Proving and disproving discrimination

by Ed Williams and Tom Coghlin


  1. This talk looks at the legal and practical tools available to employment lawyers to prove or disprove direct discrimination and harassment, exploring in particular three areas:
    1. How judges apply the burden of proof s136(2)(3) EA 2010.
    2. What is the role of Comparators in light of the Supreme Court decision of Hewage v Grampian Health Board [2012] ICR 1054, SC.
    3. What role does knowledge of the protected characteristic now play in light of IPC Media Ltd v Millar [2013] IRLR 707.

What we already know

  1. Recent figures from the HMCTS reveal two things:Winning a discrimination case, particular a direct discrimination case, is notoriously difficult.  Very few people, certainly in the witness box, will admit to discriminating someone consciously.
    1. The percentage of successful discrimination cases to trial is less than 10%: AND
    2. The median award is £6,000.
  2. Legal costs often outweigh any award for damages.

It’s simple isn’t it?

  1. To succeed a claimant needs to establish on the balance of probabilities less favourable treatment because of a protected characteristic.
  2. The less favourable treatment should be judged against a comparator whose circumstances are materially similar AND the reason must in some way be tainted by a prohibited characteristic.
  3. ET’s are obliged to consider the ‘reverse’ burden of proof as set out in s136(2)(3) EA 2010 when considering whether discrimination has been made out.
  4. These provisions adopt the burden of Proof Directive, Council Directive 97/80/EC, which, attempted to make it easier for complainants to establish discrimination.

Burden of Proof


  1. The wording of s136 EA 2010 remains key. In Igen v Wong [2005] ICR 9311, CA, para 16, Gibson LJ said that while it is possible to offer practical help, there was no substitute for the statutory language. The point was re-emphasised by Mummery LJ in Madarassy v Nomura International plc [2007] ICR 867, para 9, that guidance was not a substitute for statutes: “courts do not supplant statutes. Judicial guidance is only guidance”.
  2. s136(2), (3) EA 2010 provides for a reverse burden of proof:
(2) If there are facts from which the court could decide, in the absence of any other explanation, that a person (A) contravened the provision concerned, the court must hold that the contravention occurred.
(3) This does not apply if A shows that A did not contravene the provision
  1. This has been interpreted as a two stage process:If a prima facie case has been made out and the explanation for that treatment is unsatisfactory then discrimination has to be found by the ET. It is mandatory. This contrasted with pre reverse burden of proof situation, which gave the ET discretion to find discrimination where there is evidence of a prima facie case.
    1. At the first stage the claimant must prove on the balance of probabilities a prima facie case of discrimination.
    2. Only if the claimant establishes a prima facie case does the burden of proof then shift to the Respondent. At this stage the Respondent must prove that there was no conscious or sub-conscious discriminatory intent behind their conduct.
  2. Given the mandatory wording of the statute it is particularly important that ET’s approach the first stage with particular care and respondents provide both the right evidence and the right witnesses.
  3. Where ETs often err is due to the fact that whilst a two stage in logic is required, this is not usually reflected in the way in which evidence is received by them. “Though the analysis must be in two parts, the evidence comes in one bit” (Birmingham City Council v Millwood UKEAT/0564/11 per Langstaff J (P) at [17]).

Case Law

  1. Mindful of the potential for confusion the current approach of the courts is to take a less technical approach to the issue of the burden of proof. In Hewage v Grampian Health Board [2012] ICR 1054, SC, Lord Hope approved the obiter comments of Underhill J (P) in Martin v Devonshires Solicitors [2011] ICR 352, para 39, that it is important not to make too much of the role of the burden of proof provisions. Lord Hope said [para 32]:
They will require careful attention where there is room for doubt as to the facts necessary to establish discrimination. But they have nothing to offer where the tribunal is in a position to make positive findings on the evidence one way or another.
  1. Hewage also affirmed that the approach to interpreting the burden of proof remains that set out by the Court of Appeal in Igen and Madarrassy. The ‘revised Barton guidance’ in Igen provides the following:
(1) Pursuant to section 63A of the 1975 Act, it is for the claimant who complains of sex discrimination to prove on the balance of probabilities facts from which the tribunal could conclude, in the absence of an adequate explanation, that the employer has committed an act of discrimination against the claimant which is unlawful by virtue of Part 2, or which, by virtue of section 41 or section 42 of the 1975 Act, is to be treated as having been committed against the claimant. These are referred to below as “such facts”.(2) If the claimant does not prove such facts he or she will fail.(3) It is important to bear in mind in deciding whether the claimant has proved such facts that it is unusual to find direct evidence of sex discrimination. Few employers would be prepared to admit such discrimination, even to themselves. In some cases the discrimination will not be an intention but merely based on the assumption that “he or she would not have fitted in”.(4) In deciding whether the claimant has proved such facts, it is important to remember that the outcome at this stage of the analysis by the tribunal will therefore usually depend on what inferences it is proper to draw from the primary facts found by the tribunal.

(5) It is important to note the word “could” in section 63A(2). At this stage the tribunal does not have to reach a definitive determination that such facts would lead it to the conclusion that there was an act of unlawful discrimination. At this stage a tribunal is looking at the primary facts before it to see what inferences of secondary fact could be drawn from them.

(6) In considering what inferences or conclusions can be drawn from the primary facts, the tribunal must assume that there is no adequate explanation for those facts.

(7) These inferences can include, in appropriate cases, any inferences that it is just and equitable to draw in accordance with section 74(2)(b) of the 1975 Act from an evasive or equivocal reply to a questionnaire or any other questions that fall within section 74(2) of the 1975 Act.

(8) Likewise, the tribunal must decide whether any provision of any relevant code of practice is relevant and, if so, take it into account in determining such facts pursuant to section 56A(10) of the 1975 Act. This means that inferences may also be drawn from any failure to comply with any relevant code of practice.

(9) Where the claimant has proved facts from which conclusions could be drawn that the employer has treated the claimant less favourably on the ground of sex, then the burden of proof moves to the employer.

(10) It is then for the employer to prove that he did not commit, or as the case may be, is not to be treated as having committed, that act.

(11) To discharge that burden it is necessary for the employer to prove, on the balance of probabilities, that the treatment was in no sense whatsoever on the grounds of sex, since “no discrimination whatsoever” is compatible with the Burden of Proof Directive.

(12) That requires a tribunal to assess not merely whether the employer has proved an explanation for the facts from which such inferences can be drawn, but further that it is adequate to discharge the burden of proof on the balance of probabilities that sex was not a ground for the treatment in question.

(13) Since the facts necessary to prove an explanation would normally be in the possession of the respondent, a tribunal would normally expect cogent evidence to discharge that burden of proof. In particular, the tribunal will need to examine carefully explanations for failure to deal with the questionnaire procedure and/or code of practice.

  1. Paragraph (7) now needs to be seen in light of the abolition of the statutory questionnaire procedure.  Since an inference of discrimination may in any event be drawn from an evasive or false explanation in a document other than a questionnaire (see Dattani v Chief Constable of West Mercia [2005] IRLR 327 and Meister v Speech Design Carrier Systems GmbH (Case C-415/10) [2012] EqLR 602 at [47]), the use of formal interrogatories by those representing Claimants is to be expected.
  1. In Igen the Court also held that:
No doubt tribunals will generally wish to hear all the evidence, including the respondent’s explanation, before deciding whether the requirements at the first stage are satisfied and, if so, whether the respondent has discharged the onus shifted to him.
  1. In Laing v Manchester City Council [2006] ICR 1519 (approved by the Court of Appeal in Madarassy), Elias J (P) held that an ET in drawing inferences (ie stage one) might have to and should legitimately consider all the evidence put before it prior to concluding whether the burden of proof had shifted to stage 2 for an explanation from the employer. Laing drew a distinction between facts – that is evidence – and explanation. Perhaps controversially Elias J noted that an ET does not necessarily fall into error merely because it failed to adopt a two-stage approach [para 73].
  2. The CA in Brown v Croydon London Borough Council [2007] IRLR 259, agreed that whilst it is generally good practice to follow the two stage test, there is no necessary error of law if it fails to do so.
  3. In Madarassy Mummery LJ held that “could conclude” (which should now read “could decide”) meant:
[…] that ‘a reasonable tribunal could properly conclude’ from all the evidence before it. This will include evidence adduced by the complainant in support of the allegations of [in that case] sex discrimination, such as evidence of a difference in status, a difference in treatment and the reason for the differential treatment.
  1. Mummery LJ went on:
It would also include evidence adduced by the respondent contesting the complaint. Subject only to the statutory ‘absence of an adequate explanation’ at this stage (which I shall discuss later), the tribunal would need to consider all the evidence relevant to the discrimination complaint; for example, evidence as to whether the act complained of occurred at all; evidence as to the actual comparators relied on by the complainant to prove less favourable treatment; evidence as to whether the comparisons being made by the complainant were of like with like as required by section 5(3) of the 1975 Act; and available evidence of the reasons for the differential treatment
  1. However, “the absence of an adequate explanation for differential treatment of the complainant is not … relevant to whether there is a prima facie case of discrimination by the respondent.” The ET does appear to be able to consider inconsistent explanations by the Respondent to shift the burden of proof, namely an explanation that “is frankly inadequate and in particular one that is disbelieved”: Millwood at [26]. The effect of this is to make Stage 2 superfluous if an inadequate explanation has been determined at Stage 1.
  2. The ET is required to make an assumption at the first stage that there was no adequate explanation: Hewage. But while an ET is prohibited at the first stage from considering an adequate explanation from R, as per Madarassy [para 71], R might adduce evidence at the first stage to show that the acts did not take place:
[…] or that, if they did, they were not less favourable treatment of the complainant; or that the comparators chosen by the complainant or the situations with which comparisons are made are not truly like the complainant or the situation of the complainant; or that, even if there has been less favourable treatment of the complainant, it was not on the ground of her sex or pregnancy.
  1. Mere difference of treatment is not enough to shift the burden of proof, something more is required: Madarassy per Mummery LJ:
The bare facts of a difference in status and a difference in treatment only indicate a possibility of discrimination. They are not, without more, sufficient material from which a tribunal ‘could conclude’ that, on the balance of probabilities, the respondent had committed an unlawful act of discrimination.
  1. “Something more” is therefore needed.  However, as Sedley LJ observed in Deman v Commission for Equality and Human Rights[2010] EWCA Civ 1279 at paragraph 19,
the “more” which is needed to create a claim requiring an answer need not be a great deal. In some instances it will be furnished by non-response, or an evasive or untruthful answer, to a statutory questionnaire. In other instances it may be furnished by the context in which the act has allegedly occurred.
  1. The fact that an employer’s behaviour calls for an explanation does not automatically get a claimant to stage 2 of Igen: There still has to be reason to believe that the explanation could be that the behaviour was “attributable (at least to a significant extent)” to the prohibited ground (see B v A [2010] IRLR 400, per Underhill P at [22]).
  2. There is perhaps a tension with the approach that had been adopted by the previous President of the EAT, Elias J., in Network Rail Infrastructure Limited v Griffiths-Henry [2006] IRLR 865, at [18] and [21]:
Ms Cunningham [Counsel for the employer] says that in order to establish a prima facie case there must always be some positive evidence that the difference in treatment is race or sex, as the case may be. That seems to us to put the hurdle too high. As the courts have frequently recognised, there are real difficulties in establishing discrimination because of the obvious fact that it is never admitted, and it has to be inferred from the circumstances. The law has tried to strike the balance between on the one hand making such claims impossible to sustain, and on the other not subjecting employers to unwarranted and unfair findings of discrimination. The statutory burden of proof, as interpreted in Igen, by which of course we are bound, directs tribunals how the issue should be approached. Provided tribunals adopt a realistic and fair analysis of the employer’s explanation at the second stage, we see no justification for requiring positive evidence of discrimination at the first stage.

Ms Cunningham says that finding a prima facie case on the evidence established here puts employers in too difficult a situation. She cites the case of somebody who may be not only female and black but perhaps an Anglican or gay, or has some other legally relevant feature which distinguishes her from the remaining members of the group from which selection is made. She says that on the analysis by the tribunal one could in all those cases infer that the reason why she was rejected was each and every one of these distinctive features which distinguished her from the other employees. We accept that the logic would indeed be that there a tribunal would be entitled to find that there was a case to answer in all these examples, if the circumstances were otherwise as in this case. But it will often be easy to rebut. A tribunal will have to have regard to all the evidence when determining whether the employer has rebutted the prima facie case. For example, in most cases the employer will be able to show that he has no interest or knowledge of the religious affiliation of the staff or perhaps their sexual orientation. In some cases it may be shown that the manager alleged to have discriminated on, say, sex grounds has frequently in the past promoted women. That will obviously be powerful evidence rebutting any inference of sex discrimination.
  1. It certainly appears from Hewage that despite reiterating the primacy of statute as an interpretive tool, the pragmatic approach of Laing/Brown still appears to be one that the ET can adopt. Namely they can go straight to the reason why the treatment happened, rather than waste time on academic niceties about the burden of proof. This also appears to link in with the current approach to comparators. Whether this approach denies Respondents the protection of stage 1 afforded to them by statute is a moot point.


  1. An often confusing and complex process for ETs is the issue of identifying a relevant comparator by which to benchmark the allegations of discrimination.
  2. At the heart of the definition of direct disability discrimination is a comparative exercise. As Rimer LJ said in Lockwood v Department of Work and Pensions and Cabinet Officer [2013] EWCA Civ 1195 at [34]:
In a race discrimination case, that means that if a black complainant is alleging discrimination at work on the ground of his race, the comparator will usually be a white person who is otherwise in the same, or a not materially different, position. It is obvious that once such a comparator has been identified, the tribunal cannot hold the “relevant circumstances” of the two cases to be different on the ground that the comparator is white and the complainant is black and so regard the comparison as invalid. The whole purpose of the comparison is as an aid to seeing whether or not the way in which the comparator was, or would have been, treated in the relevant circumstances supports the Claimant’s allegation that he was subjected to less favourable treatment on the ground of the protected characteristic.
  1. In Lockwood the ET erred by using age differences amongst over 35s and under 35s to reject a relevant straightforward comparator in age (ie an over 35).  The ET had held that the younger comparator belonged to a generation whose members could adjust more easily and rapidly to the loss of jobs than their colleagues in their late thirties. This meant that her circumstances were materially different from those of the comparator.
  2. The Government explained its decision to retain the concept of a comparator in the definition of direct discrimination during the consultation process concerning the EA 2010 on the basis that, “… discrimination is principally about equal rather than fair treatment and courts and tribunals have flexibility on how to define comparators in each case.  Removing the comparator would make it harder to ascribe actions to inequitable treatment based on a protected characteristic, which is a key and long-standing principle governing discrimination law”.[1]
  3. s23(1) EA 2010 provides that:
On a comparison of cases for the purposes of section 13 [direct discrimination]… there must be no material difference between the circumstances relating to each case
  1. As per Hewage the question whether a comparator is appropriate is one of “fact and degree”, indeed the circumstances of the complainant and the comparator need not be identical. See also Lord Hoffman’s comments in Watt v Ashan [2008] ICR 82, [para 36-37] “It is probably uncommon to find a real person who qualifies… as a statutory comparator” as a materially similar comparator will be “rare in ordinary life”.
  2. There is an increasing tendency amongst ETs to go straight to the reason for the treatment, rather than look at complex issues of comparators, real or hypothetical. This reflects the belief that a comparator, hypothetical or otherwise, is not a necessary requirement in a direct discrimination case.  That is, a comparator may not be required where there is evidence which points to direct discrimination even without the benefit of the comparative exercise, for example, discriminatory comments.  See the dicta of Lord Scott in Shamoon v Chief Constable of the RUC [2003] ICR 377 at [116].
  3. See also Elias P in Islington London Borough Council v Ladele [2009] ICR 387 [EAT] [2] at [41]:
The logic of Lord Hoffmann’s analysis is that if the Tribunal is able to conclude that the respondent would not have treated the comparator more favourably, then again it is unnecessary to determine what are the characteristics of the statutory comparator? This chimes with Lord Nicholls’ observations in Shamoon to the effect that the question whether the claimant has received less favourable treatment is often inextricably linked with the question why the claimant was treated as he was. Accordingly:“employment tribunals may sometimes be able to avoid arid and confusing disputes about the identification of the appropriate comparator by concentrating primarily on why the claimant was treated as she was.” (para 10)…The construction of the statutory comparator has to be identified at the first stage of the Igen principles. But it may not be necessary to engage with the first stage at all.
  1. Ladele has recently been approved by the EAT in Dr G Kalu v Brighton & Sussex University Hospitals NHS Trust & Others UKEAT 0609/12/BA (Langstaff P).
  2. The prevailing view then in the ET and the appellate courts is that attempts to identify an actual comparator or to construct a hypothetical comparator often distract the parties and the Employment Tribunal from focusing on the “reason why” question. It is that latter approach which is usually the more useful.
  3. In Kalu, the EAT held that whilst the ET may have erred in its application of the comparator, that was not a sufficient ground of appeal, where that had been a clear finding as to the reason for the treatment (ie the approach of going straight to the “reason why”):
The ultimate question to be addressed in the case of unfavourable treatment said to be discriminatory is to ask why the treatment was as it was. It may be that an answer cannot be given directly to that question. There may be room, in such a case, for inferring what the true reason was – and the burden of proof provisions would generally lead to a conclusion that the Respondent could not satisfy the Tribunal that there has been no discrimination of the type alleged. Where, however, the reason for the treatment is established, on balance of probability, to the satisfaction of the Tribunal it becomes unnecessary to ask for a real or a hypothetical comparator.
  1. As Kalu seems to suggest, in relation to hypothetical comparators ETs can bypass this complex exercise completely, taking their cue from Shamoon v Chief Constable of the Royal Ulster Constabulary [2003] ICR 337, HL and JP Morgan Europe Ltd v Chwiedan [2012] ICR 268. This might be thought on the face of it to ignore the statutory language of s13 EA 2010 which requires the comparative exercise: that “A treats B less favourably than A treats or would treat others”.
  2. Notwithstanding the language of s13 EA 2010, certainly during his presidency of the EAT, Underhill P was trenchant in his criticism of hypothetical comparators.
  3. In D’Silva v NATFHE [2008] IRLR 412 the claimant alleged that he had been discriminated against in two respects: the failure of NATFHE (now UCU) to provide him with legal representation and their failure to reply to correspondence.  The Employment Tribunal concluded that the union had declined to provide legal representation because the claimant had explicitly expressed a lack of trust and confidence in the legal team and that the problems with correspondence had been due to a genuine oversight by the solicitor in question.
  4. Despite these findings, the claimant appealed to the EAT to challenge the Employment Tribunal’s conclusion that there had been no race discrimination by contending that it had failed to properly consider the position of a hypothetical comparator.
  5. Underhill P was critical of the claimant’s analysis and his reliance on a hypothetical comparator as it was plainly unnecessary in light of the Employment Tribunal’s unchallengeable findings of fact as to the reason for the union’s actions, which wholly excluded any link with discrimination.  He stated at [30]:
It might reasonably have been hoped that the Frankensteinian figure of the badly-constructed hypothetical comparator would have been clumping his way rather less often into discrimination appeals since the observations of Lord Nicholls in Shamoon v Chief Constable of the Royal Ulster Constabulary [2003] IRLR 285 (see in particular paragraph 11 at p.289) and the decision of this tribunal, chaired by Elias J, in Law Society v Bahl [2003] IRLR 640, at paragraphs 103–115 (pp.652–654). We regard it as clear, taking the reasons as a whole, that the tribunal made an express finding that the only reason why the union acted in the way complained of was that (as regards the initial decision and the first review decision) the appellant had expressed a lack of trust and confidence in his legal team and (as regards the subsequent review) that Mr Bryan had genuinely overlooked the appellant’s further correspondence. Those findings necessarily exclude the possibility that the acts complained of were done, even in part, on racial grounds (or on grounds which would constitute victimisation). If that finding is unassailable it necessarily answers also the question whether he would have been treated more favourably if he had been white or if he had not previously supported Mr Deman or complained of racial discrimination. It is accordingly unnecessary to consider in detail the passages in which the tribunal referred to the nature of the hypothetical comparator. We would however say that we can see no sign that it failed to appreciate any essential feature of the necessary comparison.
  1. Where an ET does select a hypothetical comparator what are the relevant comparable circumstances of that comparator?  In Stockton Tees BC v Aylott [2010] ICR 1278,  Mummery LJ held at [44]:
The relevant circumstances and attributes of an appropriate comparator should reflect the circumstances and attributes relevant to the reason for the action or decision, which is complained of.

What might be observed about this approach is, once again, the primacy of the “reason why” question: it arguably amounts, in effect, to a matter of firstly identifying the reason for the treatment and then reverse-engineering the characteristics of the relevant comparator from there. On that basis, comparators become little more than a “sense check” rather than the principal component of a direct discrimination case, which they were once thought to be.

Knowledge of the protected characteristic

  1. A claimant still has to establish that the alleged discriminator knew that they had a protected characteristic. A recent application of this can be found in IPC Media Ltd v Millar [2013] IRLR 707 (a case involving discrimination arising from disability) C was not considered for two alternative posts during a redundancy exercise. The ET drew an inference that this was because of her past and anticipated future absences due to her disability. Underhill J says that “as with other species of discrimination, an act or omission can occur ‘because of’ a proscribed factor as long as that factor operates on the mind of the putative discriminator (consciously or subconsciously) to a significant extent.”  He accepted the submission that “you cannot be influenced (consciously or unconsciously) by something of which you are unaware” and he held that “the starting-point … is to identify the individual(s) responsible for the act or omission in question.”
  2. The effect of this is to highlight yet another hurdle that a claimant has to overcome to succeed in a discrimination case, as employers may be able to escape liability because the claimant was unable to show that a particular individual decision-maker had the requisite knowledge – in this case, of C’s history of absences (or disability or pregnancy, or sexual orientation) – even though it may be clear that the organisation itself via its senior management employees does have knowledge of the proscribed factor.  However it is arguable that Millar is not authority for a principle of universal application, and that there will still be cases of direct discrimination (or victimisation) where the protected characteristic (or as the case may be the protected disclosure or protected act), although unknown to an ultimate individual decision-maker, continued to be sufficiently causally relevant to the decision in question to allow the claimant’s claim to succeed.


  1. Whilst discrimination law offers practitioners a complex array of tools in the form of shifting burdens of proof and statutory comparators to choose from, the simple fact is that many tribunals are reluctant to use them. Taking their cue from the appellate courts they prefer instead to go straight to what the reason for the treatment was. While this may be seen as an application of pragmatic judicial common sense over the complexities of the law, whether this approach is actually warranted by statute is perhaps a moot point.
  2. It is also debatable whether it is Claimants or Respondents who benefit more from the simplified approach which is now in favour.  So far as the two-stage test is concerned, Elias P in Laing held (at [75] – [76]) that to go straight to the  “reason why” question – in other words to assume that the burden of proof may have shifted – causes no prejudice to an employee but rather risks, if anything, prejudice to the employer.  That may be so in many cases, but in others, Claimants may be disadvantaged by ignoring the two-stage test, which was, after all, introduced in the first place in order to recognise and mitigate the effect of the difficulty of proving discrimination.


[1] “The Equality Bill – Government response to the Consultation”, July 2008 (Cm 7454) at [7.10].

[2] Affirmed by CA

August 2014