Terminology

Before the science is discussed, it is helpful to define terms.

Sex refers to a person’s physical characteristics, including their reproductive system (ie, whether they have ovaries or testes), hormones, chromosomes (classically male XY and female XX) and external genitalia. Sex is most commonly determined on the basis of external genitalia. Intersex people (ie, those with differences of sexual development (DSD)) are born with reproductive anatomy, chromosomes and/or hormones that cannot be straightforwardly categorised as male or female.

The conflation of transgender and intersex people leads to confusion around effective testosterone versus testosterone levels.3 This article solely discusses transwomen, and therefore our arguments are not complicated by androgen insensitivities in which receptors are not sensitive, or only partially sensitive to testosterone.3 11–13 This review only requires discussion of testosterone levels and prior male physiology.

In the scientific literature, the terms ‘male’ and ‘female’ are used to refer to biological sex. Outside of science, ‘male’ and ‘man’ are used interchangeably, as are ‘female’ and ‘woman’. We follow both these conventions, except in cases in which doing so does not read well in everyday English.

Gender refers to one’s sense of self as a man or a woman (or something else such as gender-neutral or gender-fluid). Cis-women are those women whose sex and gender align—they are born female and identify as such. Equally, a cis-man’s sex and gender are both male. For a minority of people, sex and gender do not align. A person may be of the male sex but identify as a woman, or be of the female sex but identify as a man. These people are transgender. ⁱ A transgender person may or may not undergo transition, including social changes (coming out to friends and family, changing one’s name, personal pronoun, style of dress) and medical intervention (hormone therapy, gender affirmation surgery).

A transwoman who has had full gender affirmation surgery (including testes removal) will have very low testosterone levels below 1 nmol/L.14 These transwomen will have much less opportunity for a performance advantage in comparison to elite cis-women athletes. Henceforth, unless specifically stated, we use the term ‘transwomen’ to refer to those elite transwomen athletes who have testosterone levels from 6 to 10 nmol/L (primarily in cases in which a transwoman has retained her testes). Equally, we use the term ‘cis-women’ to refer to elite cis-women athletes.

Unless stated otherwise, this article solely concerns elite sport. We use Swann et al’s 15 definition of elite athlete which includes athletes competing at national, professional or university (in USA) levels, either paid or unpaid7 15 16.

Finally, inclusion refers to the idea that all athletes should be included.7 Fairness in sport can be used in a broad sense in which inclusion is considered an aspect of fairness.16 However, to clearly illustrate the conflict between the principles of fairness and inclusion in sport, for the purposes of this article we will not consider inclusion to be an aspect of fairness. Instead, we use fairness more narrowly to refer to the idea that all athletes must begin from roughly the same starting point.

Difficulty talking about this topic

Discussing transwomen’s inclusion in elite women’s sport is a difficult conversation. Even asking whether transwomen have an unfair performance advantage over cis-women might be offensive to some. Any discussion questioning the inclusion of transwomen may be dismissed as transphobic, prejudiced or coming from a lens of cis-normativity. We recognise the importance of including transpeople in society, including sport, and acknowledge the difficulties faced by many transpeople in establishing safe spaces in everyday life (Bagger17 18). However, it is critical that the guidelines attend to both fairness and inclusion to meet the needs of both transwomen and cis-women.

Guidelines

Science

Two fundamental assumptions emerge from the IOC guidelines.

1. High testosterone levels provide an all-purpose benefit in sport.

2. Transwomen with testosterone levels under 10 nmol/L for 1 year have mitigated the performance advantage of their former male physiology.

Arguments for and against inclusion

We now present the arguments for inclusion of transwomen in the women’s category in elite sport, and each counterargument:

1. Those transwomen athletes who meet the criteria set by the IOC should compete in the women’s division.

2. Transwomen identify as women and so should be able to compete in the women’s division.

3. The advantage that transwomen have is a tolerable unfairness.

4. As transpeople face discrimination in many spheres of life, they should compete in the gender division in which they identify.

5. The science has been developed using a cis-normative lens.

6. The male/female binary is socially constructed, so athletes ought to be able to switch between men’s and women’s divisions.

­

(1) Transwomen athletes who have met accepted IOC criteria should compete in the women’s division. These criteria should be followed either because (A) they are fair. Or (B) because they are the established eligibility criteria.

The first point could be countered by showing that the eligibility criteria for the women’s category are unfair because, as explained in the science section, transwomen have a significant performance advantage from being able to compete with testosterone levels just under 10 nmol/L and because there has been no thorough exploration of the advantage of having a prior male physiology (for the elite athlete). As Teetzel points out, ‘adding a rule or eligibility requirement in a policy or rulebook does not make the rule fair, even if the sports organisation can legally enforce the addition’.59

The claim that transwomen have a performance advantage is supported by evidence that testosterone in men is a driving force for the striking dominance of men in elite sport. However, currently women with hyperandrogenic states (having excessive levels of the androgenic hormones including testosterone, such as DSD) compete in the women’s division without restrictions. This is despite evidence, although contested, that the high testosterone levels lead to a performance advantage.60–62 Both DSD (excess hyperandrogenism) and PCOS (mild hyperandrogenism) are overrepresented among elite women athletes when compared with the wider population.60 63 64 Recently, based on evidence that hyperandrogenism is associated with performance advantage, the IAAF have lowered the acceptable level of testosterone to 5 nmol/L for some ‘restricted’ events.65 66 Hence, the eligibility criteria for transwomen may also be unfair to cis-women, and therefore the guidelines need further attention.

Advocates of inclusion could then argue that transwomen athletes who meet the criteria should be able to compete even if the criteria are unfair and afford transwomen a performance advantage because hyperandrogenism already exists in the women’s division. As elite sport has eligibility criteria for transwomen, we agree that until a change is made, transwomen who meet such criteria should compete in the women’s division.

­

(2) Inclusion is an important principle in society. Transwomen identify as women and as such should be included within the women’s division in elite sport.

The most common counterargument used by the popular press and cis-gender athletes and their coaches is that the inclusion of transwomen is unfair because they have a physiological advantage. As noted above, we take fairness to refer to the idea that all athletes must begin from roughly the same starting point. As Simon explains, ’if one individual has a competitive advantage over another due to differences in initial circumstances that were under the control of neither, the favoured individual can claim no credit for the successes that flow from that undeserved head start’ (18 67; see also Sher 1979, Loland, 1999). For example, it would be unfair to allow a competitive cyclist to use a hidden electric motor, where others do not. The principle of fairness is the reason why there are sporting regulations prohibiting certain technological advancements (swimsuits using sharkskin technology to reduce drag) or standardising equipment for all (fibreglass pole vaults). The purpose of such standardisation is to level the playing field with regards to that particular concern so the most skilful athlete wins.

The principle of fairness relies on acceptance of the ‘skill thesis’ which states that the purpose of competitive sport is to identify who is the most skilful.3 67 Neither Simon nor Bianchi define skill with respect to sport. It is incredibly difficult to pin down the meaning of skill. However, we take the most skilful athlete to be the one who maximises natural talent via training).3 We recognise that this definition is problematic because, as discussed below, some natural talents (advantages) are deemed fair while others are not.

Nor do Simon or Bianchi explain how to determine the most skilful athlete. The most skilful athlete is not always the winner of the competition in question. This is because who wins a competition is also influenced by luck,67 68 such as the bounce of a ball or gust of wind. The eligibility criteria also have a part to play. For example, if a heavyweight boxer was eligible to compete against a bantamweight fighter, the heavyweight would almost invariably win yet not necessarily be the most skilful. If the eligibility criteria are unfair, the winner may not be the most skilful.59 As stated above, fairness requires that all athletes begin from roughly the same starting point. If a transwoman has a performance advantage over her cis-women competitors then she may win the competition yet not be the most skilful.

­

(3) Sport is not a level playing field as there are significant physical and psychological differences between people. Furthermore, economic and social factors give rise to differences in funding and resources available to athletes. Both these mean that sport is not fair. If sport is not fair, then the advantage transwomen have should be accepted as, to use Devine’s terminology, ‘tolerable unfairness’.69

It is often claimed that sport is uneven due to socioeconomic factors; that athletes from a wealthy country will have resources available that provide a competitive advantage.70 This is considered a tolerable unfairness. In addition, many physical differences between people are due to the genetic lottery. Some women are very short and some very tall, some have shorter limbs and some longer. Some of these differences are examples of tolerable unfairness, such as the advantage that tall people have in basketball and the advantage that short-limbed people have in weightlifting. The Finnish skier Eero Mäntyranta had a genetic mutation that enabled him to carry more oxygen in his blood which is known to be performance enhancing.71 Mäntyranta was able to compete despite his significant advantage—this was a tolerable unfairness. This suggests ‘prevailing conceptions of fair play seem unconcerned about the effect of the ‘natural lottery’’.72 Equally, those with PCOS with higher than average standard testosterone levels for cis-women compete within the women’s division suggesting their advantage is considered a tolerable unfairness. By logical extension, it could be argued that transwomen levels should also be accepted within women’s sport—as a further tolerable unfairness.

It is not entirely clear how to distinguish between a tolerable and an intolerable unfairness. Bianchi uses high testosterone levels as an example of something that may be an unfair advantage (ie, an intolerable unfairness). Bianchi says this advantage may be unfair because cis-women cannot attain the advantages afforded by high testosterone levels via endogenous means.3 Nor are cis-women permitted to take exogenous testosterone to raise their levels to those of transwomen73 in the women’s division. This suggests that an advantage is unfair if no member of the category (eg, cis-women) can attain that advantage.ⁱⁱⁱ

Another way to distinguish between tolerable and intolerable advantages is to consider whether the property under consideration provides an all-purpose benefit. High testosterone levels are beneficial in almost all sports (an all-purpose benefit) whereas other advantages are beneficial in only some. For example, a short person is highly unlikely to become an elite basketballer but may excel in gymnastics. Individuals of varying height can self-select into sports that suit their physiological make-up, whereas people with average or low testosterone levels cannot. Hence, it is arguable that height is a tolerable unfairness, whereas high testosterone levels (being an all-purpose benefit) is an intolerable unfairness. High testosterone levels and prior male physiology provide an all-purpose benefit, and a substantial advantage. As the IAAF says, "To the best of our knowledge, there is no other genetic or biological trait encountered in female athletics that confers such a huge performance advantage".66

Combined, these three factors (cis-women cannot attain the advantage, all-purpose benefit and magnitude of the advantage) provide a strong argument that transwomen have an intolerable advantage over cis-women. In turn, the impact testosterone has on performance might lead us to reconsider the divisions currently used in competitive sport. In addition, if Mäntyranta’s genetic mutation or coming from a wealthy nation are also examples of an intolerable unfairness, these would need to be accounted for to maintain the skill thesis. Inconsistency may arise if we restrict transwomen from competing in the women’s division but allow those from wealthy nations to compete without restriction.72 This is not to suggest that transwomen (and others such as Mäntyranta or those from wealthy nations) have an easy ride in elite sport. All elite athletes possess a great deal of skill, enhanced by hard work and training. Instead, our point is that some have a head start on their competitors and that the head start ought to be accounted for.

­

(4) As discussed earlier, transpeople, indeed many groups outside the binary distinctions of male and female, have consistently experienced discrimination and exclusion in all spheres of life. To address these concerns transpeople should compete in the gender division in which they identify.

We fully accept that transpeople have experienced discrimination and transphobia in many aspects of life, and this is unconscionable. While there is a need to ensure there is space for transwomen in elite sport, it does not follow that the space afforded to transwomen should automatically be in women’s sport. Cis-women have faced, and still face, difficulties being recognised in top-level sport. The 2020 Olympics will be the first in which all events in all sports will have a women’s category.74 Despite this, some cis-women still struggle for space in elite sport. The first year all 206 Olympic nations sent women to the Olympics was 2012, although many nations do not send women in equal numbers. While there has been dramatic growth in women’s sport, ‘women of this century (the twentieth century) have occupied only a marginal space in the sports world’.75 Elite women athletes still face discrimination—they are paid less, receive less media coverage, and when they do, discussion is often about their appearance, rather than athletic ability, and so they contend with ‘the ordeal of not being taken seriously as … athlete[s]’74 75. In short, women have not achieved gender equity in most sports76 so care should be taken with any guidelines that may undermine this.

Gender separation provides a dedicated and safe space in which women can compete77–79 although not all agree that this is desirable for women’s sport.80–82 Cis-women may also be discouraged from direct competition against transwomen. For example, New Zealand weightlifter Tracey Lambrechs reportedly dropped a weight class to avoid competing against Laurel Hubbard because she perceived that Hubbard had a performance advantage because of Hubbard’s previous male physiology.83 ,^iv Having transwomen compete against cis-women might erode the space painstakingly built for cis-women. Clearly, transwomen have not had an easy ride in engaging in elite sport, and so we need to negotiate a space in which they can compete fairly. However, that place is not the women’s category.

 It could be argued that the science has been interpreted using a cis-normative lens meaning that it has been carried out by people who have a (perhaps unconscious) bias towards finding that transwomen have a performance advantage, and so are drawn to results that confirm this bias.84

As Ritz explains, employing this binary understanding of sex in experimental systems inherently creates a situation in which we seek difference and which thereby plays into and reinstantiates essentialist and biologically determinist biases. At a fundamental level, experiments are implicitly (and statistically) geared to seeking difference because difference is mechanistically informative in a way that sameness is not. … The upshot is that in treating sex as a unitary dichotomous variable and comparing women and men, we are allowing stereotyped thinking about sex and gender to substitute for the mechanistic understanding that is presumably the goal of experimental research, doing a disservice to people of all genders with respect to our understanding of biology and health.85

If a cis-normative lens has affected the scientific approach, and if it is found that transwomen do not have a performance advantage, then this would create a counterargument to the fairness claim. However, there may not be a cis-normative lens. Even if there is such a lens, its existence does not necessarily mean that transwomen do not have a performance advantage. That is, science carried out in the absence of a cis-normative lens might also show that transwomen have a performance advantage. Finally, it might be impossible to do science without a lens, be it cis-normative or something else.86 If so, and if non-cis-normative science shows that transwomen do not have a performance advantage, then it does not straightforwardly follow that the science that shows ‘transwomen do have a performance advantage’ is wrong. Instead, it means that claims about truth become much more complex. In short, until there is evidence that the cis-normative lens has generated inaccurate findings, it is justifiable to claim there is strong evidence (although some of it is indirect) to suggest that transwomen have a performance advantage over cis-women. This argument requires us to have solid data on elite transwomen athletes, so we do not have to interpret and extrapolate from male data from which a cis-normative bias could emerge.

­

(5) It might be argued that the categories of man and woman are socially constructed—that is, they are not natural kinds.87 ,^vSocial constructionists might claim that male and female only appear to be natural kinds because the world is interpreted through a cis-normative lens which is entrenched in our everyday lives. If the categories of male and female are socially constructed then this supports an athlete switching from the men’s to women’s division and vice versa.

There are two possible counterarguments. The first would be to show that the male/female binary does exist in the natural world. This is outside the scope of the paper. The second is to claim that elite athletes should not be able to switch between divisions because the binary is a pragmatic categorisation as it preserves space for women. In other words, the binary might be socially constructed, yet not arbitrary. If the binary serves a pragmatic purpose, then some might claim that it should be retained despite the presence of transgender and intersex people.88 To reject the binary in sport, it would need to be shown that (A) It is not pragmatic; and/or (B) there are fairer, more respectful ways of categorising athletes.

To summarise the arguments and counterarguments, the issue of whether transwomen should compete in the women’s division is particularly difficult because elite sport ought to be both inclusive and fair. The 2015 IOC guidelines allow a transwoman to compete, if, inter alia, her testosterone levels have been under 10 nmol/L for 1 year. The science section showed that the IOC 2015 guidelines allow transwomen to have an advantage over their cis-women competitors. While the IOC 2015 guidelines do not adequately address the principle of fairness, it does not follow that transwomen should not compete in the women’s division. It could be argued that the advantage that transwomen have is a tolerable unfairness,69 given the wide range of testosterone levels currently permitted in the woman’s division (DSD and to a lesser extent PCOS). We dispute that this is a tolerable unfairness as (A) These high testosterone levels are not available to cis-women (except via doping3). (B) Testosterone provides an all-purpose benefit for most sports. (C) Hormone therapy does not mitigate all the benefits from a previous male physiology. Hence, we claim that the IOC guidelines as they stand are poorly drawn. At the same time as being inclusive, the guidelines also need to ensure fairness.

Possible solutions

There is a range of possible responses that could be taken to resolve this conflict between fairness and inclusion.

1. Exclude transwomen from competing in the women’s division. This might go some way to levelling the playing field (as evidenced by the science above) but is inconsistent with the principle of inclusion.

2. Transwomen compete in the gender with which they identify. While consistent with the principle of inclusion, this option does not satisfy the principle of fairness given the scientific evidence of a performance advantage. Having unregulated testosterone levels would be an intolerable unfairness. This remains even if testosterone levels are kept below 10 nmol/L.

3. Further restrict the IOC’s eligibility criteria for transwomen, by setting the maximum testosterone to a lower level. This approach is problematic because physical harm may result from requiring a transwoman to reduce her testosterone levels further.89 Further, it would not account for the performance advantage transwomen retain from earlier male physiology.

These three options suggest that the principles of inclusion and fairness are incompatible if the male/female binary is retained. The following solutions are more radical, challenging the male/female binary in sport:

1. Have a single category for all athletes regardless of gender.80 81 This option is problematic as elite women athletes would rarely win against elite male athletes,79 and so would not satisfy the principle of fairness. Moreover, it may erode the space carved out for women’s sport.77

   1. Have a single category for all athletes regardless of gender and allow exogenous testosterone to be taken to a single safe level.11 In addition to the problems noted by Foddy and Savulescu,11 it would not account for the prior male physiology of transwomen.

2. Change or increase the categories for participation.

   1. Create a third division for transwomen and intersex women (who retain their testes).78

   2. Create many more categories, such as cis-man, cis-woman, transwoman, transman, transwoman who has had gender affirmation surgery and so on.

Having multiple categories (in 5a and 5b) is not novel as there are weight divisions in sports such as boxing and weightlifting. Weight divisions provide safety to competitors, as well as give athletes in each category a real chance of winning. Approach 5b can also recognise and account for the disadvantage that transwomen without testes have against cis-women competitors, and the disadvantage that transmen have in comparison with cis-men.

A problem with 5a and 5b is that transgender athletes may not get to compete in the category in which they identify.38 69 Separating out gender-variant people may be perceived as stigmatising or as denying the dignity of transgender people.^vi This problem might be solved by using a system that does not classify athletes on the basis of whether they are cis-gender or transgender, but on their testosterone levels (as suggested in the following potential solution). A further potential problem with 5a and 5b could be the number of athletes is so small in some categories that it may render the competition meaningless.38

(A) Calculate a handicap for transwomen based on their testosterone levels, similar to that used in golf. This approach is suggested by Bianchi.3 79

While Bianchi’s approach accounts for the disadvantage that transwomen without testes and transmen have, there are two problems with her account. It does not account for the performance advantages transwomen have by virtue of their previous male physiology. Second, Bianchi’s system might also be considered offensive as it does not account for gender identity.

Our proposed approach is an extension of Bianchi’s.

(B) An algorithm that accounts for (a) Social parameters including gender identity and socioeconomic status. (b) Physiological parameters.

Such an algorithm would be analogous to the divisions in the Paralympics, and may also include paralympians. First, there would be multiple divisions rather than simply male and female. Second, based on the results of the algorithm, athletes would be placed into a division which best mitigates unfair physical and social parameters.

The physiological parameters could include

• Size, for example, height and weight.

• Haemoglobin levels.

• Maximal oxygen uptake (VO² max).

• Whether the athlete transitioned before, during or well after puberty.

• Past and present testosterone levels, and the presence of testes.

• Previous characteristics of physiology that are not changed via hormone therapy, that is, bone strength or structure, lung capacity and heart size.

The algorithm would need to be tailored to specific sports as a physiological factor will be more advantageous in some sports than others. For example, weight would need to be accounted for in weightlifting, and perhaps wingspan for swimming.

While it would be difficult to generate such an algorithm, a considered response that adapts to evolving scientific evidence is required. Further discussion regarding this algorithm is beyond the scope of this article but is being developed by the authors.