This was not a court case that delivered any earth shattering
judgements or precedences, but exposed how a major
employer in
Australia had rigged an employment recruitment exercise in order to prevent older applicants from winning positions as
airline flight attendants.
The low-cost carrier Virgin Blue was established in Australia in 2000 as a partnership between Sir Richard Branson and Patricks, an Australian transport and logistics company that had a pugilistic industrial relations history. Aside from offering fares more competitive than what Qantas and Ansett offered, the business presented itself as a young and dynamic airline, a fresh face in the Australian skies.
And they are fun to fly with. Whereas a four hour hop across the Nullabor on Qantas would involve watching a film you have either already seen or would not be interested in, in Virgin Blue the cheery flight attendants organise fun activities, like seeing which aisle can pass an unravelling roll of toilet paper from one end of the aircraft to the other in the fastest time. Maybe the fact they are young is what gives them their bonhomie, Or perhaps it is mandated in their AWAs (Australian Workplace Agreement).
September 12, 2001 was a fateful day in Australian aviation history. Australia's second largest airline, Ansett, went into voluntary administration and later liquidation, following years of mismanagement by both its Australian and New Zealand owners. Thousands of staff members were made redundant, including many cabin crew members who then sought positions at Virgin Blue. Both airlines were expanding into the markets left void after Ansett's collapse, and both airlines were recruiting for staff.
Unfortunately, a number of older ex-Ansett workers failed to obtain employment at Virgin Blue. This was in spite of them having years of experience as flight attendants, which seemed to have been discounted by the assessors in favour of more subjective measures of competence. Applicants were measured according to 'Virgin flair' ( “a desire to create a memorable, positive experience for customers. The ability to have fun, making it fun for the customer.” ), which was measured by observing how the applicants behaved in various contrived fun activities, like singing and dancing. Recruiters not suprisingly made a beeline to the younger, prettier applicants.
Eight ex-Ansett workers, Nicole Hopper, Maureen Mulherin, Carol Dowling, Keely Bill, Alma Frank, Theresa Stewart, Lynley Boyes and Virginia Jefferies, who all unsuccessfully applied for employment with Virgin Blue, took the airline to court. They believed that they were discriminated against on the basis of age; all eight were aged 36 or older when they first applied to join Virgin Blue, and as later transpired in the course of the hearing, out of the entire 750-strong Virgin Blue cabin crew workforce, the oldest flight attendant is a sole 36 year old. Statistics aside, the process of recruitment and selection was put under the judicial scrutiny of the Queensland Anti-discrimination Tribunal, and while some of the more emotive conspiracies the plantiffs held were dismissed, the selection procedure itself was found to have violated the Anti-discrimination Act 1991.
Member Douglas Savage SC's verdict, delivered on 10 October, 2005, neatly surmises the case:
The Complaint
1. The respondent Virgin Blue Airlines Pty Ltd (“Virgin Blue”) conducts a passenger airline service between ports in Australia. Each of the female complainants are flight attendants by occupation. All applied for, but were refused employment by Virgin Blue as “cabin crew” in the period September 2001 to September 2002 (“the relevant period”). Each now contends that Virgin Blue unlawfully discriminated in their recruitment of cabin crew in the relevant period against applicants who were or appeared to be (as each of the complainants was) over the age of 35. At material times the complainants were aged variously 36, 39, 40, 48, 49, 50, 53 and 56. The discrimination alleged in the pleading is said to be direct or indirect, intentional or unintentional. Despite the width of that complaint, in truth this is a relatively unusual and quite confined case. It is not alleged that the respondent expressly imposed any age limitation on applicants for employment (other than that each be over the age of 18). Rather, it is alleged that in its recruitment process, the respondent during the relevant period has consciously or unconsciously through a subjective assessment process, adopted an age criteria in fact which prevented the complainants being fairly considered for employment.
The Airline Industry Workforce prior to September 2001
2. Virgin Blue commenced operation as a domestic passenger airline in Australia in mid 2000. It was then in competition (albeit as a so-called budget or low cost carrier) with the major established Australian domestic carriers Ansett and Qantas.
3. Both Qantas and Ansett until it ceased operation, employed staff on far more favourable conditions than Virgin Blue offered. Staff were paid at a higher rate and the requirements of their position – the duties, flight schedules etc. were less onerous (see p.442 of the transcript).
4. Until shortly before the events in question here, it was thus relatively unusual for people employed by Ansett or Qantas to seek out employment with Virgin Blue because of the more favourable conditions offered by the former.
5. The airline industry in Australia suffered in the relevant period the usual (or more than usual) decline with age of female participation in the workforce (compared with the male population). Amongst other reasons ascribed for this is the requirement for travel away from a home base and the consequent difficulty with scheduling of work to meet family commitments – all of which (albeit less today than in previous years) still fall disproportionately on the female workforce.
6. A very significant number of female applicants for employment with Virgin Blue before the relevant period were thus young people under the age of 35. The number of older female applicants for employment with Virgin Blue was much lower.
7. Statistical evidence about participation by age in the Respondent’s workforce prior to (say) September 2001 is thus of very little use in detecting the presence of age discrimination in the selection process for Virgin Blue in the relevant period.
Recruitment by Virgin Blue in mid-2001
8. The respondent’s recruitment process consisted of a number of stages. First the applicant was required to complete an application. Second the applicant was required to attend a group assessment. There were other later requirements but since no complainant here was considered further after completion of the group assessment, the further steps in the process are unnecessary to describe.
9. By January 2001 Virgin Blue was receiving many hundreds of posted applications for employment . That caused it (sensibly) to cease inviting such applications and require lodgement of electronic applications.
10. The “on-line” applications were made to an employment services provider “Staff CV Inc”. By logging on to the Virgin Blue website, access could be gained to the Staff CV Inc site. The application form available on-line from Staff CV for use by applicants for employment as cabin crew with Virgin Blue, had as a “compulsory field” (that is, a question which was required to be answered before the employment application would be accepted) an age question. Although Staff CV Inc. required the nomination of an age for each job applicant for Virgin Blue, Virgin Blue did not select potential interviewees for employment by age – the first substantive step in the employment application process.
11. Each of the complainants received an invitation to attend a group assessment. Each attended a group assessment in the relevant period.
The Assessment Process
12. It is apparently an established recruitment practice in Australia to select potential employees from a large group of applicants by a “group assessment” procedure rather than by individual assessment. That was commonly done by the respondent for groups of 30 or less applicants with one or two assessors per group and a number of assessors assessing each group participant.
13. The members of the select smaller group obtained by this process were then subject to a more rigorous individual assessment.
14. The respondent’s group assessments of the complainants were conducted at the respondent’s premises at Brisbane. Those premises contained a large room with a seating capacity for approximately 60 people. The relevant assessment groups were not, as the complainants suggest, groups in excess of 120 people. Whilst the process would commonly see 120 people at the respondent’s premises, that was merely because the assessments were at times conducted one after another. Thus up to 120 people may have been present at the premises, but only 60 or thereabouts were assessed at any one time.
15. Virgin Blue at the relevant times employed up to six assessors – a maximum of 1 assessor for every 10 applicants. Each 60 person group was divided for the assessments relevant here into 6 groups of 10 with at least one assessor. At these times it was not possible to have any more than one assessor per group and it was rarely possible for more than one assessor to assess an individual group member. The time the assessor had to assess at these times was, I infer, less than usual.
16. The selection process depended upon an assessment of what was termed in the evidence as “behavioural competencies” – defined as an underlying characteristic (demonstrated by an applicant in the group assessment) that results in effective or superior performance of the employee at work. The use of such behavioural competencies is a recognised recruitment technique.
17. “Behavioural competencies” are assessed throughout the respondent’s assessment process. Certain specific “competencies” were however particularly and appropriately assessed at the group assessment process – namely “assertiveness”, “team work”, “communication”, and “Virgin flair”. The last of those was defined as “a desire to create a memorable, positive experience for customers. The ability to have fun, making it fun for the customer.”
18. The nominated “competencies” are relevant criteria for selection of employees. Those competencies can be ascertained in the course of a group assessment. The process adopted (which was the preparation and performance of a dramatic routine under the scrutiny of the assessors) was an appropriate way of testing for such competencies.
19. The complainants initially appeared to challenge this process as an elaborate ruse to mask an intentional choice in fact made by the assessors of the most physically attractive employees (male or female). There was no factual basis to support such a claim. That case of intentional direct discrimination was abandoned in addresses.
20. The complainants also expressly contended that the imposition of the requirement of satisfactory behavioural competency testing as a prerequisite to employment per se detrimentally affected the prospect of older applicants compared to younger applicants, and particularly the prospects of people over the age of 35 years compared to those younger than that. There is no substance in this contention. The uncontested evidence of every witness that spoke of behavioural competency testing was that it was intended to, and did in fact, produce an age neutral result. In fact as I understood it, the testing procedure is adopted in other parts of the respondent’s business and those of associated airlines internationally and does produce an age neutral result. There is simply no proper factual basis for any other conclusion. It follows that no intentional indirect discrimination can be established.
21. It follows from the above that the discrimination alleged, if it can be made out, arises because of the respondent’s use of a partly subjective process for selecting employees – namely the subjective views of assessors about the behavioural competencies of applicants for employment. That may be viewed as a case of indirect discrimination made out by demonstrating disparate impact upon particular classes of applicants – cf. and cn. Hemmings v. Tidy Man’s Inc 255 F 3d 1171 at 1190. However I prefer the view that a natural person subjectively applying a lawful criteria in an unlawful and discriminating manner (as the complainants’ case must be) discriminates directly rather than indirectly. It is thus unnecessary to enter upon any discussion concerning whether or not there can be direct and indirect discrimination demonstrated from the same facts in any given case. If I was wrong to give attention to whether or not there was direct discrimination the same result as below would be achieved by considering (as indicative of indirect discrimination) whether the subjective evaluation process had a disparate impact on those 35 years or younger compared to those 36 years or older.
The Evidence
22. The complainants led evidence to support their contention of discrimination:-
(a) of a magazine article containing photographs depicting various female employees of Virgin Blue in sexually provocative poses with an accompanying text;
(b) of a Virgin Blue magazine “Voyeur” which depicted various employees of the magazine who were or appeared to be “young” and mentioning them in the relevant article in a commendatory way;
(c) of the complainants themselves describing the group assessment as one in which it appeared to each of them that the assessors were interested in looks rather than ability; and
(d) a statistical analysis of the age of the respondent’s workforce in the relevant period.
23. None of the
evidence in categories (a) to (c) above is of any assistance in resolving the present complaint.
24. The first magazine article (referred to in (a) above) demonstrates merely that (at best) the public relations department of
Virgin Blue considered it appropriate to publicise the
airline in the media by reference to certain of its then female employees. But the choice of the employees was responsive to the requirements of the magazine editor, not the respondent. It was the
editor of the
magazine who proposed the article and photographic piece. It was that editor who contracted the photographers used and the article writer. These requirements would have plainly remained unchanged whatever the age composition of the respondent’s employees – the article and
photographs do not show anything about the age makeup of the respondent’s workforce.
25. Further, I do not ascribe any supposed discriminatory intention of the publicity department of
Virgin Blue to the employment recruiting team. That was not suggested to any of the assessors in the course of cross-examination.
26. Last, as I set out above, no direct intentional
discrimination was seriously argued, nor has it been made out. The real case, as I understood the complainants’ submission, was that
Virgin Blue encouraged a work culture that equated
youth and its outward physical manifestations with (most directly) the ability to have
fun – part of the “behavioural competency” of “Virgin
flair”. Thus the substantive case was (in contrast to a case of intentional
discrimination) that the assessors unconsciously but invariably preferred younger people (largely of their own age) to older people in such assessment of competency not that the assessors were consciously only interested in magazine
models.
27. Similarly, I reject the
evidence of the “
Voyeur” magazine. It assumes that people of various races, creeds, age etc. should be depicted by the respondent as indicative of the makeup of its workforce. Unlike
Dorian Gray and his picture, the depiction of employees in the magazine is an entirely accurate portrayal of the airline’s cabin crew. In most of the relevant period none were (for example) over the age of 36 years. That fact is not one in the making of the publicity department.
28. The third category of
evidence is the subjective, self-serving
evidence of the complainants about another person’s (that is, the assessor’s) state of mind. It is in consequence, entirely unconvincing as a basis for a finding of prohibited
discrimination. I have already rejected the (potentially more probative)
evidence of the complainants about the way they each in almost identical terms (no doubt framed for them by their solicitors) said the group assessment procedure was conducted. Had I found that 120 people were assessed over a period of a few minutes by one or two assessors, that may have persuaded me that the behavioural competency testing was a mere facade. The
evidence (particularly the uncontested
evidence) does not support such a finding – cf. & cn. Hartley v. Wisconsin Bell 124 F3d 887at 893.
The Supportive Evidence
29. In truth there are but two substantive bases to support the complainants’ challenge in the instant cases:-
(a) statistical evidence of the makeup by age of the workforce selected by the selection process in the relevant period; and
(b) evidence concerning possible flaws in the selection process which may explain an apparent age bias in the statistics in (a).
The Statistical Evidence
30. There is an inherent, but dangerous, attractiveness to lawyers in the apparently value free science of statistics, as a means to proving a discriminatory bias in areas such as employment recruitment by substantive commercial organisations. Significant errors can be made in accepting contentions about either the statistical significance of particular results or more generally the significance of statistics in supporting a finding of discrimination.
31. An example of the former is the attempt by Virgin Blue to demonstrate in evidence that statistically, the respondent gave more favourable consideration to the over 35 year age group in its selection process, than to the under 35 year group. Those supposed “statistics” relied upon a very small minority of applicants who identified their age (and then I infer not entirely truthfully). It is not possible to find such a contention made out when in fact in the relevant period only one person of 36 years of age was employed from over 750 people at or above that age who applied.
32. That is not to say that the complainants’ cases should be upheld. It remains to determine the significance of the statistics that no person over the age of 36 years was employed in the relevant period by Virgin Blue as cabin crew, despite a significant number of over 35 year old former Ansett employees (including all the complainants) applying for employment. The respondent conceded that up until the end of 2002, no person over the age of 36 was appointed (see p.441 of the transcript).
Statistics
33. One hundred and eight years ago the American jurist Oliver Wendell Holmes noted in the gender specific language of the day “for the rational study of law the black letter man may be the man of the present but the man of the future is the man of statistics...”. (The Path of the Law – (1897) 10 Harvard LR 457 at 469).
34. Today as that prescient passage suggests, the jurisprudence of US courts dictates benchmarks for determination of whether statistical evidence supports a valid inference of discrimination – cf. & cn. Hazelwood School District v United States 433 US 299 (1977); Ottaviani v State University of New York 875 F2d 365 (2nd cir. 1989).
35. That approach has recently found favour in European Courts – Bilka-Kaufhaus v Weber von Hartz 1986 IRLR 317; Stadt Lengerich v Helmig 1995 IRLR 216; Enderby v French Health Authority 1993 IRLR 591; R v Secretary of State for Employment ex p Seymour-Smith 1999 IRLR 253.
36. The Tribunal did not have the benefit of any evidence of a statistician about the statistical significance or the significance of the statistics in the instant case. Nevertheless, it seems to me that there is neither need nor warrant to displace a judgment about the legal significance of various statistical results with expert scientific evidence about those same results. Nor do I think that it is appropriate to adopt here some statistical standard by which prima facie discrimination is demonstrated or which determines what is significant enough to call for an explanation of the prima facie evidence.
37. What is required in each case is a judgment about whether all the evidence, including statistical evidence, requires a proper inference (in accordance with established methods of legal reasoning) of discrimination. The use of inference in this jurisdiction has been considered in cases such as D v G and O Pty Ltd 1997 QADT 8. Clearly, where there is a more probable explanation available, then the inference must not be drawn.
38. Here the relevant factors may be shortly stated.
39. First, the behavioural competency testing is designed to produce (and in fact produced in other applications) an age neutral result.
40. Secondly, the behavioural competency testing in the instant case did not produce an age neutral result. The proportion of people over age 35 who were selected from applicants in the period September 2001 to September 2002 were far lower than the proportion of people under 35 years who were selected from the group assessment procedure. I do not accept the respondent’s statistical analysis which seeks to demonstrate the opposite.
41. Thirdly, the respondent did not persuade me by the bland assertion made in evidence, that this statistic was merely a consequence of the age make up of the applicants for employment. That simply is not so. Based on the respondent’s pleadings, there were about 90 per cent of applicants in the 18 to 35 age bracket, and 10 per cent were aged 36 or over. However, only one of approximately 750 applicants who were in fact successful in obtaining employment in the relevant period was aged over 36 years. The evidence does not disclose precisely how many unsuccessful older applicants passed the group assessment but ultimately failed in the further stages of the recruitment, but that cannot effect the result so much as would be necessary to dispel the otherwise obvious finding that the assessments were not being conducted as designed. It should be remembered that if applicants passed the group assessment stage, behavioural competency was also individually assessed at a further stage of the recruitment process.
42. Fourthly, the respondent other than by the contention mentioned immediately above, did not suggest any other reason for the apparent statistical disparity.
43. Fifthly, the respondent’s alternative submission that discrimination should not be inferred because of the insignificant age disparity between one of the complainants (Ms Hopper, who was 36 years at the time of assessment) and at least one 36 year old applicant who was in fact employed, should also be rejected (cf. & cn. Isenbergh v Knight Rider Newspaper Inc. 97 F3d 436, 440). That ignores the fact that the disparity is much more generally between all the complainants and all the successful applicants 35 years or younger. The fact that these two groups coincide only at age 36 appears more likely to suggest discrimination than to demonstrate the reverse.
44. Sixthly, although the testing procedure for the respondent’s group assessment was independently verified by Dr Gudmundsson- whose evidence I otherwise accept – all that her report demonstrated (at best for the respondent) was that the employees chosen were people who arguably displayed the behavioural competencies. That proposition was not in contest. The fact is that no expert evidence suggested that the rejected older applicants – including the present complainants, did not display those competencies.
45. The respondent did not undertake any cross check at any relevant time to show whether or not older people with the requisite competencies were being rejected. Thus the respondent was not able to demonstrate the system in fact worked as it was intended.
46. The statistical evidence thus supports a finding that the group assessment procedure did not work as it should have.
47. I do not accept the respondent’s explanation for this statistical result. This anomaly indicates in fact, that older applicants were treated less favourably than younger ones. Thus the complainants make out their complaints of discrimination based upon age.
48. Although it is unnecessary to do so having regard to my findings above, a number of factors lead me to conclude that the assessors applying the otherwise age neutral system were unconsciously discriminating on the basis of age and that this caused the statistical variance. Those factors may be summarised thus:-
(a) between September 2001 and September 2002 the respondent’s recruitment system was placed under a severe and unanticipated strain by reason of the collapse of Ansett and the influx of a large number of Ansett’s former employees, including female flight attendants many of whom were older than the hitherto historical norm and older than 35 years;
(b) the group assessment system was designed for much smaller groups of applicants. It was required to deal with up to 60 people per group assessment with a maximum of 6 assessors – 1 for every 10 applicants. This was less assessors than the system had been designed to handle;
(c) most of the assessors had had comparatively little training in the psychological assessment procedure (cf. & cn. Holloway at p.456-7 of the transcript) and no occasion had arisen, nor had any system been established to test the results. Since the intake of employees was adequate, no need for that was identified. This is not to say that the assessors had not had “adequate” training to assess applicants, but merely that defects in assessment were not the subject of safeguard.
(d) Formal behavioural interview training undertaken by senior recruitment staff through an external training organisation did not include training in the assessment of the behavioural competency “Virgin flair” (see the evidence of Holloway at p. 435 of the transcript);
(e) at least the assessors who gave evidence at the hearing were comparatively young - most must have only recently joined the company at the time of these events. All had little experience in what was for most, a new system;
(f) inevitably a danger of employing the behavioural competencies system, especially as it required an assessment of “Virgin flair” was to identify with persons of the same age and experience as the assessors, or what the assessors regarded as, if not of the same age, a “fun” person. That person was I think likely to be a person of the same age, social class and life experience as the assessor. This was explored with the expert witness, Dr Gudmundsson as follows:-
Yes. And so does that find its way into these sorts of processes? I mean, if we had quite young assessors would there be a bias which might find its way into the results if uncorrected in assessing much older people?..There is a potential for a similar-to-me effect to bias someone's perception regardless of whether they're young or old. I mean, there's just that similarity between the assessor and the candidate.
Mmm?... That's, once again, why you try and have very detailed descriptives of what the positive and negative indicators are. So that you always need to bring it back to the behavioural observations. That's the way you could start to overcome some of our perceptual biases. The same things occur in performance management, performance reviews and so on. There are a number of human perceptual viruses that can impact on how we'd view someone's performance. So you need to have strong behavioural indicators in order to help to overcome that particular bias. And if you have multiple raters and then you're averaging your scores across multiple raters
Then it's better because ... ? ... also helps to alleviate that; that's correct.
There were no “multiple raters” in these cases.
49. Experience, a lessening work load and perhaps a wide-spread express concern by people such as the complainants, I think resulted in significant correction of the anomaly after September
2002.
50. Given the above findings, I find that the case of direct
discrimination on the basis of age made by each of the complainants is made out.
51. I order that the respondent pay the complainants’ costs of these proceedings, including reserved costs in accordance with the District Court scale.
52. I will hear the parties further concerning the appropriate relief including individual assessments of damages. I note that because the parties failed at the hearing to complete their
evidence about
quantum, I have not thought it necessary to deal presently with submissions concerning whether the
discrimination caused any of the complainants to lose an opportunity for employment or whether, had I concluded that they would not in any event have been employed, any of the complainants would be entitled to any substantive relief.
Douglas Savage SC
Member
Anti-discrimination Tribunal Queensland