Abstract
Previous research has shown that during her monthly peak fertile window, a woman competes with other women for a suitable mate. Drawing upon research on ovulation and socially constructed meanings of the color red, we examine how a woman’s fertility status and red clothing worn by a target woman change perceptions of the target, as well as behaviors toward the target. Following previous research on the ovulatory status and color red effects, we rely on both hormonal and self-reported fertility data. Across six studies, our research fails to provide support for the prediction that an ovulating woman is less likely to trust another woman wearing red compared with a nonovulating woman.
There are only a few days in a woman’s ovulatory cycle when she is fertile. According to the ovulatory shift hypothesis, in this time women’s mating psychology, behaviors, attitudes, preferences, and motivations change in ways that enhance their mating outcomes (e.g., Gangestad & Thornhill, 1998, 2008). Although controversial (e.g., Gildersleeve, Haselton, & Fales, 2014; Harris, 2011, 2013; Harris, Chabot, & Mickes, 2013; Wood, Kressel, Joshi, & Louie, 2014), the literature has documented shifts in women’s mate preferences at ovulation. For example, ovulating women prefer men who appear to have genetic superiority over other men, manifesting in more masculine facial features (Johnston, Hagel, Franklin, Fink, & Grammer, 2001; Penton-Voak & Perrett, 2000), a deeper voice (Puts, 2005), and certain kinds of scents (e.g., Gangestad & Thornhill, 1998; Havlicek, Roberts, & Flegr, 2005; Rikowski & Grammer, 1999; Thornhill et al., 2003).
Emerging literature, however, shows that ovulation also corresponds with women’s motivation to compete with other women for mates (e.g., Durante, Griskevicius, Cantú, & Simpson, 2014; Durante, Griskevicius, Hill, Perilloux, & Li, 2011). For example, ovulation increases women’s desire to dress sexy when they learn that attractive rivals are nearby (Durante et al., 2011). In addition to self-promotion, rival derogation is another strategy used to increase mating success (e.g., Buss, 1998; Buss & Dedden, 1990). It appears that the effects related to fertility status of a woman may be observed not only in the context of mating choice and preferences but also in interpersonal perception and behavior such as the general attitudes toward the same-sex rival (Durante et al., 2014; Maner, DeWall, & Gailliot, 2008).
Contributing to this literature, we explore how trust as well as perceptions of other women might change throughout a woman’s ovulatory cycle. We further examine conditions under which the change in behaviors and perceptions during a woman’s fertility window is more likely to happen. Although, as we discuss below, the fertility literature on intrasex competition would predict that an ovulating woman would treat other women generally as rivals, we expect that there are signals in other women’s appearance or behavior that ovulating women might find especially threatening, thus leading ovulating women to treat these other women poorly. Specifically, we rely on extensive literature on effects of color (Elliot & Maier, 2012, 2014) and focus on effects of another woman wearing red (relative to other colors) on ovulating women behavior and attitude toward her.
We examine ovulating women’s behaviors and perceptions of another woman dressed in red, insofar as making a case for an interactive effect between fertility status and the color of another woman’s clothes. We test our predictions in a set of studies, including hormonal and preregistered studies, and conclude by discussing implications of our research and future directions.
Trust and Related Perceptions
From an evolutionary standpoint, it is rational for an ovulating woman to avoid a threatening rival who is likely to compete with her for desirable mates (e.g., Benenson, 2009). According to Benenson (2009), avoidance can serve as a useful strategy in response to the rival because it helps avoid risky fights and defensive aggression that would deplete one of the resources necessary for interacting and mating with a potential partner. In the current research, we focus on resource allocation and trust as one manifestation of caution around the rival. Evolutionary perspective posits that trusting behavior can often be adaptive; individuals and mammals often mutually benefit with a greater likelihood of survival and procreation by cooperating than by not cooperating with each other (e.g., Bateson, 2000). Examples include penguins huddling to conserve warmth, primates grooming each other, and male lions protecting their female partners.
Bateson (2000) does note, however, that because life often involves conflict and intense competition, the balance between conflicting pressures of cooperating (and therefore trusting) and competing is delicate and may favor competition under certain conditions. Accordingly, at times of intense motivation to attract a mate, evolutionary processes guide a woman to stay away from a female rival and thus trust her less and allocate fewer resources to her. In addition to trust, evolutionary literature suggests that perceptions of another woman might also change throughout a woman’s ovulatory cycle. To that end, an important feature of our research involves examining a set of perceptions (mating-related and general) of another woman dressed in red.
First, we focus on mating-related perceptions of attractiveness and dominance. These are the two primary mating characteristics that women and men tend to focus on, as identified by evolutionary psychology research. Specifically, because men are attracted to physically attractive women (Buss, 1988, 1994; Kenrick & Keefe, 1992; Li, Bailey, Kenrick, & Linsenmeier, 2002), women are more likely to be threatened by an attractive rival (Dijkstra & Buunk, 1998). Compared with attractiveness, dominance perceptions have been found to play less of a role in women’s mating success (e.g., Gutierres, Kenrick, & Partch, 1999; Maner et al., 2008), but dominance has been traditionally shown to affect men’s mating success as women are not only attracted to dominant men but the latter are also able to outcompete rivals (e.g., Gutierres et al., 1999; Maner et al., 2008).
However, there might be an exception to these generalizations during a woman’s ovulation: The increased competitiveness with other women during a woman’s ovulation may manifest in her selectively attending to dominance cues displayed by other women. Previous research suggests that a female’s detection of a rival’s dominance may be critical for ensuring reproductive success, an important goal during ovulation. In nonhuman primates, for example, dominant females occupy more central positions in their groups (Öst, Jaatinen, & Steele, 2007; Ron, Henzi, & Motro, 1996) and enjoy greater status: The dominant female consistently wins contests, while the less dominant one loses regularly and may use formal signals of submission to indicate her subordinate position (de Waal, 1989). Similarly, research shows that female dominance is associated with greater reproductive success, more rapid production of the young, and greater infant survival rate (Pusey, Williams, & Goodall, 1997). As such, during the time of intense competition for mates, women might be more attuned to both other women’s attractiveness and dominance.
Furthermore, in addition to the mating perceptions, we seek to examine general person-perception dimensions. Warmth and competence are the two primary dimensions that underlie person perception (Fiske, Cuddy, & Glick, 2007; Fiske, Cuddy, Glick, & Xu, 2002). Due to evolutionary pressures, in social encounters, individuals must immediately determine the intentions of others (i.e., warmth perception) and their ability to act on those intentions (i.e., competence perception). Warmth captures traits that are related to perceived intent (e.g., friendliness, helpfulness, sincerity) and competence captures traits that are related to perceived ability (e.g., intelligence, skill, efficacy) (Fiske et al., 2007). “These dimensions provide fundamental social structural answers about competition and status” and elicit positive or negative emotions and behavior (Fiske et al., 2007, p. 77). Accordingly, in addition to attractiveness and dominance, perceptions of warmth and competence might similarly matter at times when a woman is competing with other women for mates.
Having delineated behaviors and perceptions of interest to our research question, we now turn to, first, reviewing the literature on intrasexual competition as predicted by the fertility status and the color red and, second, hypothesizing the sign of the effects of fertility status and color red on trust and related perceptions.
Fertility Status
The literature on ovulation has many controversies associated with it. These include, for example, proneness to experimenter degrees of freedom wherein experimenters increase the likelihood of finding their predicted effects by varying methodological and analytical approaches across the studies (e.g., Harris et al., 2013) as well as a lack of results’ replicability (e.g., Harris, 2011, 2013). Nevertheless, this literature has been very consistent in demonstrating an ovulating woman treating other women. For example, ovulating women have been shown to give less money to other women during a dictator game (Durante et al., 2014), elevate their status by focusing more on their appearance in the presence of other women (Durante et al., 2011), and rate other women lower in attractiveness (Fisher, 2004) and less likely to share a reward (Lucas, Koff, & Skeath, 2007). As such, due to the salience of the mating motive (a nonconscious motive) for an ovulating woman, it is very likely that she shows a preference for rival derogation strategies when facing other women (e.g., Buss, 1998; Buss & Dedden, 1990). Consistent with this body of literature, we test for a main effect of a woman’s fertility status on resource allocation and trust as well as perceptions toward other women, such that ovulating women are less likely to trust other women and have lower perceptions of their warmth, competence, and attractiveness as well as higher perceptions of their dominance compared with nonovulating women.
Red in Competitive Contexts
Whereas the link between a woman’s fertility status and intrasexual behavior is largely straightforward, the same cannot be said about the color red and intrasexual behavior. Over the past two decades, research on how color affects psychological functioning affectively, cognitively, and behaviorally has boomed (for a review, see Elliot & Maier, 2014). The color-in-context theory (Elliot & Maier, 2012) essentially posits that color carries not only aesthetic value but also symbolic value. In other words, in addition to evoking “like/dislike” perceptions, colors have associations to everyday experiences and, accordingly, carry meaning.
Despite some controversies associated with this research (e.g., Elliot & Maier, 2013; Francis, 2013), one of the most commonly studied colors is red. Importantly, the color is displayed by many animals in reproductive contexts as well as worn by humans in many contexts. Nonhumans, for example, are believed to have red coloration to attract mates (Deschner, Heistermann, Hodges, & Boesch, 2004). Indeed, many primates display red on their chest and genitalia specifically when they are in their ovulatory period (e.g., Dixson, 1983; Gerald, 2003; Setchell & Wickings, 2005). Similarly, previous research shows that women, in particular, wear red to signal motivation for sex and romance (e.g., Beall & Tracy, 2013; Eisenbruch, Simmons, & Roney, 2015; Elliot & Niesta, 2008; Guéguen, 2012a; Pazda, Elliot, & Greitemeyer, 2012; Roberts, Owen, & Havlicek, 2010).
Importantly, perceptions and behaviors toward the person displaying or wearing the color red differ depending on the context in which the color red is encountered. The context can be either affiliative—resulting in sexual desire—or threat-inducing—motivating avoidance (Elliot & Maier, 2012). In the case of the former, for example, research showed that red has been associated with higher perception of female attractiveness among male perceivers (e.g., Elliot & Niesta, 2008; Guéguen & Jacob, 2014; Niesta Kayser, Elliot, & Feltman, 2010; Pazda et al., 2012; Pazda, Elliot, & Greitemeyer, 2014). For instance, men are more likely to contact a woman wearing red on a dating website (Guéguen & Jacob, 2013) and approach a woman wearing red lipstick at a bar (Guéguen, 2012b). What explains this red-attractiveness effect for men is their perception of a woman’s sexual receptivity (Guéguen, 2012b; Pazda, Elliot, & Greitemeyer, 2014).
On the contrary, in competitive situations—as is the case when an ovulating woman competes with others for a mate—the color-in-context theory (Elliot & Maier, 2012) posits that the color red is associated with imminent danger and avoidance motivations, associations that are both biologically based and learned. For example, Khan, Levine, Dobson, and Kralik (2011) found that a sample of rhesus macaques, a species of monkey with humanlike color vision, avoided a human experimenter wearing red when given the opportunity to steal food, regardless of the sex of that experimenter. Similarly, in humans, research shows that in competitive contexts, opponents wearing red are perceived to be more dominant and assertive (Feltman & Elliot, 2011; Little & Hill, 2007; Sorokowski & Szmajke, 2007). For instance, when Feltman and Elliot asked participants to imagine having a taekwondo match against another same-sex person dressed in either red or blue, both male and female participants rated those wearing red as more dominant and threat-inducing. Importantly, it appears that the definition of “competitive context” spans quite wide, as recently Pazda, Prokop, and Elliot (2014) have demonstrated that even when women view a picture of another woman wearing red, they derogate her by calling her sexually receptive, questioning her fidelity, and guarding their own mate from her. As such, a woman–woman interaction context proves also to be competitive. Thus, we expect an effect of the color red on resource allocation and trust as well as perceptions of other women, such that another woman wearing red would trigger lower trust behaviors and lower perceptions of warmth, competence, attractiveness as well as high perceptions of dominance, compared with when the other woman is wearing any other color.
Interactive Effects of Fertility Status and Color Red
Relying on past literature, we expect main effects of fertility status and the color red on a woman’s perception of another woman. We now turn to making a case for interactive effect between the two. Specifically, previous research indicates that one way an ovulating woman makes herself visible to males and her mating motives salient is by wearing red or pink clothing (Beall & Tracy, 2013; Eisenbruch et al., 2015). Men will interpret this signal as an indication of sexual readiness and fertility (Pazda et al., 2012). Women’s interpretations, on the contrary, would likely depend on their own fertility status. To the extent that women are ovulating and thus are interested in attracting a man, they will likely perceive the color red on another woman as more threatening than nonovulating women. We thus argue that red clothing on another woman as compared with another color is likely to trigger stronger negative perceptions and avoidant behaviors toward that woman from an ovulating woman (vs. a nonovulating woman). In sum, we argue that ovulation—as an example of contexts or internal states wherein mating motives are salient—influences behaviors toward and perceptions of another woman wearing red.
For example, with respect to trust, in competitive contexts the color red has been linked to avoidance motivation (Maier, Elliot, & Lichtenfeld, 2008; Mehta & Zhu, 2009), suggesting that people tend to prefer not to befriend the rivals wearing red. As such, we argue that red clothes worn by another woman can serve as a threat cue, leading an ovulating woman to treat the woman in red poorly. Specifically, we expect an ovulating woman (vs. nonovulating) to be more likely to allocate fewer resources and display lower level of interpersonal trust toward another woman dressed in red compared with another woman dressed in any other color.
Similarly, provided our earlier discussion of another female derogation as suggested by the color red literature and research on fertility, we argue that during the period of high fertility, any cue (such as the color red) in other woman signaling their appeal to men (and thereby a greater threat to the ovulating woman) would lead the ovulating woman to derogate the female rival by rating her lower in attractiveness because one common strategy females use to derogate a female rival is to say “rival is ugly” and point out flaws in her physical appearance (Buss, 1998; Buss & Dedden, 1990; Fisher & Cox, 2011).
With respect to dominance, given that dominance proves advantageous for the rivals and detrimental to women around them in terms of reproduction success, it seems logical that being aware of and attuned to signs of dominance in another woman would help a woman’s own intrasexual competitive interests, as being able to outcompete rivals might be more possible if she is aware of them by avoiding them. Thus, we expect that ovulating women would be more sensitive to red clothing on another woman and would identify that woman as more dominant, and thus we expect that an ovulating woman (vs. nonovulating) would be more likely to designate another woman dressed in red (vs. another color) as more dominant.
Finally, it is conceivable, then, that to the extent trust toward the woman wearing red (vs. another color) is affected during ovulation by the change in mating-related perceptions, so will be the general person perceptions, warmth and competence. More specifically, support for this prediction comes from research on intrasexual competition, which suggests that one way a woman can derogate her rival is by negatively evaluating her intelligence and ability, as well as by “pointing out rival is dumb” (Buss, 1998; Buss & Dedden, 1990; Fisher & Cox, 2011, p. 36). As such, we argue that an ovulating woman (vs. nonovulating) would be more likely to designate another woman dressed in red (vs. another color) as less competent and less warm.
Overview of Studies
We tested our predictions in six studies in which we presented female participants with a picture of another female. In Study 1, we examined whether the color of another woman’s clothes determines ovulating (vs. nonovulating) woman’s trusting behavior toward her. In Study 2, we attempted to replicate our results in a different context by examining the resource allocation of an ovulating (vs. nonovulating) woman in a managerial role to an applicant wearing red (vs. blue) clothes. Both studies, therefore, explored the interactive and separate effects of fertility status and color red on trust-related behavior. Next, in Studies 3 and 4, we explored perceptions affected by the color red (vs. blue) and fertility status, as well as a general attitude measured by liking. In Study 5, we further examined whether the color of the other woman’s clothes (red vs. gray) influenced ovulating (vs. nonovulating) women’s perceptions as well as interpersonal trust. Finally, Study 6 is a replication of Study 5, following our preregistration at Open Science Framework.
We report all participants recruited, all conditions, and all measures in each of the studies. Across studies, we report how we determined our sample sizes prior to conducting each study. In every study, we test for main effects of the ovulatory status and color red, as well as the interactive effect of ovulatory status and the clothes’ color the target woman is wearing on participants’ behaviors toward the target woman and perceptions of the target woman.
Study 1
The main goal of Study 1 was to test main effects and interactive effect of fertility status and color red, by exploring whether ovulating women behave differently toward women wearing red (vs. another color). Specifically, we measured women’s behavior toward another woman in a standard trust game with a real behavioral outcome. In addition, provided the recent controversy regarding the use of self-report methods of detecting ovulation (see Wood et al., 2014), we began the test of our predictions with hormonal data.
Method
Participants and design
We recruited 82 female U.S. college students (Mage = 23.3, SD = 4.8) for a US$10 payment. The study employed a 2 (participant’s fertility in an ovulatory cycle: high vs. low) by 2 (other player’s shirt color: red vs. blue) between-subjects design. We chose to compare red color with blue because blue is a neutral color and is not culturally associated with dominance (Valdez & Mehrabian, 1994), thus serving as a good comparison.
We calculated our sample size based on an estimate of an effect size, f = .35 (based on the effect size reported by prior hormonal studies using urinalysis to determine fertility status; for example, Durante et al., 2014), requiring a sample size of approximately 67 participants for a study powered at 80%.
Procedure
An earlier online survey screened female participants who responded to our advertisement (see Durante et al., 2014) to help us identify their expected day of ovulation. Of the women who met our screening criteria and were available during scheduled days, we randomly assigned them a high- or low-fertility group, independent of where they were in their ovulatory cycle by following a reverse-cycle method (Durante et al., 2014; Durante et al., 2011). To do that, we first placed all women on a standard 28-day cycle (Durante et al., 2014; Haselton & Gangestad, 2006). Then, based on the information they provided with respect to their last onset of menstruation and the anticipated start of their next menstruation, we estimated their high- and low-fertility periods. The method assumes a standard luteal phase of 14 days from the day of ovulation until the end of the cycle and a follicular phase (beginning of cycle to ovulation) of varied length which is standardized to be precisely 14 days (for a standardized 28-day cycle). Typically, for nonhormonal, self-report methods, fertile period is classified as being between Days 8 and 14, while nonfertile period is classified as being between Days 1-7 and 15-28 from the onset of the last period. We used this method in this and the following studies (Studies 2-6). To ensure that women assigned to high-fertility condition were as close to their fertile period as possible upon their arrival to the laboratory, we scheduled a lab session around their expected ovulation date (12-13 days after the start date of their last period during the ovulatory phase). Following Durante et al. (2014), these days were picked with the intent of inviting women in their 8- to 14-day window where ovulation hormone (luteinizing hormone [LH]) is gradually increasing up until Day 14 (a typical ovulation day), as close as possible to the ovulation (with not wanting to miss a woman’s ovulation). Forty-five women (assigned to high fertility) came to the lab on average –0.02 days (SD = 0.72) before their expected day of ovulation.
For the women assigned to the low-fertility group, a lab session was scheduled for after or before their expected ovulation (Days 18-28 after onset of their menstruation during luteal phase or Days 1-7 after the onset of their menstruation during the menstrual phase). We avoided scheduling women’s lab sessions on their Days 15 to 17 to reduce the possibility that we would displace women whose ovulation came later than Day 14. We scheduled lab sessions for five women on Days 1 to 7 after onset of menstruation with an average of 4.4 days (SD = 1.67) after the onset of their menstruation, and for 32 women on Days 18 to 28 after onset of menstruation with an average of 9.0 days after their expected day of ovulation (SD = 2.64).
Women reported to the laboratory individually and were greeted by the experimenter. To keep the conditions as similar as possible and to minimize the influence of potential confounding variables across experimental sessions, the same female experimenter conducted all the sessions, and she never wore red clothes to the sessions. In a private room with a personal computer, the experimenter informed participants that they would play a short decision-making game with another participant. Then, the experimenter took their picture in the laboratory and told them that she would upload it into another participant’s survey, who was ostensibly sitting in another private room. She also told them she had already taken and uploaded the picture of the other participant. The photograph—a female student in her early 20s—was taken in the same room where the actual participants’ pictures were taken. Participants were randomly assigned to view the target female in either a red or a blue shirt as the other player; aside from the color of the shirt, the pictures were identical (see the appendix).
Participants completed a two-person one-shot trust game (Berg, Dickhaut, & McCabe, 1995) in which they received an initial endowment of US$10 (apart from their participation payment) and were informed that they could transfer any portion of this amount to the other player. They were told researchers would triple any dollar amount transferred by them. Then the other player would specify how much of the tripled amount she would send back to the participant. Participants were informed some participants would be randomly selected to receive money earned in the task. After the game description, participants were asked to indicate how much of their US$10 they would transfer to the other player.
Urinalysis was performed at the conclusion of the study after participants responded to the questions comprising the dependent variable. We did not want to run the risk of women deducing the study’s intent. Participants were instructed to use an over-the-counter urine applicator test, which tests for LH. Following the instructions provided with the test, a trained researcher completed the reading and recording of test results.
A positive test result signifying that an LH surge has occurred typically indicates ovulation will occur shortly (within 24-36 hr of the test), though anovulatory cycles are still possible. Any woman with a positive test result was assigned to a high-fertility group (n = 33). Twelve out of 45 participants who were initially randomly assigned to a high-fertility group had a negative test result and were thus excluded from the analyses; the remaining participants in a high-fertility group experienced an LH surge on average 0.06 days (SD = 0.70 days) after their expected day of ovulation on a standard 28-day cycle. All women initially randomly assigned to the low-fertility group with negative test results were coded as a low-fertility group (n = 37). On average, women in the low-fertility condition completed the study 7.05 days (SD = 5.59) after their expected day of ovulation on a standard 28-day cycle.
Results and Discussion
Data exclusions
Four participants who personally knew the female target whose picture was taken for the purpose of this study were excluded from all the analyses, based on the decision we made prior to data collection (including those participants in the analysis, however, made no significant differences to the results). We conducted analyses on the remaining 66 female participants.
Trust game
Including fertility and the shirt color as independent measures and the amount of money sent to the other female player as the dependent measure in the ANOVA revealed no effect of fertility status, F(1, 62) = 0.004, p = .949, ηp2 = .000; a marginal effect of shirt color, F(1, 62) = 3.213, p = .078, ηp2 = .049; and a marginally significant interaction effect, F(1, 62) = 3.234, p = .077, ηp2 = .050. Examining the interaction term, women in the high-fertility group sent significantly less money to the female participant in the red shirt (M = 6.12, SE = 0.61, 95% confidence interval [CI] = [4.908, 7.327]) than to the female participant in the blue shirt (M = 8.33, SE = 0.64, 95% CI = [7.046, 9.621]), F(1, 62) = 6.29, p = .015, ηp2 = .092. However, women in the low-fertility group sent similar amounts of money to the female participant in the red shirt as they did to the female participant in the blue shirt (M = 7.27, SE = 0.64, 95% CI = [6.119, 8.407] vs. M = 7.26, SE = 0.57, 95% CI = [5.979, 8.554]), F(1, 62) = 0.00, p = .997.
Using urinalysis to assess fertility status, 1 we found that ovulating participants transferred marginally significantly less money to a woman when she was dressed in red compared with when she was dressed in blue, signaling lower trust toward women dressed in red. No difference in the amount of money transferred was found when participants were not ovulating.
Before concluding this section, it is important to discuss the discrepancy we observed with respect to classifying participants in the high- and low-fertility groups using self-report and hormonal methods. Specifically, although we found that all women whom we originally classified as being in their low-fertility period using self-report data were indeed in their low-fertility period as evidenced by the hormonal data, 25% of women whom we originally classified as being in their high-fertility period using self-report data were in their low-fertility period. There are several reasons for why that may be the case. First, as mentioned earlier, women occasionally experience anovulatory cycles where the egg is not released and LH, therefore, is not produced to be detected by the test. In addition, some women may not diligently record the start of their past period and may not accurately estimate the start of the next one. For that reason, although, based on the self-report data, we would expect an ovulation to occur between Days 12 and 13 following the start of their previous period, the inaccuracy in recording women’s cycle-related data may lead to errors in estimating their fertile periods. Finally, certain illnesses, such as cold and flu, may delay ovulation by a few days. These points demonstrate that although self-report is a convenient method of estimating high- and low-fertility periods, it may lead to some misclassification of women between these periods.
Study 2
In Study 2, we attempted to examine our predictions with a different population in an organizational setting. In addition, given that past work has primarily focused on self-reported measure of fertility status, we switched to self-reported fertility status data to test our predictions.
Method
Participants and design
We recruited 300 women (Mage = 32.7, SD = 5.1) from an online panel (ClearVoice) for a US$5 pay. 2 The study employed a 2 (fertility in an ovulatory cycle: high vs. low) by 2 (shirt color: red shirt vs. blue shirt) between-subjects design.
We calculated our sample size based on an estimate of an effect size, f = .2, from previous studies that determined a woman’s fertility status using self-report method (Durante et al., 2014). This effect size fell in the medium-to-small range, requiring a sample size of approximately 200 participants for a study powered at 80%. We asked the online panel to start with a screening survey and restrict participation to women under 41 years old (as women after that age are at greater risk of experiencing perimenopause; American Pregnancy Association, 2017), those who are not using hormonal birth control, and are not cigarette smokers. However, previous studies have also excluded women who were sick, pregnant, breastfeeding, or experiencing irregular menstrual cycles (see Beall & Tracy, 2013; Durante et al., 2014). Given the fact that we had a list of ex-ante exclusion criteria, we recruited more women in the study to ensure a sufficient number of participants.
Procedure
Participants were told that this was a study on managerial decision making and were asked to put themselves in the shoes of a manager in an organization. All participants were then presented with the same entry-level position and list of responsibilities and requirements. Afterward, they were presented with a CV and picture of a job applicant (Sarah Whitman). Participants were randomly assigned to view the same picture participants viewed in Study 1: a woman in her early 20s wearing either a red shirt or a blue shirt. For our dependent variable, we measured salary conferral (adapted from Moss-Racusin, Dovidio, Brescoll, Graham, & Handelsman, 2012). Participants were asked, “If you had to choose one of the following starting salaries for this applicant, what would it be?” Responses were measured on a scale from US$20,000 to US$60,000 in US$5,000 increments. At the end, they answered questions related to their ovulatory cycle (Durante et al., 2014).
Fertility status
In this and the next studies (Studies 2-6), we used two methods to measure fertility status. Consistent with the most common method—dichotomous (Gangestad et al., 2016)—we asked participants to respond to questions related to their ovulatory cycle. From participants’ responses to these questions, we used a reverse-cycle method for estimating fertility status (Durante et al., 2014; Haselton & Gangestad, 2006) and therefore placed women on a standard 28-day cycle. Given that we were interested in comparing perceptions of ovulating women with perceptions of nonovulating women, based on participants’ answers to the questions, we divided women into two groups (Durante et al., 2014): a low-fertility group (Days 1-7 or Days 15-28 of their cycle) and a high-fertility group (Days 8-14).
In addition to the dichotomous measure, we created a continuous measure of fertility status by assigning the likelihood of conception to the cycle day of each woman, as calculated by subtracting the day of the participants’ last period from the day on which the survey was taken. The likelihood of conception for each cycle day was referenced from Wilcox, Dunson, Weinberg, Trussell, and Baird (2001).
Results and Discussion
Data exclusions
Ninety-one women who did not meet the inclusion criteria were excluded from all the analyses according to a decision made prior to conducting the study. Specifically, based on the criteria established by previous studies (e.g., Durante et al., 2014), women who reported having irregular menstrual cycles, were pregnant, breastfeeding, or were sick were excluded. We also excluded any participant who did not respond to ovulatory cycle-related questions (e.g., participants who provided a future date to the question about their last menstruation onset). We conducted analyses on the remaining 209 women using both dichotomous and continuous measure of fertility.
Salary conferral
A 2 × 2 between-subjects ANOVA using the average recommended salary as the dependent measure revealed no effect of fertility status, F(1, 205) = 2.52, p = .114, ηp2 = .012; no effect of shirt color, F(1, 205) = 0.52, p = .470, ηp2 = .003; and a marginally significant interaction effect, F(1, 205) = 3.68, p = .057, ηp2 = .018. Examining the interaction term, the women in the high-fertility group did not offer different salary to the woman wearing red (M = US$31,500, SE = 2,161, 95% CI = [27,240, 35,760]) than the woman in blue (M = US$35.909, SE = 2,060, 95% CI = [31,847, 39,971]), F(1, 205) = 3.41, p = .141, ηp2 = .011. Similarly, women in the low-fertility group did not offer different salaries to women wearing red (M = US$37,349, SE = 1,060, 95% CI = [35,258, 39,441]) or blue (US$35,357, SE = 1,054, 95% CI = [33,278, 37,436]), F(1, 205) = 1.77, p = .184, ηp2 = .009.
Next, we tested our predictions using the continuous fertility status measure. We conducted hierarchal regression on the salary with color of shirt, continuous fertility measure (mean centered), and their interaction. This analysis revealed a significant effect of fertility status (B = 62,355, SE = 29,245, p = .034, 95% CI = [4,707, 120,003]), no effect of shirt color (B = 920, SE = 1,310, p = .483, 95% CI = [–1,661, 3,501]), and a significant interaction effect (B = −121,610, SE = 42,758, p = .005, 95% CI = [–205,892, –37,327]). Participants 1 SD above the ovulation average did not differ in their salary offer to the woman wearing red than the woman in blue (B = −2,833, SE = 1,859, p = .129, 95% CI = [–6,497, 831]), while those 1 SD below the ovulation average offered a higher salary to the woman wearing red than the woman in blue (B = 4,647, SE = 1,852, p = .013, 95% CI = [995, 8,299]).
Results of this study demonstrate a significant interactive effect of fertility status and color red on women’s behavior toward another female using continuous method of assessing fertility but only a marginally significant interactive effect using dichotomous method of assessing fertility. The correlations between the dichotomous and continuous measures of fertility status as well as other variables are presented in Table 1. It is important to note that, with the continuous measure, the effect is different from the one we predicted, as it is driven by the low-fertility and not high-fertility women. The discrepancy between the two methods likely stems from continuous measure being more sensitive and accurate than the dichotomous measure (Gangestad et al., 2016).
Descriptive Statistics and Correlations in Study 2.
Note. Shirt color is coded as 0 = nonred, 1 = red.
Correlation is significant at the .001 level (two-tailed).
Study 3
In Studies 1 and 2, we have examined women’s behavioral reactions to another woman; the results so far have been inconclusive. Study 3 examined whether the color of another woman’s clothes influences an ovulating woman’s mating-related perceptions (of her dominance and attractiveness) as well as overall liking as a behavioral reaction.
Method
Participants and design
We recruited 220 women (Mage = 29.3, SD = 5.9) from an online panel (Instantly) for a US$5 pay. The study employed 2 (fertility: high vs. low) by 2 (shirt color: red vs. blue) design. We used the same power analysis procedure and the same set of exclusion criteria as in Study 2.
We asked the online panel to start with a screening survey and restrict participation to women under 41 years old, those who were not using hormonal birth control, those who did not smoke cigarettes, and were not pregnant. However, previous studies have also excluded women who were sick, breastfeeding, or experiencing irregular menstrual cycles (see Durante et al., 2014). Given that we had a few ex-ante exclusion criteria, we recruited a few more women in the study.
Procedure
Participants who met our selection criteria were told that the study dealt with person perception, and thus a photograph of a person would be presented to them followed by a short questionnaire. Participants were randomly assigned to view a photograph of a woman in her early 20s wearing either a red or a blue shirt; aside from the color of the shirt, the pictures were identical (the photographs can be found in the appendix).
Participants then were asked to rate the person on a number of attributes. Afterward, they answered the same questions related to their ovulatory cycle as participants answered in our Study 2.
Measures
All items were rated on a 7-point scale (1 = not at all, 7 = extremely). For dominance perception, we relied on the dominance scale (dominant, assertive, and forceful, α = .77) from the Revised Interpersonal Adjective Scales (IAS-R; Wiggins, Trapnell, & Phillips, 1988). To measure attractiveness perception, we used two items (attractive and pretty, α = .91; Elliot & Niesta, 2008). At the end, we included two items to form a general liking index: “How much do you want to be friends with this person?” and “How much do you want to get to know this person?” (α = .96).
Fertility status
We assessed participants’ dichotomous fertility status and continuous conception likelihood using the same procedure as in Study 2.
Results
Data exclusions
We used the same set of inclusion criteria as we did in Study 2, and 91 women who did not provide complete data or did not meet the inclusion criteria were excluded based on the ex-ante exclusion criteria. We conducted analyses on the remaining 129 women.
Liking
A 2 × 2 between-subjects ANOVA using the liking score as the dependent measure revealed no effect of fertility status, F(1, 125) = 0.54, p = .464, ηp2 = .004; no effect of shirt color, F(1, 125) = 1.72, p = .193, ηp2 = .014; and no significant interaction effect, F(1, 125) = 0.01, p = .930, ηp2 = .000. Examining the interaction term, the women in the high-fertility group did not like the woman wearing red (M = 3.35, SE = 0.48, 95% CI = [2.400, 4.300]) differently than the woman in blue (M = 3.82, SE = 0.35, 95% CI = [3.126, 4.505]), F(1, 125) = 0.62, p = .434, ηp2 = .005. Similarly, women in the low-fertility group did not like the women wearing red (M = 3.14, SE = 0.21, 95% CI = [2.718, 3.551]) or blue (M = 3.54, SE = 0.22, 95% CI = [3.108, 3.975]) differently, F(1, 125) = 1.79, p = .183, ηp2 = .014.
Next, we tested our predictions using the continuous fertility status measure using regression. This analysis revealed no effect of fertility status (B = 7.96, SE = 15.97, p = .619, 95% CI = [–23.647, 39.561]), no effect of shirt color (B = 0.49, SE = 0.36, p = .183, 95% CI = [–0.232, 1.204]), and no significant interaction effect (B = −2.04, SE = 9.42, p = .829, 95% CI = [–20.68, 16.593]). Both women 1 SD above ovulation average (B = 0.37, SE = 0.39, p = .346, 95% CI = [–0.403, 1.143]) and those 1 SD below ovulation average (B = 0.49, SE = 0.36, p = .183, 95% CI = [–0.232, 1.204]) did not rate the woman wearing red or blue differently.
Dominance perception
A 2 × 2 between-subjects ANOVA using the dominance score as the dependent measure revealed no effect of fertility status, F(1, 125) = 0.00, p = .998, ηp2 = .000; no effect of shirt color, F(1, 125) = 0.88, p = .349, ηp2 = .007; and no significant interaction effect, F(1, 125) = 0.22, p = .643, ηp2 = .002. Examining the interaction term, the women in the high-fertility group did not rate the woman wearing red (M = 4.07, SE = 0.46, 95% CI = [3.151, 4.983]) differently in dominance than the woman in blue (M = 4.22, SE = 0.34), 95% CI = [3.555, 4.884]), F(1, 125) = 0.07, p = .790, ηp2 = .001. Similarly, women in the low-fertility group did not rate the women wearing red (M = 3.92, SE = 0.20, 95% CI = [3.515, 4.318]) or blue (M = 4.37, SE = 0.21, 95% CI = [3.950, 4.786]) differently in dominance, F(1, 125) = 2.37, p = .126, ηp2 = .019.
Using the continuous fertility status measure, the analysis revealed no significant effect of fertility status (B = 3.98, SE = 15.46, p = .797, 95% CI = [–26.599, 34.558]), no effect of shirt color (B = 0.47, SE = 0.35, p = .180, 95% CI = [–0.221, 1.168]), and no significant interaction effect (B = −1.48, SE = 9.11, p = .871, 95% CI = [–19.516, 16.547]). Both women 1 SD above ovulation average (B = 0.39, SE = 0.38, p = .306, 95% CI = [–0.359, 1.137]) and those 1 SD below ovulation average (B = 0.47, SE = 0.35, p = .180, 95% CI = [–0.221, 1.168]) did not rate the woman wearing red or blue differently.
Attractiveness perception
A 2 × 2 between-subjects ANOVA using the attractiveness score as the dependent measure revealed no effect of fertility status, F(1, 125) = 1.37, p = .244, ηp2 = .011; no effect of shirt color, F(1, 125) = 0.21, p = .646, ηp2 = .002; and no significant interaction effect, F(1, 125) = 0.00, p = .993, ηp2 = .000. For the interaction term, the women in the high-fertility group did not rate the woman wearing red (M = 4.60, SE = 0.44, 95% CI = [3.737, 5.463]) differently in attractiveness than the woman in blue (M = 4.74, SE = 0.32, 95% CI = [4.111, 5.363]), F(1, 125) = 0.07, p = .800, ηp2 = .001. Similarly, women in the low-fertility group did not rate the women wearing red (M = 4.95, SE = 0.19, 95% CI = [4.574, 5.330]) or blue (M = 5.09, SE = 0.20, 95% CI = [4.700, 5.488]) differently in attractiveness, F(1, 125) = 0.27, p = .608, ηp2 = .002.
Using the continuous fertility status measure, the analysis revealed no significant effect of fertility status (B = 3.96, SE = 14.59, p = .786, 95% CI = [–24.896, 32.820]), no effect of shirt color (B = 0.29, SE = 0.33, p = .403, 95% CI = [–0.377, 0.933]), and no significant interaction effect (B = −3.63, SE = 8.60, p = .674, 95% CI = [–20.643, 13.390]). Both women 1 SD above ovulation average (B = 0.07, SE = 0.36, p = .841, 95% CI = [–0.634, 0.778]) and those 1 SD below ovulation average (B = 0.28, SE = 0.33, p = .403, 95% CI = [–0.377, 0.933]) did not rate the woman wearing red or blue differently.
In sum, we found no main effects or interactive effect in this study, with either dichotomous or continuous measures of fertility status, for any dependent variables we employed in this study. A summary of correlations between all variables included in this study is presented in Table 2. In light of the marginally significant findings in Studies 1 and 2 and no significant in Study 3, we set out to rerun this study with another sample.
Descriptive Statistics and Correlations in Study 3.
Note. Shirt color is coded as 0 = nonred, 1 = red.
Correlation is significant at the .10 level (two-tailed). *Correlation is significant at the .05 level (two-tailed). ***Correlation is significant at the .01 level (two-tailed).
Study 4
Given that we found no main or interaction effects between fertility status and color red on dominance and attractiveness perceptions as well as liking of another woman in Study 3, with Study 4 we aimed to examine whether the absence of significant results was a function of the sample or whether the null results would be replicated. To this end, we conducted Study 4 with another sample and included multiple perception measures as well as liking as dependent variables.
Method
Participants and design
In total, 250 women (Mage = 29.8, SD = 6.9) from Amazon Mechanical Turk (MTurk) participated in the study for US$1 payment. The study employed a 2 (fertility: high vs. low) by 2 (shirt color: red vs. blue) between-subjects design.
Employing the same design, we used the same power analysis procedure as in previous studies to estimate the required sample size. In Studies 2 and 3, we asked the online panel administration to ask participants for their method of contraception in the beginning of the survey; those on hormonal contraception were disqualified from continuing with our survey. In Study 4, which was our first MTurk study, we did not use this selection criterion in recruitment and restricted participation ex-ante only to females living in the United States and between the ages of 18 and 40 years old, and those who do not smoke cigarettes. Taking into account the data collected for 2006 to 2010, among U.S. women aged 15 to 44 years, about 28% of women use pills as their method of contraception (National Center for Health Statistics, 2012). We thus oversampled by recruiting 250 women. We used the same selection criteria (i.e., not pregnant, not breastfeeding, regular cycles, nonhormonal birth control users, and not sick) ex post.
Procedure
The same procedure (and pictures) was used as in Study 3 (see the appendix). One difference was that the person-perception measures. After rating their perceptions of a woman on the picture, participants answered questions related to their ovulatory cycle.
Measures
All items were rated on a 7-point scale (1 = not at all, 7 = extremely). To measure dominance-related traits, we relied on a set of attributes (dominating, intimidating, and arrogant, α = .72) proposed by Rudman, Moss-Racusin, Phelan, and Nauts (2012). In addition, we included perceptions of communality (helpful, warm, and friendly, α = .89; Rudman et al., 2012) as well as perceptions of competence (competent and intelligent, α = .73) and warmth (warm, sincere, and good-natured, α = .92) (Fiske et al., 2002). At the end, we included the same two items to form a general liking index: “How much do you want to be friends with this person?” and “How much do you want to get to know this person?” (α = .78).
Fertility status
We assessed participants’ dichotomous fertility status and continuous conception likelihood using the same procedure as in previous Studies 2 and 3.
Results and Discussion
Data exclusions
One hundred twenty-one women who did not provide complete data or did not meet the inclusion criteria were excluded from all the analyses according to a decision made prior to conducting the study based on the criteria established by prior studies (e.g., Durante et al., 2014) and used in Study 2. We conducted analyses on the remaining 129 women.
Liking
A 2 × 2 between-subjects ANOVA using the average of two liking items as the dependent measure revealed no effect of fertility status, F(1, 125) = 0.24, p = .626, ηp2 = .002; no effect of shirt color, F(1, 125) = 2.30, p = .132, ηp2 = .018; and a marginally significant interaction effect, F(1, 125) = 3.86, p = .052, ηp2 = .030. Examining the interaction term, women in the high-fertility group liked the woman wearing red marginally significantly less (M = 2.86, SE = 0.40, 95% CI = [2.074, 3.653]) than the woman in blue (M = 3.88, SE = 0.33, 95% CI = [3.221, 4.529]), F(1, 125) = 3.81, p = .053, ηp2 = .030. However, women in the low-fertility group did not like the woman differently across shirt-color conditions (M = 3.58, SE = 0.20, 95% CI = [3.190, 3.962] vs. M = 3.45, SE = 0.18, 95% CI = [3.097, 3.796]), F(1, 125) = 0.24, p = .623, ηp2 = .002.
Next, we tested our predictions using the continuous fertility status measure. This analysis revealed no effect of fertility status (B = 3.52, SE = 4.65, p = .451, 95% CI = [–5.684, 12.714]), no effect of shirt color (B = 0.24, SE = 0.33, p = .477, 95% CI = [–0.416, 0.886]), and no significant interaction effect (B = −12.28, SE = 8.17, p = .135, 95% CI = [–28.454, 3.889]). Both women 1 SD above ovulation average (B = −0.51, SE = 0.36, p = .157, 95% CI = [–1.222, 0.200]) and those 1 SD below ovulation average (B = 0.24, SE = 0.33, p = .477, 95% CI = [–0.416, 0.886]) did not rate the woman wearing red or blue differently.
Dominance perception
Including fertility status and shirt color as the independent measures and the average score of dominance items included as the dependent measure in the ANOVA revealed no effect of fertility status, F(1, 125) = 0.013, p = .910, ηp2 = .000; a significant effect of shirt color, F(1, 125) = 4.03, p = .047, ηp2 = .031; and a significant interaction effect, F(1, 125) = 4.16, p = .043, ηp2 = .032. Examining the interaction term, women in the high-fertility group were significantly more likely to rate the woman wearing red as more dominant (M = 5.06, SE = 0.37, 95% CI = [4.335, 5.786]) than the woman in blue (M = 3.98, SE = 0.30, 95% CI = [3.377, 4.581]), F(1, 125) = 5.15, p = .025, ηp2 = .040. However, women in low-fertility group viewed the woman wearing red or blue similarly (M = 4.49, SE = 0.18, 95% CI = [4.131, 4.840] vs. M = 4.49, SE = 0.16, 95% CI = [4.172, 4.816]), F(1, 125) = 0.001, p = .972, ηp2 = .000.
For the continuous fertility status measure, the analysis revealed no effect of fertility status (B = −6.14, SE = 4.90, p = .212, 95% CI = [–15.833, 3.552]), no effect of shirt color (B = −0.07, SE = 0.30, p = .808, 95% CI = [–0.662, 0.517]), and no significant interaction effect (B = 10.49, SE = 1.48, p = .141, 95% CI = [–3.540, 24.525]). Women 1 SD above ovulation average rated the woman in red as marginally more dominant (B = 0.55, SE = 0.31, p = .078, 95% CI = [–0.062, 1.160]), while those 1 SD below ovulation average did not rate the woman wearing red or blue differently (B = −0.07, SE = 0.30, p = .808, 95% CI = [–0.662, 0.517]).
Communality perception
A 2 × 2 between-subjects ANOVA using the communality score as the dependent measure revealed no effect of fertility status, F(1, 125) = 0.01, p = .919, ηp2 = .000; no effect of shirt color, F(1, 125) = 0.50, p = .479, ηp2 = .004; and no significant interaction effect, F(1, 125) = 1.30, p = .256, ηp2 = .010. Examining the interaction term, the women in the high-fertility group did not rate the woman wearing red (M = 3.18, SE = 0.37, 95% CI = [2.455, 3.909]) differently in communality than the woman in blue (M = 3.68, SE = 0.30, 95% CI = [3.074, 4.280]), F(1, 125) = 1.08, p = .301, ηp2 = .009. Similarly, women in the low-fertility group did not rate the women wearing red (M = 3.51, SE = 0.18, 95% CI = [3.159, 3.870]) or blue (M = 3.40, SE = 0.16, 95% CI = [3.077, 3.721]) differently in communality, F(1, 125) = 0.23, p = .634, ηp2 = .002.
For the continuous fertility status measure, the analysis revealed no significant effect of fertility status (B = 4.34, SE = 0.89, p = .375, 95% CI = [–5.306, 13.978]), no effect of color (B = 0.03, SE = 0.30, p = .917, 95% CI = [–0.555, 0.617]), and no significant interaction effect (B = −1.29, SE = 7.05, p = .855, 95% CI = [–15.254, 12.666]). Both women 1 SD above ovulation average (B = −0.05, SE = 0.31, p = .881, 95% CI = [–0.555, 0.617]) and those 1 SD below ovulation average (B = 0.03, SE = 0.30, p = .917, 95% CI = [–0.654, 0.562]) did not rate the woman wearing red or blue differently.
Competence perception
A 2 × 2 between-subjects ANOVA using the competence score as the dependent measure revealed no effect of fertility status, F(1, 125) = 0.56, p = .457, ηp2 = .004; a significant effect of shirt color, F(1, 125) = 5.65, p = .019, ηp2 = .043; and no significant interaction effect, F(1, 125) = 1.31, p = .255, ηp2 = .010. Examining the interaction term, the women in the high-fertility group rated the woman wearing red (M = 5.27, SE = 0.28, 95% CI = [4.725, 5.821]) marginally more in competence than the woman in blue (M = 4.56, SE = 0.23, 95% CI = [4.108, 5.017]), F(1, 125) = 3.90, p = .051, ηp2 = .030. Those women in the low-fertility group did not rate the women wearing red (M = 4.89, SE = 0.14, 95% CI = [4.623, 5.159]) or blue (M = 4.64, SE = 0.12, 95% CI = [4.400, 4.886]) differently in competence, F(1, 125) = 1.85, p = .176, ηp2 = .015.
For the continuous fertility status measure, the analysis revealed no significant effect of fertility status (B = 0.80, SE = 5.12, p = .88, 95% CI = [–0.555, 0.617]), no effect of color (B = 0.80, SE = 5.12, p = .88, 95% CI = [–0.555, 0.617]), and no interaction effect (B = 0.80, SE = 5.12, p = .88, 95% CI = [–0.555, 0.617]). Both women 1 SD above ovulation average (B = 0.80, SE = 5.12, p = .88, 95% CI = [–0.555, 0.617]) and those 1 SD below ovulation average (B = 0.80, SE = 5.12, p = .88, 95% CI = [–0.555, 0.617]) did not rate the woman wearing red or blue differently.
Warmth perception
A 2 × 2 between-subjects ANOVA using the warmth score as the dependent measure revealed no effect of fertility status, F(1, 125) =1.28, p = .260, ηp2 = .010; no effect of shirt color, F(1, 125) = 1.16, p = .283, ηp2 = .009; and no significant interaction effect, F(1, 125) = 1.66, p = .200, ηp2 = .013. Examining the interaction term, the women in the high-fertility group did not rate the woman wearing red (M = 2.85, SE = 0.39, 95% CI = [2.076, 3.621]) differently in warmth than the woman in blue (M = 3.52, SE = 0.32, 95% CI = [2.881, 4.161]), F(1, 125) = 1.76, p = .187, ηp2 = .014. Similarly, women in the low-fertility group did not rate the women wearing red (M = 3.54, SE = 0.19, 95% CI = [3.159, 3.914]) or blue (M = 3.48, SE = 0.17, 95% CI = [3.134, 3.818]) differently in warmth, F(1, 125) = 0.05, p = .816, ηp2 = .000.
For the continuous fertility status measure, the analysis revealed no significant effect of fertility status (B = 1.17, SE = 5.22, p = .824, 95% CI = [–9.173, 11.503]), no effect of color (B = −0.01, SE = 0.32, p = .977, 95% CI = [–0.638, 0.619]), and no significant interaction effect (B = −2.32, SE = 7.56, p = .759, 95% CI = [–17.288, 12.647]). Both women 1 SD above ovulation average (B = −0.15, SE = 0.33, p = .657, 95% CI = [–0.798, 0.505]) and those 1 SD below ovulation average (B = −0.01, SE = 0.32, p = .977, 95% CI = [–0.638, 0.619]) did not rate the woman wearing red or blue differently.
In sum, we found that the fertility status and the color red do not change perception of other women universally and only affect perceptions that deal with dominance and general liking (at the significant level using dichotomous measure of fertility and at the marginally significant level using the continuous measure of fertility). Correlations between all variables in this study are presented in Table 3. The discrepancy in the significance between the two measures may have to do with continuous measure being more sensitive and accurate compared with the dichotomous measure (Gangestad et al., 2016). By contrast, fertility status and color red did not have significant or marginally significant effects on competence, warmth, or communality. This suggests that during their peak fertility, women may be sensitive to potential rivals and their appearance, leading the ovulating women to perceive the other women as more intimidating when latter are dressed in red.
Descriptive Statistics and Correlations in Study 4.
Note. Shirt color is coded as 0 = nonred, 1 = red.
Correlation is significant at the .05 level (two-tailed). ***Correlation is significant at the .01 level (two-tailed).
From this study, however, we are unable to conclude whether it was the fact that the target woman was dressed in red or the color red itself that may have triggered the change in trust perceptions. Moreover, it is important to note that other properties of the color red such as its arousal properties (Wilson, 1966) could have led to ovulating women rating other women as more dominant. Therefore, in Study 5, we examined the phenomenon more carefully by manipulating the color of the background as well as the color of the woman’s clothes in the picture. In addition, we changed our comparison clothes color from blue to gray. Gray is another good contrast to red because it is a neutral color widely worn by women.
Study 5
In Study 5, we continued our investigation of how an ovulating woman treats another woman dressed in red versus another color. In this study, we included a measure of interpersonal trust as well as person-perception measures. In addition, to be able to conclude whether the color red itself affects women’s perceptions or whether it is the fact that a potential rival was dressed in red that caused the change, we manipulated the color of the background as well as the color of the woman’s clothes.
Method
Participants and design
Four hundred women (Mage = 29.2, SD = 5.6) from Amazon MTurk participated in the study for a US$1 payment. The study employed a 2 (fertility in an ovulatory cycle: high vs. low) by 2 (shirt color: red on a gray background vs. gray on a red background) between-subjects design.
We used the same power analysis procedure as in previous studies. We restricted participation ex-ante only to females living in the United States and between the ages of 18 and 40 years old. We recruited more women in Study 5 to ensure sufficient number of participants given the fact that a high number of women in this sample (MTurk) were hormonal birth control users or had irregular menstrual cycles based on responses from our earlier studies.
We thus oversampled by recruiting 400 women instead of 250 in Study 4. Similar to previous studies, we used the remaining selection criteria (i.e., not pregnant, nonsmoker, not breastfeeding, regular cycles, nonhormonal birth control users, and nonsick) ex post.
Procedure
The same procedure was used as in Studies 3 and 4. The only difference was that participants were randomly assigned to view the picture of a woman in her early 20s wearing either a red jumpsuit on a gray background or a gray jumpsuit on a red background (see the appendix). Then, participants rated the target on a list of person perceptions in a random order. Next, they completed an interpersonal trust scale, followed by questions related to their ovulatory cycle.
Measures
All items for person perception were rated on a 7-point scale (1 = not at all, 7 = extremely). To measure attractiveness perception, we used two items (attractive and pretty, α = .93; Elliot & Niesta, 2008). For dominance perception, we relied on the dominance scale (dominant, assertive, and forceful, α = .84) from the IAS-R (Wiggins et al., 1988). We also measured perceptions of warmth (warm, sincere, and good-natured, α = .89; Fiske et al., 2002) and competence (competent and intelligent, α = .87; Fiske et al., 2002). In addition, we used an interpersonal trust inventory (α = .90, Dunn & Schweitzer, 2005). Participants rated each item (e.g., “I could rely on information this person provides to me”) on a 7-point scale (1 = strongly disagree, 7 = strongly agree).
Fertility status
We assessed participants’ dichotomous fertility status and continuous conception likelihood using the same procedure as in our previous studies (2-4).
Results and Discussion
Data exclusions
Participants who did not provide complete data or did not meet our a priori inclusion criteria (same as in Studies 2-4) were excluded from the analyses. We conducted analyses on the remaining 192 participants.
Interpersonal trust
A 2 × 2 between-subjects ANOVA using the average of interpersonal inventory as the dependent measure revealed no effect of fertility status, F(1, 188) = 0.25, p = .619, ηp2 = .001; no effect of shirt color, F(1, 188) = 0.44, p = .510, ηp2 = .002; and a marginally significant interaction effect, F(1, 188) = 3.17, p = .076, ηp2 = .017. Examining the interaction term, women in the high-fertility group were not significantly different on trusting the woman wearing red (M = 4.56, SE = 0.21, 95% CI = [4.143, 4.983]) than the woman in gray (M = 4.94, SE = 0.16, 95% CI = [4.624, 5.261]), F(1, 188) =2.02, p = .157, ηp2 = .011. Women in the low-fertility group viewed the woman wearing red or gray similarly (M = 4.92, SE = 0.10, 95% CI = [4.714, 5.121] vs. M = 4.74, SE = 0.12, 95% CI = [4.505, 4.981]), F(1, 188) = 1.20, p = .274, ηp2 = .006.
For the continuous fertility status measure, the analysis revealed no significant effect of fertility status (B = 1.83, SE = 2.63, p = .488, 95% CI = [–3.365, 7.022]), no effect of color (B = 0.20, SE = 0.19, p = .299, 95% CI = [–0.179, 0.580]), and no significant interaction effect (B = −5.14, SE = 4.11, p = .212, 95% CI = [–13.233, 2.961]). Both women 1 SD above ovulation average (B = −0.15, SE = 0.20, p = .457, 95% CI = [–0.539, 0.244]) and those 1 SD below ovulation average (B = 0.20, SE = 0.19, p = .299, 95% CI = [–0.179, 0.580]) did not rate the woman wearing red or blue differently.
Attractiveness perception
A 2 × 2 between-subjects ANOVA revealed a marginally significant effect of ovulatory status, F(1, 188) = 3.73, p = .055, ηp2 = .019, and of color red, F(1, 188) = 3.72, p = .055, ηp2 = .019. We also found a significant interaction effect, F(1, 188) = 10.16, p = .002, ηp2 = .051. Examining the interaction term, women in the high-fertility group rated the woman wearing red as significantly less attractive (M = 4.58, SE = 0.26, 95% CI = [4.076, 5.082]) than the woman wearing gray (M = 5.53, SE = 0.19, 95% CI = [5.149, 5.912]), F(1, 188) = 8.85, p = .003, ηp2 = .045. However, women in the low-fertility group viewed the woman similarly regardless of whether she was wearing red or gray (M = 5.53, SE = 0.12, 95% CI = [5.287, 5.774] vs. M = 5.30, SE = 0.15, 95% CI = [5.011, 5.582]), F(1, 188) = 1.52, p = .219, ηp2 = .008.
For the continuous fertility status measure, the analysis revealed no significant effect of fertility status (B = 0.90, SE = 3.19, p = .778, 95% CI = [–5.394, 7.198]), no effect of color (B = 0.26, SE = 0.23, p = .276, 95% CI = [–0.206, 0.715]), and a significant interaction effect (B = −9.98, SE = 4.98, p = .056, 95% CI = [–19.792, –0.160]). Women 1 SD above ovulation average viewed the woman in red marginally significantly different from the one in blue (B = −0.42, SE = 0.24, p = .081, 95% CI = [–0.986, 0.053]), while those 1 SD below ovulation average (B = 0.26, SE = 0.23, p = .276, 95% CI = [–0.206, 0.715]) did not rate the woman wearing red or blue differently.
Dominance perception
A 2 × 2 between-subjects ANOVA revealed no main effect of ovulatory status, F(1, 188) = 0.05, p = .829 ηp2 = .000, and no main effect of color red, F(1, 188) = 0.72, p = .398, ηp2 = .014. We did, however, find a marginally significant interaction effect, F(1, 188) = 3.53, p = .062, ηp2 = .018. Examining the interaction term, women in the high-fertility group did not rate the woman wearing red different on dominance (M = 4.33, SE = 0.27, 95% CI = [3.802, 4.864]) than the woman wearing gray (M = 3.80, SE = 0.20, 95% CI = [3.395, 4.201]), F(1, 188) = 2.51, p = .115, ηp2 = .013. Women in the low-fertility group viewed the woman similarly on dominance regardless of whether she was wearing red or gray (M = 3.92, SE = 0.13, 95% CI = [3.665, 4.179] vs. M = 4.12, SE = 0.15, 95% CI = [3.823, 4.426]), F(1, 188) = 1.02, p = .315, ηp2 = .005.
For the continuous fertility status measure, the analysis revealed no significant effect of fertility status (B = −1.59, SE = 3.34, p = .635, 95% CI = [–8.176, 5.004]), no effect of color (B = −0.15, SE = 0.24, p = .539, 95% CI = [–0.632, 0.331]), and no significant interaction effect (B = 4.45, SE = 5.21, p = .394, 95% CI = [–5.822, 14.726]). Both women 1 SD above ovulation average (B = 0.15, SE = 0.25, p = .548, 95% CI = [–0.345, 0.648]) and those 1 SD below ovulation average (B = −0.15, SE = 0.24, p = .539, 95% CI = [–0.632, 0.331]) did not rate the woman wearing red or blue differently.
Competence perception
A 2 × 2 between-subjects ANOVA on the competence perception revealed a marginally significant effect of ovulatory status, F(1, 188) = 2.87, p = .092, ηp2 = .015; no main effect of color, F(1, 188) = 1.47, p = .227, ηp2 = .008; and a significant interaction effect, F(1, 188) = 9.17, p = .003, ηp2 = .047. Examining the interaction term, women in the high-fertility group rated the woman wearing red as significantly less competent (M = 4.55, SE = 0.22, 95% CI = [4.113, 4.992]) than the woman in gray (M = 5.24, SE = 0.17, 95% CI = [4.909, 5.576]), F(1, 188) = 6.08, p = .015, ηp2 = .031. Women in the low-fertility group rated the woman’s competence as marginally higher when she was wearing red (M = 5.32, SE = 0.11, 95% CI = [5.108, 5.534]) than gray (M = 5.03, SE = 0.13, 95% CI = [4.776, 5.275]), F(1, 188) = 3.16, p = .077, ηp2 = .017.
For the continuous fertility status measure, the analysis revealed no significant effect of fertility status (B = −0.78, SE = 2.80, p = .781, 95% CI = [–6.310, 4.746]), no effect of color (B = 0.19, SE = 0.21, p = .354, 95% CI = [–0.214, 0.594]), and no significant interaction effect (B = −4.71, SE = 4.37, p = .282, 95% CI = [–13.328, 3.909]). Both women 1 SD above ovulation average (B = −0.13, SE = 0.21, p = .542, 95% CI = [–0.546, 0.288]) and those 1 SD below ovulation average (B = 0.19, SE = 0.21, p = .354, 95% CI = [–0.214, 0.594]) did not rate the woman wearing red or blue differently.
Warmth perception
A similar 2 × 2 between-subjects ANOVA using the average of warmth items as the dependent measure revealed no effect of fertility status, F(1, 188) = 2.01, p = .158, ηp2 = .011; no effect of shirt color, F(1, 188) = 0.13, p = .718, ηp2 = .001; and a significant interaction effect, F(1, 188) = 7.67, p = .006, ηp2 = .039. Examining the interaction term, women in the high-fertility group rated the woman wearing red as marginally significantly less warm (M = 4.74, SE = 0.21, 95% CI = [4.318, 5.156]) than the woman in gray (M = 5.22, SE = 0.16, 95% CI = [4.904, 5.540), F(1, 188) = 3.32, p = .020, ηp2 = .029. Women in the low-fertility group rated the woman’s warmth as marginally significantly higher when she was wearing red (M = 5.39, SE = 0.10, 95% CI = [5.183, 5.589]) than gray (M = 5.01, SE = 0.12, 95% CI = [4.775, 5.250]), F(1, 188) =3.32, p = .070, ηp2 = .017.
For the continuous fertility status measure, the analysis revealed no significant effect of fertility status (B = −1.70, SE = 2.64, p = .522, 95% CI = [–6.908, 3.515]), no effect of color (B = 0.29, SE = 0.19, p = .137, 95% CI = [–0.093, 0.669]), and no significant interaction effect (B = −4.98, SE = 4.11, p = .228, 95% CI = [–13.103, 3.149]). Both women 1 SD above ovulation average (B = −0.05, SE = 0.20, p = .805, 95% CI = [–0.442, 0.344]) and those 1 SD below ovulation average (B = 0.29, SE = 0.19, p = .137, 95% CI = [–0.093, 0.669]) did not rate the woman wearing red or blue differently.
In sum, the results of this study demonstrate a significant or marginally significant interaction effect between fertility status and the color red on perceptions—mating-related and general, as well as trust using the dichotomous measure of fertility but not the continuous measure (with the exception of significant interaction effect on attractiveness using the continuous measure). Correlations between all variables in this study are presented in Table 4. Importantly, because the color red was present in both conditions either as the color of the model’s jumpsuit or as the background, our results demonstrate that it is not the color per se that elicits the significant or marginally significant effects but rather the positioning of the color. Only when the model was wearing red, and not when she was presented against a red background, did we find different perceptions and behaviors, with dichotomous measure, from women in the high-fertility group.
Descriptive Statistics and Correlations in Study 5.
Note. Shirt color is coded as 0 = nonred, 1 = red.
Correlation is significant at the .10 level (two-tailed). *Correlation is significant at the .05 level (two-tailed). **Correlation is significant at the .01 level (two-tailed). ***Correlation is significant at the .01 level (two-tailed).
Study 6
Given the inconsistent findings across studies, we aimed to test our predictions one more time with a larger sample. We increased the sample size and preregistered our hypotheses, methods, and analysis strategy at Open Science Framework.
Method
Participants and design
In total, 1,500 women (Mage = 30.2, SD = 5.6) from Amazon MTurk participated in the study for a US$1 payment. The study employed a 2 (fertility in an ovulatory cycle: high vs. low) by 2 (shirt color: red on a gray background vs. gray on a red background) between-subjects design.
In previous studies, we calculated our sample size based on an estimate of an effect size, f = .2; however, our studies so far have shown a small effect size. Therefore, we calculated sample size based on an estimate of an effect size, f = .1, requiring a sample size of approximately 787 participants for a study powered at 80%. Moreover, given that in the previous MTurk studies about half of women were excluded based on our selection criteria (i.e., not pregnant, nonsmoker, not breastfeeding, regular cycles, nonhormonal birth control users, and nonsick) ex post, this time we aimed for 1,500 participants. Similar to Studies 4 and 5, we restricted participation ex-ante only to females living in the United States and between the ages of 18 and 40 years old.
Procedure
The same procedure and materials were used as in Study 5.
Measures
The same measures were used as in Study 5: interpersonal trust inventory (α = .91), dominance (α = .74), attractiveness (α = .89), competence (α = .78), and warmth (α = .86).
Fertility status
We assessed participants’ dichotomous fertility status and continuous conception likelihood using the same procedure as in our previous studies (2-5).
Results
Data exclusions
We used the same set of inclusion criteria as we did in previous studies and 963 participants were excluded bases on the ex-ante exclusion criteria. We conducted analyses on the remaining 537 women.
Trust
A 2 × 2 between-subjects ANOVA using the trust score as the dependent measure revealed no effect of fertility status, F(1, 533) = 0.37, p = .542, ηp2 = .001; a marginally significant effect of shirt color, F(1, 533) = 3.29, p = .070, ηp2 = .006; and a marginal significant interaction effect, F(1, 533) = 3.15, p = .076, ηp2 = .006. Examining the interaction term, the women in the high-fertility group trusted the woman wearing red (M = 4.85, SE = 0.16, 95% CI = [4.625, 5.078]) less than the woman in gray (M = 5.18, SE = 0.11, 95% CI = [4.967, 5.401]), F(1, 533) = 4.34, p = .038, ηp2 = .008. However, women in the low-fertility group did not trust the women wearing red (M = 4.96, SE = 0.07, 95% CI = [4.831, 5.088]) or gray (M = 4.96, SE = 0.07, 95% CI = [4.831, 5.095]) differently, F(1, 533) = 0.001, p = .970, ηp2 = .000.
Next, we tested our predictions using the continuous fertility status measure. This analysis revealed a significant effect of fertility status (B = 4.02, SE = 1.80, p = .026, 95% CI = [0.477, 7.556]), no effect of color (B = 0.71, SE = 0.12, p = .538, 95% CI = [–0.157, 0.299]), and a marginally significant interaction effect (B = −4.77, SE = 2.52, p = .058, 95% CI = [–9.1713, 0.169]). Women 1 SD above ovulation average trust the woman in red less (B = −0.33, SE = 0.15, p = .030, 95% CI = [–0.635, –0.033]), while those 1 SD below ovulation average (B = 0.05, SE = 0.11, p = .631, 95% CI = [–0.162, 0.267]) did not rate the woman wearing red or blue differently.
Dominance perception
A 2 × 2 between-subjects ANOVA using the dominance score as the dependent measure revealed no effect of fertility status, F(1, 533) = 0.03, p = .853, ηp2 = .000; no effect of shirt color, F(1, 533) = 0.05, p = .828, ηp2 = .000; and no significant interaction effect, F(1, 533) = 0.03, p = .866, ηp2 = .000. Examining the interaction term, the women in the high-fertility group did not rate the woman wearing red (M = 4.01, SE = 0.13, 95% CI = [3.748, 4.262]) differently in dominance than the woman in gray (M = 4.00, SE = 0.13, 95% CI = [3.754, 4.246]), F(1, 533) = 0.001, p = .978, ηp2 = .000. Similarly, women in the low-fertility group did not rate the women wearing red (M = 4.04, SE = 0.07, 95% CI = [3.897, 4.188]) or gray (M = 4.00, SE = 0.08, 95% CI = [3.852, 4.152]) differently in dominance, F(1, 533) = 0.15, p = .703, ηp2 = .000.
For the continuous fertility status measure, the analysis revealed no significant effect of fertility status (B = 0.65, SE = 2.05, p = .752, 95% CI = [–3.371, 4.665]), no effect of color (B = 0.01, SE = 0.13, p = .940, 95% CI = [–0.249, 0.269]), and no significant interaction effect (B = 0.72, SE = 2.86, p = 0.252, 95% CI = [–4.888, 6.329]). Both women 1 SD above ovulation average (B = 0.07, SE = 0.17, p = .683, 95% CI = [–0.271, 0.413]) and those 1 SD below ovulation average (B = 0.01, SE = 0.12, p = .918, 95% CI = [–0.271, 0.413]) did not rate the woman wearing red or gray differently.
Attractiveness perception
A 2 × 2 between-subjects ANOVA using the attractiveness score as the dependent measure revealed no effect of fertility status, F(1, 533) = 0.69, p = .406, ηp2 = .001; no effect of shirt color, F(1, 533) = 0.46, p = .496, ηp2 = .001; and no significant interaction effect, F(1, 533) = 0.008, p = .928, ηp2 = .000. Examining the interaction term, the women in the high-fertility group did not rate the woman wearing red (M = 5.52, SE = 0.12, 95% CI = [5.277, 5.753]) differently in attractiveness than the woman in gray (M = 5.59, SE = 0.12, 95% CI = [5.362, 5.818]), F(1, 533) = 0.20, p = .655, ηp2 = .000. Similarly, women in the low-fertility group did not rate the women wearing red (M = 5.60, SE = 0.07, 95% CI = [5.470, 5.740]) or gray (M = 5.66, SE = 0.07, 95% CI = [5.523, 5.801]) differently in dominance, F(1, 533) = 0.34, p = .560, ηp2 = .001.
For the continuous fertility status measure, the analysis revealed no significant effect of fertility status (B = −1.37, SE = 1.90, p = .470, 95% CI = [–5.104, 2.355]), no effect of color (B = −0.12, SE = 0.12, p = .335, 95% CI = [–0.358, 0.122]), and no significant interaction effect (B = 1.73, SE = 2.65, p = .514, 95% CI = [–3.477, 6.935]). Both women 1 SD above ovulation average (B = 0.03, SE = 0.16, p = .858, 95% CI = [–0.289, 0.346]) and those 1 SD below ovulation average (B = −0.11, SE = 0.12, p = .334, 95% CI = [–0.337, 0.115]) did not rate the woman wearing red or gray differently.
Competence perception
A 2 × 2 between-subjects ANOVA using the competence score as the dependent measure revealed no effect of fertility status, F(1, 533) = 0.59, p = .442, ηp2 = .001; no effect of shirt color, F(1, 533) = 0.39, p = .531, ηp2 = .001; and no significant interaction effect, F(1, 533) = 0.74, p = .390, ηp2 = .001. Examining the interaction term, the women in the high-fertility group did not rate the woman wearing red (M = 5.11, SE = 0.11, 95% CI = [4.889, 5.338]) differently in competence than the woman in gray (M = 5.25, SE = 0.11, 95% CI = [5.035, 5.465]), F(1, 533) = 0.74, p = .389, ηp2 = .001. Similarly, women in the low-fertility group did not rate the women wearing red (M = 5.12, SE = 0.07, 95% CI = [4.995, 5.249]) or gray (M = 5.10, SE = 0.07, 95% CI = [4.970, 5.231]) differently in competence, F(1, 533) = 0.05, p = .818, ηp2 = .000.
For the continuous fertility status measure, the analysis revealed no significant effect of fertility status (B = 2.50, SE = 1.79, p = .162, 95% CI = [–1.006, 6.014]), no effect of color (B = 0.10, SE = 0.12, p = .401, 95% CI = [–0.129, 0.323]), and no significant interaction effect (B = −3.52, SE = 2.94, p = .159, 95% CI = [–8.417, 1.382]). Both women 1 SD above ovulation average (B = −0.20, SE = 0.15, p = .184, 95% CI = [–0.501, 0.097]) and those 1 SD below ovulation average (B = 0.08, SE = 0.11, p = .445, 95% CI = [–0.130, 0.295]) did not rate the woman wearing red or gray differently.
Warmth perception
A 2 × 2 between-subjects ANOVA using the warmth score as the dependent measure revealed no effect of fertility status, F(1, 533) = 0.34, p = .563, ηp2 = .001; no effect of shirt color, F(1, 533) = 1.83, p = .177, ηp2 = .003; and no significant interaction effect, F(1, 533) = 1.85, p = .174, ηp2 = .003. Examining the interaction term, the women in the high-fertility group did not rate the woman wearing red (M = 4.96, SE = 0.12, 95% CI = [4.725, 5.210]) differently in warmth than the woman in gray (M = 5.23, SE = 0.12, 95% CI = [5.004, 5.468]), F(1, 533) = 2.47, p = .116, ηp2 = .005. Similarly, women in the low-fertility group did not rate the women wearing red (M = 5.05, SE = 0.07, 95% CI = [4.907, 5.182]) or gray (M = 5.04, SE = 0.07, 95% CI = [4.902, 5.185]) differently in warmth, F(1, 533) = 0.00, p = .993, ηp2 = .000.
For the continuous fertility status measure, the analysis revealed significant effect of fertility status (B = 4.23, SE = 1.93, p = .029, 95% CI = [0.441, 8.014]), no effect of color (B = 0.09, SE = 0.12, p = .466, 95% CI = [–0.153, 5.128]), and a marginal significant interaction effect (B = −4.744, SE = 2.69, p = .078, 95% CI = [–10.029, 0.541]). Women 1 SD above ovulation average marginally rated the woman in red less warm (B = −0.31, SE = 0.16, p = .057, 95% CI = [–0.635, 0.010]), while those 1 SD below ovulation average did not rate the woman wearing red or gray differently (B = 0.07, SE = 0.12, p = .540, 95% CI = [–0.158, 0.301]).
In sum, we were not able to replicate significant interaction effects on our perception variables (with the exception of attractiveness, using the continuous measure of fertility status). Nevertheless, we replicated the interaction effect for trust at the marginal significance level using both dichotomous and continuous measures of fertility status, our variable indicating behavioral intentions. Correlations for all variables in this study are presented in Table 5.
Descriptive Statistics and Correlations in Study 6.
Note. Shirt color is coded as 0 = nonred, 1 = red.
Correlation is significant at the .05 level (two-tailed). **Correlation is significant at the .01 level (two-tailed). ***Correlation is significant at the .01 level (two-tailed).
General Discussion
Across six studies using both self-reported and hormonal fertility data, we did not find consistent evidence of a difference in how an ovulating and a nonovulating woman perceives and acts toward another woman wearing red clothes. The most consistent of our results, though of very low effect size, was the interactive effect of fertility status (assessed with dichotomous and/or continuous measure of fertility) and the color red on trust at either the significant or the marginally significant level (Studies 1, 2, 5, and 6). The summary of our results using the dichotomous measure of fertility is presented in Table 6.
Means and Standard Deviations of Responses by Condition for Studies 1 to 6.
Note. Numbers in parentheses are standard deviations. CI = confidence interval.
We believe there are three reasons for why we found no strong support for our hypothesized effects. First, it is possible that the interactive effect of the fertility status and color red is small than nothing, but very large sample sizes would be able to detect it. Second, it is possible that the self-report methodology we used in our Studies 2 to 5 created further difficulties in observing the hypothesized results. Specifically, hormonal methods are currently recognized as the most accurate at detecting LH. Self-report methods, however, while less accurate, are more convenient and therefore more common (e.g., Gangestad et al., 2016). Self-report methods are further subdivided into dichotomous and continuous, both of which have advantages and disadvantages. Dichotomous measures of fertility, for example, are generally less sensitive and have more methodological concerns, including median split problems associated with them than the continuous ones (Irwin & McClelland, 2003; MacCallum, Zhang, Preacher, & Rucker, 2002; Maxwell & Delaney, 1993). By contrast, continuous measures make it harder to detect interactions (McClelland & Judd, 1993). As such, due to the differences between the two types of self-report measures, inconsistencies between them were found, leading to inconsistent results. The third reason for no results may actually have to do with the theory we relied on. Although from previous color red research on mate-guarding we would expect at least a main effect of color red on perceptions of another woman (Pazda, Prokop, & Elliot, 2014), upon careful consideration, the absence of a consistent effect may at least partially corroborate previous findings. Whereas in Pazda and colleagues’ research a woman dressed in red elicited a change in perceptions that were specifically related to participants’ partners, in our current research perceptions might not have been uniformly affected because they did not directly pertain to a specific partner. As such, it appears that the color red (as well as ovulation) might have an effect on very specific rather than general perceptions. Thus, the general absence of significant results may have to do with the fact that we did not prime a mating mind-set of our participants before administering our dependent variable measures. Specifically, although most recently Durante and colleagues (2014) have demonstrated effects of fertility status on competitive behavior in the absence of such prime, it is possible that inducing women to think about potential mates would have strengthened the effects we hypothesized.
The current research is not without limitations. One limitation has to do with our choice of gray color as a comparison with red color—the primary interest of our research. As pointed out by one of our reviewers, although we chose gray due to its plainness, gray could also be associated with age and declining fertility. As such, it is possible that gray may have served as a confounding factor in our results. Another limitation pertains to the pictures differing in hue used in our studies not being balanced on saturation and brightness. Previous research (e.g., Camgöz, Yener, & Güvenç, 2004) demonstrates that people pay a lot more attention to stimuli that are at their maximum brightness and saturation. Even though we have used not one but multiple sets of pictures where the hue together with saturation and brightness were manipulated, nonetheless, future studies should examine the properties of the color used more carefully. In addition, we are aware of problems associated with lack of stimulus sampling (i.e., not providing greater variety of pictures of women), including statistical decisions being in the opposite direction (Wells & Windschitl, 1999). It would be worthwhile for future research to include greater number of pictures to mitigate the concern. Our final limitation has to do with potentially underpowered Studies 1 to 5. We originally conducted power analyses for our studies on the basis of Durante and colleagues’ (2014) work. Since then, the new meta-analyses on ovulatory effects have demonstrated varying effect size (Gildersleeve et al., 2014; Wood et al., 2014), suggesting that assuming a medium effect size may have been incorrect.
Our results raise questions for future exploration. First, though some calls have been made at comparing methods of assessing fertility status (e.g., Gangestad et al., 2016), it would be important to understand just what the difference is in sensitivity between the three methods (i.e., hormonal, dichotomous, and continuous). We found inconsistent findings depending on the methods used. In spite of very weak effects we find, it would be worth investigating whether the ovulating women’s attitudes and behaviors toward women wearing red manifests in mating contexts.
Supplemental Material
netchaeva_online_appendix – Supplemental material for The Woman in Red: Examining the Effect of Ovulatory Cycle on Women’s Perceptions of and Behaviors Toward Other Women
Supplemental material, netchaeva_online_appendix for The Woman in Red: Examining the Effect of Ovulatory Cycle on Women’s Perceptions of and Behaviors Toward Other Women by Ekaterina Netchaeva and Maryam Kouchaki in Personality and Social Psychology Bulletin
Footnotes
Appendix
The pictures used in Studies 1 to 6 (the faces of the female target were intact in all studies but are blurred here to protect privacy).
Authors’ Note
Please refer to online version of article to view figures in color detail.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Notes
Supplemental Material
Supplementary material is available online with this article.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
