Are women more likely to wear red and pink at peak fertility? What about on cold days? Conceptual, close, and extended replications with novel clothing colour measures

Abstract

Evolutionarily minded researchers have hypothesized that women advertise their ovulatory status by wearing red or pink clothing on relatively cold days. Many of these studies have been based on samples of women who have self-reported their clothing choices, a practice that raises questions about accuracy. In two studies, we evaluated the relationship between women's fertility and their clothing choices using four methods for measuring clothing colour: self-reports; trained raters' judgements of garment coloration in outfits that women drew onto mannequins to represent what they would wear to a party with single attractive people in attendance; automated colour coding of the mannequins; and trained raters' judgements of garment coloration as evinced in photographs that women took of themselves. Using these four measures of clothing choice along with measures of women's fertility and outside temperature, we did not find compelling evidence that women are particularly inclined to wear red or pink during peak fertility, even on relatively cold days. In humans and numerous other species, female fertility is cyclical, and conception hinges on mating around the time of ovulation. In some non-human animals, phenotypic markers and behavioural changes have evolved to indicate when females are ovulating (Dixson, 2012). Such markers include increased sexual receptivity, genital swellings, and oestrus vocalizations or copulation calls (

Investigating cycle shifts in women's clothing style and grooming

“…However, the effect of wearing more sexy clothes in the luteal phase is in the opposite direction as expected and was not supported by hormonal associations. While the reported null findings are in line with other studies (Arslan et al, 2021; Blake, Bastian, et al, 2017; Hone & McCullough, 2020; Schleifenbaum et al, 2021), they need replication given the limitations of Study 1: First, poor inter‐coder reliabilities for some dimensions suggest that they were very hard to detect based on the photographs, and that the measures were thus not reliable (e.g. wearing high heels).…”

“…While the null results of wearing red clothes are in line with results of Study 1 and previous studies (Blake,Dixson,10 Please note that eyeglasses were coded as 1 = yes and 2 = no. Hone & McCullough, 2020), this study did not replicate that women show less skin when fertile, as reported in Study 1. Neither did it replicate previous studies reporting that women wear sexier or more skin revealing clothes when fertile (in line with Arslan et al, 2021).…”

“…In each testing session, participants were given coloured pencils and a sheet of paper displaying an outline of a woman's figure, on which the participants were asked to draw outfits they would wear to a social gathering that same night (a party with a lot of attractive singles; Figure S4). This procedure was directly adapted from previous studies (Durante et al, 2008; Hone & McCullough, 2020). We excluded illustrations from one participant because of insufficient quality (drawing was so light that clothes were hardly recognizable), resulting in a total of 1024 drawings from 256 participants.…”

“…More precisely, the authors found a robust significant red effect predicted by the estradiol‐to‐progesterone ratio, and a fertile window effect (but not estradiol and progesterone separately) and interpret their results as being in line with the red effect. Two studies failed to find any compelling evidence for the red effect (Blake, Dixson, et al, 2017; Hone & McCullough, 2020), and two pre‐registered large‐scale diary studies did not find compelling evidence for cycle shifts in self‐reported sexy clothing (Arslan et al, 2021) or grooming (Schleifenbaum et al, 2021), casting doubt on previous findings.…”

“…The median sample size for within‐subject studies that reported (at least some) evidence for cycle shifts in clothing style or grooming was N = 43 ( N = 17 in Saad & Stenstrom, 2012, to N = 96 in Blake, Bastian, et al, 2017). Some studies even used between‐subjects designs to investigate a within‐subjects effect, leading to low test power despite generally large sample sizes (between N = 100 in Hone & McCullough, 2020, Study 1 and N = 617 in Study 2).…”