I. Introduction
In common law jurisdictions, direct discrimination claims have long required the claimant prove that their attribute or “ground”, such as their race or sex, was the reason for the adverse treatment that they received. Since the late 1980s, courts in the UK have framed this as a question of causation and have utilised two different constructions of the causal test to determine a direct discrimination claim: the “but-for” test, akin to that developed in tort law, and the “reason-why” test. The use of the language of “but-for” could be because, as was highlighted by Baroness Hale in R. (on the Application of E) v Governing Body of JFS, direct and indirect discrimination are “statutory torts”.Footnote 1 These two forms of discrimination can be considered statutory torts because the Equality Act 2010 operates to give individuals a remedy for wrongs committed against them by other private actors.Footnote 2
This paper interrogates the use and effectiveness of these causal tests in developing a doctrinally coherent approach to direct discrimination. We consider whether the use of terms such as “but-for” to determine the causative element of a direct discrimination claim appropriately reflects the nature of the courts’ inquiry and its effect on the coherency of discrimination law. That causal inquiry interrogates the relationship between a discriminator’s conduct and the detrimental effect on the complainant. We argue that the use of terminology such as “but-for” obfuscates the nature of the inquiry that the courts are conducting in a direct discrimination claim, because it focuses solely on the factual inquiry and overlooks the normative inquiry required in evaluating causation. A more open and clear approach to causation would recognise the need for both a factual and a normative inquiry: the former considering narrower questions as to whether conduct was based on a claimant’s protected ground and the latter considering broader questions about the meaning of equality and non-discrimination and whether legal responsibility should extend to the respondent.
We make this argument in four sections. In Section II, we explain the nature of the relationship between tort and discrimination and explore theoretical linkages between the two in the context of fault, intention and causation. We argue that a determination of direct discrimination does require a causal inquiry but does not necessarily require an analysis of the duty-bearer’s intention or fault, both of which are amalgamated into the singular causal “but-for” test. In Section III, we interrogate and explore the case law on direct discrimination and causation from the UK House of Lords and the Supreme Court from R. v Birmingham City Council, ex parte Equal Opportunities Commission Footnote 3 onwards, tracing this jurisprudence from 1989 to the present day. In doing so, we develop an account of the prominent terminologies utilised to determine whether a detriment suffered by the complainant was “because of” an attribute they held, as required under the relevant statutory test.
In Section IV, we then evaluate the use of the “but-for” test and assess its value as the sole causative test for direct discrimination claims. We argue that the “but-for” test alone cannot account for the factual and normative limbs required to determine whether discrimination was “because of” an attribute and consider three key conceptual challenges posed by the use of the “but-for” test in direct discrimination claims, which weaken the consistency and coherency of discrimination law. In so doing, we outline the practical problems with adopting a solely factual approach to causation: namely, that controversial cases with inexplicit and shaky normative foundations undermine the capacity of discrimination law to change behaviour and make society less discriminatory. In Section V, we briefly propose a preferred path forward: that the existing “but-for” test for direct discrimination claims be reinterpreted to ensure it is a solely factual and not normative inquiry, and a second element akin to the legal causation test found in tort be added to address normative inquiries. A neater and clearer separation between these factual and normative questions would ensure greater consistency and coherency in direct discrimination claims.
Overall, we advocate for an approach to causation in discrimination law matters that clearly separates the factual and normative inquiries. Our central argument is that discrimination law claims should not be determined purely on a factual basis, because discrimination involves inherently normative questions about the meaning of equality and non-discrimination. Whichever test is ultimately adopted for causation in direct discrimination should not be used to obfuscate or ignore the integral normative questions that are inherent in discrimination claims. Such a failure to adequately consider fundamental concepts of discrimination law, such as substantive equality or underlying justifications for when and why certain conduct should be prohibited, only serves to weaken the coherency and moral force of discrimination law.
II. Fault, Intent and Cause in Direct Discrimination
A. The Relationship between Tort Law and Discrimination Law
The Equality Act 2010 as well as its historical predecessors create two different statutory torts: direct discrimination and indirect discrimination.Footnote 4 Direct discrimination occurs where the discriminator (“Person A”) treats a complainant (“Person B”) in a manner which is unfavourable or less favourable “because of” a characteristic or ground (such as race or sex) that they have.Footnote 5 In the Equality Act 2010, there is no general defence to direct discrimination.Footnote 6
Direct discrimination is often considered the simpler or archetypal form of discrimination. Utilising Khaitan’s distinction between the lay and the legal models of discrimination, direct discrimination represents the classic “lay” understanding of discrimination: it is direct, intentional and comparative.Footnote 7 Similarly, Moreau describes this type of discrimination as the “paradigmatic” discrimination claim in that it involves directly singling someone out for inferior treatment.Footnote 8 However, the perceived simplicity of this form of discrimination belies the challenging questions as to standards of fault, intention and causation which are necessarily embedded in a direct discrimination claim.
It is through these questions of fault, intention and causation that discrimination law’s similarity to tort law has been considered. The possible relationship between tort and discrimination law is not a new observation either in the UK or in North American jurisprudence, where the link is more explicitly articulated by the courts.Footnote 9 In part this is because, as Reaume has commented, both discrimination and tort law involve the imposition of non-voluntary duties to prevent specific kinds of individual harms.Footnote 10 In developing a normative account of discrimination law’s legitimacy, scholars have regularly drawn upon theories from tort law. In particular, two doctrines have often been utilised: the harm principle and distributive justice.Footnote 11 Specifically, direct discrimination has been understood as akin to an intentional tort and is therefore justified based upon the harm principle.Footnote 12 In his early work, Gardner understood direct discrimination as a conventional harm-based tort.Footnote 13 In doing so he focused on the definitional aspects of direct discrimination, specifically the requirement for intention as opposed to motive and the need for a nexus of wrongdoing between the complainant and the defendant.Footnote 14 It is through these similarities that Gardner, as well as others such as Foran,Footnote 15 have understood discrimination law’s legitimacy as being based upon the harm principle. Such an understanding is consistent with Smith, Schleiger and Elphick’s argument that discrimination law is a system inherently focused on individual fault and moralistic wrongfulness.Footnote 16 In this understanding of discrimination law, obligations can and should be placed upon individual duty-bearers to prevent individual incidents of discrimination.
Fault, as a concept, is considered a fundamental aspect of tort law in the sense that fault explains why someone should be liable for their conduct. In tort law, fault is determined on the basis of range of varying standards.Footnote 17 These include intent, recklessness, negligence and even the absence of fault, commonly known as strict liability. As Murphy identifies, malice can also play a role in determining whether and why someone should be liable for their conduct in tort, particularly in lesser-known tortious actions.Footnote 18 The varying standards of fault in tort law raise their own complexities. For example, there remains contested academic debate over the imposition of strict liability for torts of vicarious liabilityFootnote 19 and over the role of intention in the tort of negligence.Footnote 20 It is this controversy over the interpretation of varying fault standards which establishes that fault, as a concept, is a prominent, visible and crucial feature of tort law – both in its legal construction and in its scholarship.
Fault is also considered in discrimination law jurisprudence and scholarship, but not to the same, pre-eminent degree as in tort law. In discrimination law, understanding fault is fundamental to understanding the underlying purpose of discrimination laws. Over time, there has been a shift from an understanding of discrimination law as being focused on malice and motive to being focused on the effects of conduct.Footnote 21 Nevertheless, the overarching structure of discrimination laws, focused on the difference between direct and indirect discrimination, remains. There has been some work in tracing the changing understanding of fault in discrimination law. Reaume, in particular, has considered the evolution of discrimination law using the tort concept of standard of fault, demonstrating the shift from a focus on malicious actions to intentional and then to negligent actions within the doctrine of discrimination law.Footnote 22 In doing so, she argues that the changing understanding of the theoretical underpinnings of discrimination law mirrors this changing fault standard. In more recent work considering the interplay between discrimination law and tort law, particularly pieces advocating for discrimination to be a standalone tort, it is notable that the requisite fault standard is often not explicitly considered.Footnote 23
In case law, there has been substantially less consideration of the rationale and theoretical underpinnings of discrimination and, in particular, what makes discrimination blameworthy. Instead, the apparent focus from the language utilised in the case law seems to be on finding a causal link between the treatment and the attribute of the complainant.Footnote 24 In the next section, we begin to unpack the relationship between the concepts of intention, fault and cause to develop our argument that the focus on cause in direct discrimination claims obscures the deeper inquiry into why the duty-bearer’s conduct is faulty or blameworthy.
B. The Collapsing of Intent into Cause in Discrimination Claims
While on a theoretical level this link between tort and discrimination has been explored by others, what is missing from this analysis is if and how this relationship with tort and tort theory is manifested in discrimination jurisprudence. In particular, little has been written on whether UK discrimination law has adopted a tort-like approach to questions of fault. It is noticeable that for both tort and discrimination, the meaning or definition of intention is broader, and often more abstract, than that utilised in ordinary parlance or in criminal law. In ordinary parlance, an understanding of intention focuses on the actual motivations of the person doing the act. In contrast, “intention” in the context of both discrimination law and tort law is a much broader notion not necessarily linked to the wrongdoer’s motivations. As Cane has noted, the role of “tortious intention” in what are commonly understood as intentional torts – trespass to land, trespass to goods, battery and assault – is minimal.Footnote 25 “Intention” in its ordinary sense plays a minimal role because the role of these intentional torts is to provide for protection from interference with bodily integrity and property rights.Footnote 26
The role of intention in direct discrimination is contested in many discrimination law jurisdictions. To prove a direct discrimination claim, does a claimant need to prove an intention to discriminate? Or do they need to prove a malicious or motivated intention to discriminate?Footnote 27 Should direct discrimination only require negligenceFootnote 28 or should there be a form of strict liability?Footnote 29 Different jurisdictions have adopted various approaches as to how to confront the question of intention in a discrimination claim. In some jurisdictions, such as in Canada and some Australian states, the focus in a discrimination claim has been on the impact on the claimant rather than the reasoning processes of the defendant, such that intention is not a requirement of the claim.Footnote 30 In contrast, in the US there has been a far greater focus on the reasoning processes of the defendant, with the case law also utilising a “but-for” framing in discrimination law claims.Footnote 31
However, even those jurisdictions that focus on the reasoning process of the defendant do not seem to adopt a requirement of “intention” as it is conceived in tort law or other areas of law. While discrimination law scholars often frame the “but-for” test with respect to questions of the defendant’s reasoning,Footnote 32 in other areas of law, such as torts, the “but-for” test is commonly understood as a question of causation rather than intention. It is through courts’ adoption of this latter understanding of the “but-for” test that causation, rather than intent, has become the predominant requirement in discrimination claims – and it is through the terminological inconsistencies associated with the framing of this “but-for” test that significant conceptual incoherency has arisen in discrimination law.
Stapleton has argued that the legal concept of causation is notoriously “afflicted with linguistic ambiguity”.Footnote 33 Causation is analytically complex and can resist traditional conceptions of “bottom-up” common law reasoning because the reasoning process adopted is often opaque.Footnote 34 Causation – at least in the law of tort from which discrimination law draws its “but-for” test, as will be discussed below in Section III – relates to two connected, though analytically distinct, inquiries: factual causation and legal causation. First, a factual inquiry is undertaken whereby the court is required to consider whether the complainant’s injury was a result of the defendant’s conduct.Footnote 35 The second causative question that the court is required to consider is whether the defendant should be legally sanctioned for their conduct. This second limb is often termed either the “scope-of-liability” test, because it ascertains whether it is appropriate for the scope of the defendant’s liability to extend to the harm caused by their acts or omissions, or the “remoteness” test, because it ascertains whether the harm caused is too remote to be captured by the defendant’s liability. Whether the terminology of “scope of liability” or “remoteness” is adopted, this second limb is a normative rather than factual inquiry and there is no clear equivalent used by judges in assessing the causative “reason” element in discrimination claims.Footnote 36
C. The Entangling of Fault and Cause in Discrimination Claims
Similarly to intent, cause and fault are tightly interconnected in a direct discrimination claim. This is because what caused the less favourable treatment will determine whether a decision or rule constitutes direct discrimination. Campbell and Smith highlight the tight connection between cause and fault by giving the example of a woman who fails in gaining a promotion.Footnote 37 If she fails in her bid for a promotion because she is a woman, then she has a direct discrimination claim on the basis of sex.Footnote 38 If she fails in her bid for promotion because she is always late for work, she does not have a direct discrimination claim on the basis of sex. In this example, the place of intention, cause and fault are tightly connected.Footnote 39 There must be a consideration of the reasoning processes of the individual decision maker or any rationale behind a rule in question. Thus, the intention behind the decision, the negative outcome caused by the decision and the determination of fault of the defendant are all captured within the singular analysis of whether or not the claimant was treated badly “because of” a protected attribute that she holds. In this regard, fault is entangled with causation when courts consider the “because of” requirement. Similarly to intent, then, fault has largely been amalgamated into what has been framed by courts as a singular question of causation.
In contrast to tort law, there is less consideration in discrimination law cases or literature of how these notions of intent, cause and fault can be understood together or within any given case. The very need to do so was rejected by the House of Lords as confusing and unnecessary for lower-level decision makers.Footnote 40 However, the problem with this lack of consideration is that frequently, in both the case law and the academic literature on discrimination law, the role of intention and causation have become intertwined such that intention and causation are considered one and the same – when in tort law doctrine, from which discrimination law judgments draw, they require separate analysis. This doctrinal gap, and the confusion stemming from it, has significant practical consequences. For example, in conceiving the fault and intention standard for direct discrimination one could utilise a range of different available frameworks and conceptions of justice: direct discrimination could be considered a statutory tort of strict liability, an intentional tort, an unintentional tort or nothing akin to tort. Each of these different understandings brings with it a different fault standard, thereby altering the jurisprudential tests applied to the prohibitions on direct discrimination in the Equality Act 2010. That both judges and discrimination law scholars seem unclear as to which of these different frames and fault standards should be applied to direct discrimination claims significantly undermines the coherency of discrimination law and its capacity to effect just and equal ends.
III. Understanding the Two Causal Tests Applied in Direct Discrimination Claims: “But For” and “Reason Why”
To prove direct discrimination in common law jurisdictions, including under the Equality Act 2010, a complainant must generally prove two elements. The first element of the inquiry involves the complainant demonstrating that they were treated less favourably than someone who does not hold their protected characteristic (also known as their “attribute” or “ground”) but is otherwise similarly situated to them. This is called the “comparator” question. The second element that a complainant must demonstrate is that their protected characteristic was the “cause” or “reason for” the treatment. This is the “reason” question. These can be considered separate elements but have often been considered as part of the same inquiry.Footnote 41 That inquiry, as articulated in Shamoon v Chief Constable of the Royal Ulster Constabulary is: why was the complainant treated in the manner in which they were?Footnote 42 Part of the reason the need to establish “motive” or “fault” in a direct discrimination claim has been diminished and rejected in the case law could be that the inappropriateness of using a certain attribute or characteristic as a grounding for a decision is explicitly articulated in the legislation. As such, the question as to whether the behaviour in utilising the attribute or characteristic should be considered legally fault-worthy does not need to be considered. Instead, the focus is on establishing that the attribute or characteristic was, in fact, the cause of the treatment in question. A determination of direct discrimination does require causal inquiry which can be understood as similar to the process in some tort claims: albeit one, through the “reason” question, that is entangled with fault. In this section, we interrogate the case law on direct discrimination and causation from the UK House of Lords and the Supreme Court. We use this jurisprudence to develop an account of the terminologies that courts have used in determining whether a detriment suffered by the complainant was “because of” a protected attribute they held, as required by the Equality Act 2010.
In determining this “reason” question, it has been left to the courts to define the appropriate test to decide whether the treatment was “because of” the attribute. As was highlighted in Section II, the question is whether the treatment was caused by a person’s protected characteristic. How the court should determine whether the treatment is “because of” the protected characteristic is not designated by the legislative text and there does not appear to be a universal answer to that question.Footnote 43 In Swigg v Nagarajan, Lord Nicholls concluded that there was no singular approach to the phrase “because of” that should be adopted.Footnote 44 Instead, case law from the UK appears to utilise two different causal tests: one which is described, both in the scholarship and some of the case law, as the “but-for” test; and the other the “reason-why” test. The “but-for” test is frequently used where there is a blanket policy or criterion which excludes all persons with a certain attribute, or alternatively creates a “proxy” for the attribute in a manner which excludes all persons with that attribute from accessing the good or service.
For example, the “but-for” test was utilised in R. v Birmingham City Council, ex parte Equal Opportunities Commission and James v Eastleigh Borough Council by Lord Goff in which he accepted that in each case, “but-for” a person’s sex they would have not suffered a disadvantage.Footnote 45 In the case of R. v Birmingham City Council, ex parte Equal Opportunities Commission the disadvantage was the lesser opportunity to attend a selective school and in the case of James v Eastleigh Borough Council to qualify for free swimming pool access. Later cases such as R. (on the Application of Coll) v Secretary of State for Justice also utilised a “but-for” construction for the causal test where the Supreme Court accepted that female prisoners had statistically a lesser opportunity to be housed in accommodation close to their support networks than male prisoners.Footnote 46 Adams-Prassl, Binns and Kelly-Lyth describe this test as utilising an if/then construction: the “if” part is the proxy characteristic and the “then” part is the relevant disadvantage.Footnote 47 In essence, the construction is: if there is a test or proxy characteristic which then creates a relevant disadvantage for an entire group, direct discrimination will be established.Footnote 48
In contrast, the “reason-why” test asks why the complainant was treated in the manner in which they were treated and if the characteristic was a “substantial factor”. The “reason-why” test is often used where an (alleged discriminatory) determination is the result of an individual decision. As will be seen below, the “reason-why” approach combines both the question of the appropriate comparator and the reason for the treatment in determining whether the characteristic was a substantial factor, with courts considering the entirety of the situation in this single step.
The difference between these two approaches was concisely outlined in R. (E) v JFS.Footnote 49 Lord Phillips concluded that where the factual criterion for a decision was overt (whether based explicitly on a prohibited ground or not), then the “but-for” test was appropriate.Footnote 50 Where the factors influencing the decision are not obvious, Lord Phillips concluded that the “reason-why” approach should be adopted.Footnote 51 While each of these tests can be articulated in the abstract to denote a difference in approach, it can be difficult in practice to identify when the court is utilising one or the other. For example, in the academic literature which considers R. (E) v JFS, some commentators consider the case an example of the “but-for” formulation while others conclude that it utilises the “reason-why” formulation.Footnote 52 A difficulty in R. (E) v JFS is that there are multiple judgments, but even those analysing the same judgment can come to different conclusions as to the test utilised by the particular judge.Footnote 53 The difficulty assessing which approach is being adopted in any given case could mean that there is not a significant difference between the “reason-why” test and the “but-for” test or that there is a lack of clarity surrounding their use.Footnote 54
In its early jurisprudence, the House of Lords, in cases utilising both the “but-for” test and the “reason-why” test, repeatedly emphasised that the question of whether a disadvantage was “because of” a protected characteristic was not a test of legal causation. However, it is not clear in James v Eastleigh Borough Council whether Lord Goff was discussing legal causation with respect of a second limb of a causation test, or simply a broader statement that the question of causation in a discrimination claim should not be one that the courts determined in a legalistic manner. Lord Goff, in particular, favoured the “but-for” construction to avoid what he considered to be legally convoluted arguments about motive and intention.Footnote 55 In Chief Constable of the West Yorkshire Police v Khan, Lord Nicholls rejected the idea that the phrase “by reason of” connotes a test of legal causation, concluding that it requires a factual finding of the defendant’s subjective motivations (though in Swigg v Nagarajan, Lord Nicholls acknowledged that subjective motivations could include those which are unconscious or unacknowledged).Footnote 56
These relatively simple conceptions of “but for” or “reason why”, in proving a claim for direct discrimination, have some attraction. They avoid the need for a complainant to prove the subjective or underlying basis for a decision made by a discriminator. The examples given by Lord Goff in R. v Birmingham City Council, ex parte Equal Opportunities Commission as to the problems of proving intention or motivation – that the discriminator could argue that decisions are made for the reasons of “costs” or “customer preference” – are compelling. They are particularly persuasive when one considers that the previous case law, which is littered with examples of exactly that kind of reasoning, had stymied the capacity of complainants to be able to prove their claim.Footnote 57
IV. The Conceptual Challenges Caused by the “But-For” Test in Discrimination Claims
However, there are significant problems in the utilisation of the language of “but-for” in determining discrimination claims, which weaken the internal coherence of discrimination law and its ability to achieve its overarching objectives. In other areas of law, such as negligence, causation has been described as a “complicated, convoluted and confused”.Footnote 58 In response to those challenges, scholars have approached questions of causation differently. Some, such as Green, have argued that the “but-for” test still has relevance and utility but could be reinterpreted using her Necessary Breach Analysis framework, which is designed to simplify and clarify the factual causation test in negligence using a pragmatic framework.Footnote 59 In Green’s view, the philosophical and theoretical criticisms of the “but-for” test may illustrate problems with the “but-for” test,Footnote 60 but they fail to appreciate the “but-for” test for what it is: a pragmatic legal tool. As a consequence, much of the theoretical debate and the use of different legal tests to accommodate complex fact patterns have over-complicated causation.Footnote 61 Instead, Green argues that the “but-for” test should be pragmatically focused on its central utility: identifying the breach that caused the plaintiff’s harm.Footnote 62 Her Necessary Breach Analysis framework utilises the “but-for” test in this way by identifying whether or not the defendant’s conduct was a necessary cause or aspect of the plaintiff’s loss and whether or not the conduct was legally relevant or, in her words, “operative”.Footnote 63
Others have eschewed the “but-for” test more robustly. The challenges with respect to causation, particularly challenges associated with the “but-for” test, have led some scholars to reject or limit the use of the “but-for” test and instead utilise different tests to expose causative relationships. In this section, we begin by outlining the problem with the way UK courts have utilised the “but-for” test in direct discrimination claims as the entire causative test rather than, as in tort, the first of a two-limbed causative inquiry. We then turn to consider underlying limitations of the “but-for” test in discrimination law, even when it is utilised in the same way as in tort, as well as some of the solutions offered in the tort law scholarship. We interrogate those critiques and their solutions in the context of discrimination law claims.
A. The Use of “But For” as the Entire Causative Test
Underlying these various problems is the way UK courts have utilised the “but-for” test in direct discrimination claims: as entirely decisive on the question of causation, rather than as one component of the question of causation. Because of this, UK courts are utilising only one side of the tortious causation coin (factual) and ignoring the other (normative): legal causation. This is especially problematic as the “but-for” test is the main tool used by UK courts to assess the “reason” element of direct discrimination, rather than the “reason-why” test, which entails a more express normative inquiry.
In tort law, from where the test is largely drawn, the “but-for” test is generally utilised to determine only factual causation: that is, did the act or omission factually cause the harm? However, in discrimination claims, rather than determining a factual question, the “but-for” test is utilised to answer causative liability as a whole – either ignoring normative questions or absorbing them implicitly into the factual inquiry. There is, as such, no separate normative inquiry undertaken at the causation stage of direct discrimination claims. This intermingling of factual and normative causative questions is problematic for the internal coherence of a direct discrimination claim because it is a misleading statement as to the nature of the causative inquiry undertaken. It also means that courts are likely reaching normative conclusions without sufficient interrogation of the underlying rationale or justification for the conclusion adopted and that they are incorrectly applying the “but-for” test in how it is intended to operate in tort law.
In addition to weakening the coherency of discrimination law, practical challenges also arise from the use of “but for” as the entire causative test in direct discrimination claims. Without a separate normative inquiry, as we outline in our analysis below, judicial reasoning in cases may be considered “legally” or “technically” correct but can generate significant controversy because their normative foundations are inexplicit and shaky. Two examples of this disconnect are the early decision in James v Eastleigh Borough Council,Footnote 64 as well as the more recent decision in Lee v Ashers Baking Co. Ltd.,Footnote 65 both of which are considered in greater detail below. In James v Eastleigh Borough Council, the use of the “but-for” test is straightforward, but fails to examine the non-discriminatory reasons why a council may decide to waive fees for pensioners. In Lee v Ashers Baking, again, the application of the causative test is straightforward but fails to contextualise the decision on whether discrimination law should apply where there are tensions between the different rights and freedoms of the parties involved. This matters because discrimination law is normative and is intended to change behaviour to make society less discriminatory. As we will highlight below, though, it is more challenging to articulate the normative underpinnings of discrimination law where the case law is inexplicit in its reasoning.
The “but-for” test is incapable of articulating these normative underpinnings. Steel and Edelman both describe the “but-for” test as a test determining the “relationship between an event and an outcome”, whether focused on the “counterfactual” or the “metaphysical” scenarios in which the defendant’s conduct or act had not occurred.Footnote 66 The question can be understood as follows: “but-for” the defendant’s conduct, would the harm or event in question not have occurred? Thus, the “but-for” test involves a theoretical inquiry into a counterfactual world where the defendant’s conduct did not exist and a consideration of whether the harm or event would have occurred regardless.Footnote 67
In tort law, the second causative limb of legal causation is intended to address normative questions, rather than the “but-for” test used to assess the first causative limb of factual causation. In tort cases with challenging factual foundations or evidentiary “black holes”, these two limbs of causation can become interlinked; some judges have remarked that the factual and legal limbs cannot always, therefore, be readily separated.Footnote 68 Hamer has also argued that the distinction between the two is overstated.Footnote 69 Nevertheless, Steel argues that the two limbs should be treated separately given they are focused on distinct issues of legal inquiry: one of fact and one of liability.Footnote 70 Turton similarly notes that the former limb relates to whether something is “a cause” of the harm, while the latter limb is “an evaluative question of the appropriate extent of liability”.Footnote 71 Indeed in some jurisdictions, notably Australia, this separation is mandated by civil liability statutes.Footnote 72
The “but-for” test, therefore, is not intended to assess causation in tort of and by itself. As remarked by Chief Justice Mason of the High Court of Australia in March v Stramare: “[T]he test, applied as an exclusive criterion of causation, yields unacceptable results and […] the results which it yields must be tempered by the making of value judgments and the infusion of policy considerations.”Footnote 73 Turton also argues that factual causation alone cannot determine liability and that other doctrines, such as legal causation or other negligence tests such as duty or breach, “enable the law to narrow liability from causal responsibility to interpersonal responsibility”.Footnote 74 However, UK courts have utilised factual causation as the sole determinative factor in assessing causation in direct discrimination claims, through the linguistic device of the “but-for” test. As Wright has argued, the failure to separate factual causation and legal causation causes conceptual and legal confusion.Footnote 75 This severing of one limb of the causation test from the other ignores the long history of tort law in developing an additional limiting, normative device to supplement the “but-for” factual test and exposes discrimination law to several key conceptual challenges.
B. Conceptual Challenges in the Absence of a Second Causation Limb
It is clear therefore that the absence of an equivalent legal causation limb to explicitly address normative considerations in direct discrimination claims, and reliance only on the “but-for” test, is problematic of and by itself. Even if a tortious approach to causation proved helpful to discrimination law, any such benefit is undermined by adopting only one of the two tortious causation limbs and ignoring the other.
This problem is further exacerbated by the unique nature of discrimination claims and their lack of reliance on other supplementary tortious tests and devices. In particular, the use of only the “but-for” test to establish causation in direct discrimination claims raises three key conceptual challenges, explored below through several key UK discrimination cases:
-
(1) The “but-for” conception of causation requires the consideration of the metaphysical or counter physical world, which is especially challenging in discrimination claims where the defendant’s conduct is not based upon an evidentially exposed set of circumstances but is instead based significantly on decisional reasoning;Footnote 76
-
(2) The “but-for” test is problematic when confronted with overdetermination, where multiple different actors’ conduct could be said to “cause” an event, which is especially prevalent in discrimination claims and where supplementary tortious tests have not been adopted to help resolve these issues;Footnote 77
-
(3) The “but-for” test is problematic when confronted with underdetermination, where there are multiple small causes of events, none of which on their own would be sufficient to be determined as a cause.Footnote 78
The unique nature of discrimination claims, in particular their reliance on decisional reasoning, both causes and exacerbates these conceptual challenges. As such, even if the “but-for” test in direct discrimination claims was utilised as it is in tort – as the first of two causative limbs and focused only on the factual inquiry – these challenges would persist.
1. Reasoning and the counterfactual
First, as the Australian Justice Edelman has concluded extrajudicially, the “but-for” test is not strictly a factual inquiry, but instead is a counterfactual inquiry.Footnote 79 At their heart, causative questions are theoretical or hypothetical in nature.Footnote 80 As such, and as Steel has argued, the conclusions on the basis of these inquiries depend on how the relationship between the “event” and the “conduct” are constructed, how these factual concepts are understood to interact with one another and how each interacts with other events happening in the circumstances and the world at large.Footnote 81 In the case of discrimination law, the inquiry depends on how the “event” – the rule or decision and its detrimental effect on the complainant – is constructed and understood, how the defendant’s conduct is explained and how these interact with other factors in the world around them. As a result of this counterfactual inquiry, in tort claims, the ability to determine causation is highly dependent on how the breach of duty is defined and articulated.Footnote 82 Different conceptions of the breach of duty will necessarily impact the way in which the “but-for” causative test operates in a counterfactual world.Footnote 83
Discrimination claims, though, pose additional challenges in undertaking this counterfactual inquiry. In discrimination claims, rather than articulating a breach of a duty, the complainant is tasked with articulating what disadvantage represents “less favourable treatment” on the basis of their protected attribute.Footnote 84 For example, in James v Eastleigh Borough Council, the “less favourable treatment” or disadvantage was the complainant having to pay an entry fee to enter a swimming pool while his similarly aged wife did not.Footnote 85 The majority of the House of Lords in James v Eastleigh Borough Council emphasised that the inquiry to be conducted was one of “but-for”: “but-for” his sex, would the complainant have had to pay to enter into the swimming pool?Footnote 86 However, while the majority concluded that “but-for” his sex he would not have had to pay to enter into the swimming pool, if one applied the “but-for” test in the counterfactual manner articulated by Justice Edelman extrajudicially, this simply could not be ascertained in this case as a question of fact. There are multiple counterfactual scenarios that could be utilised in the facts of James v Eastleigh Borough Council. The one utilised by the majority was: had Mr. James been a similarly aged woman, would he have had to pay an entry fee?Footnote 87 In this counterfactual the answer was no and thus sex discrimination was proven.Footnote 88 But there are other equally plausible counterfactual scenarios, not all of which would come to the same conclusion. As is noted in the dissenting judgment of Lord Griffiths, there is another answer or outcome which is equally plausible: that the council removed the free entry for pensioners entirely.Footnote 89 Utilising this counterfactual scenario, the complainant would still have been paying an entry fee and would still be suffering from the disadvantage, though that disadvantage would now equally be applied to his wife.Footnote 90 The focus on “less favourable treatment”, as opposed to “breach”, can therefore make the counterfactual inquiry more difficult to prove.
Exacerbating this difficulty is that the “but-for” test fails to account for decisional reasoning processes, which are especially prevalent in discrimination claims. As Bant and Paterson explain, the “but-for” test can be useful in cases of scientific inquiries – such as whether a bullet from a gun would have hit the plaintiff – because the court can rely on scientific evidence to determine mechanically what led to a particular result or outcome.Footnote 91 In those cases, the counterfactual inquiry can be tested technically by considering the impact and the causative force of a range of factors.Footnote 92 Where the “but-for” test is far less useful, however, is in cases of decisional causation.Footnote 93 Unlike questions that can be scientifically proven, the reasons for a person’s decision to act cannot be subjected to the same kind of scientific inquiry. People make decisions for a variety of reasons, some of which are conscious and some of which are subconscious; these reasons, and their weight and bearing, cannot be tested in any scientific or rigorous manner.Footnote 94 The “but-for” test requires a consideration of wholly hypothetical circumstances in which a particular person’s reasoning may never be adequately determined.Footnote 95 As Bant and Paterson argue, this is arguably not a solid foundation to determine any parties’ rights or liabilities.Footnote 96 In many cases, plaintiffs would simply be unable to prove causation on a “but-for” basis because each party’s decisions and actions could not be determined, even on the balance of probabilities.Footnote 97 Bant and Paterson argue that as a consequence, a different test for factual causation should be used, one that focuses on if the decision was a “factor” or a “contributing cause”.Footnote 98
Many, if not the vast majority of, discrimination claims involve this kind of decisional causation. In many discrimination claims in employment, the question the court or tribunal is required to answer is why the defendant treated a particular claimant in the way in which they did. “But-for” causative construction appears to have clearer and more obvious applicability in determining direct discrimination cases which focus on stated rules or articulated reasons for acting rather than decisional claims in which respondents could have a range of motivations. UK courts have responded to this problem in a handful of cases involving single, human decision-making such as Hewage v Grampian Health Board,Footnote 99 where they have fallen back to the causative test articulated in Shamoon v Chief Constable of the Royal Ulster Constabulary and Swigg v Nagarajan: why was the complainant treated in the way in which they were?Footnote 100 This “reason-why” test is preferable in determining decisional reasoning, with its explicit focus on the decision maker’s motivation. Despite this, there remain many UK cases where decisional reasoning is a central premise of the dispute and where, still, the “but-for” test is utilised. Bull v Hall is one example, in which two men were refused accommodation on the basis of their sexual orientation.Footnote 101 Considering the prevalence of decisional reasoning in direct discrimination claims which focus on “less favourable treatment”, rather than a breach of a particular duty, the “but-for” test remains ill-equipped to be the sole determinative test of causation.
2. Overdetermination
The second conceptual problem with the “but-for” test in direct discrimination claims is that it fails to operate effectively where there are multiple competing possible reasons for the event in question. This is classically referred to as the problem of “overdetermination”. One famous example of overdetermination is where two people shoot loaded weapons at the same time and both bullets hit the complainant.Footnote 102 On the basis of the two possible shooters, the complainant cannot feasibly argue that “but-for” one of the shooters they would not have been injured. A solution to overdetermination, popularised by Wright, is the Necessary Element of a Sufficient Set (“NESS”) test.Footnote 103 The NESS test of causation was developed and adapted from Hart and Honoré’s “causally relevant factor” account, which they initially proposed in 1959.Footnote 104 The purpose of the NESS test is to provide a test to determine causation where overdetermination renders the “but-for” test unhelpful and inapplicable to establish legal liability.Footnote 105 Thus, the use of NESS justifies the conclusion that a certain event necessarily contributed to a particular outcome, notwithstanding the fact that it was only one event amongst multiple.Footnote 106 As Miller identifies, an important role of NESS is that “it reveals the logical skeleton” that lies beneath the empirical flesh which comprises the “facts” of the case.Footnote 107 In this sense, NESS is fundamentally concerned with factual rather than legal liability in tort cases. It is used to determine the sufficiency and significance of individual factors which lead to a specific factual event.Footnote 108
While problems of overdetermination are therefore nothing new under the “but-for” test used to assess factual causation in tort law, they are exacerbated in discrimination claims for two main reasons. First, overdetermination is more likely to occur in discrimination claims than in tort claims because of the prominence of decisional reasoning and the possible range and breadth of counterfactuals based upon the motivations of alleged discriminators, as discussed above. Second, unlike in tort law, scholars have not yet developed, and courts have not yet applied, a test such as NESS to rectify this fault with “but-for” reasoning in direct discrimination claims.
Two particular examples of overdetermination can be found in R. (E) v JFS and in R. v Birmingham City Council, ex parte Equal Opportunities Commission. In R. (E) v JFS, the complainant’s claimed disadvantage was not being given priority access to the school because he was not considered Jewish by the Office of the Chief Rabbi of the United Congregation of the Commonwealth (“OCR”).Footnote 109 The question asked by the court was: “but-for” the rule to give preference to students who were considered to be Orthodox Jewish by the OCR, would the complainant be treated less favourably?Footnote 110 While again the majority of the Supreme Court answered this question in the negative, namely finding that the rule was discriminatory, a closer interrogation of the facts of R. (E) v JFS finds that this was a case of overdetermination. As was highlighted in the judgment of Lord Phillips, the fundamental problem at the heart of the dispute was that the school was oversubscribed.Footnote 111 Due to the fact that more parents wanted to send their children to the school than there were places, the school was required to employ a system to determine which children would gain admission and which would not.Footnote 112 As Lord Phillips highlighted, even though he concluded that the school’s policy was racially discriminatory, he also emphasised that the school must nevertheless employ rules to manage its oversubscription problem.Footnote 113
The problem with using the “but-for” test in this case is that it was unclear that the complainant child would have gained a place at the school under an alternative enrolment policy. The child’s disadvantage could have remained, notwithstanding the removal of the purportedly discriminatory policy. The various factors in this case which led to the child’s inability to gain a place cannot be appropriately isolated for the “but-for” test to operate effectively. The reasons that the child did not gain a place at the school were multifaceted and interrelated and, in this context, it is difficult when traditionally applying a “but-for” test to demonstrate that the child’s ethnic origin was the deciding factor in his disadvantage. In this case, though overtly applying a “but-for” test, it is possible that the way in which the majority of the Supreme Court conceived the relationship between the child’s attribute and the disadvantage suffered was one of sufficiency. While the disadvantage was not only attributable to the child’s ethnic origin, it was a factor which provided a necessary contribution to that disadvantage.
A similar analysis can be applied to R. v Birmingham City Council, ex parte Equal Opportunities Commission. The issue at the heart this case was a discrepancy in places in selective schools for girls and boys.Footnote 114 The less favourable treatment of female pupils could be articulated in two different ways. First, the female pupils could have a statistically smaller chance of obtaining entrance to a selective school given the smaller number of places available. Second, the less favourable treatment could be understood as the fact that a boy and girl who received the same grade or mark on the placement exam could experience different outcomes, with the boy gaining entry due to the higher number of places available and the girl missing out on a place at a selective school. In determining the dispute, Lord Goff (with unanimous agreement from the House of Lords) ostensibly used a “but-for” test: “but-for” the discrepancy in the number of places available, male and female students would have equal chances of gaining entry into the selective school system.Footnote 115
However, when considered in a broader context, that simple “but-for” construction cannot equate to that factual finding; there are simply too many different and competing variables at play. As the schools were single-sex education facilities, these discrepancies which form the basis of the “less favourable treatment” could remain regardless of whether there were the same number of places for girls and boys. For example, if more female students took the entrance examination than male students, female students would still face statistically smaller odds of gaining a place in a selective school. Similarly, even with the same number of places available, female students may still be required to perform better on entrance examinations if female students as a group scored higher on such examinations. In this case, like R. (E) v JFS, the problem was not simply the discrepancy in school places based on gender but a variety of different factors relating to the fact these were single-sex educational facilities. On a strict application of a “but-for” test, the complainant would be unable to show that sex was the deciding factor in their less favourable treatment. Nevertheless, the House of Lords found direct discrimination.
Again, as in R. (E) v JFS, while this finding cannot be factually workable or provable under a “but-for” conception of causation, it is possible to come to such a finding using the NESS test for causation. The sex of the pupils was one reason it was more difficult to obtain entry into a grammar school, but it was not necessarily the only reason. As such, though the outcomes in R. (E) v JFS and R. v Birmingham City Council, ex parte Equal Opportunities Commission would likely be the same if a NESS test for causation was applied, it is their conceptual underpinning and the muddying of overdetermination issues that remains problematic: a “but-for” test on its own cannot come to the conclusion reached in those two decisions and utilising it to come to those conclusions undermines the coherency of the prohibition on direct discrimination.
3. Underdetermination
The third conceptual problem with using the “but-for” test in direct discrimination claims is that it can also be under-determinative: it can fail to appropriately weight the significance of particular factors where no single factor alone will be able to account for the event, a weakness again exacerbated by the prevalence of decisional reasoning in discrimination claims. Paterson and Bant argue that cases of underdetermination are particularly pertinent in decisional causation claims because human beings will make decisions for a variety of reasons and these reasons are difficult to appropriately weigh, particularly after the decision has already been made.Footnote 116 In the context of a discrimination claim, these problems can be identified through the case of Lee v Ashers Baking. Footnote 117
In Lee v Ashers Baking, the complainant brought an action for direct discrimination where the respondents, who owned a bakery, refused to bake a cake containing a message in support of same-sex marriage due to their religious beliefs.Footnote 118 The complainant argued that this refusal of services was on the basis of his sexual orientation. This was rejected by the Supreme Court, which concluded that the complainant could not show his sexual orientation was the reason that he was refused the cake.Footnote 119 In so finding, Lady Hale, with whom the other four justices agreed, held that the complainant could not demonstrate an exact correlation between holding particular political views about same-sex marriage and sexual orientation:Footnote 120 “In a nutshell, the objection was to the message and not to any particular person or persons.”Footnote 121 While the Supreme Court did not explicitly outline the causative test it applied in Lee v Ashers Baking, there are notable mentions of the “proxy” claims in James v Eastleigh Borough Council and Bull v Hall, which did apply a “but-for” construction,Footnote 122 and its underlying approach strongly suggests a “but-for” test. It was implicit in the court’s reasoning that the complainant would have received the same treatment but for his sexual orientation and his association with others of particular sexual orientations.
In so doing, Lee v Ashers Baking demonstrates the problem of underdetermination in discrimination claims. The Court concluded that the decision to refuse to sell a cake promoting same-sex marriage was not based upon the complainant’s sexual orientation or his political opinion or his association with people of a particular sexual orientation or political opinion.Footnote 123 Instead, the decision not to provide the cake to the complainant was based on the respondent owners’ sincerely held religious beliefs. Thus, the complainant could not demonstrate that “but for” his sexual orientation or his political opinion he would have been sold the cake. However, this result failed to consider the interaction between the various factors that account for the decision that was made. In particular, it failed to give any weight to the interaction between the defendant’s “sincerely held religious beliefs”, which were about the legitimacy of a particular sexual orientation and the organisation for which the complainant was buying the cake. Being unable to appropriately weigh or understand the interaction between these various beliefs and thoughts, or indeed combine their compounding effect on the motivational decision-making of the defendant bakery, the implied use of the “but-for” test was unable to adequately unpack the causal links present in the circumstances of Lee v Ashers Baking.
Lee v Ashers Baking has been widely considered and critiqued on a range of grounds. In particular, as Brown highlights,Footnote 124 while many have concluded that the decision is “technically correct”, it did not yield an appropriate substantive or normative result because it did not adequately consider important aspects of discrimination law, such as associative discrimination or substantive equality.Footnote 125 This is why the decision can be considered “technically correct”, but still leaves many unanswered questions about how discrimination law should operate when different rights and interests have to be balanced.
This problem is not just exacerbated by the prevalence of decisional reasoning in discrimination claims, as in Lee v Ashers Baking, but also by the lack of a normative limb used to assess causation – in contrast to tort, which has a separate, normative limb to assess legal causation. Had an equivalent legal causation test been utilised as a second limb of causation by the Supreme Court, the compounding interactions between sexual orientation, political opinion and religious beliefs could have been more adequately addressed. The Court could have asked: is it appropriate, in these circumstances, for the defendant owner’s liability to extend to refusing to bake a cake where their religious beliefs conflict with the political opinion expressed on the cake and/or the sexual orientation to which that political opinion may attach? Instead, the factual “but-for” test was used as a blunt tool through which to resolve this complex normative and values-laden question. Notwithstanding Hamer’s sound argument that factual questions will at times include normative inquiries,Footnote 126 evaluative determinations of this complexity and depth should be left to a legal causation, or “scope-of-liability”, test.Footnote 127
Some scholars, such as Green, suggest an approach that combines both the factual and normative inquiries into a singular inquiry, still conceptually designed based on the traditional “but-for” approach.Footnote 128 She contends that there is little need for a two-step test of causation in negligence, as it tends to unnecessarily prolong the inquiry and that judges have the capacity to undertake both the factual and normative inquiries simultaneously.Footnote 129 Green’s test considers whether or not the defendant’s conduct is “operative” or should be legally sanctioned, a requirement which she argues does the work otherwise provided by the two-stage test.Footnote 130 Given that the main purposes of causation are to establish that the defendant should be legally liable for their conduct and to achieve corrective justice for the plaintiff, there is therefore no reason to separate “factual” and “legal” causation in Green’s test.Footnote 131 However, our analysis demonstrates that in the discrimination law context, where the factual and the normative are being answered together under the current “but-for” formulation, the normative considerations do not tend to be explicitly discussed or considered. Normative questions should be explicitly identified as such and answered accordingly.Footnote 132
V. The Path Forward: More Tort or Less Tort?
It seems, therefore, that many of the UK decisions on direct discrimination which have purportedly applied the “but-for” test from R. v Birmingham City Council, ex parte Equal Opportunities Commission onwards are not correctly applying the “but-for” test as it is conceived in tort – namely, as merely the factual component of a traditionally two-limbed causation test. It is possible that in some cases, the approach adopted by the courts could have been influenced by a NESS-like analysis which identifies the attribute as a “necessary element” of a sufficient set of other competing factors, or alternatively a “common sense” approach to causation. However, this was not explicitly identified in any of the above decisions; their framing of the requisite test to prove the reason for direct discrimination remains predominantly that of “but for”. This “but-for” approach requires far greater clarification and modification to better elucidate the manner in which the causal link in discrimination claims can, and should, be identified. Until that time, the above analysis displays that the current “but-for” test, by and of itself, is inadequate to ground claims of direct discrimination.
What, then, should be the test for causative inquiries in direct discrimination claims in the UK? While our focus has been on the problems posed by the current test for causation in direct discrimination claims, our research provides lessons for considering and developing possible solutions. We propose two alternatives to UK courts’ use of the “but-for” test in direct discrimination claims: one which departs almost entirely from tort law and the other which draws far more heavily from tort law. We argue the latter proposal is the best path forward and that more tort is preferable to less tort in determining causation in direct discrimination claims.
The first alternative has already been elucidated by UK courts in several other decisions in the early-2000s: the “reason-why” test. The “reason-why” test adopted by the victimisation decisions in Swigg v Nagarajan and Chief Constable of the West Yorkshire Police v Khan were later affirmed in direct discrimination decisions in Shamoon v Chief Constable of the Royal Ulster Constabulary Footnote 133 and in R. (on the Application of European Roma Rights Centre) v Immigration Officer, Prague Airport.Footnote 134 The main benefit of the “reason-why” test is that it does not obfuscate factual and normative questions. Unlike the “but-for” test, it focuses explicitly on the intention and motivations of the alleged discriminator, ensuring that normative questions are not hidden underneath the veil of a supposedly “neutral” factual question.Footnote 135 In this regard it is a more transparent test than “but for”. The “reason-why” test is also a relatively simple and accessible jurisprudential solution considering it has already been utilised by appellate courts in the UK in victimisation and discrimination decisions. Furthermore, the “reason-why” test is a more coherent fit for the decisional reasoning inherent in discrimination claims, owing to its explicit focus on the decision maker’s motivation.
However, this test has its own issues which may be impossible to resolve. It relies heavily on normative questions to the exclusion of factual inquiries, meaning that it effectively removes the need to prove “causation”, as understood in tort law, from claims of direct discrimination. In this regard, it contains the opposite flaw to the “but-for” test; it undertakes a normative inquiry and ignores the factual inquiry. As discussed by Lord Goff in R. v Birmingham City Council, ex parte Equal Opportunities Commission, this could mean that alleged discriminators can evade liability simply by establishing that there was an alternative reason for the adverse treatment, such as costs or customer preferences.Footnote 136 The raising of these alternative reasons as defences would likely become pervasive, which would significantly undermine the operation and strength of discrimination law protections. Additionally, as highlighted with respect to the commentary surrounding R. (E) v JFS, it is not always clear where the court is using a “but-for” construction and a “reason-why” construction, indicating there is not necessarily a significant difference between the two. If the factual tests of causation used in discrimination law claims are changed, supplementary tort tests, such as the NESS test, could also be utilised to resolve the other conceptual challenges raised above.
Our second proposal is preferable: to adopt a two-stage tortious-style causation test. This would require a reinterpretation of the “but-for” test in direct discrimination to comprise only a factual requirement; and to add a second element, akin to the legal causation test found in tort law, to provide the normative requirement. As in tort, this would ensure that questions of fact (“but for”) and normative inquiries (e.g. “scope of liability”) are both addressed in causation in direct discrimination claims, but through separate mechanisms. This would ensure that a broad range of potentially discriminatory acts are still captured factually by the “but-for” test, rectifying the potential issues identified by Lord Goff, while ensuring courts can still explicitly address broader normative questions in depth where appropriate. As courts are likely already addressing these questions in an implicit, but obscured, manner through the “but-for” test, a clearer separation would ensure greater consistency and coherency in direct discrimination claims.
However, this approach would not resolve the three key conceptual difficulties associated with the use of the “but-for” test in discrimination law identified in Section IV: decisional reasoning, overdetermination and underdetermination. While tortious principles may be useful to discrimination law, these hurdles need to be acknowledged and overcome in order for the “but-for” test, even as a first of two causative limbs, to prove beneficial in direct discrimination claims. This could be mitigated through an adaptation of the test used to assess factual causation in the first stage of the inquiry, by modifying the “but-for” test to the NESS test. The NESS test has significant benefits over the “but-for” test as a first-limb factual causation requirement in discrimination law. As detailed above, NESS solves the problem of overdetermination, which is more likely to occur in discrimination claims because of the prominence of decisional reasoning and would resolve the incoherent conceptual underpinning of the “but-for” test seen in cases such as R. (E) v JFS and R. v Birmingham City Council, ex parte Equal Opportunities Commission. NESS may not, though, resolve all of the challenges in determining causation in claims such as Lee v Ashers Baking, which involve both questions of cause as well as the balancing of different rights and interests.
While adapting the factual causation test used in tort may risk undermining the argument that discrimination law should utilise the tortious causation test in full rather than in piecemeal form, this proposal would still entail adopting a tortious-style two-stage causation inquiry with separate factual and normative limbs. It is also not a new idea in tort itself, where various scholars have already proposed adapting the factual causation test to move away from the “but-for” test.Footnote 137 This test still captures scenarios in which the person’s protected characteristic was both factually necessary and legally and normatively operative, regardless of the underlying rationale for making a decision.Footnote 138 This would, thereby, cover archetypal examples of discrimination where an employer refuses to hire a person because of customer preferences – such as a pub landlord not employing a Black bartender owing to perceived racial biases amongst their clientele.
Adopting either of these proposals, while acknowledging our preference for the second proposal, would make the use of legal mechanisms drawn from tort law far clearer. In the first, discrimination law would depart from its previous use of tortious tests and, indeed, from “causation” entirely as it is understood in the law of obligations; in the second, discrimination law would more coherently draw from the complete two-stage tortious test for causation rather than simply adopting one of these two stages and ignoring the other. Other proposals, such as a separate, bespoke causation test for direct discrimination, may also develop over time. Legislative reform may prove the most promising method through which a bespoke causation test could develop. Legislators could consult with affected communities, legal advocates and other stakeholders to consider proposals for new causation tests. Such a test is less likely to develop through the common law, bound as it is by the doctrine of precedent and the challenges posed by the various cases considered above.
While future work should be done on the best path forward, and in particular on how a “scope-of-liability”-type test could best be adapted to direct discrimination to address normative queries, it is clear that the current approach in the UK – which picks and chooses from tort rather than understanding it as a system that operates, and is at its most effective, holistically – is flawed.
VI. Conclusion: Direct Discrimination and the Co-Mingling of Factual and Normative Inquiries
Courts’ use of the “but-for” test is now the dominant feature of the causative inquiry under UK discrimination law in establishing the Equality Act 2010 requirement that the reason for the treatment in a direct discrimination claim is “because of” a person’s protected attribute. Though there are merits to a simplistic “but-for” formulation, our analysis demonstrates that, overall, it has hindered the development of discrimination law.
This leads us to three important conclusions. First, the “but-for” test of and by itself is insufficient to ground the overarching causative inquiry in a direct discrimination claim, because it fails to explicitly address normative concerns. A clearer and more coherent normative understanding of how fault, intent and causation apply in the unique realm of discrimination claims is required. Second, the “but-for” test causes particular conceptual challenges in discrimination law because of the inherent prevalence of decisional reasoning in discrimination claims, which remain unresolved. Third, tort law scholarship on, and theories of, causation establish that the “but-for” test as it is applied in direct discrimination claims does not adequately reflect the factual causation test as is commonly known in tort. This is especially so considering the absence of the associated legal liability test, as well as other tests drawn from tort such as NESS, which help moderate the effect of the “but-for” test. The use and labelling of the “reason” element in direct discrimination claims as a “but-for” test obscures its application and undermines its value in discrimination claims.
It is possible that the failure to apply a strictly tortious “but-for” test is because, as Lord Nicholls emphasised in both Swigg v Nagarajan and Chief Constable of the West Yorkshire Police v Khan,Footnote 139 the terminology of “but for” in the context of a direct discrimination claim is not intended to denote a legal test, but instead to develop a framework to understand or reveal the factual reason why a decision was taken by a defendant. This may be why, when properly applied and understood, the “but-for” test does not adequately explain causation in numerous direct discrimination claims considered by the House of Lords or the Supreme Court. However, this explanation is not a justification and it allows for the court to hide its normative function in determining a direct discrimination claim by shrouding that inquiry in the language of factual causation. Necessarily, discrimination claims require normative determinations to be made; the capacity of discrimination law to change behaviour and make society less discriminatory depends on such determinations. Clearer and more coherent doctrine will surely emerge if courts explain and illuminate those normative reasons rather than hiding those in the language of factual causation. Whether this is to be achieved through closer ties with tort law or a severance of those ties is the crucial next question to be answered.
Open access
