Intrasexual competition predicts willingness to engage in risky weight loss behaviors: sex differences and similarities

Abstract

Intrasexual Competitiveness has been theorized to be related to the development of anorexia nervosa (AN) and bulimia nervosa (BN) in women. A university sample was used to test the relationship between gender and intrasexual competition (ISC) on the willingness to engage in risky weight-loss behaviors symptomatic of AN or BN. A gender-neutral scale of intrasexual competition was created, allowing for a direct comparison of male and female respondents. The results revealed that greater intrasexual competitiveness in women (but not men) was related to an increased self-reported willingness to engage in risky dieting behaviors (e.g., going on a crash diet, fasting, and taking diet pills). Women and individuals higher in ISC expressed a greater willingness to vomit and/or use laxatives. These findings support the notion that ISC may in part contribute to the development of AN and BN in women.

Introduction

Exploring the etiology of eating disorders (ED) is widely studied, but may yet be missing crucial predictive components. Factors currently hypothesized to be related to the development of EDs include media exposure¹, internalization of the thin ideal², personality³, perfectionism⁴, neuroanatomical differences⁵, abnormal endocrine levels⁶, genetics⁷, and sexual abuse⁸. In the clinical literature, understanding the proximate mechanisms underpinning EDs and treating the symptoms has been given priority over exploring ultimate or evolutionary factors that might also contribute to their development. Ignoring ultimate factors risks missing valuable insights into how these EDs develop as well as potential interventions⁹.

Anorexia nervosa (AN) and bulimia nervosa (BN) are pernicious psychiatric disorders characterized by an intense fear of fatness, body dissatisfaction, and a persistent drive for thinness (DSM-V¹⁰). People with AN restrict food intake, exercise excessively, and maintain a very low body mass index (< 17.5), whereas those with BN engage in binge-purge cycles that includes “uncontrolled” eating followed by self-induced vomiting, laxative abuse, or prolonged exercise (DSM-V¹⁰). These eating disorders (EDs) carry a high mortality rate¹¹, and individuals may suffer severe physical (e.g., organ failure, dental problems), mental (e.g., depression, low self-esteem), and social consequences (e.g., stigma, social isolation).

We argue that evolutionary theory can help elucidate various enigmatic aspects related to AN and BN, including why they are more common in women and more prevalent in industrialized societies. While this topic has been debated within evolutionary psychology^12–16, it has received limited empirical investigation^17–20. Ostensibly, the symptoms of AN and BN (like the cessation of the menstrual cycle) should limit an individual’s ability to survive and reproduce. Accordingly, some researchers have theorized EDs are not adaptive for the symptomatic individual, rather the beneficiary is a dominant female that uses indirect aggression (leading in EDs) to suppress the reproduction of other females^14,15. Other researchers have suggested EDs are due to a mismatch between the current environment and our evolved tendencies^9,12. Indeed, behaviors shaped by evolution may not be adaptive in modern environments or when taken to an extreme. For example, the human taste for sugar now leads to a variety of health problems, but it was (and is) quite adaptive for hunter-gatherers who do not have access to manufactured carbohydrate-dense foods. We suggest that the underlying evolved processes that lead women to engage in social comparison, monitor their bodies, and attempt to manipulate their body’s shape, may push some individuals to engage in risky dieting practices, and thus increase their likelihood of developing an ED like AN or BN—especially in industrialized societies.

We hypothesized that AN and BN are at least in part a function of female intrasexual competition^9,12,21. While human intrasexual competition is often associated with males competing for reproductive access to females, females also compete with each other for access to high quality mates^22,23. Intrasexual competition in women, however, tends to manifest itself differently: through indirect—rather than direct—aggression (e.g., gossiping, social exclusion, and derogation)^22,24. Since physical attractiveness and body shape is an important component of a women’s mate value^25,26, it follows that appearance could play a central role in female competitive tactics. Predictably then, women (more so than men) seeking to out-compete their mating rivals should be more likely to take risks in order to lose weight to improve their appearance. Supporting this, research reveals that mating and competitive motives increased women’s willingness to tan and take dangerous diet pills²⁷. Given that being slender is generally considered more beautiful (especially in Western/industrialized societies), intrasexual competition may thus drive women to engage in risky dieting behaviors.

Why is thinness related to attractiveness? Part of the answer resides in the perception that slimmer bodies are more youthful²⁶. Youthfulness is a desirable trait in women²⁵ because it indicates that many years of reproductive fertility remain. Many lines of research in evolutionary psychology suggest that this leads to a desire to appear youthful^25,28. According to the Sexual Competition Hypothesis (SCH¹²), young women have an adaptive drive for thinness because it is a signal of youthfulness. This predisposes them to take actions in order to maintain a relatively more slender—and thus a more youthful appearance—than the other women in their environment. In pre-industrialized societies, an adolescent girl would generally live in an environment surrounded by adult women who were relatively heavier (especially around the hips and midsection) than the adolescent girls due to multiple pregnancies¹². Consequently, a young women’s set-point would be in the normal BMI range.

In industrialized societies, however, young women are often exposed to adult women (both in their immediate environments and the media) who have slender hourglass figures. Not only are industrialized societies more densely populated leading to greater exposure to other women, but the birth rates are also significantly lower than in traditional hunter-gather and agrarian societies. Consequently, older women are able to maintain a slender figure for longer than what would have historically been the norm. This results in a runaway process of internalizing progressively lower set-points across the generations, and in young women’s desire for increasingly thinner bodies¹². Additionally, an uptick in the use of cosmetic surgery and novel sociocultural forces, such as the use of social media, dating applications, and changing norms related to marriage and sex, may have increased the degree of female competition ^9,14,21], which could intensify this process.

Such evolutionary theories suggest that in order to be competitive, some women might engage in risky and dysfunctional dieting practices, including behaviors associated with AN and or BN, in an effort to attain an unrealistically slender body¹². Some percentage of these women may fully develop AN or BN (or both) as the physiological and psychological changes associated with extreme dieting may be difficult to escape^9,13. For example, Rantala et al. (2019) in their psychoneuroimmunological model, suggested that women living in sexually competitive and obesogenic environments (where hyper-rewarding, calorically dense ‘junk’ food is ubiquitous) leads to stringent dieting (to be thinner than other women) and consequently chronic stress. Chronic stress can increase cortisol levels and neuroinflammation, which may promote the development of OCD-like behaviors and EDs⁹. Thus, these EDs, while imposing a powerful fitness cost, are hypothesized to be—at least in part—an unintended outcome of female intrasexual competition in the realm of appearance.

The SCH-ED connection has received some preliminary support. Studies conducted by Abed and colleagues found that intrasexual competition was related to disordered eating behaviors in a non-clinical sample of female undergraduates^17,18 and to BN in a clinical sample²⁰. While these studies are suggestive, the Intrasexual Competition for Mates Scale (ISCM:¹⁸) used to measure intrasexual competition was potentially problematic. Example items include: “The skinnier I am, the more attractive I am to men,” “I work out or watch what I eat because I want a body that will impress men,” and “I exercise because I want to impress men.” These, and other items pertaining to clothing choices and attractiveness, overlap with concepts being measured in the dependent variables of these studies, such as body satisfaction, exercising, and dieting behaviors on the Eating Disorder Inventory, which could give rise to a spurious correlation between these scales.

Here, we use a university sample to test, in both men and women, the relationship between intrasexual competition and the willingness to engage in risky weight-loss behaviors that are often exhibited in individuals diagnosed with either AN or BN. We adapted previous measures to develop a gender-neutral measure of intrasexual competition. Critically, this allows a direct comparison of male and female respondents, and so affords a test of whether there is a stronger relationship between intrasexual competition and risky dieting inclinations in women versus men.

This article has two broad aims. First, we aim to assess the relationship between intrasexual competition and women’s risky weight-loss (dieting) behaviors in a way that overcomes the methodological shortcoming of the previous work. We predict that women scoring higher on intrasexual competition will also report more risky weight-loss behaviors. Second, we aim to assess the same relationship between men’s intrasexual competition and risky dieting behaviors, an important comparison given that the SCH is sex-specific in its implications. Across a variety of risky weight-loss behaviors, we predict both main effects of gender, with women showing greater levels of risky dieting behaviors than men, but more specifically interactions of ISC and gender, with ISC having a bigger effect for women than men. Western cultural norms and media depictions of male attractiveness emphasize lean muscularity, rather than overall thinness. Thus, males higher in ISC might increase their muscle mass by exercising. Although we see no a priori reason why men might not show some evidence of ISC motivating them to engage in disordered eating, the strong prediction is that risky weight loss behaviors, and especially those with the goal of calorie restriction, should be much more pronounced in women.

Method

Participants

Survey items were administered to a convenience sample of Introductory Psychology 101 students at a public university in the Southwestern U.S (n = 356). We used a department-wide survey to maximize the representativeness of the sample—although it is admittedly young and educated (M = 19.23, SD = 2.72; 98.5% were under the age of 23)—to attain a sample size that was equivalent to previous examinations of ISC (e.g.,²⁷) to maximize power. Participants were asked to select their gender from the following options: "Male”, “Female”, or “Other” (the “Other” option allowed participants to write in a response). Note that the survey only asked about “gender”, which could be interpreted by the participant as their biological sex or their gender identity since this was not specified. Most of the sample identified as heterosexual (n = 324) with a smaller portion identifying as bisexual (n = 18). We excluded respondents that reported homosexual (n = 11) or other sexual orientations (n = 3) from the sample because our scale asked about same sex competition. The scale could easily be adapted for measurement in these populations, but we failed to do so here.

Of those who participated, a smaller, randomly selected sample of the participants viewed the variables related to this study (n = 252). One additional person was excluded from the analysis due to missing data on the GNISC scale. Our final sample consisted of 251 participants that reported binary gender identities (male = 113, female = 138). Participants were able to select multiple ethnic identities which resulted in the following frequencies: White/Caucasian 66.3% (n = 167), 18.3% Hispanic/Latino (n = 46), 11.5% Asian/Asian-American (n = 29), 7.5% Black/African American (n = 19), 5.2% Middle Eastern/Middle Eastern-American (n = 13), and 2.0% Native American (n = 5). The relationship status of participants was mostly single 59.8% (interested in dating n = 103; not interested in dating n = 47), while 35.5% were currently in a relationship (dating exclusively n = 82; married/ engaged n = 7). A small minority of participants reported causally dating one or more people (n = 11, 4.4%) and one person selected “Other”.

Measures

Gender neutral intrasexual competition scale (GNISC)

Intrasexual competition was measured using 5 items that were adapted from Buunk & Fisher’s Intrasexual Competition Scale²⁹. Buunk & Fisher’s original scale is problematic, because it includes items that measure behavioral inclinations that may be the outcome of a person having an intrasexually competitive orientation (i.e., “I wouldn’t hire a very attractive man/women as a colleague”). Furthermore, 6 of the 12 individual items contained within their scale differed significantly between men and women in at least one of their samples . Note that, we are not suggesting that it is necessary for men and women to score equally on each item, since men and women may differ in their intrasexual competitiveness. Any observed differences between men and women, however, should not be driven by competitive tactics that have been found or hypothesized to be gender specific^30,31. For example, in Buunk and Fisher’s report²⁹, men and women differed in their responses to items such as: “I just don’t like very ambitious men/women” (men scored significantly higher than women in the Dutch sample), and “I can’t stand it when I meet another man/woman who is more attractive than I am” (women scored higher than men in Dutch and Canadian sample).

We refined the Buunk and Fisher scale to create a more gender-neutral version—the Gender Neutral IntraSexual Competition (GNISC) scale. Like Buunk and Fisher’s scale²⁹, the GNISC measures how strongly individuals perceive members of the same sex as rivals for romantic attention—however, we have removed items that assess competitive tactics that are hypothesized to be gender-specific. The aim was to improve the applicability of Buunk and Fisher’s original scale, ensuring it consistently captures this construct across different genders. The items: “I see other women/men as rivals,” “If I’m in a group of women/men, it is important for me to stand out”, “I compete with other women/men for male/female attention”, “I always want to beat other women/men”, and “I want to be better than other women/men”. All items were phrased to match the participant’s gender and were answered using a 7-point scale ranging from 1 (Strongly Disagree) to 7 (Strongly Agree). The 5 scale items were presented in randomized order.

Critically, the GNISC makes no explicit mention of physical appearance, body shape, status, or other domains in which people may compete, or of traits that members of the opposite sex might find desirable (e.g., attractiveness, status, a sense of humor, etc.). This is important for two reasons. First, it eliminates the overlap between our independent and dependent variables—e.g., we avoid using an item such as “I can’t stand it when I meet another man/woman who is more attractive than I am,” to predict appearance-based outcome variables. Second, our scale was designed to be gender neutral (Note that gender can in principle refer to biological sex or gender identity. In the current study we did not distinguish between the two.). Significant differences in men’s and women’s scores on our scale will not be due to differences in how men and women may compete against members of their own gender but will instead be due to differences in viewing members of the same gender as mating rivals—i.e., differences will be due to gender differences in intrasexual competition intensity, not strategy.

In the present data set, the five items load onto a single factor accounting for 66.0% of the variance in women (with principal axis factor loadings ranging from 0.707 to 0.792) and 62.4% of the variance in men (with principal axis factor loadings ranging from 0.645 to 0.755). Across gender, a preliminary series of measurement invariance analyses reveled that the GNISC items show both configural and metric (equivalence of factor-loadings) invariance—establishing weak invariance—but do not show strong invariance (invariance of intercepts). Measurement invariance analysis is a statistical method used to test whether a scale or set of items measures the same underlying construct across different groups (e.g., men and women). Strong invariance requires the item intercepts to be equal across groups, but in our data, this was not observed. This means that while men and women may report different mean levels of ISC, the underlying structure and relationships between the scale items remain consistent. Despite these mean differences, the evidence suggests that the scale measures the same construct in both groups. This is expected given that there is a mean gender difference, with men reporting 0.95 more ISC, on average, than women, t(249) = 5.84, p < 0.001. Despite these mean differences, because the factor loadings are the same, we can use the factor score as a covariate and make stronger inferences about differences in how this variable predicts disordered eating behaviors. Welzel and colleagues have argued that strong invariance (and even weaker varieties) can obscure important results in research proposing to explore group differences³². It is also important to note that although such unequal group means on the covariate could weaken the power of the significance tests in our analyses, significant results are still interpretable in the standard ways³³.

Risky dieting behaviors

To measure risky dieting behaviors, participants self-reported their willingness “to engage in the following behaviors in order to look good for an important upcoming event” on a 7-point scale ranging from 1 (Not at all willing) to 7 (Completely willing). There were six behaviors based on potentially risky dieting practices that are often associated with eating disorders, including: “Go on a crash diet,” “Exercise over 90 min per day,” “Vomit after eating,” “Fast (not eating)” “Take laxatives after eating,” and “Use diet (weight loss) pills.” The six items were presented in randomized order.

Notably, other problematic weight-loss strategies exist beyond what was included in our questionnaire. Furthermore, the risky dieting behaviors in this study are not sufficient nor specifically diagnostic of an ED as outlined by the DSM-V. In the DSM-V, criteria for AN includes restriction of energy intake leading to a significantly low body weight body, an intense fear of weight-gain or becoming fat, being disturbed by one’s body weight or shape or a lack of recognition of the seriousness of one’s low bodyweight. Criteria for BN includes, frequent episodes of binge eating that’s associated with a loss of self-control followed by inappropriate compensatory behaviors (e.g., vomiting, exercise, laxatives, diuretics, enemas, or diet pills). AN-purging type includes elements of both. Our short selection of items neither captures the full range of the behavioral criteria (e.g., binging), the accompanying psychological components (e.g., fear of weight gain, etc.), nor the duration that behaviors must be exhibited for diagnosis (e.g., purging once a week over a period of 3 months). Even so, we do feel our items are a reasonable proxy for some of the problematic behaviors associated with AN and BN. In addition, three of our items relating to vomiting, laxative use, and exercising are similar to those on the Eating Attitudes Test (EAT-26 [Part C; the behavioral component³⁴), which is a widely used scale that produces valid and reliable data to measure whether a person might have an ED.

Study design and procedure

A correlational study using self-report measures was conducted to investigate whether gender, individual differences in ISC, and the interaction (predictor variables) predicted a participant’s willingness to engage in six risky dieting practices (dependent variables). The items were embedded in a larger survey that was administered to the Introductory Psychology participation pool. Students completed the questionnaire online on their own personal computers after providing consent. Participants completed the GNISC scale before the Risky Dieting Behaviors. The Risky Dieting Behaviors questionnaire asked participants, “How willing would you be to engage in the following behaviors in order to look good for an important upcoming event”. Notably, participants were not given specific examples of what kind of event this would be or when it would take place—they were able to interpret the prompt freely. Following the prompt, participants indicated their willingness to engage in the six different risky behaviors “Go on a crash diet,” “Exercise over 90 min per day,” “Vomit after eating,” “Fast (not eating)” “Take laxatives after eating,” and “Use diet (weight loss) pills” in a randomized order. All experimental protocols were approved by Arizona State University Institutional Review Board and all methods were carried out in accordance with the Declaration of Helsinki and we obtained informed consent from all participants.

Results

The descriptive statistics and correlations for all the GNISC and Risky Dieting Behavior items are presented in Tables 1, 2 and 3 respectively. Multiple regression analyses were conducted to see if ISC (standardized across both genders), gender, and their interaction predicted an individual’s willingness to engage in the 6 risky dieting behaviors “in order to look good for an upcoming event”. Significant interactions were then examined using simple slopes analyses. Figure 1 displays these results as interaction plots.

Table 1.

Descriptive statistics for GNSC and risky dieting behavior items and means by gender.

Variable	M (SD)	95% CI		Mdn
		(LB)	(UB)
GNISC
1. I see other (men/women) as rivals
Male	3.28 (1.48)	3.01	3.56		4.00
Female	2.57 (1.58)	2.30	2.83		2.00
2. If I’m in a group of (men/women), it is important for me to stand out
Male	3.42 (1.46)	3.15	3.70		4.00
Female	3.17 (1.60)	2.90	3.44		3.00
3. I compete with other (men/women) for (men/women) attention
Male	3.38 (1.62)	3.08	3.68		4.00
Female	2.52 (1.60)	2.25	2.79		2.00
4. I always want to beat other (men/women)
Male	3.33 (1.71)	3.01	3.65		3.00
Female	2.06 (1.46)	1.81	2.30		2.00
5. I want to be better than other (men/women)
Male	4.53 (1.77)	4.20	4.86		5.00
Female	2.90 (1.78)	2.60	3.20		2.00
GNISC scale mean
Male	3.59 (1.25)	3.36	3.82		3.80
Female	2.64 (1.31)	2.42	2.86		2.40
Risky dieting behaviors
1. Fast (not eating)
Male	2.78 (1.69)	2.46	3.09		2.00
Female	3.50 (1.96)	3.17	3.84		3.00
2. Exercise over 90 min/day
Male	5.01 (1.69)	4.69	5.32		5.00
Female	4.27 (2.06)	3.92	4.62		5.00
3. Vomit after eating
Male	1.57 (1.16)	1.35	1.79		1.00
Female	1.66 (1.33)	1.43	1.88		1.00
4. Take laxatives
Male	1.72 (1.32)	1.48	1.97		1.00
Female	1.78 (1.47)	1.53	2.03		1.00
5. Use diet (weight loss) pills
Male	1.89 (1.40)	1.63	2.16		1.00
Female	2.38 (1.93)	2.05	2.70		1.00
6. Go on a crash diet
Male	2.27 (1.57)	1.97	2.56		2.00
Female	2.47 (1.90)	2.15	2.80		1.00
Risky dieting behaviors mean
Male	2.54 (1.00)	2.35	2.73		2.17
Female	2.68 (1.28)	2.46	2.89		2.50

Table 2.

Correlations between GNISC and the 6 risky dieting behaviors (Men, n = 113).

	1	2	3	4	5	6
1. GNISC	–
2. Fast (not eating)	−0.04
3. Exercise over 90 min	0.11	−0.01
4. Vomit after eating	0.27**	0.45***	0.07
5. Take laxatives after eating	0.24**	0.45***	0.04	0.88***
6. Use diet (weight loss) pills	0.14	0.32***	0.01	0.54***	0.61***
7. Go on a crash diet	0.18	0.33***	0.16	0.53***	0.61***	0.68***

*p ≤ 0.05 (two-tailed), **p ≤ 0.01(two-tailed), ***p ≤ 0.001(two-tailed).

Table 3.

Correlations between GNISC and the 6 risky dieting behaviors (women, n = 138).

	1	2	3	4	5	6
1. GNISC	–
2. Fast (not eating)	0.21*
3. Exercise over 90 min	0.10	0.30***
4. Vomit after eating	0.23**	0.48***	0.16
5. Take laxatives after eating	0.35***	0.46***	0.23**	0.67***
6. Use diet (weight loss) pills	0.33***	0.49***	0.19*	0.50***	0.56***
7. Go on a crash diet	0.40***	0.55***	0.27***	0.46***	0.53***	0.66***

*p ≤ 0.05 (two-tailed), **p ≤ 0.01(two-tailed), ***p ≤ 0.001(two-tailed).

Fig. 1.

Interaction between gender and ISC (standardized; plotting from −1 to + 1 SD) with Willingness to engage in risky dieting behaviors.

Fast (not eating)

A significant regression equation was found, F(3, 247) = 5.91, p = 0.001, R² = 0.067. There was a significant effect of gender (β = −0.227, 95% CI [−0.36, −0.11], p < 0.001, with female respondents’ mean endorsement 0.74 points higher than males’). There was no effect of GNISC (β = 0.097, 95% CI [−0.03, 0.23], p = 0.144), but this was qualified by a significant interaction (β = −0.127, 95% CI [−0.25, −0.004], p = 0.043). Simple slopes within gender revealed a significant effect of GNISC for women (β = 0.213, 95% CI [0.05, 0.38], F[1,136] = 6.45, p = 0.012, R² = 0.045): for every standard deviation unit-increase in the standardized score, endorsement of the use of fasting increased 0.435 points, but no effect for men (F[1, 111] = 0.167, p > 0.2). Women thus show both a greater tendency to fast in order to look good for an important event, and that those with higher levels of ISC (as indexed by GNISC) show a greater propensity to do so.

Exercise over 90 min per day

A significant regression equation was found, F(3, 247) = 4.15, p = 0.007, R² = 0.048. There was a significant effect of gender (β = 0.160, 95% CI [0.03, 0.29], p = 0.017, with male respondents’ mean endorsement 0.74 points higher than females’). There was no effect of GNISC (β= .105, 95% CI [-.03, .24], p= .119) or the interaction (p > 0.2).

Vomit after eating

A significant regression equation was found, F(3, 247) = 5.35, p = 0.001, R² = 0.061. There was a significant effect of gender (β = −0.133, 95% CI [−0.26, 0.003], p = 0.044) and a significant effect of GNISC (β = 0.261, 95% CI [0.13, 0.39], p < 0.001), with female respondents’ mean endorsement 0.09 points higher than males’. There was no significant interaction, p > 0.2. Thus, women show evidence of a greater propensity to vomit after eating relative to men, but higher GNISC scores predict a greater propensity as well and accounts for substantially more variance.

Take laxatives after eating

A significant regression equation was found, F(3, 245) = 8.86, p < 0.001, R² = 0.098. There was a significant effect of gender (β = −0.128, 95% CI [−0.26, 0.00], p = 0.05), with female respondents’ mean endorsement 0.06 points higher than males’. There was also a significant effect of GNISC (β = 0.317, 95% CI [0.19, 0.45], p < 0.001). There was no significant interaction, (p > 0.2). Thus, women show a greater propensity than men to take laxatives after eating, but higher GNISC scores also predict a greater propensity for both men and women.

Take diet (weight loss) pills

A significant regression equation was found, F(3, 247) = 9.05, p < 0.001, R² = 0.099. There were significant effects of both gender (β = −0.2230, 95% CI [−0.35, −0.10], p <0.001, with female respondents’ mean endorsement 0.49 points higher than males’) and GNISC (β = 0.252, 95% CI [.12, .38], p < 0.001). However, these were qualified by a significant interaction (β = −0.122, 95% CI [−0.24, −0.002], p = 0.047). Simple slopes within gender revealed a significant effect of GNISC for women (β = 0.328, 95% CI [0.17, 0.48], F[1, 136] = 16.43, p < 0.001, R² = 0.108): for every standard deviation unit-increase in the standardized score, endorsement of the use of diet pills increased 0.66 points. There was no effect for men (F[1, 111] = 2.14, p = 0.146). Women thus show both a greater tendency to take diet pills to look good for an upcoming event, and that those with higher levels in the GNISC show an increased willingness to do so.

Go on a crash diet

A significant regression equation was found, F(3, 247) = 11.03, p < 0.001, R² = 0.118. There were significant effects of both gender (β = −0.163, 95% CI [−0.29, −0.04], p = 0.011, with female respondents’ mean endorsement 0.20 points higher than males’) and GNISC (β = 0.314, 95% CI [0.19, 0.44], p < 0.001). However, these were qualified by a significant interaction (β = −0.129, 95% CI [−0.25, −0.01] p = 0.034). Simple slopes within gender revealed a significant effect of GNISC for women (β = 0.401, 95% CI [0.25, 0.56], F[1, 136] = 26.04, p < 0.001, R² = 0.161): for every standard deviation unit-increase in the standardized score, endorsement of crash diets increased 0.79 points. There was a trend toward this effect for men (β = 0.181, 95% CI [−0.004, 0.37], F[1, 111] = 3.77, p = 0.055, R² = 0.033). Women thus show a greater willingness to go on a crash diet in order to look good for an upcoming event, and that those with higher levels of ISC (as indexed by GNISC) show an increased willingness to do so, an effect that accounts for substantially more variance in the outcome than we see for men.

Discussion

Anorexia nervosa (AN) and bulimia nervosa (BN) in women have previously been linked to several proximate social and biological explanations, but empirical investigations exploring ultimate or evolutionary explanations are sparse in the research literature. We sought to test the evolutionarily-grounded hypothesis that intrasexual competition—rivalry between members of the same gender for access to mates—is related to a willingness to engage in behaviors commonly associated with AN and BN. We created a gender neutral intrasexual competitive scale (GNISC) to measure individual differences and then directly compared the strength of this relationship in female and male respondents. The results of the study largely confirm the predictions.

As hypothesized, greater intrasexual competitiveness in women (but not men) was related to an increase in self-reported willingness to engage in risky dieting behaviors, such as going on a crash diet, fasting (not eating), and taking diet pills to look good for an upcoming event. Furthermore, women’s willingness to use laxatives and/or vomiting was greater than men’s. In general, people scoring higher in ISC also reported a greater willingness to use laxatives and/or vomiting after eating. Intriguingly, the interaction between gender and ISC was not significant for a willingness to take laxatives or vomit after a meal, which are common symptom of BN or purging type AN. Whether men tend to use these tactics to improve their physique should be investigated further.

Our results lend support to the theorized connection between female ISC and eating disorder behaviors. These results are in line with studies investigating this relationship^15,17–20, as well as research showing that priming sexually competitive motives may increase women’s willingness to tan, take diet pills, and to minimize the potential negative consequences of doing so²⁷. Our study, however, is the first to use a gender-neutral scale devoid of references to sex-specific competitive tactics, and the first to find evidence of some of these effects in heterosexual men.

Generally, young men tend to be more tolerant of and engage in risker behaviors than women, including performing stunts (e.g., attempting risky skateboard tricks³⁵ and aggressing against others [male-male homicide, ^36,37). While ISC may lead women to take appearance-based risks associated with weight loss and disordered eating, we recognize that men are also concerned about their appearance. In the current study, women reported a greater willingness to engage in all risky dieting behaviors except for exercising for over 90 min per day, which was reported more by men. Although the interaction was non-significant—i.e., men higher in ISC did not endorse exercising more than those lower in ISC—it is a reasonable hypothesis that male-male ISC might inspire certain appearance-based risks. Research by Frederick et al. found that a high percentage (> 90%) of undergraduate US men surveyed desired greater muscularity—a result that was echoed by samples from Ukraine and Ghana³⁸. Conceivably, individual differences in male ISC may predict a preoccupation or dissatisfaction with physical traits that signal dominance or “virility”, such as height, muscularity, body shape (shoulder-to-hip ratio), and penis size. Men can attempt to change their body shape and composition via exercise (e.g., weightlifting), diet/ supplementation, hormonal injections, anabolic steroids, as well as through cosmetic surgery—including pectoral implants, abdominal etching, jaw/ chin enhancement, liposuction, penis enlargement, and even limb lengthening. Whether individual differences in ISC among men and women predicts a willingness to modify one’s body surgically (even if it poses a physical or financial risk) ought to be tested. The association between ISC and a desire for cosmetic surgery in women has been hypothesized³⁹ and has received preliminary support in both women and men⁴⁰.

Notably, our results do not suggest that men lack a “fear of fatness” or never engage in eating disordered behaviors. In this study, ISC correlated with a willingness to use laxatives and vomiting in women and men. Eating disorders in men are currently increasing⁴¹, and there has been a call for greater empirical investigations and designing different ED assessment tools for men^42,43. Interestingly, homosexual men are overrepresented in clinical samples of men with EDs⁴⁴. Intrasexual competition primes (i.e., individual profiles that had competitive and high-status cues) led to worse body image and eating attitudes for gay men (which was similar to the results among heterosexual women), but not lesbian women¹⁹. Given that homosexual men also emphasize the importance of looks in mate choice⁴⁵ and that EDs in homosexual men may be driven by body dissatisfaction⁴⁴, future studies should examine whether ISC moderates the relationship between body dissatisfaction and EDs in this population.

The present study had several limitations. Our sample is young and WEIRD—Western, Educated, Industrialized Rich, and Democratic⁴⁶. While a college population can be ideal for ISC research—university-aged men and women are often seeking romantic partners and campuses are replete with potential mates and rivals—the results might fail to generalize to the broader population or to other cultures. Although ISC may predict a women’s desire to be relatively slender in modern, Western/industrialized societies, this outcome of female-female competition is malleable. By refining the Buunk and Fisher ISC scale²⁹] to tap into pure, same-sex rivalry (without sex-specific—and perhaps culturally specific— methods of competing), we feel this gender-neutral version can help propel research on ISC in underrepresented samples and observe how ISC manifests in different cultures.

Evolutionary psychologists, while often interested in human universals, recognize that environmental factors, individual differences, and social conventions can influence how evolved tendencies present. For example, a recent set of studies⁴⁷ found that US state-level operational sex ratios predicted Google search terms for cosmetic surgery while experimentally manipulated (female biased) sex ratios predicted a women’s desire to purchase appearance-enhancing products in a Chinese sample. Cultures may also have different beauty standards. For example, fair skin is associated with beauty and social status in several Asian cultures. Consequently, the Asian-Pacific market for skin lightening products accounts for roughly 54% of all global revenue (US$7.05 billion in 2021), and the market base is overwhelmingly female (approximately 80%)[⁴⁸](). Accordingly, ISC would be hypothesized to be related to dissatisfaction with one’s skin and a greater willingness to purchase such products. Even within a country, such as the United States—which is a highly individualist and industrialized country—ethnic background and socioeconomic status might be related to differential preferences for an ideal body type⁴⁹. Given this, women from varying backgrounds may predictably desire to “improve” certain physical aspects over others (e.g., weight, buttocks, breasts, etc.).

The operationalization of the outcome variable whereby participants self-reported their willingness to engage in six risky dieting behaviors (fasting, exercising for 90 min, taking laxatives, vomiting after eating, taking diet pills, and going on a crash diet) in order to “look good for an important upcoming event” also had several limitations. Firstly, self-report responses may differ from actual behavior, and secondly, the prompt of an “important upcoming event” is ambiguous. The vagueness of the prompt allowed participants to imagine a scenario that was personally relevant and realistic, however, not all events will evoke competitive and mating related motives. For example, a family reunion compared to a party (where potential mates and rivals might be), should influence the degree of appearance-based risk-taking. Future research should explore how the type of event moderates the effect.

Thirdly, the risky dieting behaviors used in this study, while being related to disordered eating, are not specifically diagnostic of AN or BN. Future studies could utilize validated scales, such as the EAT-26[³⁴] or the Eating Disorder Inventory (EDI-3, [⁵⁰). Given that our questions were embedded in a larger, department-wide survey for introductory psychology students, we could only administer a small number of minimally invasive questions. The EAT-26 and EDI-3 are longer questionnaires that provide a more comprehensive assessment of whether a person may have an ED by measuring concepts such as body image and dysmorphia, preoccupation with food, drive for thinness, guilt associated with eating, self-control, and more. While we suspect that ISC is related to several of these constructs, the current study focused on a willingness to engage in the behavioral aspects of EDs rather than the more complex psychological components since we had a non-clinical sample and were specifically interested in how evolutionary based motives can lead people down a risky pathway.

Fourthly, the current study measured whether a participant’s gender and ISC score predicted their willingness to engage in behaviors representative of disordered eating to “look good for an upcoming event”—not that they actually have an ED or that they would be willing to continue these behaviors over a long period of time. Critically, people who are willing to, for example, fast to meet a short-term goal are not directly comparable to those diagnosed with an ED. EDs are complex psychiatric disorders that tend to co-occur with other mental health issues, including anxiety, depression, substance use disorders, obsessive–compulsive disorder (OCD), and posttraumatic stress disorder (PTSD). In addition, according to the DSM-V, ED symptoms must be displayed for a certain length of time (e.g., for BN, binge/purging at least once week for 3 months). Our study cannot differentiate between a person’s short-term versus long-term willingness to engage in risky dieting behaviors. Given the significant relationship between gender and ISC on the willingness to engage in certain risky weight loss practices, it could be hypothesized that in a subset of more suspectable individuals, a higher degree of ISC may increase the odds of developing a subclinical or clinically diagnosable ED. Whether it does or does not in practice merits further research. Relatedly, future studies should investigate whether being diagnosed with an ED moderates the relationship between ISC and the behavioral (as well as the psychological) components of EDs. Likewise, the type of eating disorder may also matter. The role ISC plays in the development or persistence of AN, BN, binge-eating disorder (BE), or other specified feeding or eating disorder (OSFED) is not clear.

Lastly, the risky dieting behaviors in this study carry different levels of harm and benefits. For example, fasting (e.g., calorie restriction and time-restricted feeding) may be associated with beneficial effects, such as reduced inflammation, improved circadian rhythm, increased stress resistance, autophagy, and promotion of beneficial gut bacteria⁵¹ when not taken to an extreme (which can lead to malnutrition and severe weight loss). On the other hand, purging (abusing laxatives and vomiting after eating) carries a higher cost. Future studies could experimentally vary the degree of perceived risk and rewards associated with certain weight loss or appearance enhancing methods to see if intrasexually competitive individuals may still be willing to try techniques that are framed as potentially dangerous.

Collectively, our findings suggest a significant new line of inquiry into the etiology of eating disorders. Future studies on eating disordered behaviors can investigate how experimentally manipulating mating and competitive motives (e.g.,^19,27,47) interact with individual differences in GNISC. Mating primes or the presence of rivals ought to increase a desire to restrict or purge food in individuals most sensitive to these cues (i.e., those higher in ISC). Future studies could also explore whether individual differences in ISC predict how women (and men) respond to seeing idealized figures. For example, Garza and Pazhoohi [52] found that more competitive women (measured by Buunk and Fisher’s scale²⁹) reported a higher likelihood to diet and exercise after viewing images of breasts generally. Additionally, when exposed to larger breast sizes (C and D-cups), participants higher in ISC expressed a greater likelihood to enhance their appearance. Future studies could manipulate other variables such as weight, waist-to-hip and shoulder-to-hip ratio, and other typically feminine (e.g., full lips) or masculine (e.g., muscularity) features to see how it interacts with GNISC to influence appearance based risk-taking in men and women. Future research could also investigate how ISC may vary overtime or between contexts. Research by Reynolds and colleagues⁵³ found that both the perception and experimental manipulation of female-skewed sex ratio corresponded to greater intrasexual competitiveness, body dissatisfaction, and dieting inclinations. Continued investigations of this in populations with highly skewed sex-ratios (e.g., female college dorms) using mixed method approaches could be insightful to see how ISC and mating motives are related to risky dieting practices and EDs. Similarly, one’s relationship status or use of hormonal birth control could influence a person’s competitive tendencies. Importantly, future research should not only examine whether different EDs are related to ISC, but also what proximate and biological mechanism help explain this relationship (e.g., individual differences in stress responsivity, up or down regulation of the serotonergic system) as suggested by Rantala et al. (2019) [⁹].

Although the generalizability of our study is limited by using a non-clinical sample and does not directly test whether ISC is related to the development of AN, BN, or other forms of EDs, it does, however, take a necessary first step. The results suggest that women higher in ISC might be more willing to engage in extreme and risky dieting behaviors to appear more attractive. Conceivably, for some women these behaviors may go too far, which could result in unhealthy eating patterns, extreme weight loss or fluctuations, or even EDs. We are not claiming that ISC in women is the only or even the predominant reason behind why women diet or develop EDs as this is likely the result of several interactions of personal motives, environmental causes, and genetic/biological influences. We do suggest, however, that ISC may be an overlooked factor that pushes some women to engage in risky behaviors, therefore increasing the likelihood of developing an ED, which merits further investigation. We agree with researchers that have called for a more comprehensive investigation of how (proximate mechanisms) and why (ultimate or evolutionary explanations) mental disorders manifest as they do in modern society⁵⁴, including EDs⁹. Given the increasing prevalence and severity of EDs like AN and BN, the results underscore the need to consider ultimate causes in addition to the widely studied proximate ones. Although only part of the complex picture, research into the hidden influence of intrasexual competition could point the way to promising new interventions and therapeutic approaches (Supplementary Information).

Author contributions

S.D. developed the research question and theoretical basis. S.D. and D.B. developed the questionnaire with D.B. and C.P. collecting the data. D.B. and C.P. conducted the main analyses with D.B. creating the figures. S.D. wrote the main manuscript (Abstract, Introduction, Method, Discussion). D.B. and C.P. wrote the Results. All authors provided feedback and revisions to the main manuscript. All authors reviewed the final manuscript and agreed to submission.

Data availability

All data generated or analyzed during this study are included in this published article (and its Supplementary Information Files).

Competing interests

The authors declare no competing interests.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-86985-x.