gangestad, 2008; Van Dongen & Gangestad, 2011).
Because researchers cannot directly measure developmental stability, they typically measure fluctuating asymmetry as a proxy
(e.g., Klingenberg, 2003; Van Dongen, 2006). Fluctuating asymmetry is the extent to which the right and left sides of the body
deviate randomly from perfect bilateral symmetry (mirror images).
To the extent that fluctuating asymmetry represents a departure
from a genetic “blueprint” for a symmetrical body, it could indicate lower developmental stability and thus lower genetic quality.
Consistent with this view, lower symmetry3 (higher fluctuating
asymmetry) has been linked to inbreeding, homozygosity, and
deleterious recessive genes in nonhuman animals (see Rhodes,
2006; Thornhill & Gangestad, 1994; see also Carter, Weier, &
Houle, 2009, for experimental evidence) and to negative health
outcomes in humans (see Thornhill & Møller, 1997; Van Dongen
& Gangestad, 2011).
In addition, fluctuating asymmetry appears to influence male
success in attracting mates. Studies of many nonhuman animal
species have found that more symmetrical individuals (lower in
fluctuating asymmetry) have a significantly greater number of
mates than less symmetrical individuals (meta-analyzed by Møller
& Thornhill, 1998). Several findings support parallel associations
in humans. For example, more symmetrical men report having had
a greater number of sex partners and having had sex at a younger
2
A related hypothesis is that women will experience elevated preferences at high fertility for characteristics in men that reflect the presence of
genes that would have been compatible with their own genes in the
ancestral past. For example, it has been hypothesized that, all else equal,
individuals who inherit different major histocompatibility complex (MHC)
alleles from each of their parents have better pathogen defense than
individuals who receive the same alleles from both of their parents (e.g.,
Chen & Parham, 1989; Hughes & Nei, 1988, 1989; Penn, Damjanovich, &
Potts, 2002). It follows that women might experience elevated attraction at
high fertility to men with different MHC alleles than their own (men with
whom they are, according to this view, genetically compatible). Our search
discovered only two studies examining cycle shifts related to MHC compatibility. One study found that women who shared a greater number of
MHC alleles with their romantic partner (less compatible) experienced a
greater increase at high fertility in their attraction to other men (GarverApgar, Gangestad, Thornhill, Miller, & Olp, 2006). A second study did not
find evidence for a cycle shift in women’s attraction to the scent of
MHC-compatible men (Thornhill et al., 2003). Although the latter of these
two studies was eligible for inclusion in this meta-analysis, we were unable
to obtain the data needed to compute an effect size for it.
3
Although it is typical in this literature to discuss effects of fluctuating
asymmetry, for ease of interpretation, in the balance of this article we
discuss effects of symmetry, by which we mean the inverse of fluctuating
asymmetry. For example, we note that the ovulatory shift hypothesis
predicts that women will demonstrate a stronger preference for more
symmetrical men (men who are low in fluctuating asymmetry) at high
fertility compared to low fertility.
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
age than less symmetrical men (Thornhill & Gangestad, 1994).
And women rate more facially symmetrical men as more attractive
than less facially symmetrical men (meta-analyzed by Rhodes,
2006, and Van Dongen & Gangestad, 2011).
in ancestral males are conjectural and that the goal of this metaanalysis is not to directly test the accuracy of such claims. Rather,
the goal is to determine whether predicted patterns of cycle shifts
are robust for the characteristics most studied to date.
Method
Masculinity
In biology, masculine characteristics refer to a number of physical and behavioral secondary sex characteristics that develop in
males around the time of sexual maturity. Masculine characteristics are costly to produce and maintain; therefore, pronounced
masculine characteristics could reflect good overall condition.
Consistent with this view, studies of nonhuman animals have
shown that food shortages bring about substantial reductions in the
size of masculine characteristics, suggesting that masculine characteristics entail energetic costs that only individuals in good
condition can afford (e.g., Wilson, Rogler, & Erb, 1979).4 Good
condition is, in turn, partially tied to genetic quality (Rowe &
Houle, 1996).
Like symmetry, masculine characteristics have been linked to
male success in attracting mates. A meta-analysis of nonhuman
lekking species, in which males engage in highly visible competitions against other males to attract females, found that males with
larger masculine characteristics (e.g., antlers) attract a larger number of mates than males with smaller masculine characteristics
(Fiske, Rintamaki, & Karvonen, 1998). Relatedly, many studies
support the idea that masculine characteristics have historically
contributed to men’s success in attracting mates, perhaps especially by increasing their success in competitive interactions with
other men. For example, studies have found that experimentally
increasing men’s vocal, facial, and body masculinity increases
others’ perceptions of their dominance even more than perceptions
of their attractiveness (see Puts, 2010). Studies also support a
direct effect of masculinity on men’s sexual attractiveness to
women. For example, women in one study reported greater attraction to hypothetical men with more masculine faces, bodies, and
voices when evaluating them as short-term sex partners than as
long-term relationship partners (Little, Connely, Feinberg, Jones,
& Roberts, 2011). Likewise, women in another study reported
greater attraction to men whose photos they rated as more masculine and who had higher measured circulating testosterone when
they evaluated those men’s desirability for a brief affair than when
they evaluated those men’s desirability for a long-term relationship
(Roney, Hanson, Durante, & Maestripieri, 2006).
In sum, although research in this area has examined cycle shifts
in women’s preferences for a broad range of characteristics (discussed in detail in the Inclusion Criteria section), to date most
studies have examined cycle shifts in women’s preferences for
symmetrical and masculine characteristics because these characteristics are widely thought to have served as cues of genetic
quality in ancestral males. In addition, a smaller number of studies
have examined cycle shifts in women’s preferences for warmth
and kindness, parenting ability, faithfulness, trustworthiness, material resources, and related characteristics because these characteristics are widely thought to have served as cues of “long-term
partner quality” in ancestral males (a term that, for brevity, we use
henceforward to refer to suitability as a long-term social partner
and coparent). Importantly, we note that claims that certain characteristics were cues of genetic quality or long-term partner quality
5
Search Strategy
As shown in Tables 1 and 2, we identified a large number of
studies that collected data relevant to examining ovulation-related
cycle shifts in women’s preferences for various characteristics in
men. We located studies through several channels, including reference sections of published articles, online databases and search
engines, conference proceedings, listserv postings, and personal
correspondence with researchers in this area. We chose several of
these strategies with the specific goal of locating unpublished data
and manuscripts not identified through other search methods. For
example, we searched through the annual conference programs of
the Society for Personality and Social Psychology (2005–2012)
and of the Human Behavior and Evolution Society (2000 –2012) to
identify researchers who had given talks or presented posters on
research related to mating and the ovulatory cycle. We e-mailed all
of these researchers a request for relevant unpublished data, including student projects. We also sent similar solicitations via
listservs operated by the Society for Personality and Social Psychology, Society for the Psychological Study of Social Issues, and
Society of Experimental Social Psychology and printed a solicitation in the summer 2010 Human Behavior and Evolution Society
newsletter. Last, we e-mailed colleagues known to have conducted
research on mating and the ovulatory cycle and requested that they
alert us to any unpublished data that might be eligible for inclusion
in the meta-analysis.
We used the following databases and search engines to locate
published journal articles and unpublished manuscripts (e.g., master’s theses and dissertations): PsycINFO, PubMed Central, Web
of Science, BIOSIS, Dissertation Abstracts Online, ProQuest Dissertations & Theses, and Google Scholar. All searches utilized
Boolean logic to search for entries that included a term related to
ovulation, the menstrual cycle, fertility, or cycling hormones in
conjunction with a term related to mate preferences—for example,
“ovulatⴱ” or “mid-cycle” or “menstrual cycle” or “cyclⴱ” or “fertilⴱ” or “high-fertility” or “low-fertility” or “conception risk” or
“hormonⴱ” or “luteal” or “follicular” or “estrogen” or “estradiol”
and “mate” or “mating” or “attractive” or “partner” or “mate
preferenceⴱ” or “good genes” or “genetic quality” or “genetic
benefits” or “fitness” or “symmetⴱ” or “masculinⴱ” or “dominanⴱ”
4
Some evidence suggests that testosterone, which is typically required
to produce and often required to maintain masculine characteristics, also
suppresses immune function. If correct, this implies that masculine characteristics entail immune costs (in addition to energetic costs) that only
individuals in good condition— owing in part to their relatively high
underlying genetic quality— can afford (see the immunocompetence handicap hypothesis, as discussed by Folstad & Karter, 1992; reviewed in
Thornhill & Møller, 1997; meta-analyzed in Roberts, Buchanan, & Evans,
2004). Whether this is a likely mechanism through which masculinity was
ancestrally associated with genetic quality has been contested. For a
critique and alternative hypothesis, see Braude, Tang-Martinez, and Taylor
(1999).
6
Table 1
Studies Assessing Ovulation-Related Cycle Shifts in Mate Preferences: Basic Characteristics, Effect Size, and Inclusion in Analyses
Inclusion in analyses
Long-term partner
quality
Sample size
Study and effects
U
92
33
59
0.00
0.05
U
92
33
59
⫺0.20
0.05
U
92
33
59
M
M
U
92
33
59
M
M
ST, LT
33
Within
participants
M
M
ST, LT
33
Within
participants
M
M
ST, LT
33
Within
participants
M
M
✓
GILDERSLEEVE, HASELTON, AND FALES
Beaulieu (2007), Study 2
Relationship skills
(composite of
kind,
understanding,
loyal, generous)
Dominance
(composite of
dominant,
powerful,
aggressive)
Education
(composite of
educated,
cultured,
intelligent)
Good financial
prospects
(composite of
wealthy, good
financial
prospects)
Beaulieu (2007), Study 4
Relationship skills
(composite of
kind,
understanding,
loyal, generous)
Dominance
(composite of
dominant,
powerful,
aggressive)
Education
(composite of
educated,
cultured,
intelligent)
Good financial
prospects
(composite of
wealthy, good
financial
prospects)
Genetic quality
HighLow- Effect
Reason
Broad
Narrow
Broad
Narrow
Scent
Facial cues
Relationship
fertility fertility Size
for
set of
set of
set of
set of
Facial
cues of
Facial
Body
Vocal
Behavioral
of
context
N
n
n
(g) Variance exclusion measures measures measures measures symmetry symmetry masculinity masculinity masculinity dominance testosterone
✓
(table continues)
Table 1 (continued)
Inclusion in analyses
Long-term partner
quality
Sample size
Study and effects
Caryl et al. (2009)
Pupil size
Arrogant
Ingenious
Aggressive
Strong
Conceited
Enterprising
Inventive
Warm
Sensitive
Sentimental
Sympathetic
Jolly
Helpful
Appreciative
Considerate
Cooperative
Friendly
Talkative
Forgiving
Emotional
Foresighted
Shrewd
Industrious
Assertive
Forceful
Timid
Dependent
Fickle
Frivolous
Opportunistic
Hardheaded
Confident
DeBruine et al. (2005)
Facial selfresemblance
HighLow- Effect
Reason
Broad
Narrow
Broad
Narrow
Scent
Facial cues
Relationship
fertility fertility Size
for
set of
set of
set of
set of
Facial
cues of
Facial
Body
Vocal
Behavioral
of
context
N
n
n
(g) Variance exclusion measures measures measures measures symmetry symmetry masculinity masculinity masculinity dominance testosterone
U
198
97
101
U
60
M
M
ST
53
Within
participants
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
U
43
21
0.00
M
0.02
✓
✓
✓
✓
✓
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
Bressan & Stranieri
(2008)
Facial masculinity
Bullock (2000),
Chapter 4
Chin length
Cárdenas & Harris
(2007)
Facial symmetry
Genetic quality
M
0.00
0.02
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
M
0.04
⫺0.22
0.02
⫺0.29
0.17
⫺0.14
0.04
⫺0.11
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
0.08
0.08
0.08
0.08
0.08
0.08
0.08
0.08
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
22
⫺0.09
0.09
✓
✓
4
✓
✓
✓
4
4
✓
4
(table continues)
7
8
Table 1 (continued)
Inclusion in analyses
Long-term partner
quality
Sample size
Study and effects
Genetic quality
HighLow- Effect
Reason
Broad
Narrow
Broad
Narrow
Scent
Facial cues
Relationship
fertility fertility Size
for
set of
set of
set of
set of
Facial
cues of
Facial
Body
Vocal
Behavioral
of
context
N
n
n
(g) Variance exclusion measures measures measures measures symmetry symmetry masculinity masculinity masculinity dominance testosterone
Feinberg et al. (2006)
Vocal masculinity
U
26
Within
participants
M
Feinberg (2012)
Vocal masculinity
ST
22
0.45
0.06
✓
✓
✓
Vocal masculinity
LT
22
Within
participants
Within
participants
⫺0.21
0.05
✓
✓
✓
U
20
⫺0.06
0.09
✓
✓
U
106
45
61
M
M
U
36
15
21
0.19
0.11
✓
ST
237
0.40
0.02
✓
✓
✓
LT
237
0.08
0.02
✓
✓
✓
ST
237
0.12
0.02
✓
✓
✓
LT
237
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
⫺0.11
0.02
✓
✓
✓
ST
237
0.17
0.02
✓
✓
✓
Muscular
LT
237
⫺0.09
0.02
✓
✓
✓
Confrontative (with
other men)
Confrontative (with
other men)
Socially respected
and influential
Socially respected
and influential
Arrogant and selfcentered
Arrogant and selfcentered
Intelligent
ST
237
0.24
0.02
✓
✓
✓
LT
237
⫺0.12
0.02
✓
✓
✓
ST
237
0.05
0.02
✓
✓
✓
LT
237
⫺0.21
0.02
✓
✓
✓
ST
237
0.14
0.02
✓
✓
✓
LT
237
⫺0.20
0.02
✓
✓
✓
ST
243
⫺0.22
0.02
4
Intelligent
LT
243
⫺0.12
0.02
4
Faithful
ST
243
⫺0.22
0.02
Fink (2012)
Facial masculinity
Social presence
Direct intrasexual
competitiveness
Direct intrasexual
competitiveness
Gangestad et al. (2007)
Muscular
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
✓
✓
✓
(table continues)
GILDERSLEEVE, HASELTON, AND FALES
Flowe et al. (2012)
Behavioral
masculinity
Frost (1994)
Darker skin tone
Gangestad et al. (2004)
Social presence
Within
participants
M
Table 1 (continued)
Inclusion in analyses
Long-term partner
quality
Sample size
Study and effects
Faithful
Gangestad & Thornhill
(1998)
Scent cues of body
symmetry
Garver-Apgar &
Gangestad (2012)
Average of social
presence and
direct intrasexual
competitiveness
Average of social
presence and
direct intrasexual
competitiveness
Harris (2011)
Facial masculinity
Haselton & Miller
(2006)
Wealth versus
creativity
Havlíček et al. (2005)
Scent cues of
dominance
(Narcissism scale
from CPI)
Hromatko et al. (2006)
Facial symmetry
HighLow- Effect
Reason
Broad
Narrow
Broad
Narrow
Scent
Facial cues
Relationship
fertility fertility Size
for
set of
set of
set of
set of
Facial
cues of
Facial
Body
Vocal
Behavioral
of
context
N
n
n
(g) Variance exclusion measures measures measures measures symmetry symmetry masculinity masculinity masculinity dominance testosterone
0.04
0.02
✓
✓
⫺0.16
0.02
✓
✓
0.05
0.02
✓
✓
⫺0.15
0.02
✓
✓
Fertility
continuous
⫺0.10
0.02
✓
✓
Fertility
continuous
Fertility
continuous
0.04
0.02
✓
✓
0.06
0.02
✓
✓
59
Within
participants
0.08
0.01
✓
✓
U
28
Fertility
continuous
1.25
0.21
✓
✓
ST
18
Within
participants
0.15
0.11
✓
✓
✓
LT
18
Within
participants
⫺0.16
0.11
✓
✓
✓
U
258
0.03
0.02
✓
✓
LT
243
ST
243
LT
243
ST
243
LT
243
ST
243
LT
243
U
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
80
178
ST, LT
41
Fertility
continuous
M
M
U
65
30
35
0.33
0.06
✓
✓
U
64
11
53
⫺0.26
0.11
✓
✓
✓
✓
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
Warm (kind and
understanding)
Warm (kind and
understanding)
Likely to be
financially
successful
Likely to be
financially
successful
Likely to be a good
parent
Likely to be a good
parent
Gangestad et al. (2011)
Facial masculinity
Genetic quality
✓
5
✓
(table continues)
9
10
Table 1 (continued)
Inclusion in analyses
Long-term partner
quality
Sample size
Study and effects
HighLow- Effect
Reason
Broad
Narrow
Broad
Narrow
Scent
Facial cues
Relationship
fertility fertility Size
for
set of
set of
set of
set of
Facial
cues of
Facial
Body
Vocal
Behavioral
of
context
N
n
n
(g) Variance exclusion measures measures measures measures symmetry symmetry masculinity masculinity masculinity dominance testosterone
ST
42
Facial masculinity
LT
42
Strong
ST
42
Strong
LT
42
Warm
ST
42
Warm
LT
42
Mature
ST
42
Mature
LT
42
Socially competent
ST
42
Socially competent
LT
42
Nurturant
ST
42
Nurturant
LT
42
Threatening
ST
42
Threatening
LT
42
Dominant
ST
42
Dominant
LT
42
Dark
ST
42
Dark
LT
42
U
29
U
328
Jones, Little, et al.
(2005), Study 2
Facial masculinity
Koehler et al. (2002)
Facial symmetry
ST, LT
29
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
169
159
Within
participants
⫺0.29
0.04
✓
✓
✓
⫺0.42
0.03
✓
✓
✓
⫺0.13
0.02
✓
⫺0.13
0.01
✓
0.05
0.02
✓
0.10
0.02
✓
⫺0.21
0.03
4
⫺0.13
0.02
4
⫺0.11
0.03
4
0.02
0.03
4
0.02
0.03
✓
0.26
0.03
✓
0.00
0.02
4
0.07
0.02
4
0.02
0.03
✓
⫺0.09
0.02
✓
0.00
0.01
✓
0.15
0.01
✓
0.40
0.19
✓
✓
✓
0.33
0.01
✓
✓
✓
M
GILDERSLEEVE, HASELTON, AND FALES
Izbicki & Johnson
(2010)
Facial masculinity
Johnston et al. (2001)
Facial masculinity
Genetic quality
M
(table continues)
Table 1 (continued)
Inclusion in analyses
Long-term partner
quality
Sample size
Study and effects
HighLow- Effect
Reason
Broad
Narrow
Broad
Narrow
Scent
Facial cues
Relationship
fertility fertility Size
for
set of
set of
set of
set of
Facial
cues of
Facial
Body
Vocal
Behavioral
of
context
N
n
n
(g) Variance exclusion measures measures measures measures symmetry symmetry masculinity masculinity masculinity dominance testosterone
U
50
Facial symmetry
U
50
Li et al. (2006)
Multiple traits
ST, LT
54
Within
participants
Little, Jones, et al.
(2007), Study 1
Facial symmetry
U
31
Within
participants
0.41
ST
LT
210
210
63
63
147
147
ST
LT
97
97
36
36
61
61
ST
17
LT
17
U
150
ST
25
Dominant
LT
25
Facial masculinity
ST
25
Facial masculinity
LT
25
Warm
ST
25
Warm
LT
25
ST
111
LT
111
Body masculinity
Little et al. (2008)
Facial masculinity
Luevano & Zebrowitz
(2006)
Dominant
Lukaszewski & Roney
(2009)
Dominant
Dominant
0.15
0.04
✓
0.04
0.04
✓
✓
✓
0.07
✓
✓
✓
0.59
0.05
0.02
0.02
✓
✓
✓
✓
✓
✓
0.59
0.05
0.05
0.04
✓
✓
✓
✓
✓
✓
Within
participants
Within
participants
0.69
0.07
✓
✓
✓
0.28
0.06
✓
✓
✓
54
0.72
0.03
✓
✓
✓
0.09
0.04
✓
0.30
0.03
✓
⫺0.23
0.04
✓
✓
✓
0.06
0.02
✓
✓
✓
⫺0.25
0.03
✓
0.00
0.03
✓
0.36
0.04
✓
0.19
0.04
✓
Within
participants
Within
participants
96
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Fertility
continuous
Fertility
continuous
M
M
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
Koehler et al. (2006)
Facial averageness
Little, Jones, et al.
(2007), Study 2
Facial symmetry
Facial symmetry
Little, Jones, & Burriss
(2007), Study 1
Body masculinity
Body masculinity
Little, Jones, & Burriss
(2007), Study 2
Body masculinity
Genetic quality
5
(table continues)
11
12
Table 1 (continued)
Inclusion in analyses
Long-term partner
quality
Sample size
Study and effects
HighLow- Effect
Reason
Broad
Narrow
Broad
Narrow
Scent
Facial cues
Relationship
fertility fertility Size
for
set of
set of
set of
set of
Facial
cues of
Facial
Body
Vocal
Behavioral
of
context
N
n
n
(g) Variance exclusion measures measures measures measures symmetry symmetry masculinity masculinity masculinity dominance testosterone
Kind
ST
111
Kind
LT
111
Trustworthy
ST
111
Trustworthy
LT
111
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
⫺0.02
0.04
✓
⫺0.05
0.04
✓
⫺0.02
0.04
✓
⫺0.05
0.04
✓
ST, LT, U
ST, LT, U
24
24
10
10
14
14
M
M
M
M
U
80
42
38
M
M
U
80
42
38
M
M
U
81
43
38
M
M
U
81
43
38
M
M
ST
45
Intelligent
LT
45
Future kids’
intelligence
Future kids’
intelligence
Mathematical
problem-solving
ability
Mathematical
problem-solving
ability
Good grades
ST
45
LT
45
ST
45
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
LT
45
ST
45
Good grades
LT
45
Creative/imaginative
ST
45
Creative/imaginative
LT
45
Future kids’ sense of
humor
ST
45
0.11
0.09
4
0.01
0.09
4
0.08
0.09
4
0.38
0.10
4
0.28
0.09
4
Fertility
continuous
0.28
0.09
4
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
0.50
0.10
4
0.31
0.09
4
⫺0.36
0.10
4
0.05
0.09
4
⫺0.22
0.09
4
GILDERSLEEVE, HASELTON, AND FALES
McClellan et al. (2007)
Body masculinity
Age
McDonald &
Navarrete (2012),
Sample 1
Body muscularity
Same-race (vs.
other-race) face
McDonald &
Navarrete (2012),
Sample 2
Body muscularity
Same-race (vs.
other-race) face
Miller (2003)
Intelligent
Genetic quality
(table continues)
Table 1 (continued)
Inclusion in analyses
Long-term partner
quality
Sample size
Study and effects
Genetic quality
HighLow- Effect
Reason
Broad
Narrow
Broad
Narrow
Scent
Facial cues
Relationship
fertility fertility Size
for
set of
set of
set of
set of
Facial
cues of
Facial
Body
Vocal
Behavioral
of
context
N
n
n
(g) Variance exclusion measures measures measures measures symmetry symmetry masculinity masculinity masculinity dominance testosterone
LT
45
ST
45
Social sensitivity
LT
45
Adaptable to
situations and
challenges
Adaptable to
situations and
challenges
Big ego
ST
45
LT
45
ST
45
Big ego
LT
45
Body muscularity
ST
45
Body muscularity
LT
45
Facial masculinity
ST
45
Facial masculinity
LT
45
Tall
ST
45
Tall
LT
45
Happy
ST
45
Happy
LT
45
Exciting/spontaneous
ST
45
Exciting/spontaneous
LT
45
Talkative/extraverted
ST
45
Talkative/extraverted
LT
45
Likelihood of being
unfaithful
(reverse-coded)
Likelihood of being
unfaithful
(reverse-coded)
Future money
making
ST
45
LT
45
ST
45
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
0.34
0.10
4
⫺0.32
0.09
4
0.03
0.09
4
⫺0.35
0.10
4
Fertility
continuous
0.15
0.09
4
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
⫺0.09
0.09
✓
0.12
0.09
✓
⫺0.35
0.10
✓
0.09
0.09
✓
⫺0.65
0.10
✓
⫺0.48
0.10
✓
⫺0.15
0.09
✓
⫺0.15
0.09
✓
⫺0.46
0.10
4
⫺0.09
0.09
4
⫺0.16
0.09
4
0.24
0.09
4
⫺0.13
0.09
4
0.26
0.09
4
0.01
0.09
✓
Fertility
continuous
⫺0.18
0.09
✓
Fertility
continuous
⫺0.06
0.09
✓
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
Future kids’ sense of
humor
Social sensitivity
(table continues)
13
14
Table 1 (continued)
Inclusion in analyses
Long-term partner
quality
Sample size
Study and effects
HighLow- Effect
Reason
Broad
Narrow
Broad
Narrow
Scent
Facial cues
Relationship
fertility fertility Size
for
set of
set of
set of
set of
Facial
cues of
Facial
Body
Vocal
Behavioral
of
context
N
n
n
(g) Variance exclusion measures measures measures measures symmetry symmetry masculinity masculinity masculinity dominance testosterone
LT
45
ST
45
LT
45
ST
45
LT
45
ST
45
Sympathetic/kind
LT
45
Constructive in
arguments
Constructive in
arguments
Neat/organized
ST
45
LT
45
ST
45
Neat/organized
LT
45
Moody/irritable
ST
45
Moody/irritable
LT
45
Fun at sex
ST
45
Fun at sex
LT
45
ST
45
LT
45
ST
45
LT
45
U
43
U
43
U
112
0.12
0.09
✓
⫺0.37
0.10
✓
Fertility
continuous
⫺0.13
0.09
✓
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
Fertility
continuous
0.05
0.09
✓
0.41
0.10
✓
⫺0.21
0.09
✓
0.12
0.09
✓
0.14
0.09
4
0.44
0.10
4
0.19
0.09
4
0.27
0.09
4
⫺0.05
0.09
4
0.13
0.09
4
⫺0.49
0.10
4
0.04
0.09
4
⫺0.23
0.09
4
⫺0.08
0.09
4
0.26
0.09
4
0.47
0.10
4
Within
participants
Within
participants
⫺0.17
0.04
1.18
0.06
4
72
⫺0.16
0.21
4
Fertility
continuous
Fertility
continuous
40
GILDERSLEEVE, HASELTON, AND FALES
Future money
making
Good at playing
with and caring
for kids
Good at playing
with and caring
for kids
Future career
success
Future career
success
Sympathetic/kind
Sexually
experienced
Sexually
experienced
Likelihood of using
threats to get sex
Likelihood of using
threats to get sex
Moore et al. (2011),
Study 2
Facial cues of
testosterone
Facial cues of
cortisol
Moore (2011)
Intelligent
Genetic quality
✓
✓
✓
(table continues)
Table 1 (continued)
Inclusion in analyses
Long-term partner
quality
Sample size
Study and effects
HighLow- Effect
Reason
Broad
Narrow
Broad
Narrow
Scent
Facial cues
Relationship
fertility fertility Size
for
set of
set of
set of
set of
Facial
cues of
Facial
Body
Vocal
Behavioral
of
context
N
n
n
(g) Variance exclusion measures measures measures measures symmetry symmetry masculinity masculinity masculinity dominance testosterone
✓
✓
✓
✓
✓
✓
0.04
✓
✓
✓
0.24
0.04
✓
✓
✓
0.39
0.03
✓
✓
✓
Within
participants
0.45
0.03
✓
✓
✓
Within
participants
Within
participants
0.23
0.08
✓
✓
✓
⫺0.01
0.07
✓
✓
✓
✓
✓
⫺0.38
0.09
✓
0.96
0.11
0.04
0.00
✓
✓
✓
Within
participants
Within
participants
0.32
0.04
✓
✓
✓
0.27
0.03
✓
✓
✓
Within
participants
M
ST
47
ST
47
U
21
9
12
M
M
U
21
9
12
M
M
ST, LT, U
38
19
19
M
M
U
16
Within
participants
⫺0.23
0.05
ST
99
37
62
0.46
LT
108
39
69
U
139
55
84
U
39
Penton-Voak et al.
(1999), Study 2
Facial masculinity
ST
23
Facial masculinity
LT
26
Perrett et al. (2013),
Study 1
Facial masculinity
Perrett et al. (2013),
Study 2
Facial masculinity
U
1290
ST
29
Facial masculinity
LT
29
Peters et al. (2008)
Face and body cues
of semen quality
ST
25
189
189
0.10
0.10
446
446
Pawlowski & Jasienska
(2005)
Taller man relative
to self
Taller man relative
to self
Penton-Voak & Perrett
(2000)
Facial masculinity
Penton-Voak et al.
(1999), Study 1
Facial masculinity
257
257
⫺0.04
⫺0.11
U
U
Fertility
continuous
Fertility
continuous
527
763
✓
4
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
Facial masculinity
Facial symmetry
Morrison et al. (2010)
Male-typical facial
movements
Flirtatious facial
movements
Navarrete et al. (2009)
Body muscularity
Same-race (vs.
other-race) face
Oinonen et al. (2008)
Facial symmetry
Oinonen & Mazmanian
(2007)
Facial symmetry
Genetic quality
M
(table continues)
15
16
Table 1 (continued)
Inclusion in analyses
Long-term partner
quality
Sample size
Study and effects
HighLow- Effect
Reason
Broad
Narrow
Broad
Narrow
Scent
Facial cues
Relationship
fertility fertility Size
for
set of
set of
set of
set of
Facial
cues of
Facial
Body
Vocal
Behavioral
of
context
N
n
n
(g) Variance exclusion measures measures measures measures symmetry symmetry masculinity masculinity masculinity dominance testosterone
ST
25
ST
25
ST
25
ST
25
Body masculinity
ST
25
Facial symmetry
ST
25
Body symmetry
ST
25
M
M
M
M
M
M
Within
participants
Within
participants
Within
participants
Within
participants
0.00
0.08
7
⫺0.11
0.08
7
0.06
0.08
7
⫺0.03
0.08
7
204
Fertility
continuous
U
20
Within
participants
0.45
0.05
✓
Puts (2005)
Vocal masculinity
ST
137
0.42
0.03
✓
✓
✓
Vocal masculinity
LT
137
Fertility
continuous
Fertility
continuous
0.47
0.03
✓
✓
✓
ST, LT
136
38
98
M
M
ST, LT
136
38
98
M
M
U
36
11
25
⫺0.18
0.13
✓
✓
U
186
62
124
⫺0.35
0.02
✓
U
40
14
26
M
M
U
40
14
26
M
M
U
74
Fertility
continuous
0.46
0.06
Vocal cues of
perceived
physical
dominance
Vocal cues of
perceived social
dominance
Rantala et al. (2006)
Scent cues of
testosterone
Rantala et al. (2010)
Torso hair
Rikowski & Grammer
(1999)
Scent cues of body
symmetry
Scent cues of facial
symmetry
Roney & Simmons
(2008)
Facial cues of
testosterone
ST, LT
Within
participants
Within
participants
Within
participants
M
M
GILDERSLEEVE, HASELTON, AND FALES
Face and body
averageness
Face and body
masculinity
Face and body
symmetry
Peters et al. (2009)
Facial masculinity
Prokosch et al. (2009)
Creativity and
intelligence
Provost et al. (2008)
Male-typical walk
Genetic quality
4
✓
✓
✓
(table continues)
Table 1 (continued)
Inclusion in analyses
Long-term partner
quality
Sample size
Study and effects
Genetic quality
HighLow- Effect
Reason
Broad
Narrow
Broad
Narrow
Scent
Facial cues
Relationship
fertility fertility Size
for
set of
set of
set of
set of
Facial
cues of
Facial
Body
Vocal
Behavioral
of
context
N
n
n
(g) Variance exclusion measures measures measures measures symmetry symmetry masculinity masculinity masculinity dominance testosterone
U
18
U
18
Rupp, Librach, et al.
(2009)
Facial masculinity
U
13
Fertility
continuous
Rupp, James, et al.
(2009)
Facial masculinity
U
12
Within
participants
ST
64
49
15
0.00
LT
130
91
39
ST
64
49
LT
130
LT
0.43
0.11
✓
✓
⫺0.23
0.10
✓
✓
✓
⫺0.64
0.39
✓
✓
✓
0.04
✓
✓
✓
0.11
0.02
✓
✓
✓
15
0.39
0.09
✓
✓
✓
91
39
⫺0.11
0.04
✓
✓
✓
52
8
44
M
M
LT
76
30
46
M
M
ST
14
0.31
0.13
✓
Kindness
LT
14
0.01
0.13
✓
Faithfulness
ST
14
0.35
0.14
✓
Faithfulness
LT
14
0.45
0.14
✓
Social status
ST
14
0.17
0.13
4
Social status
LT
14
⫺0.03
0.13
4
Financial resources
ST
14
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
0.11
0.13
Singh & Bailey (2006)
Male-typical
shoulder-to-hip
ratio
Male-typical
shoulder-to-hip
ratio
Male-typical waistto-hip ratio
Male-typical waistto-hip ratio
Soler et al. (2003),
Study 1
Facial cues of semen
quality
Soler et al. (2003),
Study 2
Facial cues of semen
quality
Teatero (2009)
Kindness
Within
participants
Within
participants
M
M
✓
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
Roney et al. (2011)
Facial cues of
testosterone
Facial masculinity
6
✓
(table continues)
17
18
Table 1 (continued)
Inclusion in analyses
Long-term partner
quality
Sample size
Study and effects
HighLow- Effect
Reason
Broad
Narrow
Broad
Narrow
Scent
Facial cues
Relationship
fertility fertility Size
for
set of
set of
set of
set of
Facial
cues of
Facial
Body
Vocal
Behavioral
of
context
N
n
n
(g) Variance exclusion measures measures measures measures symmetry symmetry masculinity masculinity masculinity dominance testosterone
LT
14
Sense of humor
ST
14
Sense of humor
LT
14
Good parent
ST
14
Good parent
LT
14
Intelligence
ST
14
Intelligence
LT
14
U
48
U
48
U
48
U
65
U
139
U
70
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
Within
participants
✓
0.26
0.13
0.20
0.13
4
0.39
0.14
4
0.39
0.14
✓
⫺0.12
0.13
✓
0.03
0.13
4
0.00
0.13
4
Fertility
continuous
0.94
0.11
✓
✓
Fertility
continuous
Fertility
continuous
0.66
0.10
✓
✓
0.55
0.09
0.55
0.09
✓
✓
79
⫺0.01
0.03
✓
✓
✓
Within
participants
0.26
0.01
✓
✓
✓
Fertility
continuous
60
GILDERSLEEVE, HASELTON, AND FALES
Financial resources
Thornhill & Gangestad
(1999b)
Scent cues of body
symmetry
Thornhill et al. (2013)
Scent cues of
testosterone
Scent cues of
cortisol
Thornhill et al. (2003)
Scent cues of body
symmetry
Vaughn et al. (2010)
Facial masculinity
Welling et al. (2007)
Facial masculinity
Genetic quality
✓
4
✓
Note. Checkmarks in the “Inclusion in analyses” columns indicate in which analyses an effect was included (a blank indicates that the effect was not included in a given analysis). For nonmissing
effects excluded from all analyses, the “Reason for exclusion” column indicates the specific inclusion criterion that the effect did not satisfy (i.e., the reason it was excluded). For example, “4” refers
to Inclusion Criterion 4. ST ⫽ short term; LT ⫽ long term; U ⫽ unspecified; M ⫽ missing data; CPI ⫽ California Psychological Inventory.
Table 2
Studies Assessing Ovulation-Related Cycle Shifts in Mate Preferences: Study and Effect Characteristics
Average
conception
probability
Study and effects
Caryl et al. (2009)
Pupil size
Arrogant
Ingenious
Aggressive
Country
Sample
type
Setting
Design
High
fertility
Low
fertility
Stimuli
Method to
determine
amount of
male trait
Stated
versus
revealed
preferences
Task
Number
of trials
U
United States
Col
L
B
0.09
0.01
Forw
SR
N/A
Stated
Ratings
1
U
United States
Col
L
B
0.09
0.01
Forw
SR
N/A
Stated
Ratings
1
U
United States
Col
L
B
0.09
0.01
Forw
SR
N/A
Stated
Ratings
1
U
United States
Col
L
B
0.09
0.01
Forw
SR
N/A
Stated
Ratings
1
ST, LT
United States
Col
L
B
0.09
0.01
Forw
SR
N/A
Stated
Ratings
1
ST, LT
United States
Col
L
B
0.09
0.01
Forw
SR
N/A
Stated
Ratings
1
ST, LT
United States
Col
L
B
0.09
0.01
Forw
SR
N/A
Stated
Ratings
1
ST, LT
United States
Col
L
B
0.09
0.01
Forw
SR
N/A
Stated
Ratings
1
U
Italy
Col
L
B
0.06
0.01
Forw
FP
SSR
Revealed
Ratings
12
U
Canada
Col
L
B
0.06
0.02
Forw
FP
Manip
Revealed
Ratings
5
ST
United States,
Chile
Col
L
W
0.09
0.01
Forw
FP
Manip
Revealed
TOFC
54
U
U
U
U
Scotland
Scotland
Scotland
Scotland
Col
Col
Col
Col
L
L
L
L
B
B
B
B
0.06
0.06
0.06
0.06
0.02
0.02
0.02
0.02
Forw
Forw
Forw
Forw
SR
SR
SR
SR
N/A
N/A
N/A
N/A
Stated
Stated
Stated
Stated
Ratings
1
Ratings
1
Ratings
1
Ratings
1
(table continues)
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
Beaulieu (2007), Study 2
Relationship skills
(composite of kind,
understanding, loyal,
generous)
Dominance (composite
of dominant,
powerful, aggressive)
Education (composite
of educated, cultured,
intelligent)
Good financial
prospects (composite
of wealthy, good
financial prospects)
Beaulieu (2007), Study 4
Relationship skills
(composite of kind,
understanding, loyal,
generous)
Dominance (composite
of dominant,
powerful, aggressive)
Education (composite
of educated, cultured,
intelligent)
Good financial
prospects (composite
of wealthy, good
financial prospects)
Bressan & Stranieri
(2008)
Facial masculinity
Bullock (2000), Chapter 4
Chin length
Cárdenas & Harris (2007)
Facial symmetry
Relationship
context
Cycle
position
estimation
method
19
20
Table 2 (continued)
Average
conception
probability
Study and effects
Flowe et al., 2012
Behavioral masculinity
Frost (1994)
Darker skin tone
Country
Sample
type
Setting
Design
High
fertility
Low
fertility
Stimuli
Method to
determine
amount of
male trait
Stated
versus
revealed
preferences
Task
Number
of trials
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Scotland
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
Col
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
U
Scotland
Col/Com
L
B
0.06
0.02
Forw
FP
Manip
Revealed
TOFC
36
U
Scotland
Col/Com
L
W
0.07
0.02
Rev, VM
VR
Manip
Revealed
Ratings
8
ST
LT
Canada
Canada
Col
Col
L
L
W
W
0.07
0.07
0.02
0.02
Rev, VM
Rev, VM
VR
VR
Manip
Manip
Revealed
Revealed
TOFC
TOFC
4
4
U
Austria, United
Kingdom
M
M
W
0.06
0.02
Forw
FP
Manip
Revealed
Slider
1
U
United Kingdom
Col
L
B
0.08
0.01
Forw
VB, D
Manip
Revealed
Ratings
1
U
Canada
Col
L
B
0.08
0.01
Forw
FP
Manip
Revealed
TOFC
3
(table continues)
GILDERSLEEVE, HASELTON, AND FALES
Strong
Conceited
Enterprising
Inventive
Warm
Sensitive
Sentimental
Sympathetic
Jolly
Helpful
Appreciative
Considerate
Cooperative
Friendly
Talkative
Forgiving
Emotional
Foresighted
Shrewd
Industrious
Assertive
Forceful
Timid
Dependent
Fickle
Frivolous
Opportunistic
Hardheaded
Confident
DeBruine et al. (2005)
Facial self-resemblance
Feinberg et al. (2006)
Vocal masculinity
Feinberg (2012)
Vocal masculinity
Vocal masculinity
Fink (2012)
Facial masculinity
Relationship
context
Cycle
position
estimation
method
Table 2 (continued)
Average
conception
probability
Study and effects
Country
Sample
type
Setting
Design
High
fertility
Low
fertility
Stimuli
Method to
determine
amount of
male trait
Stated
versus
revealed
preferences
Task
Number
of trials
ST
LT
United States
United States
Col
Col
L
L
B
B
Continuous
Continuous
Forw
Forw
VB
VB
Meas
Meas
Revealed
Revealed
Ratings
Ratings
38
38
ST
United States
Col
L
B
Continuous
Forw
VB
Meas
Revealed
Ratings
38
LT
United States
Col
L
B
Continuous
Forw
VB
Meas
Revealed
Ratings
38
ST
LT
ST
United States
United States
United States
Col
Col
Col
L
L
L
B
B
B
Continuous
Continuous
Continuous
Forw
Forw
Forw
VB
VB
VB
SSR
SSR
SSR
Revealed
Revealed
Revealed
Ratings
Ratings
Ratings
38
38
38
LT
United States
Col
L
B
Continuous
Forw
VB
SSR
Revealed
Ratings
38
ST
United States
Col
L
B
Continuous
Forw
VB
SSR
Revealed
Ratings
38
LT
United States
Col
L
B
Continuous
Forw
VB
SSR
Revealed
Ratings
38
ST
United States
Col
L
B
Continuous
Forw
VB
SSR
Revealed
Ratings
38
LT
ST
LT
ST
LT
United
United
United
United
United
States
States
States
States
States
Col
Col
Col
Col
Col
L
L
L
L
L
B
B
B
B
B
Continuous
Continuous
Continuous
Continuous
Continuous
Forw
Forw
Forw
Forw
Forw
VB
VB
VB
VB
VB
SSR
SSR
SSR
SSR
SSR
Revealed
Revealed
Revealed
Revealed
Revealed
Ratings
Ratings
Ratings
Ratings
Ratings
38
38
38
38
38
ST
United States
Col
L
B
Continuous
Forw
VB
SSR
Revealed
Ratings
38
LT
United States
Col
L
B
Continuous
Forw
VB
SSR
Revealed
Ratings
38
ST
United States
Col
L
B
Continuous
Forw
VB
SSR
Revealed
Ratings
38
LT
United States
Col
L
B
Continuous
Forw
VB
SSR
Revealed
Ratings
38
ST
United States
Col
L
B
Continuous
Forw
VB
SSR
Revealed
Ratings
38
LT
United States
Col
L
B
Continuous
Forw
VB
SSR
Revealed
Ratings
38
U
United States
M
L
W
LH
FP
Manip
Revealed
Slider
U
United States
Col/Com
L
B
Forw
SS
Meas
Revealed
Ratings
42
(table continues)
0.23
0.02
Continuous
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
Gangestad et al. (2004)
Social presence
Social presence
Direct intrasexual
competitiveness
Direct intrasexual
competitiveness
Gangestad et al. (2007)
Muscular
Muscular
Confrontative (with
other men)
Confrontative (with
other men)
Socially respected and
influential
Socially respected and
influential
Arrogant and selfcentered
Arrogant and selfcentered
Intelligent
Intelligent
Faithful
Faithful
Warm (kind and
understanding)
Warm (kind and
understanding)
Likely to be financially
successful
Likely to be financially
successful
Likely to be a good
parent
Likely to be a good
parent
Gangestad et al. (2011)
Facial masculinity
Gangestad & Thornhill
(1998)
Scent cues of body
symmetry
Relationship
context
Cycle
position
estimation
method
1
21
22
Table 2 (continued)
Average
conception
probability
Study and effects
Haselton & Miller (2006)
Wealth versus
creativity
Havlíček et al. (2005)
Scent cues of
dominance
(Narcissism scale
from CPI)
Hromatko et al. (2006)
Facial symmetry
Izbicki & Johnson (2010)
Facial masculinity
Facial masculinity
Strong
Strong
Warm
Warm
Mature
Mature
Socially competent
Socially competent
Nurturant
Nurturant
Threatening
Threatening
Dominant
Dominant
Dark
Dark
Johnston et al. (2001)
Facial masculinity
Country
Sample
type
Setting
Design
High
fertility
Low
fertility
Stimuli
Method to
determine
amount of
male trait
Stated
versus
revealed
preferences
Task
Number
of trials
ST
United States
Col
L
W
0.26
0.02
LH
VB
Meas
Revealed
Ratings
38
LT
United States
Col
L
W
0.26
0.02
LH
VB
Meas
Revealed
Ratings
38
U
Canada, United
States
Com
F
B
0.07
0.01
Forw
FP
Manip
Revealed
MOFC
5
ST, LT
United States
Col
L
B
Forw
D
Manip
Revealed
TOFC, Ratings
4
U
Czech Republic
Col
L
B
0.08
0.02
Forw
SS
Meas
Revealed
Ratings
10
U
Croatia
M
L
B
0.09
0.01
Forw & Rev
FP
Manip
Revealed
Ratings
40
ST
LT
ST
LT
ST
LT
ST
LT
ST
LT
ST
LT
ST
LT
ST
LT
ST
LT
United
United
United
United
United
United
United
United
United
United
United
United
United
United
United
United
United
United
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
0.08
0.08
0.08
0.08
0.08
0.08
0.08
0.08
0.08
0.08
0.08
0.08
0.08
0.08
0.08
0.08
0.08
0.08
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
Rev
Rev
Rev
Rev
Rev
Rev
Rev
Rev
Rev
Rev
Rev
Rev
Rev
Rev
Rev
Rev
Rev
Rev
FP
FP
FP
FP
FP
FP
FP
FP
FP
FP
FP
FP
FP
FP
FP
FP
FP
FP
PR
PR
PR
PR
PR
PR
PR
PR
PR
PR
PR
PR
PR
PR
PR
PR
PR
PR
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
U
United States
Col
L
W
0.06
0.02
Rev, VM
FP
Manip
Revealed
Slider
States
States
States
States
States
States
States
States
States
States
States
States
States
States
States
States
States
States
Continuous
1
(table continues)
GILDERSLEEVE, HASELTON, AND FALES
Garver-Apgar &
Gangestad (2012)
Average of social
presence and direct
intrasexual
competitiveness
Average of social
presence and direct
intrasexual
competitiveness
Harris (2011)
Facial masculinity
Relationship
context
Cycle
position
estimation
method
Table 2 (continued)
Average
conception
probability
Study and effects
Country
Sample
type
Setting
Design
High
fertility
Low
fertility
Stimuli
Method to
determine
amount of
male trait
Stated
versus
revealed
preferences
Task
Number
of trials
U
United Kingdom
Com
F
B
0.07
0.02
Rev
FP
Manip
Revealed
TOFC
7
ST, LT
Australia
Col
L
W
0.09
0.01
Forw
FP
Manip
Revealed
Ratings
24
U
U
Australia
Australia
Col
Col
L
L
W
W
0.09
0.09
0.01
0.01
Forw
Forw
FP
FP
Manip
Manip
Revealed
Revealed
TOFC
TOFC
24
24
ST, LT
United States
Col
L
W
M
M
Forw
SR
N/A
Stated
Mate dollars
M
U
United Kingdom
M
L
W
0.08
0.01
Rev
FP
Manip
Revealed
TOFC
6
ST
LT
United Kingdom
United Kingdom
Col/Com
Col/Com
L, F
L, F
B
B
0.06
0.06
0.02
0.02
Forw
Forw
FP
FP
Manip
Manip
Revealed
Revealed
TOFC
TOFC
15
15
ST
LT
United Kingdom
United Kingdom
Com
Com
F
F
B
B
0.06
0.06
0.02
0.02
Forw
Forw
BP
BP
Manip
Manip
Revealed
Revealed
TOFC
TOFC
10
10
ST
LT
United Kingdom
United Kingdom
M
M
L
L
W
W
0.06
0.06
0.02
0.02
Forw
Forw
BP
BP
Manip
Manip
Revealed
Revealed
TOFC
TOFC
10
10
U
United Kingdom
Com
F
B
0.06
0.02
Forw
FP
SSR
Revealed
TOFC
10
ST
LT
ST
LT
ST
LT
United
United
United
United
United
United
States
States
States
States
States
States
M
M
M
M
M
M
L
L
L
L
L
L
W
W
W
W
W
W
0.08
0.08
0.08
0.08
0.08
0.08
0.02
0.02
0.02
0.02
0.02
0.02
Rev
Rev
Rev
Rev
Rev
Rev
FP
FP
FP
FP
FP
FP
SSR
SSR
SSR
SSR
SSR
SSR
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
153
153
153
153
153
153
ST
LT
ST
LT
ST
LT
United
United
United
United
United
United
States
States
States
States
States
States
Col
Col
Col
Col
Col
Col
L
L
L
L
L
L
B
B
B
B
B
B
Forw
Forw
Forw
Forw
Forw
Forw
SR
SR
SR
SR
SR
SR
N/A
N/A
N/A
N/A
N/A
N/A
Stated
Stated
Stated
Stated
Stated
Stated
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
1
1
1
1
1
1
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
Jones, Little, et al. (2005),
Study 2
Facial masculinity
Koehler et al. (2002)
Facial symmetry
Koehler et al. (2006)
Facial averageness
Facial symmetry
Li et al. (2006)
Multiple traits
Little, Jones, et al. (2007),
Study 1
Facial symmetry
Little, Jones, et al. (2007),
Study 2
Facial symmetry
Facial symmetry
Little, Jones, & Burriss
(2007), Study 1
Body masculinity
Body masculinity
Little, Jones, & Burriss
(2007), Study 2
Body masculinity
Body masculinity
Little et al. (2008)
Facial masculinity
Luevano & Zebrowitz
(2006)
Dominant
Dominant
Facial masculinity
Facial masculinity
Warm
Warm
Lukaszewski & Roney
(2009)
Dominant
Dominant
Kind
Kind
Trustworthy
Trustworthy
McClellan et al. (2007)
Relationship
context
Cycle
position
estimation
method
(table continues)
23
24
Table 2 (continued)
Average
conception
probability
Study and effects
Country
Sample
type
Setting
Design
High
fertility
Low
fertility
Stimuli
Method to
determine
amount of
male trait
Stated
versus
revealed
preferences
Task
Number
of trials
ST, LT, U
ST, LT, U
United States
United States
Col
Col
L
L
B
B
0.05
0.05
0.02
0.02
Forw
Forw
BPa
BPa
Meas, PR
Meas, PR
Revealed
Revealed
Rank
Rank
M
M
U
United States
Col
L
B
0.09
0.01
Forw
BAv
Manip
Revealed
Ratings
10
U
United States
Col
L
B
0.09
0.01
Forw
FAv
Manip
Revealed
Ratings
10
U
United States
Col
L
B
0.09
0.01
Forw
BAv
Manip
Revealed
Ratings
M
U
United States
Col
L
B
0.09
0.01
Forw
FAv
Manip
Revealed
Ratings
M
ST
LT
United States
United States
Col
Col
L
L
B
B
Continuous
Continuous
Forw
Forw
D
D
PR
PR
Revealed
Revealed
Ratings
Ratings
3
3
ST
United States
Col
L
B
Continuous
Forw
D
PR
Revealed
Ratings
3
LT
United States
Col
L
B
Continuous
Forw
D
PR
Revealed
Ratings
3
ST
United States
Col
L
B
Continuous
Forw
D
PR
Revealed
Ratings
3
LT
ST
LT
ST
LT
United
United
United
United
United
States
States
States
States
States
Col
Col
Col
Col
Col
L
L
L
L
L
B
B
B
B
B
Continuous
Continuous
Continuous
Continuous
Continuous
Forw
Forw
Forw
Forw
Forw
D
D
D
D
D
PR
PR
PR
PR
PR
Revealed
Revealed
Revealed
Revealed
Revealed
Ratings
Ratings
Ratings
Ratings
Ratings
3
3
3
3
3
ST
United States
Col
L
B
Continuous
Forw
D
PR
Revealed
Ratings
3
LT
ST
LT
United States
United States
United States
Col
Col
Col
L
L
L
B
B
B
Continuous
Continuous
Continuous
Forw
Forw
Forw
D
D
D
PR
PR
PR
Revealed
Revealed
Revealed
Ratings
Ratings
Ratings
3
3
3
ST
United States
Col
L
B
Continuous
Forw
D
PR
Revealed
Ratings
3
LT
ST
LT
ST
LT
ST
LT
ST
United
United
United
United
United
United
United
United
Col
Col
Col
Col
Col
Col
Col
Col
L
L
L
L
L
L
L
L
B
B
B
B
B
B
B
B
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
D
D
D
D
D
D
D
D
PR
PR
PR
PR
PR
PR
PR
PR
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Ratings
3
Ratings
3
Ratings
3
Ratings
3
Ratings
3
Ratings
3
Ratings
3
Ratings
3
(table continues)
States
States
States
States
States
States
States
States
GILDERSLEEVE, HASELTON, AND FALES
Body masculinity
Age
McDonald & Navarrete
(2012), Sample 1
Body muscularity
Same-race (vs. otherrace) face
McDonald & Navarrete
(2012), Sample 2
Body muscularity
Same-race (vs. otherrace) face
Miller (2003)
Intelligent
Intelligent
Future kids’
intelligence
Future kids’
intelligence
Mathematical problemsolving ability
Mathematical problemsolving ability
Good grades
Good grades
Creative/imaginative
Creative/imaginative
Future kids’ sense of
humor
Future kids’ sense of
humor
Social sensitivity
Social sensitivity
Adaptable to situations
and challenges
Adaptable to situations
and challenges
Big ego
Big ego
Body muscularity
Body muscularity
Facial masculinity
Facial masculinity
Tall
Relationship
context
Cycle
position
estimation
method
Table 2 (continued)
Average
conception
probability
Study and effects
Country
Sample
type
Setting
Design
High
fertility
Low
fertility
Stimuli
Method to
determine
amount of
male trait
Stated
versus
revealed
preferences
Task
Number
of trials
LT
ST
LT
ST
LT
ST
LT
United
United
United
United
United
United
United
States
States
States
States
States
States
States
Col
Col
Col
Col
Col
Col
Col
L
L
L
L
L
L
L
B
B
B
B
B
B
B
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
Forw
Forw
Forw
Forw
Forw
Forw
Forw
D
D
D
D
D
D
D
PR
PR
PR
PR
PR
PR
PR
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
3
3
3
3
3
3
3
ST
United States
Col
L
B
Continuous
Forw
D
PR
Revealed
Ratings
3
LT
ST
LT
United States
United States
United States
Col
Col
Col
L
L
L
B
B
B
Continuous
Continuous
Continuous
Forw
Forw
Forw
D
D
D
PR
PR
PR
Revealed
Revealed
Revealed
Ratings
Ratings
Ratings
3
3
3
ST
United States
Col
L
B
Continuous
Forw
D
PR
Revealed
Ratings
3
LT
ST
LT
ST
LT
United
United
United
United
United
States
States
States
States
States
Col
Col
Col
Col
Col
L
L
L
L
L
B
B
B
B
B
Continuous
Continuous
Continuous
Continuous
Continuous
Forw
Forw
Forw
Forw
Forw
D
D
D
D
D
PR
PR
PR
PR
PR
Revealed
Revealed
Revealed
Revealed
Revealed
Ratings
Ratings
Ratings
Ratings
Ratings
3
3
3
3
3
ST
United States
Col
L
B
Continuous
Forw
D
PR
Revealed
Ratings
3
LT
ST
LT
ST
LT
ST
LT
ST
LT
United
United
United
United
United
United
United
United
United
States
States
States
States
States
States
States
States
States
Col
Col
Col
Col
Col
Col
Col
Col
Col
L
L
L
L
L
L
L
L
L
B
B
B
B
B
B
B
B
B
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
Forw
D
D
D
D
D
D
D
D
D
PR
PR
PR
PR
PR
PR
PR
PR
PR
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Revealed
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
Ratings
3
3
3
3
3
3
3
3
3
ST
United States
Col
L
B
Continuous
Forw
D
PR
Revealed
Ratings
3
LT
United States
Col
L
B
Continuous
Forw
D
PR
Revealed
Ratings
3
U
U
Scotland
Scotland
Col
Col
M
M
W
W
Rev
Rev
FP
FP
Manip
Manip
Revealed
Revealed
Ratings
16
Ratings
16
(table continues)
0.07
0.07
0.02
0.02
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
Tall
Happy
Happy
Exciting/spontaneous
Exciting/spontaneous
Talkative/extraverted
Talkative/extraverted
Likelihood of being
unfaithful (reversecoded)
Likelihood of being
unfaithful (reversecoded)
Future money making
Future money making
Good at playing with
and caring for kids
Good at playing with
and caring for kids
Future career success
Future career success
Sympathetic/kind
Sympathetic/kind
Constructive in
arguments
Constructive in
arguments
Neat/organized
Neat/organized
Moody/irritable
Moody/irritable
Fun at sex
Fun at sex
Sexually experienced
Sexually experienced
Likelihood of using
threats to get sex
Likelihood of using
threats to get sex
Moore et al. (2011),
Study 2
Facial cues of
testosterone
Facial cues of cortisol
Relationship
context
Cycle
position
estimation
method
25
26
Table 2 (continued)
Average
conception
probability
Study and effects
Country
Sample
type
Setting
Design
High
fertility
Low
fertility
0.08
0.08
0.08
0.01
0.01
0.01
U
U
U
M
M
M
Com
Com
Com
L
L
L
B
B
B
ST
United Kingdom
Col
L
B
ST
United Kingdom
Col
L
B
U
United States
Col
L
B
0.09
U
United States
Col
L
B
ST, LT, U
Canada
Col/Com
L
U
Canada
Col/Com
ST
Poland
LT
Stimuli
Method to
determine
amount of
male trait
Stated
versus
revealed
preferences
Task
Number
of trials
Forw
Forw
Forw
FP
FP
FP
Manip
Manip
Manip
Revealed
Revealed
Revealed
Ratings
Ratings
Ratings
6
16
6
Continuous
Forw
MFO
SSR
Revealed
Ratings
30
Continuous
Forw
MFO
SSR
Revealed
Ratings
30
0.01
Forw
BAv
Manip
Revealed
Ratings
4
0.09
0.01
Forw
FAv
Manip
Revealed
Ratings
4
W
0.07
0.01
Forw & Rev
FP
Manip
Revealed
Ratings
80
L
W
0.07
0.02
Rev
FP
Manip
Revealed
TOFC
80
Com
L
B
0.04
0.03
Rev, VM
BOD
Manip
Revealed
TOFC
45
Poland
Com
L
B
0.04
0.03
Rev, VM
BOD
Manip
Revealed
TOFC
45
U
United Kingdom
Com
F(mag)
B
0.06
0.02
Forw
FP
Manip
Revealed
MOFC
1
U
Japan
Col/Com
L
W
0.04
0.02
Rev
FP
Manip
Revealed
MOFC
10
ST
LT
United Kingdom
United Kingdom
Col
Col
M
M
W
W
0.04
0.04
0.02
0.02
Rev
Rev
FP
FP
Manip
Manip
Revealed
Revealed
MOFC
MOFC
1
1
U
M
Com
F
B
0.04
0.02
Rev
FP
Manip
Revealed
TOFC
3
ST
LT
United Kingdom
United Kingdom
Col
Col
L
L
W
W
M
M
M
M
M
M
FP
FP
Manip
Manip
Revealed
Revealed
TOFC
TOFC
6
6
ST
Australia
M
L
W
0.32
M
LH
BP, FP
Meas
Revealed
Ratings
101
(table continues)
GILDERSLEEVE, HASELTON, AND FALES
Moore (2011)
Intelligent
Facial masculinity
Facial symmetry
Morrison et al. (2010)
Male-typical facial
movements
Flirtatious facial
movements
Navarrete et al. (2009)
Body muscularity
Same-race (vs. otherrace) face
Oinonen et al. (2008)
Facial symmetry
Oinonen & Mazmanian
(2007)
Facial symmetry
Pawlowski (2005)
Taller man relative to
self
Taller man relative to
self
Penton-Voak & Perrett
(2000)
Facial masculinity
Penton-Voak et al.
(1999), Study 1
Facial masculinity
Penton-Voak et al.
(1999), Study 2
Facial masculinity
Facial masculinity
Perrett et al. (2013),
Study 1
Facial masculinity
Perrett et al. (2013),
Study 2
Facial masculinity
Facial masculinity
Peters et al. (2008)
Face and body cues of
semen quality
Relationship
context
Cycle
position
estimation
method
Table 2 (continued)
Average
conception
probability
Study and effects
Country
Sample
type
Setting
Design
High
fertility
Low
fertility
Stimuli
Method to
determine
amount of
male trait
Stated
versus
revealed
preferences
Task
Number
of trials
ST
Australia
M
L
W
0.32
M
LH
BP, FP SSR
Revealed
Ratings
116
ST
Australia
M
L
W
0.32
M
LH
BP, FP SSR
Revealed
Ratings
116
ST
Australia
M
L
W
0.32
M
LH
BP, FP SSR
Revealed
Ratings
116
ST
ST
ST
ST
Australia
Australia
Australia
Australia
M
M
M
M
L
L
L
L
W
W
W
W
0.32
0.32
0.32
0.32
M
M
M
M
LH
LH
LH
LH
FP
BP
FP
BP
SSR
SSR
SSR
SSR
Revealed
Revealed
Revealed
Revealed
Ratings
Ratings
Ratings
Ratings
117
117
117
117
ST, LT
United States
Col
L
B
Continuous
Rev
VB
PR & Meas
Revealed
Ratings
5
U
Canada
Col
L
W
0.08
Forw, Saliv
PLW
Manip
Revealed
Slider
1
ST
LT
ST
United States
United States
United States
Col
Col
Col
L
L
L
B
B
B
Continuous
Continuous
0.07
0.02
Rev
Rev
Rev
VR
VR
VR
Manip
Manip
SSR
Revealed
Revealed
Revealed
Ratings
Ratings
Ratings
30
30
33
LT
United States
Col
L
B
0.07
0.02
Rev
VR
SSR
Revealed
Ratings
31
ST
United States
Col
L
B
0.07
0.02
Rev
VR
SSR
Revealed
Ratings
34
LT
United States
Col
L
B
0.07
0.02
Rev
VR
SSR
Revealed
Ratings
32
U
Finland
Col
L
B
0.08
0.02
LH
SS
Meas
Revealed
Ratings
19
U
Finland
Com
L
B
0.08
0.02
Rev
BP
Manip
Revealed
TOFC
20
U
Austria
Col
L
B
0.06
0.01
Forw
SS
Meas
Revealed
Ratings
8
U
Austria
Col
L
B
0.06
0.01
Forw
SS
Meas
Revealed
Ratings
8
(table continues)
0.01
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
Face and body
averageness
Face and body
masculinity
Face and body
symmetry
Peters et al. (2009)
Facial masculinity
Body masculinity
Facial symmetry
Body symmetry
Prokosch et al. (2009)
Creativity and
intelligence
Provost et al. (2008)
Male-typical walk
Puts (2005)
Vocal masculinity
Vocal masculinity
Vocal cues of
perceived physical
dominance
Vocal cues of
perceived physical
dominance
Vocal cues of
perceived social
dominance
Vocal cues of
perceived social
dominance
Rantala et al. (2006)
Scent cues of
testosterone
Rantala et al. (2010)
Torso hair
Rikowski & Grammer
(1999)
Scent cues of body
symmetry
Scent cues of facial
symmetry
Relationship
context
Cycle
position
estimation
method
27
28
Table 2 (continued)
Average
conception
probability
Study and effects
Country
Sample
type
Setting
Design
High
fertility
U
United States
Col
L
B
U
U
United States
United States
Col
Col
L
L
W
W
U
United States
Col
F
B
U
United States
Col
L
W
0.08
ST
United States
M
M
B
LT
United States
M
M
ST
United States
M
LT
United States
LT
Low
fertility
Continuous
Stimuli
Method to
determine
amount of
male trait
Stated
versus
revealed
preferences
Task
Number
of trials
Forw
FP
Meas
Revealed
Ratings
37
Rev, VM
Rev, VM
FP
FP
Manip
Manip
Revealed
Revealed
TOFC
TOFC
7
7
Forw
FP
SSR
Revealed
Ratings
510
0.02
Forw
FP
Manip
Revealed
Ratings
224
0.06
0.01
Forw
BOD
Manip
Revealed
Ratings
6
B
0.06
0.01
Forw
BOD
Manip
Revealed
Ratings
6
M
B
0.06
0.01
Forw
BOD
Manip
Revealed
Ratings
6
M
M
B
0.06
0.01
Forw
BOD
Manip
Revealed
Ratings
6
Spain
Col
L
B
M
M
Forw
FP
Meas
Revealed
Ratings
66
LT
Spain
Col
L
B
M
M
Forw
FP
Meas
Revealed
Ratings
12
ST
LT
ST
LT
ST
LT
ST
LT
ST
LT
ST
LT
ST
LT
Canada
Canada
Canada
Canada
Canada
Canada
Canada
Canada
Canada
Canada
Canada
Canada
Canada
Canada
Col/Com
Col/Com
Col/Com
Col/Com
Col/Com
Col/Com
Col/Com
Col/Com
Col/Com
Col/Com
Col/Com
Col/Com
Col/Com
Col/Com
F
F
F
F
F
F
F
F
F
F
F
F
F
F
W
W
W
W
W
W
W
W
W
W
W
W
W
W
0.09
0.09
0.09
0.09
0.09
0.09
0.09
0.09
0.09
0.09
0.09
0.09
0.09
0.09
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
Rev, VM
Rev, VM
Rev, VM
Rev, VM
Rev, VM
Rev, VM
Rev, VM
Rev, VM
Rev, VM
Rev, VM
Rev, VM
Rev, VM
Rev, VM
Rev, VM
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
SR
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Stated
Ratings
1
Ratings
1
Ratings
1
Ratings
1
Ratings
1
Ratings
1
Ratings
1
Ratings
1
Ratings
1
Ratings
1
Ratings
1
Ratings
1
Ratings
1
Ratings
1
(table continues)
0.08
0.08
0.02
0.02
Continuous
GILDERSLEEVE, HASELTON, AND FALES
Roney & Simmons (2008)
Facial cues of
testosterone
Roney et al. (2011)
Facial cues of
testosterone
Facial masculinity
Rupp, Librach, et al.
(2009)
Facial masculinity
Rupp, James, et al. (2009)
Facial masculinity
Singh & Bailey (2006)
Male-typical shoulderto-hip ratio
Male-typical shoulderto-hip ratio
Male-typical waist-tohip ratio
Male-typical waist-tohip ratio
Soler et al. (2003), Study 1
Facial cues of semen
quality
Soler et al. (2003), Study 2
Facial cues of semen
quality
Teatero (2009)
Kindness
Kindness
Faithfulness
Faithfulness
Social status
Social status
Financial resources
Financial resources
Sense of humor
Sense of humor
Good parent
Good parent
Intelligence
Intelligence
Relationship
context
Cycle
position
estimation
method
Table 2 (continued)
Study and effects
Thornhill & Gangestad
(1999b)
Scent cues of body
symmetry
Thornhill et al. (2013)
Scent cues of
testosterone
Scent cues of cortisol
Thornhill et al. (2003)
Scent cues of body
symmetry
Vaughn et al. (2010)
Facial masculinity
Welling et al. (2007)
Facial masculinity
Relationship
context
Country
Sample
type
Setting
Design
High
fertility
Low
fertility
Cycle
position
estimation
method
Stimuli
Method to
determine
amount of
male trait
Stated
versus
revealed
preferences
Task
Number
of trials
U
United States
Col
L
B
Continuous
Forw & Rev
SS
Meas
Revealed
Ratings
78
U
U
United States
United States
Col
Col
L
L
B
B
Continuous
Continuous
Forw & Rev
Forw & Rev
SS
SS
Meas
Meas
Revealed
Revealed
Ratings
Ratings
46
46
U
United States
Col
L
B
Continuous
Forw & Rev
SS
Meas
Revealed
Ratings
56
U
United States
Col
L
B
0.08
0.01
Forw
FP
Manip
Revealed
Slider
1
U
M
M
L
W
0.09
0.01
Forw
FP
Manip
Revealed
TOFC
20
Note. ST ⫽ short term; LT ⫽ long term; U ⫽ unspecified; Col ⫽ college/university students; Com ⫽ community; L ⫽ lab; F ⫽ field (online unless specified as “Mag” [magazine survey]); W ⫽
within participants; B ⫽ between participants; Forw ⫽ forward counting method; Rev ⫽ reverse counting method; Forw & Rev ⫽ average from forward and reverse counting methods; VM ⫽ verified
benchmark date of menstrual onset; Saliv ⫽ salivary ferning method to verify ovulation; LH ⫽ luteinizing hormone tests to verify ovulation; SR ⫽ self-reported preference; FP ⫽ facial photos; BP ⫽
body photos; BP & FP ⫽ body photos and facial photos (ratings averaged); VR ⫽ vocal recordings, VB ⫽ videotaped behavior; SS ⫽ scent samples; FAv ⫽ face avatars; BAv ⫽ full body avatars;
MFO ⫽ moving facial outlines; BOD ⫽ body outline drawings; D ⫽ descriptions of hypothetical men; PLW ⫽ point-light walker; Manip ⫽ manipulated; Meas ⫽ measured; SSR ⫽ ratings by a separate
sample of participants; N/A ⫽ not applicable; PR ⫽ ratings by cycle study participants; Ratings ⫽ rated stimuli on a scale; TOFC ⫽ two-option forced choice; MOFC ⫽ multiple-option (3⫹) forced
choice; Mate dollars ⫽ allocated “mate dollars” from a fixed budget to purchase more of a characteristic in a hypothetical mate; M ⫽ missing data; CPI ⫽ California Psychological Inventory.
a
Degraded to conceal identity of stimulus men.
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
Average
conception
probability
29
30
GILDERSLEEVE, HASELTON, AND FALES
or “dimorphⴱ” or “father” or “parentⴱ.” We also identified articles
from the reference lists of empirical articles and earlier reviews of
cycle shifts in women’s sexual motivations and mate preferences
(e.g., Gangestad & Thornhill, 2008; Gangestad, Thornhill, &
Garver-Apgar, 2005a; Jones et al., 2008). We discontinued our
literature search in December 2012.5
Inclusion Criteria
Studies have assessed ovulation-related cycle shifts in women’s
preferences for a variety of male characteristics using a variety of
measures and have reported results and effect sizes in a variety of
formats. We designed inclusion criteria that would retain a large
and diverse sample of effects, while also limiting the sample to
those effects that would facilitate a coherent evaluation of the
evidence for the ovulatory shift hypothesis. In the following, we
outline each of the specific inclusion criteria. For thoroughness,
Tables 1 and 2 present all studies (and effects within studies) that
met basic inclusion criteria (Criteria 1, 2, and 3), regardless of
whether they were ultimately included in the meta-analysis. If a
study assessed women’s preferences for a variety of different
characteristics, we included in the meta-analysis whichever effects
were relevant to testing the ovulatory shift hypothesis and excluded those that were not.
Criterion 1: Naturally cycling women. The effect must have
come from a study that included only naturally cycling women— by which we mean reproductive-aged women not using
hormonal contraception— or collected information about hormonal contraception use so that it was possible to examine naturally cycling women’s data separately.6
Criterion 2: Assessed ovulatory cycle position. The effect
must have come from a study that collected information that could
be used to estimate participants’ position in the ovulatory cycle
(e.g., date of last menstrual onset; see supplemental materials for
a more detailed description of cycle position estimation methods).
Criterion 3: Assessed women’s preference for a specific
characteristic in men. The effect must have assessed a cycle
shift in women’s preference for a specific characteristic in men.
For example, “facial masculinity” refers to a single, specific characteristic. In contrast, a man’s relationship status or feelings about
a current relationship partner could reflect a number of specific
characteristics, as well as circumstances unrelated to those characteristics. It is unclear which specific characteristics women infer
on the basis of a man’s relationship status. Therefore, we excluded
effects that assessed women’s preferences for men depicted as
single, in love, having a girlfriend, or married (Bressan & Stranieri, 2008). In addition, we excluded effects that assessed women’s
attraction to real men whose characteristics were unknown to the
researcher (e.g., a current relationship partner or celebrity; Gangestad, Thornhill, & Garver, 2002; Laeng & Falkenberg, 2007).
Physical attractiveness reflects a number of more specific characteristics and their interactions. It is unclear which specific characteristics women infer in men described as “physically attractive.” Furthermore, the ovulatory shift hypothesis posits that the
characteristics women find physically attractive vary systematically across the ovulatory cycle. For example, a physically attractive face could be a face high in masculinity for a woman at high
fertility within the cycle but average in masculinity for the same
woman at low fertility within the cycle. In other words, the
ovulatory shift hypothesis posits that women’s standards for what
is physically attractive themselves shift across the cycle, making
predictions about cycle shifts in women’s attraction to men described as physically attractive unclear. For these reasons, we
excluded effects that assessed women’s preferences for physical
attractiveness and handsomeness (e.g., Beaulieu, 2007; Caryl et
al., 2009; Gangestad, Garver-Apgar, Simpson, & Cousins, 2007;
Gangestad et al., 2010a).
Criterion 4: Assessed preferences pertinent to the ovulatory
shift hypothesis. The effect must have assessed a cycle shift in
women’s preference for a specific characteristic for which the ovulatory shift hypothesis makes a clear prediction—namely, a characteristic for which the extant literature provides a clear and widely
accepted rationale for why it is likely to have been reliably associated
with either genetic quality or long-term partner quality in ancestral
males. Along these lines, we excluded effects measuring women’s
preference for social status, social competence, social sensitivity, and
other social status-related characteristics (e.g., Izbicki & Johnson,
2010; Miller, 2003; Teatero, 2009); intelligence, inventiveness, creativity, academic achievement, and other intelligence-related characteristics (e.g., Caryl et al., 2009; Miller, 2003; Prokosch, Coss, Scheib,
& Blozis, 2009); and cues of good health (e.g., a healthy-looking
appearance; Jones, Perrett, et al., 2005). Extant findings suggest that
all of these characteristics were associated with both genetic quality
and partner quality in ancestral males, making predictions unclear
(see, e.g., Miller, 2000; Prokosch, Coss, Scheib, & Blozis, 2009; von
Rueden, Gurven, & Kaplan, 2011).
Furthermore, the leading hypothesis pertaining to cycle shifts in
women’s preferences for cues of good health predicts that women
will experience an elevated preference to affiliate with individuals
in general (not only mates) displaying cues of good health when
progesterone levels are highest within the cycle (e.g., in the luteal
phase—the portion of the cycle following ovulation and extending
to next menstrual onset; Jones, Perrett, et al., 2005). Progesterone
dampens immune function in preparation for possible pregnancy,
enabling the implantation of an embryo that is only partially
genetically related to the mother and could otherwise be attacked
by her immune system. Because of immune suppression associated
with high progesterone levels, women might prefer to avoid potentially contagious individuals and instead affiliate with healthy
individuals during the luteal phase. Fertility levels are also low
during the luteal phase. Therefore, women could experience stronger preferences for cues of good health at low than at high fertility.
However, these progesterone-related cycle shifts in women’s general social preferences would reflect different psychological mechanisms from those posited by the ovulatory shift hypothesis to
produce ovulation-related cycle shifts in women’s mate preferences. A meta-analysis evaluating evidence for progesteronerelated cycle shifts would test a different hypothesis and require a
5
Some researchers sent us unpublished data that have since been published (e.g., Thornhill, Chapman, & Gangestad, 2013). Thus, although the
references of some studies included in the meta-analysis indicate a later
date, we had in fact collected all data by December 2012.
6
Most studies in this meta-analysis also reported having excluded
women who were pregnant (or suspected pregnancy), breastfeeding, menopausal or postmenopausal, or who reported a highly irregular cycle or other
cycle abnormalities. However, we did not eliminate studies that did not
report having collected and excluded women on the basis of this information.
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
different analysis strategy from that of the present meta-analysis
(e.g., it would require comparing high-progesterone to lowprogesterone days of the cycle, rather than high-fertility to lowfertility days of the cycle; therefore, we did not include these
health effects in Tables 1 and 2).
Finally, because the ovulatory shift hypothesis only makes predictions about cycle shifts in women’s preferences for cues of
genetic quality and cues of long-term partner quality, we excluded
a number of effects measuring preferences for characteristics that
are not thought to have been associated with genetic quality or
long-term partner quality in ancestral males (e.g., a mature appearance, a threatening appearance, same-race versus other-race facial
appearance, adaptability, etc.; Izbicki & Johnson, 2010; McDonald
& Navarrete, 2012; Miller, 2003).
Criterion 5: Assessed preference for more over less of one
characteristic, rather than for one characteristic over another.
The effect must have assessed women’s preference for more of a
characteristic over less of that same characteristic (e.g., wealthy
men over poor men), rather than women’s preference for one
characteristic over another characteristic (e.g., wealthy men over
creative men). The latter confounds preference for one characteristic with preference for another, rendering effects from such
studies incomparable with the other effects in the meta-analysis
sample. For this reason, we excluded two studies: one that used a
forced-choice paradigm to examine women’s relative preference
for creativity versus wealth in a prospective partner (Haselton &
Miller, 2006) and another that used a “mate dollars” paradigm (Li,
Bailey, Kenrick, & Linsenmeier, 2002) to examine the extent to
which women traded off certain characteristics to “purchase” more
of other characteristics in a hypothetical prospective partner (e.g.,
intelligence, social status, fit body, compatible interests, etc.; Li,
Pillsworth, & Haselton, 2006).
Criterion 6: Common mate preference measure. The effect
must have been provided by a study that used a relatively common
measure of mate preferences. We excluded studies that used highly
uncommon measures of mate preferences in order to ensure that
there was sufficient conceptual overlap among the measures included in the meta-analysis to yield interpretable mean effect sizes.
Specifically, we excluded one study that measured women’s selfreported perceived romantic compatibility with stimulus men
(Flowe, Swords, & Rockey, 2012) and one study that measured
women’s self-reported likelihood of having sex with stimulus men
(Rupp et al., 2009).7 We would have also excluded one study that
used women’s pupil dilation as a measure of attraction to male
stimuli, but we had already excluded it on the basis of Criterion 3
(Laeng & Falkenberg, 2007).
Criterion 7: Provided information to compute appropriate
Hedges’s g. The article, poster, or study author must have provided the information needed to compute an appropriate Hedges’s
g effect size, as described below (see Computing Effect Sizes). If
an article or poster did not report the needed information, we
contacted study authors to request this information. If the information was unavailable, we excluded the effect from the metaanalysis. Of those effects that were otherwise eligible for inclusion, we were unable to obtain effect size information for 11
effects from three studies: face and body averageness (one effect),
face and body masculinity (one effect), and face and body symmetry (one effect; Peters, Rhodes, & Simmons, 2008); facial
masculinity (one effect), body masculinity (one effect), facial
31
symmetry (one effect), and body symmetry (one effect; Peters,
Simmons, & Rhodes, 2009); vocal cues associated with perceived
physical dominance (two effects) and vocal cues associated with
perceived social dominance (two effects; Puts, 2005).
Analyses Conducted on Broad Versus Narrow Sets of
Mate Preference Measures
Even after removing effects that assessed cycle shifts in women’s preferences for male characteristics for which the ovulatory
shift hypothesis does not make a clear prediction (Criterion 4) and
effects assessed with highly uncommon measures (Criterion 6), the
remaining sample of effects was still very heterogeneous. A benefit of including all of these effects in the meta-analysis is that
weighted mean effect sizes would reflect diverse male characteristics and measures. However, a cost is that weighted mean effect
sizes would reflect male characteristics for which predictions are
relatively weak (e.g., characteristics that are not yet widely accepted in this area as cues of genetic quality or long-term partner
quality) and measures that are likely to be relatively insensitive to
the fleeting, relationship context-dependent cycle shifts predicted
by the ovulatory shift hypothesis. To resolve these trade-offs, we
created two nested samples of effects and conducted separate
analyses on each. The first sample included a relatively “broad” set
of male characteristics and measures, whereas the second sample
included the relatively “narrow” subset of male characteristics and
measures that we reasoned would be provide the strongest test of
the ovulatory shift hypothesis.
The first, broad sample included effects examining cycle shifts
in women’s preferences for the following characteristics hypothesized to have served as cues of genetic quality in ancestral males:
facial symmetry, body symmetry, scents associated with body
symmetry, structural facial masculinity, male-typical facial movements, facial darkness, structural body masculinity (including, in
addition to general body masculinity, muscularity, height, maletypical shoulder-to-hip ratio, male-typical waist-to-hip ratio, and
strength), male-typical body motion (walking stride), torso hair,
vocal masculinity (lower vocal pitch), behavioral dominance (including, in addition to general dominance, social presence, social
respect and influence, direct intrasexual competitiveness, confrontativeness with other men, aggressiveness, arrogance and selfcenteredness, egotism, and conceitedness), scents associated with
behavioral dominance (specifically, scents associated with narcissism as assessed using the California Personality Inventory; see
Havlíček, Roberts, & Flegr, 2005), facial cues associated with
circulating testosterone, scents associated with circulating testosterone, and facial averageness. Although we might have excluded
“social respect and influence” from this analysis for the same
reason that we excluded social status (see Criterion 4), we chose to
include it because a factor analysis in that study showed that social
respect and influence had a very high loading on an Intrasexual
Competitiveness factor (in fact, it had the highest loading of all
characteristics rated in that study) and only a modest loading on a
Good Investing Mate Qualities factor (see Gangestad et al., 2007).
7
Reported likelihood of having sex is conceptually different from attraction because it also entails attitudes toward casual sex and constraints
on sexual behavior (e.g., having a current partner, risks associated with sex,
taboos against casual sex, etc.).
32
GILDERSLEEVE, HASELTON, AND FALES
Excluding this characteristic had a negligible impact on the mean
weighted effect sizes we report below and did not impact the
statistical significance of any effects. The broad sample also included effects examining cycle shifts in women’s preferences for
the following characteristics hypothesized to have served as cues
of long-term partner quality in ancestral males: relationship skills,
parenting skills, nurturance, sympathy, warmth, kindness, trustworthiness, faithfulness (including likelihood of being unfaithful,
reverse-coded), financial success, and career success.
In terms of measures, the broad sample included studies in
which women were asked to evaluate men or male stimuli as
prospective short- or long-term relationship partners; to evaluate
their attractiveness, physical attractiveness, sexual attractiveness,
or sexiness without reference to a specific relationship context; or
to evaluate a characteristic (e.g., “relationship skills”) on its importance or on how positive or negative they would feel about it in
a prospective partner.
The second, narrow sample included the same studies and
effects as the first sample, with three exceptions. First, we excluded effects measuring cycle shifts in women’s preferences for
characteristics that are not yet widely accepted as cues of ancestral
genetic quality: specifically, male-typical facial movements, maletypical walk, chest hair, skin darkness, and facial averageness
(Frost, 1994; Izbicki & Johnson, 2010; Koehler, Rhodes, & Simmons, 2006; Morrison, Clark, Gralewski, Campbell, & PentonVoak, 2010; Provost, Troje, & Quinsey, 2008; Rantala, Polkki, &
Rantala, 2010). Some researchers in this area have argued that the
fact that a characteristic is more typical of men than of women
suggests that that characteristic was linked to genetic quality in
ancestral males. However, others have argued against this claim,
noting an absence of strong theoretical or empirical reasons to
posit that certain sex-differentiated characteristics were linked to
male genetic quality ancestrally or that these characteristics play a
role in male–male competition or in men’s sexual attractiveness to
women. In addition, some researchers in this area have argued that
to the extent that averageness by definition indicates an absence of
atypical features that might result from genetic mutations, rare
alleles, homozygosity, or other potentially deleterious genetic factors, averageness might have served as a reliable indicator of
genetic quality in ancestral males (see, e.g., Thornhill & Gangestad, 1999a). However, recent evidence that extreme features are
more attractive than average features for many dimensions of
facial attractiveness poses a potential challenge to this view (Said
& Todorov, 2011).
Second, we excluded studies that used measures of stated preferences. These measures involve women explicitly reporting how
important or desirable a characteristic is in a prospective partner.
Excluding these measures limited the sample to studies that used
measures of revealed preferences. These measures involve women
rating the attractiveness of (or choosing the most attractive among)
male stimuli known by the researcher to vary on a characteristic.
This allows the researcher to infer women’s preferences on the
basis of their ratings (see supplemental materials for more detail).
We excluded studies using measures of stated preferences because
we reasoned that such measures might tend to elicit women’s
reports of their general preferences, rather than in-the-moment
preferences that might shift across the cycle. Thus, measures of
stated preferences might be relatively insensitive to the temporally
localized cycle shifts predicted by the ovulatory shift hypothesis.
Furthermore, given that several studies have found that stated
preferences are only weakly predictive of real-life dating behavior
(see, e.g., Eastwick & Finkel, 2008; Eastwick, Luchies, Finkel, &
Hunt, 2013; Todd, Penke, Fasolo, & Lenton, 2007), it remains an
open question whether women have explicit knowledge of and can
accurately report on the mate preferences that influence their
real-life attractions. Finally, we reasoned that measures of stated
preferences might not be as ecologically valid as measures of
revealed preferences. That is, responding to a questionnaire about
one’s mate preferences might be less likely than directly evaluating male stimuli to bring online the evolved psychological mechanisms that are hypothesized to produce cycle shifts.
Third, we excluded studies that used stimuli that did not enable
women to directly observe (see, hear, or smell) the characteristic of
interest. For example, in one study, women viewed facial photos
(no bodies) and rated the pictured men on attractiveness and
physical strength (Izbicki & Johnson, 2010). Information relevant
to judging men’s physical strength is present to some extent in
their facial appearance (Sell et al., 2009); therefore, the association
between these two sets of ratings likely provides at least a rough
measure of women’s preference for strength. Nonetheless, women’s ratings of body photos would likely have provided a more
precise measure of their strength preferences. As a more extreme
example, in another study, women read verbal descriptions of
hypothetical men that varied only in the quality of their sense of
humor and rated the men on attractiveness and body muscularity
(Miller, 2003). In this case, the verbal descriptions contained little
to no information relevant to judging body muscularity. To the
extent that women envisioned more or less muscular men when
rating the attractiveness of the hypothetical men, the association
between their attractiveness ratings and body muscularity ratings
could provide a rough measure of their preference for body muscularity. However, similar to strength ratings, body muscularity is
a characteristic of the body; therefore, collecting women’s ratings
of body photos would likely have provided a more precise measure
of their body muscularity preferences.
Computing Effect Sizes
The studies that we identified as potentially eligible for inclusion in this meta-analysis varied substantially in the type of data
they produced and in the format in which they reported results. We
used Hedges’s g effect size metric for this meta-analysis because
it could be computed for most of the studies in the sample, and its
interpretation intuitively maps onto the predictions of the ovulatory shift hypothesis. In this meta-analysis, g represents the standardized mean difference between high and low fertility in women’s preference for a characteristic (greater attraction to more
versus less of the characteristic). A larger (more positive) g indicates that women’s preference for a characteristic was stronger at
high fertility than at low fertility. For example, a g of 0.2 would
indicate that women’s preference was, on average, two tenths of a
standard deviation stronger at high fertility than at low fertility.
Hedges’s g is mathematically identical to Cohen’s d, except that it
includes an adjustment that reduces bias in small samples (Borenstein, Hedges, Higgins, & Rothstein, 2009). Hedges’s g also has
the same interpretation as Cohen’s d; in psychology, effect sizes of
0.2, 0.5, and 0.8 are typically considered small, moderate, and
large, respectively (Cohen, 1988).
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
Women’s “preference” for a male characteristic was operationalized as one of the following: the proportion of forced-choice
trials on which a woman chose stimuli with more of a characteristic over stimuli with less of that same characteristic, a woman’s
mean rating of the strength of her preference for stimuli with more
of a characteristic over stimuli with less of that characteristic (in
some studies, in each trial women completed a forced choice
between two options and then rated the strength of their preference
for the option they chose), the difference between a woman’s mean
rating of the attractiveness of stimuli with more of a characteristic
and her mean attractiveness rating of stimuli with less of that
characteristic, the correlation between a woman’s attractiveness
ratings of stimuli and the amount of a characteristic those stimuli
possessed, the amount of a characteristic a woman perceived as
most attractive (in some studies, women used a slider to manipulate a characteristic in a male stimulus until they had created what
they perceived to be the most attractive version of the stimulus), or
a woman’s rating (or mean rating, if multiple items were used to
assess a given preference) of the importance or desirability of a
characteristic in a prospective partner.
If a study treated fertility as dichotomous (comparing highfertility women to low-fertility women or the same women at high
versus low fertility), computing Hedges’s g to represent the difference between high and low fertility in women’s preference for
a male characteristic was straightforward. If a study treated fertility
as continuous (assigning each woman a conception probability
estimate based on her day in the cycle), computing Hedges’s g
entailed first computing the correlation between the continuous
fertility variable and preference for the male characteristic across
all women and then converting this correlation to g. We computed
a Hedges’s g for each preference assessed in each study; thus,
studies that assessed multiple preferences contributed multiple
effects (gs) to the meta-analysis.
Importantly, studies using measures of revealed preferences to
assess women’s mate preferences produce data that can be analyzed treating raters (women) or targets (men or male stimuli) as
units of analysis. In this meta-analysis, all Hedges’s gs were
computed based on analyses that treated women as units of analysis. Thus, we can expect any statistically significant effects to
generalize to new samples of women rating the stimuli that were
included in this meta-analysis (rather than generalizing to new sets
of male stimuli rated by the sample of women included in this
meta-analysis). For example, for studies in which women rated the
attractiveness of multiple male stimuli varying on a characteristic,
we first computed for each woman the correlation between her
attractiveness ratings of the male stimuli and the amount of the
characteristic those stimuli possessed, then computed the mean
correlation across all high-fertility women and the mean correlation across all low-fertility women, and finally computed a g
representing the standardized difference between those two means.
If available information could not be used to compute an effect
size based on raters (women) as units of analysis but could be used
to compute an effect size based on targets (men) as units of
analysis, we report the latter effect size in Table 1 for thoroughness; however, we excluded effects based on targets as units of
analysis from all analyses (Peters et al., 2009; Puts, 2005).
Several pieces of data identified as eligible for inclusion in this
meta-analysis had not yet been analyzed to examine cycle shifts. In
such cases, we asked the researcher to use the following guidelines
33
to analyze the data or, if the researcher preferred, we used these
guidelines to analyze their data. We developed these guidelines
with the intent of retaining a large number of observations while
providing a precise test of cycle shifts and giving researchers
options to accommodate the format of their data while minimizing
the potential for researchers to select among methods in order to
obtain significant results (Simmons, Nelson, & Simonsohn, 2011).
First, we asked the researcher to exclude women who reported
using hormonal contraception at the time of their participation.
Next, if the researcher had collected this information, we asked the
researcher to exclude women who, based on their self-reports, had
irregular ovulatory cycles (typically operationalized as varying
substantially in length from one cycle to the next), had used
hormonal contraception at any time in the past 3 months (Nassaralla et al., 2011), were currently experiencing symptoms of or had
experienced menopause, had an average cycle length shorter than
24 days or longer than 35 days (Harlow, 2000), suspected that they
might be pregnant, or were over the age of 35 (and were therefore
at an elevated likelihood of experiencing anovulatory cycles; Hale
et al., 2007). Last, if a study included subsamples of women tested
at both high and low fertility and women tested only at high or low
fertility, we asked the researcher to limit the sample to women who
had been tested at high and low fertility to enable withinparticipants comparisons.
If the researcher had already categorized women or observations
as high- and low-fertility based on predetermined window definitions or had assigned each woman a conception probability estimate, we asked the researcher to retain their operationalization of
fertility in effect size computations.8 If the researcher had not yet
defined high- and low-fertility windows or assigned conception
probability estimates, we recommended that the researcher do so
as follows: For studies using a between-participants design in
which each woman completed a session at a single point in her
cycle, we asked the researcher to assign each woman a conception
probability estimate (Wilcox, Dunson, Weinberg, Trussell, & Day
Baird, 2001) and to treat fertility as a continuous variable in effect
size computations. If this was not possible, we asked researchers
instead to categorize women who participated on forward cycle
days 9 –15 as high fertility and women who participated on forward cycle days 21–35 as low fertility and to exclude women
falling outside of these windows. Likewise, for studies using a
within-participants design in which each woman completed at least
one session at high fertility and at least one session at low fertility,
we asked researchers to categorize observations on forward cycle
days 9 –15 as high fertility and observations on forward cycle days
21–35 as low fertility and to exclude observations falling outside
of these windows. We chose these particular high- and low-fertility
window definitions in order to maximize and minimize, respectively, the associated average conception probabilities (Wilcox et
al., 2001), while still retaining a large number of observations in
the analysis.
8
One study (Harris, 2011) reported multiple sets of results based on
different high-fertility windows. For that study, we computed an effect size
using the results based on the high-fertility window with the highest
estimated average conception probability according to the values reported
by Wilcox et al. (2001).
34
GILDERSLEEVE, HASELTON, AND FALES
Coding Study Characteristics
Studies that have aimed to examine ovulation-related cycle
shifts in women’s mate preferences have varied in a number of
ways—including, for example, characteristics of the sample of
participants, researcher control over the research setting, methods
for assessing women’s fertility and mate preferences, and the
specific characteristics for which preferences were assessed. Some
of these methods have permitted greater researcher control and
internal validity but limited sample size and external validity,
whereas others have limited researcher control and internal validity but permitted a larger sample size and greater external validity.
See supplemental materials for a detailed discussion of the many
sources of variation in this literature.
As shown in Table 2, we coded each study for a variety of
characteristics. This included (a) relationship context (short-term,
long-term, or unspecified), (b) country from which the sample of
participants was drawn, (c) sample type (college/university
women, community women, or both), (d) study setting (lab vs.
“field,” which included online studies and one magazine survey
with a mail-in response), (e) study design (within participants vs.
between participants), (f) estimated average conception probability
associated with the high- and low-fertility scheduling windows, (g)
cycle position estimation method (forward counting method vs.
reverse counting method vs. average from forward and reverse
counting methods vs. luteinizing hormone tests to verify impending ovulation vs. salivary ferning method to verify impending
ovulation, noting for studies that used counting methods whether
the benchmark date of menstrual onset was verified), (h) type of
stimuli (e.g., self-reported preferences vs. facial photos vs. body
photos vs. average across face and body photos vs. vocal recordings vs. videotaped behavior vs. scent samples vs. face avatars vs.
full-body avatars vs. moving facial outlines vs. body outline drawings vs. verbal descriptions of hypothetical men vs. point-light
walkers), (i) method of determining the amount of the characteristic of interest possessed by the male stimuli (direct manipulations
by the researcher vs. measured or coded by the researcher vs. rated
by the participants in the cycle shift study vs. rated by a separate
sample of participants), (j) type of preference measure (stated
preference vs. revealed preference), (k) rating task (ratings of
individual stimuli vs. two-option forced choice vs. multiple-option
[three or more] forced choice vs. used a slider to manipulate the
characteristic of interest), (l) number of trials, and (m) study
publication status. Two researchers independently coded each
study and then cross-checked their codes. In the case of discrepancies (which were rare), the researchers referred back to the
article or contacted the authors to verify the correct code. Thus, all
codes were verified as correct.
Coding study characteristics was generally straightforward. As
an exception, coding relationship context required additional considerations. In many studies examining cycle shifts, women were
asked to complete two sets of ratings— one in which they evaluated men or male stimuli as potential “short-term” partners (typically defined as someone with whom they would consider having
a brief sexual affair) and another in which they evaluated men or
male stimuli as potential “long-term” partners (typically defined as
someone with whom they would consider having a long-term
dating or marriage/marriage-like relationship; e.g., Little, Jones, &
Burriss, 2007)— or less commonly, just one or the other. When
studies explicitly specified a short-term and/or long-term relationship context, we coded them as such. Notably, however, in many
studies, women were asked to evaluate men or male stimuli on
“attractiveness” (e.g., Rupp, Librach, et al., 2009), “physical attractiveness” (e.g., Roney & Simmons, 2008), “sexual attractiveness” (e.g., Rantala et al., 2010), or “sexiness” (e.g., Thornhill &
Gangestad, 1999b), or less commonly, to evaluate the importance
or desirability of a characteristic in a potential partner (e.g., Caryl
et al., 2009) without reference to a specific relationship context.
When studies did not explicitly specify a short- or long-term
relationship context, we coded them as “unspecified.”
Analyses
We used multilevel modeling for all analyses. Meta-analysis can
be viewed as a special case of a multilevel model involving effects
nested within studies (Raudenbush & Bryk, 1985). The multilevel
modeling approach offers a range of benefits over traditional
meta-analytic methods, including the ability to properly include
multiple, nonindependent effects from the same sample within a
single analysis and to test effect-level and study-level predictors of
effect size and their cross-level interactions. As is conventional in
meta-analysis, we weighted each effect by its inverse variance in
order to give more precisely measured effects— often, those from
larger studies—more “pull” on weighted mean effect sizes and
regression coefficients (Raudenbush & Bryk, 2002). We estimated
fixed effects and variance components using restricted maximumlikelihood estimation procedures, which tend to reduce downward
bias in variance components compared with full maximumlikelihood estimation procedures (O’Connell & McCoach, 2008).
We conducted all analyses in HLM 7.0 and used the weighting and
known-variance options to weight effects by their inverse variances.
As indicated in the “Inclusion in analyses” section of Table 1,
we conducted two sets of analyses: one to examine cycle shifts in
women’s preferences for hypothesized cues of genetic quality in
ancestral males and another to examine cycle shifts in women’s
preferences for hypothesized cues of long-term partner quality in
ancestral males. Within each set of analyses, we first conducted
analyses on the broad sample of effects and then conducted analyses on the narrow subset of effects described above (see Analyses
Conducted on Broad Versus Narrow Sets of Mate Preference
Measures).
As described below, focal analyses examining cycle shifts in
women’s preferences for all hypothesized cues of ancestral genetic
quality revealed robust cycle shifts. These analyses included large
but heterogeneous samples of effects. Therefore, they were sufficiently powered to provide clear results regarding the robustness of
cycle shifts across all hypothesized cues of genetic quality but
could not provide insight into how the magnitude and robustness
of these cycle shifts differed across different kinds of studies (e.g.,
using different methods) or across different specific male characteristics (e.g., facial vs. body masculinity). To address this issue,
we conducted two additional sets of analyses. First, in both the
broad and narrow samples of effects, we ran a series of moderation
analyses. These analyses examined associations between specific
study characteristics and the magnitude of cycle shifts across all
hypothesized cues of genetic quality. Second, in the narrow sample
of effects, we examined cycle shifts separately for each specific
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
hypothesized cue of genetic quality for which the sample contained at least three effects. These analyses included small but
relatively homogeneous samples of effects. Consequently, they
were often underpowered and sometimes contained effects in only
one or two relationship contexts. Nonetheless, their results provide
insight into the specific male characteristics for which cycle shifts
in women’s preferences are or are not robust and highlight areas
still in need of more research.
In the following, we describe the models used in these analyses
in more detail. Results from the key analyses are presented in
Figure 1.
Step 1. In each sample of effects, we first specified an unconditional random-effects model to compute the weighted mean g as
an estimate of the “true” (population) mean standardized mean
difference between high and low fertility in women’s preference
for a characteristic across all relationship contexts:
35
effects, and u0j is a study-level random error. Specifying ␦0j as
random entails conceiving of g as varying randomly over the
population of studies, thus allowing g to vary both as a function of
sampling error and as a function of true between-studies variance
(whereas specifying ␦0j as fixed would allow g to vary as a
function of sampling error alone; see Raudenbush & Bryk, 2002).
This approach is appropriate given that the studies included in this
set of analyses are diverse in terms of sample characteristics,
methods, and measures.
Step 2. In each sample of effects, we next specified three
models to compute the weighted mean g in a short-term context
(where the cycle shift is predicted to be largest), unspecified
context (where the cycle shift is predicted to be intermediate
between a short-term and long-term context), and long-term context (where the cycle shift is predicted to be smallest or absent),
respectively, and to compare the weighted mean g across these
three contexts. We created several variables to represent relationship context. These included short, a dummy-coded dichotomous
variable taking on a value of 1 for effects measured in a short-term
context and 0 for effects measured in a long-term or unspecified
context; long, a dummy-coded dichotomous variable taking on a
value of 1 for effects measured in a long-term context and 0 for
effects measured in a short-term or unspecified context; and unspecified, a dummy-coded dichotomous variable taking on a value
Level 1 model 共effects兲: gij ⫽ ␦0j ⫹ eij .
Level 2 model 共studies兲: ␦0j ⫽ ␥00 ⫹ u0j .
In the model above, gij is the observed standardized mean difference i for study j, ␦0j is the “true” mean g in the population of
effects, eij is the sampling error associated with gij as an estimate
of ␦0j, ␥00 is the observed weighted mean g in the sample of
1.1
1
0.9
0.8
0.7
Overall
Short-term
Unspecified
Long-term
Hedges’s g
0.6
0.5
*
*
0.4
***
0.3
0.2
***
***
**
**
†
***
***
***
†
*
*
*
†
**
0.1
0
-0.1
†
-0.2
Genetic Quality
(Broad)
†
Genetic Quality
(Narrow)
Facial Symmetry
Scent Cues of
Symmetry
Facial Masculinity Body Masculinity Vocal Masculinity
Behavioral
Dominance
Facial Cues of
Testosterone
†
LT Partner Quality LT Partner Quality
(Broad)
(Narrow)
p ≤ .10. * p ≤ .05. ** p ≤ .01. *** p ≤ .001.
Figure 1. Summary of results from all analyses examining cycle shifts in women’s preferences for hypothesized cues of genetic and long-term partner quality in ancestral males. For each sample, the weighted mean
Hedges’s g is presented overall (across relationship contexts) and separately for short-term, unspecified, and
long-term relationship contexts. Errors bars represent standard error. LT ⫽ long-term.
GILDERSLEEVE, HASELTON, AND FALES
36
of 1 for effects measured in an unspecified context and 0 for
effects measured in a short-term or long-term context. Starting
with the unconditional model described in Step 1, we added
dummy-coded relationship context variables, two at a time, as
effect-level predictors. For example, in the following model, we
have added long and unspecified. This establishes a short-term
context as the comparison group, thereby enabling us to compute
the weighted mean g in a short-term context and to estimate the
magnitude of the difference between the weighted mean g in a
short-term versus long-term context and between the weighted
mean g in a short-term versus unspecified context. Although we
report the results from all three models (with each of the three
contexts as a comparison group) in the text of the Results section,
for brevity, we present the complete results from only the models
in which a short-term context was the comparison group in Tables
3–13.
Level 1 model 共effects兲: gij ⫽ ␦0j ⫹ ␦1j(long)
⫹ ␦2j(unspecified) ⫹ eij .
Level 2 model 共studies兲: ␦0j ⫽ ␥00 ⫹ u0j
␦1j ⫽ ␥10
␦2j ⫽ ␥20.
In the model above, gij is the observed standardized mean difference i for study j, ␦0j is the “true” mean g in a short-term
relationship context, ␦1j is the “true” difference between g in a
long-term versus short-term context, ␦2j is the “true” difference
between g in an unspecified versus short-term context, eij is the
residual sampling error associated with gij as an estimate of ␦0j
unexplained by relationship context, ␥00 is the observed weighted
mean g in a short-term relationship context, u0j is a study-level
random error, ␥10 is the regression coefficient representing the
expected difference between g in a long-term versus short-term
context (a negative value indicates that g is larger in a short-term
context than in a long-term context), and ␥20 is the regression
coefficient representing the expected difference between g in an
unspecified versus short-term context (a negative value indicates
that g is larger in a short-term context than in an unspecified
context).
We specified ␦1j and ␦2j as fixed in the above model. This
assumes that any effect of relationship context on g varies across
studies as a function of sampling error alone (and not as a function
of true between-studies variance). Studies differed in how they
defined short-term and long-term relationships and, if no relationship context was specified, in whether they asked women to
evaluate male stimuli on physical attractiveness, attractiveness,
sexual attractiveness, sexiness, or another variable. Thus, any
effect of relationship context on g could vary as a function of true
between-studies variance in addition to sampling error. For many
of the analyses, we were working with relatively small samples of
effects and therefore had insufficient power to specify relationship
context effects as random. However, when possible, we tested
these effects as both fixed and random and found that this did not
change the pattern of results. For consistency, in the text and
tables, we report results based on models in which relationship
context effects were fixed.
Step 3. In the genetic quality analyses only, we then ran
numerous analyses to test whether specific study characteristics
were associated with between-studies variance in effect size (g)
after controlling for relationship context and, if there was sufficient
power, to test whether specific study characteristics were associated with between-studies variance in the effect of relationship
context (short-term vs. unspecified and short-term vs. long-term)
on effect size. When possible, we ran moderation analyses for each
of the study characteristics displayed in Table 2, with the exception
of sample country. This included study publication status, sample
composition, setting, design, estimated difference in the average
conception probability of the high- versus low-fertility windows,
counting method used to estimate ovulatory cycle position,
whether the benchmark date of menstrual onset had been verified,
type of stimuli, method used to determine the amount of a characteristic male stimuli possessed, whether the study used a stated
or revealed preference measure to assess mate preferences, type of
rating task, and number of trials.
Notably, moderation analyses were limited in several ways.
Because power was often low, we tested one study characteristic at
a time. Thus, if analyses revealed an association between a study
characteristic and effect size, other correlated study features could
account for this association. Indeed, many study characteristics
were highly intercorrelated. For example, nearly all studies using
a between-participants design also used the forward counting
method to estimate women’s position in the ovulatory cycle. In
addition, very few studies used rigorous methods to determine
women’s position in the ovulatory cycle (e.g., few studies verified
ovulation by using luteinizing hormone tests). Thus, analyses
examining associations between the use of these methods and
effect size were underpowered. As in all research literatures, many
factors influence the extent to which studies provide precise measures of effects. Even if this meta-analysis cannot examine all of
the many sources of variation in cycle shifts, it can still examine
key sources of variation and determine whether robust patterns of
cycle shifts emerge despite this variation.
Results
As explained in detail above, the ovulatory shift hypothesis
posits that women experience a relationship context-dependent
cycle shift in their attraction to characteristics that reliably indicated genetic quality in ancestral males. Specifically, the ovulatory
shift hypothesis predicts that women’s attraction to these characteristics is stronger at high fertility than at low fertility and that this
shift will be most pronounced when women evaluate prospective
partners in a short-term relationship context and least pronounced
when they evaluate prospective partners in a long-term relationship context. Most studies categorized as “unspecified” in this
meta-analysis asked women to evaluate men or male stimuli on
attractiveness. As noted above, previous research has shown that
women value physical attractiveness more in short-term sex partners than in long-term relationship partners (e.g., Li & Kenrick,
2006; Regan, 1998); therefore, we further predict that women will
exhibit a pattern of cycle shifts in an unspecified relationship
context that more closely resembles the pattern of cycle shifts in a
short-term context than in a long-term context. Although an overall
cycle shift in women’s preferences for cues of genetic quality
could emerge across the three relationship contexts, this is not a
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
requirement of the ovulatory shift hypothesis. Rather, the more
precise prediction is that any such cycle shift will be strongly
moderated by relationship context.
Last, the ovulatory shift hypothesis posits that, regardless of
relationship context, women do not experience a cycle shift in
their preferences for characteristics that reliably indicated suitability as a long-term social partner and coparent in ancestral
males.
37
Step 3 revealed several moderation effects. All of the following
study characteristics were associated with a larger cycle shift after
controlling for the effect of relationship context: using scent stimuli, rather than any other type of stimuli (p ⫽ .01); direct measurement, rather than any other method to determine the amount of
a characteristic possessed by male stimuli (p ⫽ .06); and the study
being published (p ⫽ .03). In contrast, the following study characteristic was associated with a smaller cycle shift after controlling
for the effect of relationship context: having participants in the
cycle study rate a characteristic in male stimuli, rather than using
any other method to determine the amount of a characteristic
possessed by male stimuli (p ⫽ .06).
All of the following were associated with a larger difference
between the magnitude of the cycle shift in a short-term and
long-term relationship context (short term ⬎ long term): a field
(usually online) setting, rather than a lab setting (p ⫽ .02); a
forward counting method, rather than a backward counting method
or an average of forward and backward counting methods, to
estimate women’s cycle position (p ⫽ .01); and the study being
published (p ⫽ .003). In contrast, the following was associated
with a smaller difference between the magnitude of the cycle shift
in a short-term and long-term relationship context: using facial
photos as stimuli, rather than any other kind of stimuli (p ⫽ .03).
All of the following were associated with a larger difference
between the magnitude of the cycle shift in a short-term and
unspecified relationship context (short term ⬎ unspecified): using
body photos as stimuli, rather than any other type of stimuli (p ⫽
.001); directly manipulating the male characteristic, rather than
using any other method to determine the amount of the male
characteristic possessed by the stimuli (p ⬍ .001); using a twooption forced-choice, rather than any other task to assess mate
preferences (p ⫽ .02). In contrast, the following was associated
with a smaller difference between the magnitude of the cycle shift
in a short-term and unspecified relationship context: using a rating
task, rather than any other task to assess mate preferences (p ⫽
.04).
Preference for All Hypothesized Cues of Ancestral
Genetic Quality: Broad Set of Measures
The first analysis examined cycle shifts in preferences for all
cues of genetic quality in the sample of effects that included a
broad set of mate preference measures. This analysis included 96
effects from 50 studies (total N ⫽ 5,471). As shown in Table 3,
Step 1 revealed that the weighted mean g estimating the true
population mean standardized mean difference between high and
low fertility in women’s preference for hypothesized cues of
ancestral genetic quality across short-term, long-term, and unspecified relationship contexts was small (g ⫽ 0.15, SE ⫽ 0.04) but
statistically significant (p ⬍ .001). Thus, in this set of effects,
women’s preference for these characteristics was approximately
0.15 of a standard deviation stronger at high fertility than at low
fertility.
Step 2 revealed that the weighted mean g in a short-term context
was small (g ⫽ 0.21, SE ⫽ 0.06) but statistically significant (p ⫽
.001); the weighted mean g in an unspecified relationship context
was small (g ⫽ 0.16, SE ⫽ 0.05) but statistically significant (p ⫽
.003); and the weighted mean g in a long-term context was near
zero (g ⫽ 0.06, SE ⫽ 0.06) and not statistically significant (p ⫽
.32). Comparing the three contexts revealed that the weighted
mean g was larger in a short-term context than in a long-term
context, and this difference was statistically significant (p ⫽ .002).
The weighted mean g did not significantly differ between a shortterm context and an unspecified context or between an unspecified
context and a long-term context (p ⫽ .54 and .19, respectively).
Table 3
All Hypothesized Cues of Ancestral Genetic Quality: Broad Set of Measures
Effect
Coefficient
SE
Variance
component
SD
t ratio
df
4.13
49
2
p
Step 1
Fixed
Overall weighted mean effect size, ␥00
Random
True mean effect size, ␦0j
0.15
0.04
0.03
0.18
49
⬍.001
141.32
⬍.001
Step 2
Fixed
Weighted mean effect size in a short-term context, ␥00
Difference between a long-term and short-term context, ␥10
Difference between an unspecified and short-term context, ␥20
Random
True mean effect size in a short-term context, ␦0j
0.21
⫺0.15
⫺0.05
0.06
0.04
0.08
3.54
⫺3.28
⫺0.62
0.03
0.18
49
93
93
49
.001
.002
.54
138.33
⬍.001
Note. Step 1: Results from unconditional multilevel model estimating the true mean standardized mean difference between high and low fertility in
women’s preference for all hypothesized cues of ancestral genetic quality in the sample of effects selected using relatively relaxed inclusion criteria. Step
2: Results from multilevel model estimating the true mean standardized mean difference between high and low fertility in women’s preference for all
hypothesized cues of ancestral genetic quality (relaxed inclusion criteria) in a short-term relationship context (compared to a long-term or unspecified
relationship context).
GILDERSLEEVE, HASELTON, AND FALES
38
design, rather than a within-participants design (p ⫽ .08); using
scent stimuli, rather than any other type of stimuli (p ⫽ .02); direct
measurement, rather than any other method to determine the
amount of a characteristic possessed by male stimuli (p ⫽ .09); and
the study being published (p ⫽ .03). In contrast, the following
study characteristic was associated with a significantly smaller
effect after controlling for relationship context: having participants
in the cycle study rate a characteristic in male stimuli, rather than
using any other method to determine the amount of a characteristic
possessed by male stimuli (p ⫽ .02). Lastly, the following characteristic was associated with a larger difference between the
effect size in a short-term and long-term relationship context: a
field (usually online) setting, rather than a lab setting (p ⫽ .09).
Preference for All Hypothesized Cues of Ancestral
Genetic Quality: Narrow Set of Measures
The next analysis examined cycle shifts in preferences for all
cues of genetic quality in the sample of effects that included a
narrow set of mate preference measures. This analysis included
68 effects from 42 studies (total N ⫽ 4,884). As shown in Table
4, Step 1 revealed that the weighted mean g estimating the true
mean standardized mean difference between high and low fertility in women’s preference for hypothesized cues of ancestral
genetic quality across short-term, long-term, and unspecified
relationship contexts was small (g ⫽ 0.17, SE ⫽ 0.04) but
statistically significant (p ⬍ .001). Thus, in this set of effects,
women’s preference for these characteristics was generally
approximately 0.17 of a standard deviation stronger at high
fertility than at low fertility.
Step 2 revealed that the weighted mean g in a short-term context
was small to moderate (g ⫽ 0.26, SE ⫽ 0.07) and statistically
significant (p ⬍ .001); the weighted mean g for attractiveness
ratings made in an unspecified relationship context was small
(g ⫽ 0.20, SE ⫽ 0.05) but statistically significant (p ⫽ .001);
and the weighted mean g in a long-term context was near zero
(g ⫽ 0.02, SE ⫽ 0.06) and not statistically significant (p ⫽ .75).
Comparing the three contexts revealed that the weighted mean
g was larger in a short-term context than in a long-term context,
and this difference was statistically significant (p ⬍ .001). The
weighted mean g did not significantly differ between a shortterm context and an unspecified context (p ⫽ .42). The
weighted mean g was significantly larger in an unspecified
context than in a long-term context (p ⫽ .04).
Step 3 revealed several moderation effects. All of the following
study characteristics were associated with a significantly or marginally significantly larger effect after controlling for relationship
context: a sample composed of women from the community or a
combination of undergraduate and community women, rather than
only undergraduate women (p ⫽ .08); a field (usually online)
setting, rather than a lab setting (p ⫽ .08); a between-participants
Preference for Facial Symmetry
The next few analyses examined cycle shifts in women’s preferences for specific hypothesized cues of genetic quality in the
sample of effects that included a narrow set of mate preference
measures. The first of these analyses examined cycle shifts in
women’s preference for facial symmetry and included eight effects
from seven studies (total N ⫽ 870). As shown in Table 5, Step 1
revealed that the weighted mean g estimating the true mean standardized mean difference between high and low fertility in women’s preference for symmetry across short-term, long-term, and
unspecified relationship contexts was near zero (g ⫽ 0.07, SE ⫽
0.10) and not statistically significant (p ⫽ .48). Thus, in this set of
effects, women’s preference for symmetry was not generally stronger at high fertility than at low fertility.
Step 2 revealed that the weighted mean g in a short-term context
was small to moderate (g ⫽ 0.30, SE ⫽ 0.20) and not statistically
significant (p ⫽ .19); the weighted mean g in an unspecified
context was near zero (g ⫽ ⫺0.02, SE ⫽ 0.16) and not statistically
significant (p ⫽ .90); and the weighted mean g in a long-term
context was small and negative (g ⫽ ⫺0.16, SE ⫽ 0.25) and not
statistically significant (p ⫽ .54). Comparing the three contexts
revealed that the weighted mean g was larger in a short-term
Table 4
All Hypothesized Cues of Ancestral Genetic Quality: Narrow Set of Measures
Effect
Coefficient
SE
Variance
component
SD
t ratio
df
4.33
41
2
p
Step 1
Fixed
Overall weighted mean effect size, ␥00
Random
True mean effect size, ␦0j
0.17
0.04
0.03
0.18
42
⬍.001
111.48
⬍.001
Step 2
Fixed
Weighted mean effect size in a short-term context, ␥00
Difference between a long-term and short-term context, ␥10
Difference between an unspecified and short-term context, ␥20
Random
True mean effect size in a short-term context, ␦0j
0.26
⫺0.24
⫺0.07
0.07
0.05
0.08
4.07
⫺4.52
⫺0.82
0.03
0.18
41
65
65
41
⬍.001
⬍.001
.42
108.23
⬍.001
Note. Step 1: Results from unconditional multilevel model estimating the true mean standardized mean difference between high and low fertility
in women’s preference for all hypothesized cues of ancestral genetic quality in the sample of effects selected using relatively relaxed inclusion
criteria. Step 2: Results from multilevel model estimating the true mean standardized mean difference between high and low fertility in women’s
preference for all hypothesized cues of ancestral genetic quality (relaxed inclusion criteria) in a short-term relationship context (compared to a
long-term or unspecified relationship context).
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
39
Table 5
Facial Symmetry
Effect
Coefficient
SE
Variance
component
SD
t ratio
df
0.75
6
2
p
Step 1
Fixed
Overall weighted mean effect size, ␥00
Random
True mean effect size, ␦0j
0.07
0.1
0.03
0.16
6
.48
10.55
.1
Step 2
Fixed
Weighted mean effect size in a short-term context, ␥00
Difference between a long-term and short-term context, ␥10
Difference between an unspecified and short-term context, ␥20
Random
True mean effect size in a short-term context, ␦0j
0.3
⫺0.46
⫺0.32
0.2
0.21
0.26
1.47
⫺2.2
⫺1.24
0.06
0.24
6
5
5
6
.19
.08
.27
13.41
.04
Note. Step 1: Results from unconditional multilevel model estimating the true mean standardized mean difference between high and low fertility in
women’s preference for facial symmetry in the sample of effects selected using relatively strict inclusion criteria. Step 2: Results from multilevel model
estimating the true mean standardized mean difference between high and low fertility in women’s preference for facial symmetry (strict inclusion criteria)
in a short-term relationship context (compared to a long-term or unspecified relationship context).
context than in a long-term context, and this difference was marginally statistically significant (p ⫽ .08). The weighted mean g
was somewhat larger in a short-term context than in an unspecified
context, but this difference was not statistically significant (p ⫽
.27). Likewise, the weighted mean g was somewhat less negative
in an unspecified context than in a long-term context, but this
difference was not statistically significant (p ⫽ .66).
Preference for Scents Associated With Face and
Body Symmetry
The next analysis examined cycle shifts in women’s preference
for scents associated with face and body symmetry and included a
small sample of three effects from three studies (total N ⫽ 141). As
shown in Table 6, Step 1 revealed that the weighted mean g
estimating the true mean standardized mean difference between
high and low fertility in women’s preference for scent cues of
symmetry was large (g ⫽ 0.83, SE ⫽ 0.20) but not statistically
significant (p ⫽ .14). We could not perform Step 2 because all of
the effects in this sample were measured in an unspecified relationship context. Thus, in this set of effects, women’s preference
for scents associated with symmetry was approximately 0.83 of a
standard deviation stronger at high fertility than at low fertility, but
more data are needed to confidently determine the robustness of
this cycle shift and to examine differences across relationship
contexts.
Preference for Structural Facial Masculinity
The next analysis examined cycle shifts in women’s preference
for structural facial masculinity and included 23 effects from 19
studies (total N ⫽ 3,335). As shown in Table 7, Step 1 revealed
that the weighted mean g estimating the true mean standardized
mean difference between high and low fertility in women’s preference for structural facial masculinity across short-term, longterm, and unspecified relationship contexts was small (g ⫽ 0.13,
SE ⫽ 0.06) but statistically significant (p ⫽ .05). Thus, in this set
of effects, women’s preference for structural facial masculinity
was generally approximately 0.13 of a standard deviation stronger
at high fertility than at low fertility.
Step 2 revealed that the weighted mean g in a short-term context
was near zero (g ⫽ ⫺0.02, SE ⫽ 0.14) and not statistically
significant (p ⫽ .91); the weighted mean g in an unspecified
context was small (g ⫽ 0.17, SE ⫽ 0.07) but statistically significant (p ⫽ .02); and the weighted mean g in a long-term context
was near zero (g ⫽ ⫺0.01, SE ⫽ 0.13) and not statistically
significant (p ⫽ .95).9 Comparing the three contexts revealed that
the weighed mean g was somewhat larger in an unspecified context than in a short-term or long-term context, but these differences
were not significant (p ⫽ .24 and .23, respectively). The weighted
mean g did not differ between a short-term and long-term context
(p ⫽ .96).
Preference for Structural Body Masculinity
The next analysis examined women’s preference for structural
body masculinity and included 12 effects from five studies (total
N ⫽ 589). As shown in Table 8, Step 1 revealed that the weighted
mean g estimating the true mean standardized mean difference
between high and low fertility in women’s preference for structural
body masculinity across short-term and long-term relationship
contexts was small (g ⫽ 0.21, SE ⫽ 0.08) and marginally statis9
Luevano and Zebrowitz (2006) and Izbicki and Johnson (2010) both
presented participants with facial photographs and asked them to rate the
pictured men for “masculinity,” as well as certain personality characteristics (e.g., dominance, warmth, maturity, etc.). Because participants were
asked to evaluate the pictured men for personality characteristics, it is
possible that participants evaluated the men on inferred personality masculinity rather than on structural facial masculinity. Excluding the four
effects (two measured in a short-term context, two measured in a long-term
context) from these two studies changed the results as follows: overall
weighted mean g ⫽ 0.18 (SE ⫽ 0.05, p ⬍ .01), short-term weighted mean
g ⫽ 0.28 (SE ⫽ 0.20, p ⫽ .19), unspecified weighted mean g ⫽ 0.18 (SE ⫽
0.06, p ⫽ .01), long-term weighted mean g ⫽ 0.17 (SE ⫽ 0.19, p ⫽ .38),
and there were no statistically significant differences between relationship
contexts.
GILDERSLEEVE, HASELTON, AND FALES
40
Table 6
Scent Cues of Face and Body Symmetry
Effect
Coefficient
Variance
component
SE
SD
t ratio
df
4.15
2
2
p
Step 1
Fixed
Overall weighted mean effect size, ␥00
Random
True mean effect size, ␦0j
0.83
0.20
0.0003
0.02
.14
2
1.82
⬎.50
Note. Step 1: Results from unconditional multilevel model estimating the true mean standardized mean difference between high and low fertility in
women’s preference for scent cues of symmetry in the sample of effects selected using relatively strict inclusion criteria.
tically significant (p ⫽ .07). Thus, in this set of effects, women’s
preference for structural body masculinity was generally 0.21 of a
standard deviation stronger at high fertility than at low fertility, but
more data are needed to determine the robustness of this cycle
shift.
Step 2 revealed that the weighted mean g in a short-term context
was small to moderate (g ⫽ 0.35, SE ⫽ 0.10) and statistically
significant (p ⫽ .04); and the weighted mean g in a long-term
context was near zero (g ⫽ 0.09, SE ⫽ 0.09) and not statistically
significant (p ⫽ .40) The weighted mean g was larger in a
short-term context than in a long-term context, and this difference
was statistically significant (p ⫽ .03). None of the effects in this
sample were measured in an unspecified context.
preference for vocal masculinity appeared to be 0.28 of a standard
deviation stronger at high fertility than at low fertility, but more
data are needed to determine the robustness of this cycle shift.
Step 2 revealed that the weighted mean g in a short-term context
was small to moderate (g ⫽ 0.40, SE ⫽ 0.20), and the weighted
mean g in a long-term context was small (g ⫽ 0.18, SE ⫽ 0.20).
Power was insufficient to test the statistical significance of either
effect. The weighted mean g was somewhat larger in a short-term
context than in a long-term context, but this difference was not
statistically significant (p ⫽ .39). None of the effects in this sample
were measured in an unspecified context.
Preference for Behavioral Dominance or Felt
Superiority Over Other Men
Preference for Vocal Masculinity (Lower Vocal Pitch)
The next analysis examined cycle shifts in women’s preference
for behavioral dominance and included 12 effects from three
studies (total N ⫽ 255). As shown in Table 10, Step 1 revealed that
the weighted mean g estimating the true mean standardized mean
difference between high and low fertility in women’s preference
for behavioral dominance across short-term and long-term relationship contexts was near zero (g ⫽ 0.04, SE ⫽ 0.06) and not
statistically significant (p ⫽ .55). Thus, in this set of effects,
women’s preference for behavioral dominance was not generally
stronger at high fertility than at low fertility.
The next analysis examined cycle shifts in women’s preference
for vocal masculinity and included a small sample of four effects
from two studies (total N ⫽ 159). As shown in Table 9, Step 1
revealed that the weighted mean g estimating the true mean standardized mean difference between high and low fertility in women’s preference for vocal masculinity (lower vocal pitch) across
short-term and long-term relationship contexts was small (g ⫽
0.28, SE ⫽ 0.18), but power was insufficient to test the statistical
significance of this effect. Thus, in this set of effects, women’s
Table 7
Structural Facial Masculinity
Effect
Coefficient
SE
Variance
component
SD
t ratio
df
2.09
18
2
p
Step 1
Fixed
Overall weighted mean effect size, ␥00
Random
True mean effect size, ␦0j
0.13
0.06
0.04
0.2
18
.05
51.25
⬍.001
Step 2
Fixed
Weighted mean effect size in a short-term context, ␥00
Difference between a long-term and short-term context, ␥10
Difference between an unspecified and short-term context, ␥20
Random
True mean effect size in a short-term context, ␦0j
⫺0.02
0.01
0.19
0.14
0.14
0.16
⫺0.11
0.05
1.2
0.04
0.19
18
20
20
18
.91
.96
.24
46.32
⬍.001
Note. Step 1: Results from unconditional multilevel model estimating the true mean standardized mean difference between high and low fertility in
women’s preference for facial masculinity in the sample of effects selected using relatively strict inclusion criteria. Step 2: Results from multilevel model
estimating the true mean standardized mean difference between high and low fertility in women’s preference for facial masculinity (strict inclusion criteria)
in a short-term relationship context (compared to a long-term or unspecified relationship context).
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
41
Table 8
Structural Body Masculinity
Effect
Coefficient
SE
Variance
component
SD
t ratio
df
2.45
4
2
p
Step 1
Fixed
Overall weighted mean effect size, ␥00
Random
True mean effect size, ␦0j
0.21
0.08
0.02
0.14
4
.07
9.52
.05
Step 2
Fixed
Weighted mean effect size in a short-term context, ␥00
Difference between a long-term and short-term context, ␥10
Random
True mean effect size in a short-term context, ␦0j
0.35
⫺0.27
0.1
0.11
3.49
⫺2.46
0.02
0.13
4
10
4
.04
.03
8.68
.07
Note. Step 1: Results from unconditional multilevel model estimating the true mean standardized mean difference between high and low fertility in
women’s preference for body masculinity in the sample of effects selected using relatively strict inclusion criteria. Step 2: Results from multilevel model
estimating the true mean standardized mean difference between high and low fertility in women’s preference for body masculinity (strict inclusion criteria)
in a short-term relationship context (compared to a long-term relationship context).
Step 2 revealed that the weighted mean g in a short-term context
was small (g ⫽ 0.19, SE ⫽ 0.07) and marginally statistically
significant (p ⫽ .09); and the weighted mean g in a long-term
context was small and negative (g ⫽ ⫺0.11, SE ⫽ 0.07) and not
statistically significant (p ⫽ .28). Comparing the two contexts
revealed that the weighted mean g was larger in short-term context
than in a long-term context, and this difference was statistically
significant (p ⫽ .01). None of the effects in this sample were
measured in an unspecified relationship context.
dardized mean difference between high and low fertility in women’s preference for facial cues of testosterone was small (g ⫽ 0.20,
SE ⫽ 0.22) and not statistically significant (p ⫽ .46). Thus, in this
set of effects, women’s preference for facial cues of circulating
testosterone was not generally stronger at high fertility than at low
fertility. All of the effects in this sample were measured in an
unspecified relationship context. More data are needed to determine whether there is any cycle shift in women’s preference for
facial cues of circulating testosterone and to examine possible
differences across relationship contexts.
Preference for Facial Cues of Testosterone
Preference for All Hypothesized Cues of Ancestral
Long-Term Partner Quality: Broad Set of Measures
The next analysis examined cycle shifts in women’s preference
for a facial appearance associated with higher levels of circulating
testosterone and included a small sample of three effects from
three studies (total N ⫽ 135). As shown in Table 11, Step 1
revealed that the weighted mean g estimating the true mean stan-
The next analysis examined cycle shifts in women’s preferences
for cues of long-term partner quality in the sample of effects that
included a broad set of mate preference measures. This analysis
Table 9
Vocal Masculinity
Effect
Coefficient
SE
Variance
component
SD
t ratio
df
2
p
Step 1
Fixed
Overall weighted mean effect size, ␥00
0.28
0.18
Random
True mean effect size, ␦0j
1.61
0.04
0.2
(Unable to
compute)
1
3.03
.08
Step 2
Fixed
Weighted mean effect size in a short-term context, ␥00
Difference between a long-term and short-term context, ␥10
Random
True mean effect size in a short-term context, ␦0j
0.4
0.2
1.98
⫺0.21
0.2
⫺1.08
0.04
0.19
(Unable to
compute)
2
1
.39
2.86
.09
Note. Step 1: Results from unconditional multilevel model estimating the true mean standardized mean difference between high and low fertility in
women’s preference for vocal masculinity in the sample of effects selected using relatively strict inclusion criteria. Step 2: Results from multilevel model
estimating the true mean standardized mean difference between high and low fertility in women’s preference for vocal masculinity (strict inclusion criteria)
in a short-term relationship context (compared to a long-term relationship context).
GILDERSLEEVE, HASELTON, AND FALES
42
Table 10
Behavioral Dominance and Felt Superiority Over Other Men
Effect
Coefficient
SE
Variance
component
SD
t ratio
df
0.71
2
2
p
Step 1
Fixed
Overall weighted mean effect size, ␥00
Random
True mean effect size, ␦0j
0.04
0.06
0.004
0.06
2
.55
2.49
.29
Step 2
Fixed
Weighted mean effect size in a short-term context, ␥00
Difference between a long-term and short-term context, ␥10
Random
True mean effect size in a short-term context, ␦0j
0.19
⫺0.3
0.07
0.08
2.65
⫺3.65
0.0004
0.06
2
10
2
.09
.01
2.57
.28
Note. Step 1: Results from unconditional multilevel model estimating the true mean standardized mean difference between high and low fertility
in women’s preference for behavioral dominance and felt superiority over other men in the sample of effects selected using relatively strict inclusion
criteria. Step 2: Results from multilevel model estimating the true mean standardized mean difference between high and low fertility in women’s
preference for behavioral dominance and felt superiority over other men (strict inclusion criteria) in a short-term relationship context (compared to
a long-term relationship context).
included 38 effects from eight studies (total N ⫽ 622). As shown
in Table 12, Step 1 revealed that the weighted mean g estimating
the true mean standardized mean difference between high and low
fertility in women’s preference for hypothesized cues of long-term
partner quality across short-term, unspecified, and long-term relationship contexts was near zero (g ⫽ ⫺0.004, SE ⫽ 0.04) and not
statistically significant (p ⫽ .91). Thus, in this set of effects,
women’s preferences for these characteristics did not generally
shift across the cycle.
Step 2 revealed that the weighted mean g was near zero and not
statistically significant in a short-term (g ⫽ ⫺0.06, SE ⫽ 0.05, p ⫽
.30), unspecified (g ⫽ ⫺0.04, SE ⫽ 0.17, p ⫽ .83), or long-term
relationship context (g ⫽ 0.05, SE ⫽ 0.05, p ⫽ .31). The weighted
mean g was somewhat more negative in a short-term context than
in a long-term context (suggesting that women’s preferences for
these characteristics are somewhat weaker at high fertility as
compared with low fertility when they evaluate men as short-term
partners), and this difference was marginally statistically significant (p ⫽ .09). The weighted mean g was somewhat more negative
in an unspecified context than in a long-term context, but this
difference was not statistically significant (p ⫽ .61). The weighted
mean g did not significantly differ between an unspecified and
short-term context (p ⫽ .92).
Preference for All Hypothesized Cues of Ancestral
Long-Term Partner Quality: Narrow Set of Measures
The next analysis examined cycle shifts in women’s preferences
for cues of long-term partner quality in the sample of effects that
included a narrow set of mate preference measures. This analysis
included eight effects from a single study (total N ⫽ 243). Because
all effects were from the same study, we used least squares
estimation procedures.
As shown in Table 13, Step 1 revealed that the weighted mean
g estimating the true mean standardized mean difference between
high and low fertility in women’s preference for cues of long-term
partner quality across short-term and long-term relationship contexts was near zero (g ⫽ ⫺0.05, SE ⫽ 0.05) and not statistically
significant (p ⫽ .28). Thus, this preliminary analysis did not reveal
any evidence that women’s preferences for these characteristics
shift across the cycle.
Step 2 revealed that the weighted mean g in a short-term context
was small and negative (g ⫽ ⫺0.12, SE ⫽ 0.07) and marginally
significant (p ⫽ .11), and the weighted mean g in a long-term
context was near zero (g ⫽ 0.01, SE ⫽ 0.07) and not statistically
significant (p ⫽ .83). The weighted mean g was somewhat more
negative in a short-term than in a long-term context, but this
Table 11
Facial Cues of Testosterone
Effect
Coefficient
Variance
component
SE
SD
t ratio
df
0.9
2
2
p
Step 1
Fixed
Overall weighted mean effect size, ␥00
Random
True mean effect size, ␦0j
0.20
0.22
0.08
0.29
2
.46
4.91
.08
Note. Step 1: Results from unconditional multilevel model estimating the true mean standardized mean difference between high and low fertility in
women’s preference for facial cues of testosterone in the sample of effects selected using relatively strict inclusion criteria.
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
43
Table 12
All Hypothesized Cues of Ancestral Long-Term Partner Quality: Broad Set of Measures
Effect
Coefficient
SE
Variance
component
SD
t ratio
df
⫺0.11
7
2
p
Step 1
Fixed
Overall weighted mean effect size, ␥00
Random
True mean effect size, ␦0j
⫺0.004
0.04
0.002
0.04
.91
7
6.97
⬎.50
Step 2
Fixed
Weighted mean effect size in a short-term context, ␥00
Difference between a long-term and short-term context, ␥10
Difference between an unspecified and short-term context, ␥20
Random
True mean effect size in a short-term context, ␦0j
0.05
0.06
0.18
⫺0.06
0.11
0.02
⫺1.13
1.76
0.1
0.002
0.05
7
35
35
.3
.09
.92
7
7.17
.41
Note. Step 1: Results from unconditional multilevel model estimating the true mean standardized mean difference between high and low fertility in
women’s preference for all hypothesized cues of ancestral long-term partner quality in the sample of effects selected using relatively relaxed inclusion
criteria. Step 2: Results from multilevel model estimating the true mean standardized mean difference between high and low fertility in women’s preference
for all hypothesized cues of ancestral long-term partner quality (strict inclusion criteria) in a short-term relationship context (compared to a long-term or
unspecified relationship context).
difference was not statistically significant (p ⫽ .19). None of the
effects in this sample were measured in an unspecified relationship
context. Ultimately, more data from a larger number of studies are
needed to determine with confidence whether women experience
relationship context-dependent cycle shifts in their preferences for
these characteristics.
Can Bias Account for the Observed Patterns
of Cycle Shifts?
Underrepresentation of small effects. When a meta-analysis
reveals robust, nonzero mean effects, and perhaps particularly
when those effects are consistent with predictions from a theory or
previously published findings, an important question is whether
these mean effects have been inflated by an underrepresentation of
small effects in the meta-analysis sample. Larger effects are more
likely to reach statistical significance, and statistically significant
findings are more likely to make their way into the published
literature (e.g., due to pressure on researchers and journals not to
publish null effects). In turn, published findings are typically easier
for meta-analysts to locate. In addition, if researchers are more
confident in or keep better track of unpublished data showing
significant effects, they might be more likely to share these data
with meta-analysts. Therefore, larger effects might be more likely
to make their way into a meta-analysis sample, whereas smaller
effects are more likely to be overlooked.
A common method for assessing whether it is likely that small
effects are underrepresented in a meta-analysis sample is to examine funnel plots. In funnel plots, effect sizes are plotted against
their standard errors, with larger effects on the right and smaller
standard errors—indicating more precise estimates (often from
larger studies)—at the top. If small effects are sufficiently well
represented, effects will be distributed symmetrically about the
mean effect size from the top to the bottom of the funnel. This is
because sampling error is equally likely to result in an overestimation as an underestimation of the true effect size. If, however,
small effects are underrepresented, more precise effects (top of the
funnel) will be symmetrically distributed about the mean effect
size, but less precise effects (bottom of the funnel) will skew to the
Table 13
All Hypothesized Cues of Ancestral Long-Term Partner Quality: Narrow Set of Measures
Effect
Coefficient
SE
t ratio
df
p
0.05
⫺1.174
7
.28
0.07
0.09
⫺1.89
1.49
6
6
.11
.19
Step 1
Fixed
Overall weighted mean effect size, ␥00
⫺0.05
Step 2
Fixed
Weighted mean effect size in a short-term context, ␥00
Difference between a long-term and short-term context, ␥10
⫺0.12
0.14
Note. Step 1: Results from unconditional multilevel model estimating the true mean standardized mean difference (g) between high and low fertility in
women’s preference for all hypothesized cues of ancestral long-term partner quality in the sample of effects selected using relatively strict inclusion criteria.
Because this sample consisted of eight effects from a single study, these are least squares estimates. Step 2: Results from multilevel model estimating the
true mean g between high and low fertility in women’s preference for all hypothesized cues of ancestral long-term partner quality (strict inclusion criteria)
in a short-term relationship context (compared to a long-term relationship context).
GILDERSLEEVE, HASELTON, AND FALES
44
right. At low precision, only large effects reach statistical significance. Therefore, the gap that forms in the lower left quadrant of
the funnel suggests that small effects are missing, perhaps due to
publication bias or some other sources of bias preventing the
inclusion of nonsignificant effects.
We used funnel plots to assess whether it was likely that small
effects were underrepresented in the sample of effects in our analysis
for which the ovulatory shift hypothesis predicts a relationship
context-dependent cycle shift—namely, effects measuring cycle shifts
in women’s preferences for hypothesized cues of ancestral genetic
quality. We predicted based on the ovulatory shift hypothesis that
women would exhibit cycle shifts in these preferences in a short-term
and unspecified relationship context but not in a long-term relationship context, and indeed this is the pattern we observed in the focal
analyses examining cycle shifts in preferences for all hypothesized
Long-term Relationship Context
Short-term and Unspecified Relationship Contexts
0
0
0.1
0.1
0.2
0.2
Standard Error
Standard Error
A
cues of ancestral genetic quality. Therefore, we plotted effects in a
short-term or unspecified context separately from effects in a longterm context. We created these plots for both the broad and narrow
samples of effects.
As shown in Figure 2A, the funnel plots did not reveal any
evidence of bias in the sample of effects that included a broad set
of mate preference measures. Observed effect sizes are roughly
evenly distributed about the mean from the top to the bottom of the
funnel in the long-term context and in the combined short-term and
unspecified context. Furthermore, Duval and Tweedie’s (2002)
“trim and fill” procedure, performed with Comprehensive MetaAnalysis software, did not indicate an absence of any putative
missing effects in either plot.
As shown in Figure 2B, the funnel plots revealed evidence of
slight bias in the sample of effects that included a narrow set of
0.3
0.4
0.5
0.6
0.3
0.4
0.5
0.6
0.7
0.7
-2
-1
0
1
-2
2
-1
Hedges’s g
Long-term Relationship Context
1
2
Short-term and Unspecified Relationship Contexts
0
0
0.1
0.1
0.2
0.2
Standard Error
Standard Error
B
0
Hedges’s g
0.3
0.4
0.5
0.6
0.3
0.4
0.5
0.6
0.7
0.7
-2
-1
0
Hedges’s g
1
2
-2
-1
0
Hedges’s g
Figure 2. Funnel plots to examine evidence for an underrepresentation of small effects among the sample of
effects for which the ovulatory shift hypothesis predicts relationship context-dependent cycle shifts—namely,
effects assessing cycle shifts in preferences for characteristics hypothesized to have reflected genetic quality in
ancestral males. Effects assessing cycle shifts in a long-term relationship context (no cycle shift predicted) are
plotted separately from effects assessing cycle shifts in a short-term or unspecified relationship context (positive
cycle shift predicted). Empty circles represent observed effects. Filled circle represents imputed putative missing
effect. (A) Sample of effects that included a broad set of mate preference measures. (B) Sample of effects that
included a narrow set of mate preference measures.
1
2
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
mate preference measures. Whereas observed effect sizes are
roughly evenly distributed about the mean from the top to the
bottom of the funnel in the long-term context, observed effect sizes
skew slightly to the right, moving from the top to the bottom of the
funnel in the combined short-term and unspecified context. Accordingly, the trim and fill procedure indicated that one effect was
missing from the short-term and unspecified plot. Imputing the
putative missing effect resulted in a negligible reduction in the
weighted mean effect size in the combined short-term and unspecified context (from g ⫽ 0.21 to 0.20, with no change in the 95%
confidence interval). Therefore, overall, the funnel plots and trim
and fill procedures did not reveal compelling evidence that the
pattern of cycle shifts observed in this meta-analysis is accounted
for by an underrepresentation of small effects in the sample.
Researcher degrees of freedom in defining high- and lowfertility windows. “Researcher degrees of freedom” refers to
ambiguity or flexibility in data collection and analysis practices
that enables researchers to try out several different methods and,
possibly, choose whichever method or analysis produces significant results (thereby dramatically increasing the Type I error rate;
Simmons et al., 2011). Most aspects of study design are determined in advance of data collection, eliminating concerns about
researcher degrees of freedom therein. However, one aspect of
study design that is relatively unique to cycle shift research and is
not always determined in advance of data collection is how to
define high- and low-fertility windows. This leaves open the
possibility that researchers could select, post hoc, high- and lowfertility windows that happen to produce predicted cycle shifts.
We initially attempted to address this potential concern by conducting a moderation analysis on the sample of effects examining
cycle shifts in women’s preferences for hypothesized cues of genetic
quality. Specifically, we examined the association between effect size
and the difference between the estimated average conception probability of the high-fertility window and the estimated average conception probability of the low-fertility window. We reasoned that if true
cycle shifts were present, effects would be larger among studies that
used a stronger fertility “manipulation” (a larger difference between
the estimated average conception probability of the high- and lowfertility windows). We did not observe any such association. However, notably, our method of estimating the average conception probability of high- and low-fertility windows had several potential
shortcomings (see supplemental materials).
Given the uninformative nature of this null finding, we next
attempted to address the issue by visually examining associations
between effect size and high- and low-fertility window definitions.
Figure 3 presents the high- and low-fertility windows used to
measure each effect that was predicted to be positive—namely,
each effect assessing cycle shifts in women’s preferences for
hypothesized cues of ancestral genetic quality in a short-term or
unspecified context. Effects are presented in ascending order by
effect size. We reasoned that if true cycle shifts are absent, and the
(spurious) cycle shifts observed in this meta-analysis resulted from
researchers selecting whichever high- and low-fertility windows
produced significant findings, larger effects would be associated
with (a) more variable high- and low-fertility window definitions,
(b) more poorly placed high- and low-fertility windows (highfertility windows that included true low-fertility days of the cycle
and/or low-fertility windows that included true high-fertility days
of the cycle), and (c) less frequent use of a continuous fertility
45
variable, which circumvents the problem of window definition
flexibility because all cycle days are included in the analysis.
Although a visual analysis cannot replace rigorous statistical tests
of associations between effect size and high- and low-fertility
window definitions, it is noteworthy that Figure 3 does not reveal
obvious evidence for the pattern just described; smaller and larger
effects do not appear to differ in a, b, or c.
Finally, we conducted an analysis examining cycle shifts in
women’s preferences for all hypothesized cues of genetic quality
but limited the analysis to those studies that used a continuous
fertility variable. As noted above, we reasoned that if cycle shifts
observed in this meta-analysis resulted from researcher degrees of
freedom in high- and low-fertility window definitions, these cycle
shifts would not be robust in the subsample of effects that is less
vulnerable to this problem (though we cannot definitively rule out
the possibility that researchers chose, post hoc, to use a continuous
fertility variable because doing so produced predicted cycle shifts).
We conducted this analysis, first, in the sample of effects that
included a broad set of mate preference measures and, then, in the
sample that included a narrow set of measures.
The first, broad sample included 31 effects from 12 studies. The
weighted mean g across contexts was small to moderate (g ⫽ 0.26,
SE ⫽ 0.12) and borderline statistically significant (p ⫽ .05). The
weighted mean g in a short-term context was small (g ⫽ 0.17,
SE ⫽ 0.11) and fell short of statistical significance (p ⫽ .14); the
weighted mean g in an unspecified relationship context was moderate to large (g ⫽ 0.62, SE ⫽ 0.17) and statistically significant
(p ⫽ .004); and the weighted mean g in a long-term context was
near zero (g ⫽ ⫺0.03, SE ⫽ 0.11) and not statistically significant
(p ⫽ .77). Comparing the three contexts revealed that the weighted
mean g was significantly larger in a short-term context than in a
long-term context and in an unspecified context than in a long-term
context (p ⫽ .005 and .003, respectively). The weighted mean g was
also significantly larger in an unspecified context than in a short-term
context (p ⫽ .01). This difference is likely due to the influence of
several particularly large positive effects included in the unspecified
subsample of effects (e.g., studies examining women’s preferences for
scents associated with symmetry) and one large negative effect included in the short-term subsample (Morrison et al., 2010).
The second, narrow sample included 20 effects from nine studies.
The weighted mean g across contexts was small to moderate (g ⫽
0.38, SE ⫽ 0.13) and statistically significant (p ⫽ .02).
The weighted mean g in a short-term context was small to moderate
(g ⫽ 0.29, SE ⫽ 0.12) and statistically significant (p ⫽ .04); the
weighted mean g in an unspecified relationship context was moderate
to large (g ⫽ 0.62, SE ⫽ 0.16) and statistically significant (p ⫽ .005);
and the weighted mean g in a long-term context was near zero (g ⫽
0.03, SE ⫽ 0.11) and not statistically significant (p ⫽ .81). Comparing
the three contexts revealed that the weighted mean g was significantly
larger in a short-term context than in a long-term context and in an
unspecified context than in a long-term context (p ⫽ .002 and .009,
respectively). The weighted mean g did not differ between a shortterm and an unspecified context (p ⫽ .12). Thus, results were largely
consistent with those observed in the full samples of effects.
In sum, we used multiple procedures to assess and adjust for
various forms of potential bias. The results of these procedures do
not suggest that these sources of bias account for the robust cycle
shifts observed in this meta-analysis.
GILDERSLEEVE, HASELTON, AND FALES
46
Study
Effect (Relationship Context)
Miller (2003)*
Rupp, Librach, et al. (2009)
Morrison et al. (2010)*
Miller (2003)*
Rantala et al. (2010)*
Facial Masculinity (ST)
Facial Masculinity (U)
Male Facial Movements (ST)
Body Muscularity (ST)
Torso Hair (U)
Caryl et al. (2009)*
Izbicki & Johnson (2010)
Hromatko et al. (2006)
Oinonen & Mazmanian (2007)
Roney et al. (2011)
Rantala et al. (2006)
Moore et al. (2011), Study 2
Miller (2003)*
Izbicki & Johnson (2010)*
Moore (2011)
Izbicki & Johnson (2010)*
Miller (2003)*
Fink (2012)
Moore (2011)
Vaughn et al. (2010)
Izbicki & Johnson (2010)*
Strong (U)
Facial Masculinity (ST)
Facial Symmetry (U)
Facial Symmetry (U)
Facial Masculinity (U)
Scent Cues of Testosterone (U)
Facial Cues of Testosterone (U)
Tall (ST)
Strong (ST)
Facial Symmetry (U)
Dominant (ST)
Big Ego (ST)
Facial Masculinity (U)
Facial Masculinity (U)
Facial Masculinity (U)
Facial Darkness (ST)
Beaulieu (2007), Study 2*
Bressan & Stranieri (2008)
Cárdenas & Harris (2007)
Singh & Bailey (2006)
Dominance (U)
Facial Masculinity (U)
Facial symmetry (ST)
Male SHR (ST)
Caryl et al. (2009)*
Harris (2011)
Aggressive (U)
Facial Masculinity (U)
Caryl et al. (2009)*
Koehler et al. (2006)
Perrett et al. (2013), Study 1
Gangestad et al. (2007)
Gangestad et al. (2011)
Luevano & Zebrowitz (2006)*
Gangestad et al. (2004)
Arrogant (U)
Facial Symmetry (U)
Facial Masculinity (U)
Socially Respected/Influential (ST)
Facial Masculinity (U)
Dominant (ST)
Competitiveness (ST)
Gangestad et al. (2007)
Garver-Apgar & Gangestad (2012)
Koehler et al. (2006)*
Arrogant/Self-Centered
p (ST)
(ST)
Facial Averageness (U)
Caryl et al. (2009)*
Gangestad et al. (2007)
Frost (1994)*
Penton-Voak et al. (1999), Study 2
Gangestad et al. (2007)
Welling et al. (2007)
Luevano & Zebrowitz (2006)
Perrett et al. (2013), Study 2
Havlíček et al. (2005)
Jones, Little, et al. (2005), Study 2
Lukaszewski & Roney (2009)*
Penton-Voak & Perrett (2000)
Singh & Bailey (2006)
Gangestad et al. (2004)
Johnston et al. (2001)
Little, Jones, et al. (2007), Study 1
Puts (2005)
Roney et al. (2011)
Feinberg (2012)
Penton-Voak et al. (1999), Study 1
Provost et al. (2008)*
Pawlowski & Jasienska (2005)
Roney & Simmons (2008)
Thornhill et al. (2003)
Little, Jones, et al. (2007), Study 2
Little, Jones, & Burriss (2007), Study 1
Thornhill et al. (2013)
Little, Jones, & Burriss (2007), Study 2
Little et al. (2008)
Thornhill & Gangestad (1999b)
Gangestad & Thornhill (1998)
Conceited (U)
Muscular (ST)
Facial Darkness (U)
Facial Masculinity (ST)
Confrontative (ST)
Facial Masculinity (U)
Facial Masculinity (ST)
Facial Masculinity (ST)
Scent Cues of Dominance (U)
Facial Masculinity (U)
Cycle Days Defined as "High fertility" and "Low fertility"
Effect
28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 RCD
Size
(g ) FCD 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
------Peak Fertility-----Ov
Dominant (ST)
Facial Masculinity (U)
Male-Typical WHR (ST)
Social Presence (ST)
Facial Masculinity (U)
Facial Symmetry (U)
Vocal Masculinity (ST)
Facial Cues of Testosterone (U)
Vocal Masculinity (ST)
Facial Masculinity (U)
Male-Typical Walk (U)
Taller Man Relative to Self (ST)
Facial Cues of Testosterone (U)
Scent Cues of Body Symmetry (U)
Facial Symmetry (ST)
Body Masculinity (ST)
Scent Cues of Testosterone (U)
Body Masculinity (ST)
Facial Masculinity (U)
Scent Cues of Body Symmetry (U)
Scent Cues of Body Symmetry (U)
-0.65
-0.64
-0.38
-0.35
-0.35
-0.29
-0.29
-0.26
-0.23
-0.23
-0.18
-0.17
-0.15
-0.13
-0.11
-0.09
-0.09
-0.06
-0.04
-0.01
0.00
0.00
0.00
0.00
0.00
0.02
0.03
0.04
0.04
0.04
0.05
0.08
0.09
0.12
0.14
0.15
0.15
0.17
0.17
0.19
0.23
0.24
0.26
0.30
0.32
0.33
0.33
0.36
0.39
0.39
0.40
0.40
0.41
0.42
0.43
0.45
0.45
0.45
0.46
0.46
0.55
0.59
0.59
0.66
0.69
0.72
0.94
1.25
Figure 3. High- and low-fertility cycle phase definitions for effects assessing cycle shifts in preferences for
hypothesized cues of genetic quality in a short-term (ST) or unspecified (U) relationship context (where a cycle
shift was predicted). Effects marked with asterisks were included only in the broad sample. Effects not marked
with asterisks were included in both the broad and narrow samples. Black boxes and light gray boxes indicate
cycle days defined as high fertility and low fertility, respectively. White (unfilled) boxes indicate days that fell
outside of high- and low-fertility windows and were therefore excluded from analysis. Dark gray boxes indicate
that fertility was treated as a continuous variable, and therefore all cycle days were included in analyses. High-
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
Discussion
Summary of Meta-Analysis Findings
We evaluated evidence for the ovulatory shift hypothesis in a
large sample of published and unpublished effects and found clear
support for the predicted pattern of relationship context-dependent
cycle shifts in women’s mate preferences. Women exhibited a
stronger preference for characteristics widely thought to have
reflected genetic quality in ancestral males on high-fertility days of
the cycle as compared with low-fertility days of the cycle. However, this cycle shift depended on the type of relationship for which
women evaluated a prospective partner. Women exhibited a robust
cycle shift in their preferences for hypothesized cues of ancestral
genetic quality when they evaluated men or male stimuli as prospective partners for a short-term relationship (e.g., a one-night
stand) or evaluated the attractiveness of male stimuli or desirability
of male characteristics without reference to a specific relationship
context. In contrast, women exhibited no such cycle shift when
they evaluated men or male stimuli as prospective partners for a
long-term relationship (e.g., marriage). Likewise, women did not
exhibit a cycle shift in their preferences for characteristics widely
thought to have reflected suitability as a long-term social partner
and coparent in ancestral males in any relationship context. This
pattern of cycle shifts was robust across both a broad sample of
effects that included a diverse set of male characteristics and
measures of mate preferences and a narrow sample of effects that
included only those characteristics and measures that we reasoned
would provide a particularly strong test of the predicted cycle
shifts. Furthermore, importantly, the observed cycle shifts do not
appear to be accounted for by an underrepresentation of small
effects in the meta-analysis sample (as could result from publication bias) or by researcher degrees of freedom in definitions of
high- and low-fertility cycle phases.
We conducted more focused analyses to examine cycle shifts in
women’s preferences for specific characteristics hypothesized to
have indicated genetic quality in ancestral males. Many of these
analyses were conducted on small samples of effects, and in such
cases, results should be considered preliminary. Among the specific characteristics we examined, body masculinity and behavioral
dominance showed the strongest support for the pattern of cycle
shifts predicted by the ovulatory shift hypothesis. Analyses revealed a significant and marginally significant cycle shift in women’s preference for body masculinity and behavioral dominance,
respectively, in a short-term relationship context, no cycle shift in
a long-term relationship context, and a significant difference in the
magnitude of this cycle shift comparing a short-term to a long-term
context. Analyses examining cycle shifts in preferences for facial
symmetry and vocal masculinity hinted at a similar pattern, but the
47
predicted cycle shifts fell short of statistical significance. However, given that these analyses were underpowered, more data are
needed to make any confident claims about the presence or absence of cycle shifts in women’s facial symmetry and vocal
masculinity preferences.
Analyses examining cycle shifts in women’s preference for
facial masculinity revealed partial support for the ovulatory shift
hypothesis. Analyses revealed a significant cycle shift in attractiveness ratings made without reference to a specific type of
relationship and no cycle shift in a long-term context. However,
analyses did not reveal a cycle shift in a short-term context (where
a cycle shift was predicted). Removing two studies that used
potentially problematic measures of women’s facial masculinity
preferences revealed a small, though still not statistically significant, cycle shift in a short-term context. Ultimately, more data are
needed to determine whether this unexpected pattern of results is
robust and in need of explanation or reflects the influence of
idiosyncratic features of the particular studies included in this
analysis.
Lastly, analyses examining cycle shifts in women’s preferences
for scents associated with symmetry and facial cues associated
with circulating testosterone both hinted at a cycle shift in attractiveness ratings made without reference to a specific type of
relationship, but these cycle shifts fell short of statistical significance. However, these analyses were underpowered, so again,
more data are needed to make any confident claims about the
presence or absence of cycle shifts in these preferences.
Interpreting Differences in Statistical Significance
Across Contexts and Characteristics
This meta-analysis revealed differences in the magnitude of
cycle shifts across relationship contexts and across specific male
characteristics, raising the question of how to properly interpret
these differences. Interpreting a single statistically significant cycle shift—for example, the high-fertility increase in short-term
body masculinity preferences—is straightforward: Although possible, the probability that a cycle shift of this magnitude and level
of statistical significance is accounted for by chance alone is very
low, and thus it is conventional to infer that the cycle shift is
probably real. Likewise, given a statistically significant difference
between relationship contexts in the magnitude of a given cycle
shift—for example, the difference between a short-term and longterm relationship context in the magnitude of the cycle shift in
body masculinity preferences—we can also straightforwardly conclude that the probability that this apparent context effect is accounted for by chance alone is very low.
In contrast, it is less clear how to properly interpret null effects
and comparisons between null and statistically significant effects.
Figure 3 (opposite).
and low-fertility windows are displayed in terms of forward cycle day (FCD; days since last menstrual onset) and reverse cycle day
(RCD; days until next menstrual onset) for studies that used the forward counting or reverse counting method, respectively. High-fertility windows are
displayed in terms of days from ovulation, and low-fertility windows are displayed in terms of FCD, for studies that used luteinizing hormone tests to verify
impending ovulation. To enable comparing high- and low-fertility windows across these three methods, we have assumed a 28-day cycle length, with
ovulation (Ov) occurring on FCD 14/RCD 15. We have demarcated a suggested “peak fertility” window with double lines. This window includes the 6
days with the highest average conception probabilities for regularly cycling women as reported by Wilcox et al. (2001). SHR ⫽ shoulder-to-hip ratio;
WHR ⫽ waist-to-hip ratio.
48
GILDERSLEEVE, HASELTON, AND FALES
For example, analyses revealed a nonsignificant cycle shift in
women’s short-term vocal masculinity preferences that was, nonetheless, comparable in magnitude to the statistically significant
cycle shift in women’s short-term body masculinity preferences.
One possible interpretation of this pattern of statistical significance
is that women’s preferences for body masculinity shift across the
cycle, whereas their preferences for vocal masculinity do not. If
the ovulatory shift hypothesis is correct, this could indicate that
body masculinity reflected genetic quality ancestrally, whereas
vocal masculinity did not. However, importantly, several other
possibilities are equally consistent with this pattern of statistical
significance. For example, it is possible that the body masculinity
analysis was sufficiently powered to detect a cycle shift, whereas
the vocal masculinity analysis was not (and, in fact, the body
masculinity analysis included 3 times as many effects as the vocal
masculinity analysis). It is also possible that researchers manipulated or measured body masculinity with greater precision than
they manipulated or measured vocal masculinity, that participants
were able to perceive variation in body masculinity in male body
photos or drawings with greater acuity than they were able to
perceive variation in vocal masculinity in vocal recordings, or that
studies examining preferences for body masculinity incidentally
used more rigorous methods (e.g., for determining women’s position in the ovulatory cycle) than studies examining preferences for
vocal masculinity. Ultimately, in the case of null effects, especially
those produced by analyses that are likely to have been underpowered, additional studies are needed to test for the presence and
magnitude of cycle shifts. In summary, whereas statistically significant effects indicate the likely presence of real phenomena
deserving of explanation, null effects based on small numbers of
effects indicate a need for more evidence.
Limitations
The focal analyses examining cycle shifts in women’s preferences for all characteristics hypothesized to have reflected genetic
quality in ancestral males contained many effects and produced a
clear pattern of results supporting the ovulatory shift hypothesis.
However, a common limitation of the more focused analyses
examining cycle shifts in preferences for specific male characteristics—for example, vocal masculinity, scents associated with
symmetry, and facial cues of testosterone—was a lack of sufficient
statistical power. Therefore, although the overall pattern of results
was typically consistent with the ovulatory shift hypothesis, the
meta-analysis findings do not compel firm conclusions regarding
the robustness of cycle shifts in preferences for these or other
specific characteristics.
In addition, although many analyses revealed significant unexplained between-studies variation in the magnitude of cycle shifts,
the moderation analyses revealed few and somewhat inconsistent
associations between study characteristics and effect size. A possible explanation is that studies in this meta-analysis varied in so
many ways that there was simply too much noise to observe true
moderation effects. In addition, despite substantial methodological
heterogeneity in the sample as a whole, there often was not enough
variation on specific moderators to obtain a precise estimate of
their effect. For example, only three of the 50 studies that contributed effects to the analysis examining cycle shifts in preferences
for all hypothesized cues of ancestral genetic quality (broad sam-
ple of effects) used luteinizing hormone tests to verify the timing
of ovulation, though this method is widely regarded as one of the
most rigorous for assessing cycle position. Therefore, moderation
analyses examining associations between the use of this particular
method and the magnitude of cycle shifts (or between the use of
this method and the moderating effect of relationship context on
cycle shifts) were underpowered. We emphasize that these null
findings do not indicate that methodological rigor has no association with effect size; rather, there currently is an absence of
evidence for such associations.
Several moderators did emerge across both the broad and narrow samples of effects as being significantly or marginally significantly associated with the pattern of cycle shifts predicted by the
ovulatory shift hypothesis. Studies that used scent stimuli, used
direct measurement to determine the amount of the characteristic
of interest possessed by the male stimuli, or were published generally showed larger predicted cycle shifts after controlling for the
effect of relationship context. In addition, studies conducted outside of the lab (usually online) generally showed larger predicted
cycle shifts in a short-term relationship context relative to a longterm relationship context. In contrast (contrary to the predictions of
the ovulatory shift hypothesis), studies in which participant ratings
were used to determine the amount of the characteristic of interest
possessed by the male stimuli generally showed smaller predicted
cycle shifts after controlling for the effect of relationship context.
Importantly, the moderation analyses tested for associations between study characteristics and effect size, rather than for causal
relationships. Nonetheless, the results provide preliminary insight
into the kinds of studies that might be better at capturing true
context-dependent cycle shifts in mate preferences if they are
present.
Also important, the finding that predicted cycle shifts were
generally larger in published studies than in unpublished studies is
consistent with several possible, nonmutually exclusive interpretations. One possibility is that the mean effect size within the
published literature overestimates the true magnitude of cycle
shifts. Upward bias in effect size among published studies could
reflect a tendency among reviewers, journal editors, or researchers
themselves to evaluate articles that report positive findings as more
worthy of publication than articles that report null or negative
findings simply by virtue of the fact that they provide support for
the hypothesis in question. It is important to note that any such
tendency did not result in a detectable underrepresentation of small
effects in the meta-analysis sample as a whole (see funnel plots
above). Another possibility is that the mean effect size within the
unpublished literature underestimates the true magnitude of cycle
shifts. Downward bias in effect size among unpublished studies
could reflect a tendency among reviewers, journal editors, or
researchers to evaluate articles that report positive findings as
more worthy of publication than articles that report null or negative findings, not because they provide support for the ovulatory
shift hypothesis but rather because these studies actually used
more rigorous methods or otherwise provided more precise tests of
predicted cycle shifts. In sum, publication status appears to be an
additional source of between-studies variation in cycle shift magnitude, but this finding should be interpreted with due caution.
An additional limitation of this meta-analysis is that the results
cannot provide insight into whether women find high levels of a
given characteristic particularly attractive at high fertility, find low
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
levels of a given characteristic particularly aversive at high fertility, or both. This limitation is in fact not unique to this metaanalysis but rather is a limitation of many of the studies in the
meta-analysis sample—including, for example, all studies that
used a forced-choice or slider task to assess women’s preference
for a characteristic. To accommodate the large number of these
studies in the meta-analysis sample, we selected an effect size that
does not differentiate between the above possibilities.
Lastly, in general, meta-analyses evaluate the strength and robustness of effects in an empirical literature, rather than provide a
direct test of the hypothesis of interest. Thus, this meta-analysis
provides a test of the ovulatory shift hypothesis only to the extent
that the set of empirical findings it synthesized provided a test of
that hypothesis. Given the challenges of estimating and verifying
women’s position in the ovulatory cycle (see supplemental materials), it is likely that some studies included in this meta-analysis
provided a relatively weak test of the ovulatory shift hypothesis.
Therefore, the weighted mean effect sizes we report here could be
conservative estimates of the true effect sizes. Despite these issues
and other limitations, the findings of the focal analyses examining
cycle shifts in women’s preferences for all hypothesized cues of
ancestral genetic quality offer clear support in the extant empirical
literature for the pattern of cycle shifts predicted by the ovulatory
shift hypothesis.
Strengths
The focal analyses examining cycle shifts in preferences for all
hypothesized cues of ancestral genetic quality included large numbers of effects from unpublished studies (e.g., 34 of the 96 effects
in the analysis that included a relatively broad set of mate preference measures) obtained through a variety of methods (e.g., listserv posts). Although unpublished studies have often yielded null
results, the key analyses revealed cycle shifts that were robust
across the entire sample of published and unpublished effects.
Furthermore, funnel plots and trim and fill procedures did not
provide compelling evidence that the statistically significant cycle
shifts observed in this meta-analysis could be accounted for by an
underrepresentation of small effects. In addition, we used several
procedures to assess whether the statistically significant cycle
shifts observed in this analysis appeared to result from bias in
researchers’ definitions of high- and low-fertility cycle phases but
did not find evidence of such bias. Thus, publication bias and
researcher degrees of freedom in high- and low-fertility definitions
do not appear to account for the cycle shifts observed in this
meta-analysis.
Another strength of this meta-analysis is that we used multilevel
meta-analytic methods. This enabled us to include multiple effects
from the same study in a single analysis, while properly accounting
for the nonindependence of these nested effects. It also enabled us
to test cross-level interactions among effect- and study-level predictors, for example, to identify study characteristics that moderated relationship context effects.
Lastly, we used carefully designed inclusion criteria to create
two samples of effects: a relatively heterogeneous, “broad” sample
of effects that we reasoned would capture the diversity of mate
preference measures used in this literature and a relatively homogeneous, “narrow” sample of effects that we reasoned would
provide a relatively strong test of the ovulatory shift hypothesis. In
49
fact, in an earlier version of this article, we had reported results
based only on the narrow sample. However, in response to suggestions from reviewers, we subsequently relaxed the inclusion
criteria twice to create two broader samples. We report the broader
of these two samples here. Although the pattern of cycle shifts
predicted by the ovulatory shift hypothesis was somewhat stronger
in the narrow sample, it remained robust in both of the broader
samples. This indicates that the pattern of cycle shifts observed in
this meta-analysis is not a mere artifact of the particular inclusion
criteria that we used to select the initial, narrow sample of effects.
Convergent Evidence for Cycle Shifts in
Mating Motivations
The key findings of this meta-analysis are consistent with a
growing body of research supporting the overarching idea that
women’s mating-related motivations, preferences, cognitions, and
behaviors shift near ovulation, leading to systematic changes
across the ovulatory cycle. For example, other lines of work have
documented cycle shifts in women’s attractions to their relationship partners and other individuals (e.g., Larson, Pillsworth, &
Haselton, 2012), opportunistic orientation toward sex (Gangestad
et al., 2010a), evaluations of their relationship partner’s flaws and
virtues and feelings of closeness and satisfaction with their partners (Larson et al., 2013), preferences for attractive and revealing
clothing (Durante, Li, & Haselton, 2008; Haselton, Mortezaie,
Pillsworth, Bleske-Rechek, & Frederick, 2007), interest in attending events where they might meet potential partners (Haselton &
Gangestad, 2006), and receptiveness to others’ attempts to initiate
romantic involvements with them (Guéguen, 2009a, 2009b).
The body of research examining cycle shifts in women’s attractions to men other than their primary partners is particularly
relevant to the idea that women’s mate preferences shift across the
cycle. This line of research aims to test the prediction that women
whose primary partners are relatively lacking in the characteristics
women particularly prefer at high fertility—namely, characteristics
thought to have reflected genetic quality in ancestral males—will
be particularly likely to experience an increase at high fertility
relative to low fertility in their attraction to other men (presumably,
men who possess higher levels of these characteristics). Consistent
with this idea, across five studies, the extent to which women
reported experiencing greater extra-pair attraction (attraction to
men other than their primary partner) at high fertility relative to
low fertility depended on their partner’s sexual attractiveness or on
the extent to which their partner possessed specific characteristics
thought to have reflected genetic quality in ancestral males (e.g.,
partner sexual attractiveness, Pillsworth & Haselton, 2006a; partner sexual attractiveness relative to investment attractiveness,
Haselton & Gangestad, 2006; partner facial masculinity, Gangestad, Thornhill, & Garver-Apgar, 2010b; facial masculinity and
partner facial attractiveness [marginally significant], Gangestad et
al., 2010b; composite partner face and body attractiveness, Larson
et al., 2012). Furthermore, in several studies, women’s reports of
their partner’s mate retention behavior (e.g., jealousy, possessiveness, and attentiveness) increased at high- relative to low-fertility,
(Gangestad et al., 2002; Haselton & Gangestad, 2006), and this
effect appeared to depend on the extent to which their partner
possessed characteristics that women are thought to particularly
prefer at high fertility (Haselton & Gangestad, 2006; Pillsworth &
50
GILDERSLEEVE, HASELTON, AND FALES
Haselton, 2006a). These findings are consistent with the notion
that, as ancestral females evolved psychological mechanisms that
produced cycle shifts in mate preferences, males coevolved psychological mechanisms that facilitated behaviors that mitigated the
risk of a mate engaging in extra-pair sex at high fertility.
Suggested Directions for Future Research
The existence of robust ovulation-related changes in women’s
mate preferences across the ovulatory cycle highlights a number of
interesting and potentially illuminating avenues for future research
and theory in this area. First, it is not yet known whether cycle
shifts in women’s mate preferences represent the output of psychological mechanisms that have been favored by selection during
human evolutionary history or psychological mechanisms that
were favored by selection in an ancestral species but are vestigial
in humans. Therefore, the specific conditions that initially gave
rise to and have maintained or modified the psychological mechanisms posited to produce cycle shifts in women’s mate preferences are not yet well understood. A phylogenetic analysis could
help to shed light on the precise evolutionary pathways that gave
rise to the posited psychological adaptations. In addition, if these
psychological mechanisms initially evolved in an ancestral species, theoretical and empirical work could help to clarify how these
mechanisms have since been modified in the context of high rates
of pair bonding among humans (Gangestad & Garver-Apgar,
2013).
Second, future research should seek to identify the hormonal
mechanisms underlying cycle shifts in women’s mate preferences.
Previous research has suggested several possible candidates for
hormonal mediators of such cycle shifts. For example, two studies
have found a positive association between women’s measured
estradiol levels within the ovulatory cycle and their preferences for
facial cues of testosterone in men (Roney & Simmons, 2008;
Roney, Simmons, & Gray, 2011). In addition, several studies have
used women’s position within the ovulatory cycle to estimate their
hormone levels and have found a negative association between
women’s estimated progesterone levels and preferences for scents
associated with symmetry and vocal masculinity (Garver-Apgar,
Gangestad, & Thornhill, 2008; Puts, 2005), a positive association
between women’s estimated luteinizing hormone and follicle stimulating hormone levels and preference for dominance in a shortterm sex partner (Lukaszewski & Roney, 2009), and a positive
association between women’s estimated levels of testosterone and
preference for facial masculinity (Welling et al., 2007). It is
possible that all of these hormones play a role in shifts in women’s
mate preferences across the cycle or that a particular hormone,
such as estradiol, is the primary hormone driving cycle shifts.
Ultimately, research directly measuring each of these potential
hormonal mediators is needed to better address the question of
which hormonal mechanisms underlie cycle shifts.
Third, future research should examine the impact of cycle shifts
in women’s mate preferences on long-term relationship functioning and longevity. As noted above, several lines of work suggest
that women whose long-term partners possess relatively low levels
of the characteristics women find most attractive at high relative to
low fertility might be particularly likely to experience a cycle shift
in their attraction to other men (e.g., Haselton & Gangestad, 2006),
in their satisfaction with their current partner (Larson et al., 2012),
and in their partner’s mate retention behaviors toward them (e.g.,
Haselton & Gangestad, 2006; Pillsworth & Haselton, 2006a),
potentially leading them to experience increased conflict with their
partner or other changes in their relationship in the fertile period of
the cycle. What remains unknown is whether such changes completely resolve, allowing relationships to return to their prior state
after each fertile period, or have a cumulative effect on relationship
functioning and longevity. Furthermore, it remains unknown how
hormonal contraceptive use, pregnancy, menopause, and other
factors that dramatically alter or eliminate cyclic variation in
women’s hormones impact relationship functioning and longevity.
Given the important and far-reaching implications of these questions, rigorous research is needed to examine the long-term impacts of cycle shifts on long-term relationships.
Fourth, research in this area has primarily involved Western
samples of educated young women. Overreliance on such samples
is common throughout psychology and not unique to this research
area (Henrich, Heine, & Norenzayan, 2010). Nonetheless, future
research should examine variation in the robustness and magnitude
of cycle shifts in mate preferences in other ecologies and cultural
contexts. For example, as a result of having more frequent pregnancies and breastfeeding for longer periods, women in traditional,
“natural-fertility” populations experience far fewer ovulatory cycles than women in Western populations (see Lancaster & Alvarado, 2010). Among the Dogon of Mali, for example, women
have about 100 ovulatory cycles in their lifetime, compared with
an estimated 400 lifetime ovulatory cycles among American
women (see Strassmann, 1997). This raises the question of
whether women who have relatively few ovulatory cycles in their
lifetime experience cycle shifts in mate preferences similar to
those experienced by women who have relatively many ovulatory
cycles, such as the women included in this meta-analysis. Furthermore, it remains unknown whether the behavioral effects of these
cycle shifts vary across different populations. Are women who
experience relatively few ovulatory cycles in their lifetime more or
less likely to act on their shifting desires?
Lastly, as noted above, there is not yet an established set of
conventions for how to best design studies to measure ovulatory
cycle shifts. At present, there is considerable variation in the
methods researchers use to examine cycle shifts (see supplemental
materials), including in whether researchers (a) use a betweenversus within-participants design, (b) obtain hormonal confirmation of women’s ovulatory cycle position versus estimate women’s
cycle position based on a “counting method,” (c) estimate women’s cycle position based on a forward versus reverse counting
method, (d) base estimates of cycle position solely on participants’
retrospectively recalled or predicted dates of menstrual onset versus dates of menstrual verified during the course of the study, (e)
treat fertility as continuous by assigning each woman a conception
probability estimate from actuarial tables versus treat fertility as
dichotomous by defining discrete high- and low-fertility cycle
phases, and so on. An important task for future research is to
empirically evaluate these methods and their relative strengths. For
example, it is reasonable to argue that studies that track women
over time, obtain verified dates of menstrual onset, and use hormone tests to confirm ovulation provide some of the most precise
tests of ovulatory cycle shifts. However, using such methods is
very costly. A key question, therefore, is how simpler methods—
for example, a between-participants design, requiring only wom-
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
en’s retrospectively recalled date of menstrual onset— compare
with more rigorous methods.
Notably, the majority of the studies included in this metaanalysis used counting methods that rely on women’s reports of
retrospectively recalled or predicted dates of menstrual onset to
estimate their position in the ovulatory cycle. Given the ease with
which these methods can be used, they are likely to continue to be
popular. As noted above, among studies using counting methods to
estimate women’s position within the ovulatory cycle, there is
considerable variation in the cycle days researchers have defined
as high and low fertility (see Figure 3). Ideally, researchers will
work to establish a convention about the best days to include in
these windows. However, a straightforward alternative, which we
recommend, is to treat fertility as continuous by assigning each
woman a conception probability estimate based on actuarial tables
(Wilcox et al., 2001). By eliminating the opportunity to select
among different high- and low-fertility windows that produce
somewhat different results, this method helps to alleviate concerns
that any observed statistically significant cycle shifts reflect researcher degrees of freedom.
Conclusions
Over the past 2 decades, there has been a surge of interest in
examining systematic shifts in women’s mate preferences across
the ovulatory cycle, with dozens of empirical articles examining
these and related effects and many more referencing the work. This
meta-analysis shows that there is robust support in the extant
published and unpublished empirical literatures for the pattern of
relationship context-dependent cycle shifts in women’s mate preferences predicted by the ovulatory shift hypothesis. Although this
meta-analysis answers the important empirical question of whether
these cycle shifts are robust, it also highlights a number of unresolved issues to be addressed by future theory and research, as
noted above. Nonetheless, the findings of this meta-analysis have
important implications for understanding the ultimate evolutionary
and proximate causes of systematic day-to-day variation in women’s attractions, motivations, and social relationships.
References
References marked with an asterisk indicate studies included in the
meta-analysis.
Beaulieu, D. A. (2007). Avoiding costly mating mistakes: The evolution of
female reproductive safeguards (Unpublished doctoral dissertation).
University of California, Santa Barbara.
Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Rothstein, H. R. (2009).
Introduction to meta-analysis. Chichester, England: Wiley. doi:10.1002/
9780470743386
Braude, S., Tang-Martinez, Z., & Taylor, G. T. (1999). Stress, testosterone,
and the immunoredistribution hypothesis. Behavioral Ecology, 10, 345–
350. doi:10.1093/beheco/10.3.345
ⴱ
Bressan, P., & Stranieri, D. (2008). The best men are (not always) already
taken: Female preference for single versus attached males depends on
conception risk. Psychological Science, 19, 145–151. doi:10.1111/j
.1467-9280.2008.02060.x
Bullock, H. L. (2000). Multiple components of facial attractiveness in
humans (Unpublished doctoral dissertation). Queen’s University, Kingston, Ontario, Canada.
ⴱ
51
Cárdenas, R. A., & Harris, L. J. (2007). Do women’s preferences for
symmetry change across the ovulatory cycle? Evolution and Human
Behavior, 28, 96 –105. doi:10.1016/j.evolhumbehav.2006.08.003
Carter, A. J. R., Weier, T. M., & Houle, D. (2009). The effect of inbreeding
on fluctuating asymmetry of wing veins in two laboratory strains of
Drosophila melanogaster. Heredity, 102, 563–572. doi:10.1038/hdy
.2009.13
ⴱ
Caryl, P. G., Bean, J. E., Smallwood, E. B., Barron, J. C., Tully, L., &
Allerhand, M. (2009). Women’s preference for male pupil size: Effects
of conception risk, sociosexuality, and relationship status. Personality
and Individual Differences, 46, 503–508. doi:10.1016/j.paid.2008.11
.024
Chen, B. P., & Parham, P. (1989). Direct binding of influenza peptides to
Class I HLA molecules. Nature, 337, 743–745. doi:10.1038/337743a0
Clarke, G. M. (1993). The genetic basis of developmental stability, I.
Relationships between stability, heterozygosity, and genomic coadaptation. Genetica, 89, 15–23. doi:10.1007/BF02424502
Cohen, J. (1988). Statistical power analysis for the behavioral sciences
(2nd ed.). Hillsdale, NJ: Erlbaum.
DeBruine, L., Jones, B. C., Frederick, D. A., Haselton, M. G., PentonVoak, I. S., & Perrett, D. I. (2010). Evidence for menstrual cycle shifts
in women’s preferences for masculinity: A response to Harris (in press)
“Menstrual cycle and facial preferences reconsidered.” Evolutionary
Psychology, 8, 768 –775.
DeBruine, L. M., Jones, B. C., & Perrett, D. I. (2005). Women’s attractiveness judgments of self-resembling faces change across the menstrual
cycle. Hormones and Behavior, 47, 379 –383. doi:10.1016/j.yhbeh.2004
.11.006
Durante, K. M., Li, N. P., & Haselton, M. G. (2008). Changes in women’s
choice of dress across the ovulatory cycle: Naturalistic and laboratory
task-based evidence. Personality and Social Psychology Bulletin, 34,
1451–1460. doi:10.1177/0146167208323103
Duval, S., & Tweedie, R. (2000). Trim and fill: A simple funnel-plot-based
method of testing and adjusting for publication bias in meta-analysis.
Biometrics, 56, 455– 463. doi:10.1111/j.0006-341X.2000.00455.x
Eastwick, P. W., & Finkel, E. J. (2008). Sex differences in mate preferences revisited: Do people know what they initially desire in a romantic
partner? Journal of Personality and Social Psychology, 94, 245–264.
doi:10.1037/0022-3514.94.2.245
Eastwick, P. W., Luchies, L. B., Finkel, E. J., & Hunt, L. J. (2013). The
predictive validity of ideal partner preferences: A review and metaanalysis. Psychological Bulletin. Advance online publication. doi:
10.1037/a0032432
Fehring, R. J., & Gaska, N. (1998). Evaluation of the Lady Free Biotester®
in determining the fertile period. Contraception, 57, 325–328. doi:
10.1016/S0010-7824(98)00039-0
Fehring, R. J., Schneider, M., & Raviele, K. (2006). Variability in the
phases of the menstrual cycle. Journal of Obstetric, Gynecological, &
Neonatal Nursing, 35, 376 –384. doi:10.1111/j.1552-6909.2006.00051.x
ⴱ
Feinberg, D. R. (2012). [Ovulatory cycle shifts in women’s preferences for
vocal masculinity]. Unpublished raw data.
Feinberg, D. R., Jones, B. C., Law Smith, M. J., Moore, F. R., DeBruine,
L. M., Cornwell, R. E., . . . Perrett, D. I. (2006). Menstrual cycle, trait
estrogen level, and masculinity preferences in the human voice. Hormones and Behavior, 49, 215–222. doi:10.1016/j.yhbeh.2005.07.004
ⴱ
Fink, B. (2012). [Ovulatory cycle shifts in women’s preferences for facial
masculinity]. Unpublished raw data.
Fiske, P., Rintamaki, P. T., & Karvonen, E. (1998). Mating success in
lekking males: A meta-analysis. Behavioral Ecology, 9, 328 –338. doi:
10.1093/beheco/9.4.328
Flowe, H. D., Swords, E., & Rockey, J. C. (2012). Women’s behavioural
engagement with a masculine male heightens during the fertile window:
Evidence for the cycle shift hypothesis. Evolution and Human Behavior,
33, 285–290. doi:10.1016/j.evolhumbehav.2011.10.006
52
GILDERSLEEVE, HASELTON, AND FALES
Folstad, I., & Karter, A. J. (1992). Parasites, bright males, and the immunocompetence handicap. American Naturalist, 139, 603– 622.
ⴱ
Frost, P. (1994). Preference for darker faces in photographs at different
phases of the menstrual cycle: Preliminary assessment of evidence for a
hormonal relationship. Perceptual and Motor Skills, 79, 507–514. doi:
10.2466/pms.1994.79.1.507
Gangestad, S. W., & Garver-Apgar, C. E. (2013). The nature of female
sexuality: Insights into the dynamics of romantic relationships. In J. A.
Simpson & L. Campbell (Eds.), The Oxford handbook of close relationships (pp. 374 –399). New York, NY: Oxford University Press. doi:
10.1093/oxfordhb/9780195398694.013.0017
ⴱ
Gangestad, S. W., Garver-Apgar, C. E., Simpson, J. A., & Cousins, A. J.
(2007). Changes in women’s mate preferences across the ovulatory
cycle. Journal of Personality and Social Psychology, 92, 151–163.
doi:10.1037/0022-3514.92.1.151
Gangestad, S. W., & Simpson, J. A. (2000). The evolution of human
mating: Trade-offs and strategic pluralism. Behavioral and Brain Sciences, 23, 573–587. doi:10.1017/S0140525X0000337X
ⴱ
Gangestad, S. W., Simpson, J. A., Cousins, A. J., Garver-Apgar, C. E., &
Christensen, P. N. (2004). Women’s preferences for male behavioral
display change across the menstrual cycle. Psychological Science, 15,
203–207. doi:10.1111/j.0956-7976.2004.01503010.x
ⴱ
Gangestad, S. W., & Thornhill, R. (1998). Menstrual cycle variation in
women’s preferences for the scent of symmetrical men. Proceedings of
the Royal Society B: Biological Sciences, 265, 927–933. doi:10.1098/
rspb.1998.0380
Gangestad, S. W., & Thornhill, R. (2008). Human oestrus. Proceedings of
the Royal Society B: Biological Sciences, 275, 991–1000. doi:10.1098/
rspb.2007.1425
Gangestad, S. W., Thornhill, R., & Garver, C. E. (2002). Changes in
women’s sexual interests and their partner’s mate-retention tactics
across the ovulatory cycle: Evidence for shifting conflicts of interest.
Proceedings of the Royal Society B: Biological Sciences, 269, 975–982.
doi:10.1098/rspb.2001.1952
Gangestad, S. W., Thornhill, R., & Garver-Apgar, C. E. (2005a). Adaptations to ovulation: Implications for sexual and social behavior. Current
Directions in Psychological Science, 14, 312–316. doi:10.1111/j.09637214.2005.00388.x
Gangestad, S. W., Thornhill, R., & Garver-Apgar, C. E. (2005b). Women’s
sexual interests across the ovulatory cycle depend on primary partner
developmental instability. Proceedings of the Royal Society B: Biological Sciences, 272, 2023–2027. doi:10.1098/rspb.2005.3112
Gangestad, S. W., Thornhill, R., & Garver-Apgar, C. E. (2010a). Fertility
in the cycle predicts women’s interest in sexual opportunism. Evolution
and Human Behavior, 31, 400 – 411. doi:10.1016/j.evolhumbehav.2010
.05.003
Gangestad, S. W., Thornhill, R., & Garver-Apgar, C. E. (2010b). Men’s
facial masculinity predicts changes in their female partner’s sexual
interest across the ovulatory cycle, whereas men’s intelligence does not.
Evolution and Human Behavior, 31, 412– 424. doi:10.1016/j.evolhumbehav
.2010.06.001
ⴱ
Gangestad, S. W., Thornhill, R., & Garver-Apgar, C. E. (2011). [Ovulatory cycle shifts in women’s preferences for facial masculinity]. Unpublished raw data.
ⴱ
Garver-Apgar, C. E., & Gangestad, S. W. (2012). [Ovulatory cycle shifts
in women’s mate preferences]. Unpublished raw data.
Garver-Apgar, C. E., Gangestad, S. W., & Thornhill, R. (2008). Hormonal
correlates of women’s midcycle preference for the scent of symmetry. Evolution and Human Behavior, 29, 223–232. doi:10.1016/j
.evolhumbehav.2007.12.007
Garver-Apgar, C. E., Gangestad, S. W., Thornhill, R., Miller, R. D., & Olp,
J. (2006). Major histocompatibility complex alleles, sexual responsivity,
and unfaithfulness in romantic couples. Psychological Science, 17, 830 –
835. doi:10.1111/j.1467-9280.2006.01789.x
Geary, D. C. (2000). Evolution and proximate expression of human paternal investment. Psychological Bulletin, 126, 55–77. doi:10.1037/00332909.126.1.55
Guéguen, N. (2009a). Menstrual cycle phases and female receptivity to a
courtship solicitation: An evaluation in a nightclub. Evolution and Human Behavior, 30, 351–355. doi:10.1016/j.evolhumbehav.2009.03.004
Guéguen, N. (2009b). The receptivity of women to courtship solicitation
across the menstrual cycle: A field experiment. Biological Psychology,
80, 321–324. doi:10.1016/j.biopsycho.2008.11.004
Guermandi, E., Vegetti, W., Bianchi, M. M., Uglietti, A., Ragni, G., &
Crosignani, P. (2001). Reliability of ovulation tests in infertile women.
Obstetrics & Gynecology, 97, 92–96.
Hale, G. E., Zhao, X., Hughes, C. L., Burger, H. G., Robertson, D. M., &
Fraser, I. S. (2007). Endocrine features of menstrual cycles in middle
and late reproductive age and the menopausal transition classified according to the Staging of Reproductive Aging Workshop (STRAW)
staging system. Journal of Clinical Endocrinology & Metabolism, 92,
3060 –3067. doi:10.1210/jc.2007-0066
Harlow, S. D. (2000). Menstruation and menstrual disorders: The epidemiology of menstruation and menstrual dysfunction. In M. B. Goldman
& M. C. Hatch (Eds.), Women and health (pp. 99 –113). San Diego, CA:
Academic Press.
ⴱ
Harris, C. R. (2011). Menstrual cycle and facial preferences reconsidered.
Sex Roles, 64, 669 – 681. doi:10.1007/s11199-010-9772-8
Haselton, M. G., & Gangestad, S. W. (2006). Conditional expression of
women’s desires and men’s mate guarding across the ovulatory cycle.
Hormones and Behavior, 49, 509 –518. doi:10.1016/j.yhbeh.2005.10
.006
Haselton, M. G., & Miller, G. F. (2006). Women’s fertility across the cycle
increases the short-term attractiveness of creative intelligence. Human
Nature, 17, 50 –73. doi:10.1007/s12110-006-1020-0
Haselton, M. G., Mortezaie, M., Pillsworth, E. G., Bleske-Rechek, A., &
Frederick, D. A. (2007). Ovulatory shifts in human female ornamentation: Near ovulation, women dress to impress. Hormones and Behavior,
51, 40 – 45. doi:10.1016/j.yhbeh.2006.07.007
Havlíček, J., Roberts, S. C., & Flegr, J. (2005). Women’s preference for
dominant male odour: Effects of menstrual cycle and relationship status.
Biology Letters, 1, 256 –259. doi:10.1098/rsbl.2005.0332
Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in
the world. Behavioral and Brain Sciences, 33, 61– 83. doi:10.1017/
S0140525X0999152X
ⴱ
Hodges-Simeon, C. R., Gaulin, S. J. C., & Puts, D. A. (2010). Different
vocal parameters predict perceptions of dominance and attractiveness.
Human Nature, 21, 406 – 427. doi:10.1007/s12110-010-9101-5
Hromatko, I., Tadinac, M., & Prizmić, H. (2006). Women’s hormonal
status and mate value influence relationship satisfaction and perceived
male attractiveness. Psychological Topics, 15, 315–330.
Hughes, A. L., & Nei, M. (1988). Pattern of nucleotide substitution at
major histocompatibility complex class I loci reveals overdominant
selection. Nature, 335, 167–170. doi:10.1038/335167a0
Hughes, A. L., & Nei, M. (1989). Nucleotide substitution at major histocompatibility complex class II loci: Evidence for overdominant selection. Proceedings of the National Academy of Sciences of the United
States of America, 86, 958 –962.
ⴱ
Izbicki, E. V., & Johnson, K. J. (2010, June). Dark, tall, and handsome:
Evidence for a female increase in openness to other-race partners
during periods of high conception risk. Poster presented at the annual
meeting of the Human Behavior and Evolution Society, Eugene, OR.
ⴱ
Johnston, V. S., Hagel, R., Franklin, M., Fink, B., & Grammer, K. (2001).
Male facial attractiveness: Evidence for hormone-mediated adaptive
design. Evolution and Human Behavior, 22, 251–267. doi:10.1016/
S1090-5138(01)00066-6
Jones, B. C., DeBruine, L. M., Perrett, D. I., Little, A. C., Feinberg, D. R.,
& Law Smith, M. J. (2008). Effects of menstrual cycle phase on face
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
preferences. Archives of Sexual Behavior, 37, 78 – 84. doi:10.1007/
s10508-007-9268-y
ⴱ
Jones, B. C., Little, A. C., Boothroyd, L., DeBruine, L. M., Feinberg,
D. R., Law Smith, M. J., . . . Perrett, D. I. (2005). Commitment to
relationships and preferences for femininity and apparent health in faces
are strongest on days of the ovulatory cycle when progesterone level is
high. Hormones and Behavior, 48, 283–290. doi:10.1016/j.yhbeh.2005
.03.010
Jones, B. C., Perrett, D. I., Little, A. C., Boothroyd, L., Cornwell, R. E.,
Feinberg, D. R., & Moore, F. R. (2005). Menstrual cycle, pregnancy and
oral contraceptive use alter attraction to apparent health in faces. Proceedings of the Royal Society B: Biological Sciences, 272, 347–354.
doi:10.1098/rspb.2004.2962
Keddy-Hector, A. C. (1992). Mate choice in non-human primates. American Zoologist, 32, 62–70. doi:10.1093/icb/32.1.62
Klingenberg, C. P. (2003). Developmental instability as a research tool:
Using patterns offluctuating asymmetry to infer the developmental origins of morphological integration. In M. Polak (Ed.), Developmental
instability: Causes and consequences (pp. 427– 442). New York, NY:
Oxford University Press.
Knott, C. D., Thompson, E. M., & Stumpf, R. M. (2007). Sexual coercion
and mating strategies of wild Bornean orangutans [Abstract]. American
Journal of Physical Anthropology, 44(Suppl.), 145.
Koehler, N., Rhodes, G., & Simmons, L. W. (2002). Are human female
preferences for symmetrical male faces enhanced when conception is
likely? Animal Behaviour, 64, 233–238. doi:10.1006/anbe.2002.3063
ⴱ
Koehler, N., Rhodes, G., & Simmons, L. W. (2006). Do cyclic changes in
women’s face preferences target cues to long-term health? Social Cognition, 24, 641– 656. doi:10.1521/soco.2006.24.5.641
Laeng, B., & Falkenberg, L. (2007). Women’s pupillary responses to
sexually significant others during the hormone cycle. Hormones and
Behavior, 52, 520 –530. doi:10.1016/j.yhbeh.2007.07.013
Lancaster, J. B., & Alvarado, L. C. (2010, June). The hormonal platform
for conception in natural fertility populations: The co-occurrence of
lactation, ovulation, and gestation. Talk presented at the annual meeting
of the Human Behavior and Evolution Society, Eugene, Oregon.
Larson, C. M., Haselton, M. G., Gildersleeve, K. A., & Pillsworth, E. G.
(2013). Changes in women’s feelings about their romantic relationships
across the ovulatory cycle. Hormones and Behavior, 63, 128 –135.
doi:10.1016/j.yhbeh.2012.10.005
Larson, C. M., Pillsworth, E. G., & Haselton, M. G. (2012). Ovulatory
shifts in women’s attractions to primary partners and other men: Further
evidence of the importance of primary partner sexual attractiveness.
PLoS ONE, 7, e44456. doi:10.1371/journal.pone.0044456
Li, N. P., Bailey, J. M., Kenrick, D. T., & Linsenmeier, J. A. W. (2002).
The necessities and luxuries of mate preferences: Testing the tradeoffs.
Journal of Personality and Social Psychology, 82, 947–955. doi:
10.1037/0022-3514.82.6.947
Li, N. P., & Kenrick, D. T. (2006). Sex similarities and differences in
preferences for short-term mates: What, whether, and why. Journal of
Personality and Social Psychology, 90, 468 – 489. doi:10.1037/00223514.90.3.468
Li, N. P., Pillsworth, E. G., & Haselton, M. G. (2006). [Cycle shifts in
women’s mate preferences assessed using a mate dollars paradigm].
Unpublished raw data.
Little, A. C., Connely, J., Feinberg, D. R., Jones, B. C., & Roberts, S. C.
(2011). Human preference for masculinity differs according to context in
faces, bodies, voices, and smell. Behavioral Ecology, 22, 862– 868.
doi:10.1093/beheco/arr061
ⴱ
Little, A. C., Jones, B. C., & Burriss, R. P. (2007). Preferences for
masculinity in male bodies change across the ovulatory cycle. Hormones
and Behavior, 51, 633– 639. doi:10.1016/j.yhbeh.2007.03.006
ⴱ
53
Little, A. C., Jones, B. C., Burt, D. M., & Perrett, D. I. (2007). Preferences
for symmetry in faces change across the menstrual cycle. Biological
Psychology, 76, 209 –216. doi:10.1016/j.biopsycho.2007.08.003
ⴱ
Little, A. C., Jones, B. C., & DeBruine, L. M. (2008). Preferences for
variation in masculinity in real faces change across the menstrual cycle:
Women prefer more masculine faces when they are more fertile. Personality and Individual Differences, 45, 478 – 482. doi:10.1016/j.paid
.2008.05.024
ⴱ
Luevano, V. X., & Zebrowitz, L. A. (2006, June). When fighters are
lovers too: Predictors of attractiveness as a function of partner-type and
fertility level. Poster presented at the annual meeting of the Human
Behavior and Evolution Society, Philadelphia, PA.
Lukaszewski, A. W., & Roney, J. R. (2009). Estimated hormones predict
women’s mate preferences for dominant personality traits. Personality
and Individual Differences, 47, 191–196. doi:10.1016/j.paid.2009.02
.019
Matsumoto-Oda, A. (1999). Female choice in the opportunistic mating of
wild chimpanzees (Pan troglodytes schweinfurthii) at Mahale. Behavioral Ecology and Sociobiology, 46, 258 –266.
McClellan, M. E., Blackwell, A. D., Garrett, E. L., & Sugiyama, L. S.
(2007, May). Women’s preferences for male body types in long-term and
short-term mating contexts. Paper presented at the annual meeting of the
Human Behavior and Evolution Society, Williamsburg, VA.
ⴱ
McDonald, M. M., & Navarrete, C. D. (2012). [Ovulatory cycle shifts in
women’s preferences for same-race versus other-race male avatars].
Unpublished raw data.
Miller, G. F. (2000). Sexual selection for indicators of intelligence. In G. R.
Bock, J. A. Goode, & K. Webb (Eds.), The nature of intelligence (pp.
260 –275). New York, NY: Oxford University Press. doi:10.1002/
0470870850.ch16
ⴱ
Miller, G. (2003, June). A great sense of humor is a good genes indicator:
Ovulatory cycle effects on the sexual attractiveness of male humor
ability. Poster presented at the annual meeting of the Human Behavior
and Evolution Society, Lincoln, NE.
Møller, A. P., & Thornhill, R. (1998). Bilateral symmetry and sexual
selection: A meta-analysis. American Naturalist, 151, 174 –192. doi:
10.1086/286110
ⴱ
Moore, F. R. (2011). [Ovulatory cycle shifts in women’s face preferences]. Unpublished raw data.
ⴱ
Moore, F. R., Cornwell, R. E., Law Smith, M. J., Al Dujaili, E. A. S.,
Sharp, M., & Perrett, D. I. (2011). Evidence for the stress-linked immunocompetence handicap hypothesis in human male faces. Proceedings of
the Royal Society B: Biological Sciences, 278, 774 –780. doi:10.1098/
rspb.2010.1678
ⴱ
Morrison, E. R., Clark, A. P., Gralewski, L., Campbell, N., & PentonVoak, I. S. (2010). Women’s probability of conception is associated with
their preference for flirtatious but not masculine facial movement. Archives of Sexual Behavior, 39, 1297–1304. doi:10.1007/s10508-0099527-1
Nassaralla, C. L., Stanford, J. B., Daly, D. K., Schneider, M., Schliep,
K. C., & Fehring, R. J. (2011). Characteristics of the menstrual cycle
after discontinuation of oral contraceptives. Journal of Women’s Health,
20, 169 –177. doi:10.1089/jwh.2010.2001
Navarrete, C. D., Fessler, D. M. T., Santos Fleischman, D., & Geyer, J.
(2009). Race bias tracks conception risk across the menstrual cycle.
Psychological Science, 20, 661– 665. doi:10.1111/j.1467-9280.2009
.02352.x
O’Connell, A. A., & McCoach, D. B. (Eds.). (2008). Multilevel modeling
of educational data. Greenwich, CT: Information Age.
Oinonen, K. A., Klemencic, N., & Mazmanian, D. (2008). The periovulatory sociosexuality tactic shift (PSTS): Activational hormonal mechanisms in two female sexual strategies. In G. A. Conti (Ed.), Progress in
biological psychology research (pp. 139 –158). Hauppauge, NY: Nova
Science.
54
ⴱ
GILDERSLEEVE, HASELTON, AND FALES
Oinonen, K. A., & Mazmanian, D. (2007). Facial symmetry detection
ability changes across the ovulatory cycle. Biological Psychology, 75,
136 –145. doi:10.1016/j.biopsycho.2007.01.003
ⴱ
Pawlowski, B., & Jasienska, G. (2005). Women’s preferences for sexual
dimorphism in height depend on menstrual cycle phase and expected
duration of relationship. Biological Psychology, 70, 38 – 43. doi:
10.1016/j.biopsycho.2005.02.002
Penn, D. J., Damjanovich, K., & Potts, W. K. (2002). MHC heterozygosity
confers a selective advantage against multiple-strain infections. Proceedings of the National Academy of Sciences of the United States of
America, 99, 11260 –11264. doi:10.1073/pnas.162006499
ⴱ
Penton-Voak, I. S., & Perrett, D. I. (2000). Female preference for male
faces change cyclically: Further evidence. Evolution and Human Behavior, 21, 39 – 48. doi:10.1016/S1090-5138(99)00033-1
ⴱ
Penton-Voak, I. S., Perrett, D. I., Castles, D. L., Kobayashi, T., Burt,
D. M., Murray, K., & Minamisawa, R. (1999). Menstrual cycle alters
face preference. Nature, 399, 741–742. doi:10.1038/21557
ⴱ
Perrett, D. I., Jones, B. C., Little, A. C., Boothroyd, L., Cornwell, R. E.,
Feinberg, D. R., . . . Stirrat, M. R. (2013). Pregnancy, the menstrual
cycle, and hormonal contraceptive use alter attraction to apparent
health in faces. Unpublished manuscript.
Peters, M., Rhodes, G., & Simmons, L. W. (2008). Does attractiveness in
men provide clues to semen quality? Journal of Evolutionary Biology,
21, 572–579. doi:10.1111/j.1420-9101.2007.01477.x
Peters, M., Simmons, L. W., & Rhodes, G. (2009). Preferences across the
menstrual cycle for masculinity and symmetry in photographs of male
faces and bodies. PLoS ONE, 4, e4138. doi:10.1371/journal.pone
.0004138
Pieta, K. (2008). Female mate preferences among Pan troglodytes schweinfurthii of Kanyawara, Kibale National Park, Uganda. International Journal of Primatology, 29, 845– 864. doi:10.1007/s10764-008-9282-5
Pillsworth, E. G., & Haselton, M. G. (2006a). Male sexual attractiveness
predicts differential ovulatory shifts in female extra-pair attraction and
male mate retention. Evolution and Human Behavior, 27, 247–258.
doi:10.1016/j.evolhumbehav.2005.10.002
Pillsworth, E. G., & Haselton, M. G. (2006b). Women’s sexual strategies:
The evolution of long-term bonds and extrapair sex. Annual Review of
Sex Research, 17, 59 –100. doi:10.1080/10532528.2006.10559837
Prokosch, M. D., Coss, R. G., Scheib, J. E., & Blozis, S. A. (2009).
Intelligence and mate choice: Intelligent men are always appealing.
Evolution and Human Behavior, 30, 11–20. doi:10.1016/j.evolhumbehav
.2008.07.004
Provost, M. P., Troje, N. F., & Quinsey, V. L. (2008). Short-term mating
strategies and attraction to masculinity in point-light walkers. Evolution
and Human Behavior, 29, 65– 69. doi:10.1016/j.evolhumbehav.2007.07
.007
ⴱ
Puts, D. A. (2005). Mating context and menstrual phase affect women’s
preferences for male voice pitch. Evolution and Human Behavior, 26,
388 –397. doi:10.1016/j.evolhumbehav.2005.03.001
Puts, D. A. (2010). Beauty and the beast: Mechanisms of sexual selection
in humans. Evolution and Human Behavior, 31, 157–175. doi:10.1016/
j.evolhumbehav.2010.02.005
ⴱ
Rantala, M. J., Eriksson, C. J. P., Vainikka, A., & Kortet, R. (2006). Male
steroid hormones and female preference for male body odor. Evolution
and Human Behavior, 27, 259 –269. doi:10.1016/j.evolhumbehav.2005
.11.002
ⴱ
Rantala, M. J., Polkki, M., & Rantala, L. M. (2010). Preferences for
human male body hair changes across the menstrual cycle and menopause. Behavioral Ecology, 21, 419 – 423. doi:10.1093/beheco/arp206
Raudenbush, S. W., & Bryk, A. S. (1985). Empirical Bayes meta-analysis.
Journal of Educational and Behavioral Statistics, 10, 75–98. doi:
10.3102/10769986010002075
Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models (2nd
ed.). Thousand Oaks, CA: Sage.
Regan, P. C. (1998). What if you can’t get what you want? Willingness to
compromise ideal mate selection standards as a function of sex, mate
value, and relationship context. Personality and Social Psychology Bulletin, 24, 1294 –1303. doi:10.1177/01461672982412004
Rhodes, G. (2006). The evolutionary psychology of facial beauty. Annual
Review of Psychology, 57, 199 –226. doi:10.1146/annurev.psych.57
.102904.190208
Rikowski, A., & Grammer, K. (1999). Human body odour, symmetry, and
attractiveness. Proceedings of the Royal Society B: Biological Sciences,
266, 869 – 874. doi:10.1098/rspb.1999.0717
Roberts, M. L., Buchanan, K. L., & Evans, M. R. (2004). Testing the
immunocompetence handicap hypothesis: A review of the evidence.
Animal Behaviour, 68, 227–239. doi:10.1016/j.anbehav.2004.05.001
Roney, J. R., Hanson, K. N., Durante, K. M., & Maestripieri, D. (2006).
Reading men’s faces: Women’s mate attractiveness judgments track
men’s testosterone and interest in infants. Proceedings of the Royal
Society B: Biological Sciences, 273, 2169 –2175. doi:10.1098/rspb.2006
.3569
ⴱ
Roney, J. R., & Simmons, Z. L. (2008). Women’s estradiol predicts
preference for facial cues of men’s testosterone. Hormones and Behavior, 53, 14 –19. doi:10.1016/j.yhbeh.2007.09.008
ⴱ
Roney, J. R., Simmons, Z. L., & Gray, P. B. (2011). Changes in estradiol
predict within-women shifts in attraction to facial cues of men’s testosterone. Psychoneuroendocrinology, 36, 742–749. doi:10.1016/j
.psyneuen.2010.10.010
Rowe, L., & Houle, D. (1996). The lek paradox and the capture of genetic
variance by condition dependent traits. Proceedings of the Royal Society
B: Biological Sciences, 263, 1415–1421. doi:10.1098/rspb.1996.0207
Rupp, H. A., James, T. W., Ketterson, E. D., Sengelaub, D. R., Janssen, E.,
& Heiman, J. R. (2009). Neural activation in women in response to
masculinized male faces: Mediation by hormones and psychosexual
factors. Evolution and Human Behavior, 30, 1–10. doi:10.1016/j
.evolhumbehav.2008.08.006
ⴱ
Rupp, H., Librach, G. R., Feipel, N. C., Ketterson, E. D., Sengelaub,
D. R., & Heiman, J. R. (2009). Partner status influences women’s
interest in the opposite sex. Human Nature, 20, 93–104. doi:10.1007/
s12110-009-9056-6
Said, C. P., & Todorov, A. (2011). A statistical model of facial attractiveness. Psychological Science, 22, 1183–1190. doi:10.1177/
0956797611419169
Sear, R., & Mace, R. (2008). Who keeps children alive? A review of the
effects of kin on child survival. Evolution and Human Behavior, 29,
1–18. doi:10.1016/j.evolhumbehav.2007.10.001
Sell, A., Cosmides, L., Tooby, J., Sznycer, D., von Rueden, C., & Gurven,
M. (2009). Human adaptations for the visual assessment of strength and
fighting ability from the body and face. Proceedings of the Royal Society
B: Biological Sciences, 276, 575–584. doi:10.1098/rspb.2008.1177
Simmons, J. P., Nelson, L. D., & Simonsohn, U. (2011). False-positive
psychology: Undisclosed flexibility in data collection and analysis allows presenting anything as significant. Psychological Science, 22,
1359 –1366. doi:10.1177/0956797611417632
ⴱ
Singh, D., & Bailey, D. (2006, June). Do women’s judgments of male
attractiveness based on testosterone-mediated body parts depend on
menstrual cycle phase? Poster presented at the annual meeting of the
Human Behavior and Evolution Society, Philadelphia, PA.
Soler, C., Núñez, M. Gutiérrez, R., Núñez, J., Medina, P. Sancho, M., &
Núñez, A. (2003). Facial attractiveness in men provides clues to semen
quality. Evolution and Human Behavior, 24, 199 –207. doi:10.1016/
S1090-5138(03)00013-8
Strassmann, B. I. (1997). The biology of menstruation in Homo sapiens:
Total lifetime menses, fecundity, and nonsynchrony in a natural-fertility
population. Current Anthropology, 38, 123–129. doi:10.1086/204592
Stumpf, R. M., & Boesch, C. (2005). Does promiscuous mating preclude
female choice? Female sexual strategies in chimpanzees (Pan troglo-
META-ANALYSIS OF CYCLE SHIFTS IN MATE PREFERENCES
dytes verus) of the Taï National Park, Côte d’Ivoire. Behavioral Ecology
and Sociobiology, 57, 511–524. doi:10.1007/s00265-004-0868-4
Teatero, M. (2009). Mating strategies across the menstrual cycle: Preferences, jealousy, and masculinity (Unpublished thesis). Lakehead University, Thunder Bay, Ontario, Canada.
Testart, J., & Frydman, R. (1982). Minimum time lapse between luteinizing hormone surge or human chorionic gonadotrophin administration
and follicular rupture. Fertility and Sterility, 37, 50 –53.
ⴱ
Thornhill, R., Chapman, J. F., & Gangestad, S. W. (2013). Women’s
preferences for men’s scents associated with testosterone and cortisol
levels: Patterns across the ovulatory cycle. Evolution and Human Behavior, 34, 216 –221. doi:10.1016/j.evolhumbehav.2013.01.003
Thornhill, R., & Gangestad, S. W. (1994). Human fluctuating asymmetry
and sexual behavior. Psychological Science, 5, 297–302. doi:10.1111/j
.1467-9280.1994.tb00629.x
Thornhill, R., & Gangestad, S. W. (1999a). Facial attractiveness. Trends in
Cognitive Sciences, 3, 452– 460. doi:10.1016/S1364-6613(99)01403-5
ⴱ
Thornhill, R., & Gangestad, S. W. (1999b). The scent of symmetry: A
human sex pheromone that signals fitness? Evolution and Human Behavior, 20, 175–201. doi:10.1016/S1090-5138(99)00005-7
Thornhill, R., & Gangestad, S. W. (2008). The evolutionary biology of
human female sexuality. New York, NY: Oxford University Press.
ⴱ
Thornhill, R., Gangestad, S. W., Miller, R., Scheyd, G., McCollough,
J. K., & Franklin, M. (2003). Major histocompatibility complex genes,
symmetry, and body scent attractiveness in men and women. Behavioral
Ecology, 14, 668 – 678. doi:10.1093/beheco/arg043
Thornhill, R., & Møller, A. P. (1997). Developmental stability, disease, and
medicine. Biological Reviews, 72, 497–548. doi:10.1111/j.1469-185X
.1997.tb00022.x
Todd, P. M., Penke, L., Fasolo, B., & Lenton, A. P. (2007). Different
cognitive processes underlie human mate choices and mate preferences.
Proceedings of the National Academy of Sciences of the United States of
America, 104, 15011–15016. doi:10.1073/pnas.0705290104
Van Dongen, S. (2006). Fluctuating asymmetry and developmental instability in evolutionary biology: Past, present, and future. Journal of
Evolutionary Biology, 19, 1727–1743. doi:10.1111/j.1420-9101.2006
.01175.x
Van Dongen, S., & Gangestad, S. W. (2011). Human fluctuating asymmetry in relation to health and quality: A meta-analysis. Evolution and
55
Human Behavior, 32, 380 –398. doi:10.1016/j.evolhumbehav.2011.03
.002
ⴱ
Vaughn, J. E., Bradley, K. I., Byrd-Craven, J., & Kennison, S. M. (2010).
The effect of mortality salience on women’s judgments of male faces.
Evolutionary Psychology, 8, 477– 491.
von Rueden, C., Gurven, M., & Kaplan, H. (2011). Why do men seek high
social status? Fitness payoffs to dominance and prestige. Proceedings of
the Royal Society B: Biological Sciences, 278, 2223–2232. doi:10.1098/
rspb.2010.2145
Wegienka, G., & Day Baird, D. (2005). A comparison of recalled date of
last menstrual period with prospectively recorded dates. Journal of
Women’s Health, 14, 248 –252. doi:10.1089/jwh.2005.14.248
ⴱ
Welling, L. L. M., Jones, B. C., DeBruine, L. M., Conway, C. A., Law
Smith, M. J., Little, A. C., & Al-Dujaili, E. A. S. (2007). Raised salivary
testosterone in women is associated with increased attraction to masculine faces. Hormones and Behavior, 52, 156 –161. doi:10.1016/j.yhbeh
.2007.01.010
Wilcox, A. J., Dunson, D. B., Weinberg, C. R., Trussell, J., & Day Baird,
D. (2001). Likelihood of conception with a single act of intercourse:
Providing benchmark rates for assessment of post-coital contraceptives.
Contraception, 63, 211–215. doi:10.1016/S0010-7824(01)00191-3
Wilcox, A. J., Weinberg, C. R., & Day Baird, D. (1995). Timing of sexual
intercourse in relation to ovulation: Effects on the probability of conception, survival of pregnancy, and sex of the baby. New England
Journal
of
Medicine,
333,
1517–1521.
doi:10.1056/
NEJM199512073332301
Wilson, E. K., Rogler, J. C., & Erb, R. E. (1979). Effect of sexual
experience, location, malnutrition, and repeated sampling on concentrations of testosterone in blood plasma of Gallus domesticus roosters.
Poultry Science, 58, 178 –186.
Wood, D., & Brumbaugh, C. C. (2009). Using revealed mate preferences
to evaluate market force and differential preference explanations for
mate selection. Journal of Personality and Social Psychology, 96, 1226 –
1244. doi:10.1037/a0015300
Received March 2, 2013
Revision received October 16, 2013
Accepted October 23, 2013 䡲