. It is regulated in
national and international law. The rationale for regulation is the
assumption that discrimination is unfair; the legal issue presupposes the
moral one. If discrimination does not measure up as a moral concept, the
legal regulation is not a regulation of “it” but of an ad hoc
set of behaviors that we dislike, possibly for good reason. I start by
trying to identify what discrimination is taken to be in the moral and
legal discourse. I will go on to set up a number of requirements an
account of discrimination should meet and assess what I understand to be
the standard view in light of these requirements.
Pinning discrimination down as a legal issue in need of regulation
makes it more difficult to account for it in a meaningful way as a moral
issue. There seems to be a conflict between these two concerns. I will end
by spelling that conflict out.
Setting the Scene
It is difficult to infer from existing legislation what discrimination
is understood to be. Consider these two legal documents:
Every effort should be made to ensure that human genetic data
and human proteomic data are not used for purposes that discriminate in a
way that is intended to infringe, or has the effect of infringing human
rights, fundamental freedoms or human dignity of an individual.
(International Declaration on Human Genetic Data, Article 7,
UNESCO; 2003)
An employer may not disadvantage a job applicant or employee
with a disability through treating him or her less favorably than the
employer treats or would have treated persons without such a disability in
a similar situation, unless the employer shows that the disadvantage is
not connected to the disability. (Act Prohibiting Work Related
Discrimination of Persons with Disabilities, SFS 1999:132 Paragraph
3; my translation from Swedish.)
In the first example, discrimination is a way of expressing that
someone's human rights are violated. In the labor law example,
discrimination is essentially a matter of interpersonal comparison. A
person is discriminated if treated less favorably than someone else
because of a property that he or she has but the other person lacks. If
they are treated equally there is no discrimination, even if both are
treated badly (there might then, of course, be a violation of something
else). In the first example, discrimination is assessed in relation to a
moral standard. In the second example, it is assessed in relation to how
other people are treated.
In the human rights example, “discrimination” seems
redundant. The article says that you may not violate anyone's human
rights, but we know that already from the meaning of right. In
the labor law example “discrimination” is used to describe a
particular type of decisionmaking, where differentiations are made between
people on the basis of a personal characteristic. Here
discrimination signifies a particular form of unfairness and is
not redundant. For this reason, this is what I will focus on.
We now have a place to start. From the semantics of the word we know
that discrimination involves making distinctions or noting differences. We
discriminate between P and Q when we note in which ways P and Q are
different, rather than in which ways they are similar or the same. The
presumption of unfairness tells us that the reason why discrimination
enters into moral and legal discourse is that making distinctions or
noting differences and letting this influence the way we treat some
relative to others is sometimes unfair.
In a sense discrimination is a morally neutral concept, because we
necessarily discriminate between people all the time. If we didn't,
we couldn't even single out an acquaintance in a crowd. I will,
however, use it as a normative concept: When an action has been correctly
described as an instance of discrimination, it has at the same time been
correctly described as unfair.1
Needless to
say, an action can be unfair for other reasons too. Discrimination is
merely one form of unfairness.
Words have different meanings in
different contexts and we are not obligated to bring general or historical
meanings of a word into our moral discourse and stick
unfair or
immoral in front of it to signify why we are concerned about it.
Because behavior is impossible without discrimination in the sense of
noting differences, we do not need to burden our moral debate with it.
On the same note, discrimination does not apply in all contexts. In
all relationships we are in, personal and nonpersonal alike, we make
decisions that disadvantage some individuals in relation to others because
of personal characteristics. If Jane dates Bob rather than Bill because
Bob has conversational skills that Bill lacks, Jane still does not
discriminate against Bill. The relationship between them is not one where
discrimination applies because it is not one where fairness applies. Jane
does not have an obligation to date a man she does not like. There is
controversy over the domain of fairness, coinciding with the controversy
over which relations are public and which private. Here it is enough to
note that only in those relations and settings where issues of fairness
apply—whatever they are—can the issue of discrimination apply,
because discrimination is a form of unfairness.
The Standard View
From the literature, and from law and legal practice, we can identify
what I call the Standard View on discrimination. The descriptions we get
tend to contain two features: “differential treatment” (or
“treating less favorably”) for “arbitrary,”
“irrational,” or “irrelevant” reasons. The
variations of the second feature amount to the same thing, because reasons
are supposed to be arbitrary or irrational because they are irrelevant.
Here are some examples that I treat as falling within the Standard View: A
classification “must be reasonable, not arbitrary, and must rest
upon some ground of difference having a fair and substantial relation to
the object of legislation, so that all persons similarly circumstanced
shall be treated alike”;2
Reed v.
Reed 404 U.S. 71; 1971.
“Discrimination is narrowly
conceived as ‘irrational’ decision making, unrelated to
differences in the productivity enhancing characteristics of particular
workers”;
3 Conaghan J. Feminism and
labour law: Contesting the terrain. In: Morris A, O'Donnell T, eds.
Feminist Perspectives on Employment Law. London: Cavendish
Publishing; 1999:31.
“Sex discrimination, as commonly
interpreted, involves the arbitrary or irrational use of gender in the
awarding of benefits or positions.”
4 Kymlicka W. Contemporary Political Philosophy.
Oxford: Oxford University Press; 1990:240.
Piecing together the operative parts of these descriptions (I doubt
whether they qualify as, or are intended to be taken as, definitions) we
get the Standard View: Discrimination is decisionmaking
representing (or resulting in) a disadvantage for someone (P) on grounds
that are irrelevant in the decisionmaking context (C). The ground (X) is a
personal characteristic.
Consequently, “P is disadvantaged because of X in C” and
“P is discriminated against because of X in C” are equivalent
if and only if X is irrelevant in C. A familiar example is an employer
choosing an able-bodied job applicant over an applicant with a disability
when the disability does not affect the person's capacity for doing
the job.
I have said that I use discrimination as a normative concept. That is
not the same as saying that the meaning of the term
discrimination contains a moral component. Even if we choose a
value-neutral definition of it, we use it in argument with the
purpose of conveying that an unfairness has been done. In saying that an
act qualifies as discrimination, we pronounce it as or convey the message
that it is unfair.5
Compare the distinction
between locution (meaning) and illocution (force) in Austin JL. How to
Do Things with Words. Oxford: Oxford University Press; 1975:94, 98,
109.
General Requirements
If an account of discrimination is to qualify as an aspect of a theory
of fairness, then it needs to satisfy certain requirements. What they are
will always be contentious. As we have seen, there is not even agreement
over the area of application for a theory of fairness. The requirements
that I suggest here seem reasonable and I hope that my discussion of them
will make the case for each. I do not claim that the list is exhaustive.
All I claim is that an account that satisfies these requirements is
stronger than one that does not. I suggest, then, that an account of
discrimination should
a. specify where the badness lies
c. not be conditioned on bad intention
d. offer a stable criterion for fairness
e. be resilient against unfair background
factors
f. contain a nonarbitrary and non-question-begging principle
for ground selection.
I will spell these requirements out and see how well the Standard View
fares. To anticipate, the Standard View can meet the first three
requirements but struggles with the last three. Moreover, the merit of
meeting item b is somewhat spoiled by its inability to meet item e.
Item a: Where the Badness Lies
We need to know what it is that makes discrimination a concern. On the
Standard View, it either represents or results in disadvantage. On the
first alternative, the disadvantage resides in the unfairness itself.
There is a widely accepted assumption that people who are relevantly
similar should be treated alike. Take John and Jerry, who are equally good
at their jobs. John is passed over for promotion solely because he is gay.
Jerry, who is not gay, gets promoted. Provided that sexual orientation is
irrelevant for the kind of work they both do, John suffers an injustice.
That injustice is the disadvantage, regardless of the effects it may have
on his life. The disadvantage condition is satisfied automatically by the
unfairness of the procedure. On the other alternative, two conditions for
unfairness have to be satisfied separately: decisionmaking on irrelevant
grounds and negative effects on the person's life resulting
from this decision, like a loss of income or emotional trauma. It is
reasonable to require that the disadvantage is not trivial.
We do not need to decide on the relative merits of these positions. It
is enough to note that the Standard View has an answer to where the
badness lies.
Item b: Context Sensitivity
An account of discrimination should be context sensitive in two
ways.
First, it should be able to explain the difference between these two
cases: (1) Sara is passed over for employment as a bus driver because she
is a woman. (2) Sara is passed over for employment as a therapist in a
male support group because she is a woman. These cases illustrate the
difference between different types of relevance: context relevance and
moral relevance. Sex is morally irrelevant—it should not be
allowed to influence our moral principles—but may still be
context relevant, as in the second case. It is perfectly possible
that a woman is unsuitable as a therapist for males with shattered egos.
If that is the case, passing Sara over is not a decision made on context
irrelevant grounds, even though sex is, as always, morally irrelevant.6
When John Rawls talks about “things that
are irrelevant from the standpoint of justice” (A Theory of
Justice. Oxford: Oxford University Press; 1971:18f), he is referring
to morally irrelevant factors.
On the Standard View, as
I understand it, “relevance” is context relevance.
It would be possible to piece together a position resembling the
Standard View (or perhaps even a variant of it) where the relevance
criterion is taken as moral relevance. Jan Narveson's position is an
example: “Discrimination is treating some people less favorably
than others for morally irrelevant reasons.”7
Narveson J. Moral Matters. Peterborough, Ontario:
Broadview Press; 1993:243.
But it is difficult to see how this
position could work. A characteristic is morally irrelevant as a matter of
principle and remains so irrespective of context. With such a criterion,
we could never make decisions that affect individuals based on factors
like sex, religion, or sexual orientation. Gay organizations would not be
allowed to favor gay people, churches would not be allowed to favor
believers, and shelters for battered women would not be allowed to hire
women only. In the Standard View, “relevance” is the relevance
of a property in a given situation, that is, context relevance. So the
Standard View satisfies the first part of this requirement.
The second part is that the context irrelevant characteristic should
be the correct explanation for the disadvantage in the case in question.
An account of discrimination should be sensitive to how a context
irrelevant characteristic features in a decision. Say that Sara is passed
over for employment at an assembly line and this has to do with the fact
that she is a woman. The way in which her sex features in the decision is
that she is 164 cm tall. The height requirement for working at the
assembly line is 170 cm, which means that women of average height, like
Sara, are excluded. Men of average height are not. If the assembly line
could be adjusted without unreasonable cost to suit women better, then the
context irrelevant property of being a woman does feature among the
reasons for the decision not to employ Sara, and the decision might amount
to discrimination. This is sometimes referred to as “indirect
discrimination,”8
European Council
Directive 97/80/EC on The Burden of Proof in Cases of
Discrimination Based on Sex, Article 2. Official Journal of the
European Community 20.1.98.
meaning that a selection
criterion that looks neutral systematically leads to disadvantage for a
particular group. If, however, the height requirement is determined by
technical factors (changing it might result in risk for employees), then
the height requirement is independently motivated.
The Standard View lives up to this second part of the context
sensitivity requirement as well. The two aspects of context sensitivity
are shown by the emphasized passages in this formulation of it: P suffers
discrimination if she is disadvantaged because of X in C and
X is irrelevant in C.
Item c: Independent of Bad Intention
The Standard View describes discrimination through describing a
selection process and its effects externally. The decisionmaker's
reason can be prejudice, ignorance, or even misguided benevolence. The
reason is not part of the ground for classifying an act as discrimination.
The disadvantage caused or represented by the decision need not be
intended. This is a strength.
Stereotypes and prejudice influence and reinforce attitudes and
behavior without us always being conscious of it. If the classification of
the action turns on an intention to disadvantage someone, this is lost
from view. People tend to believe that their actions are justified. An
employer may sincerely believe that women make bad workers or are better
off at home. Prejudice and misconceptions result in disadvantage, but the
intention may be to secure production. The Standard View does well to
resist including a condition of bad intention.
Item d: A Stable Criterion for Fairness
The Standard View relies on the intuition that decisions affecting
individual persons should be made on relevant grounds. But here is a
potential conflict with other intuitions about fairness.
Let me exemplify. In some countries, insurance companies are
prohibited from asking individuals seeking health or life insurance to
undergo or disclose results from genetic tests. The reason for regulating
“genetic discrimination” is to ensure that people are not
disadvantaged because of their genetic makeup. There is a strong intuition
at work here as well: It is not fair to be disadvantaged because of
something one cannot control, and one cannot control one's genetic
makeup. But this intuition collides with the relevance criterion in the
Standard View. In assessing health risks, genetic information is certainly
relevant.
Many characteristics that are irrelevant in other contexts are
relevant in insurance decisions. As far as it is statistically true that
women live longer than men, sex is relevant for calculating premiums on
pension insurance. A disability is relevant in calculating the risk of
illness and premature death. According to the second intuition, allowing
such information to be used against an individual's interests is
unfair. According to the first intuition—the one that feeds the
Standard View—it is not.
Still, we might believe that allowing genetic information (or any
medical information for that matter) to be used in insurance decisions is
unfair because it makes insurance unavailable for those people who need it
the most. If we believe that, then in all likelihood we will find
regulation of insurance decisions legitimate on that ground.9
It might be argued, not unreasonably, that the
best solution to this problem is a general public health insurance rather
than regulation of the private sector. Indeed, the effects of an
unregulated private insurance market may not be very serious in welfare
states where there is protection available for everyone. This is a
consideration applicable to that variety of the Standard View that
includes a condition of nontrivial negative effects (see item a). However,
the intuition feeding the Standard View still stands. If I want to top up
my public health insurance with private insurance, then fairness requires
that decisions taken in my case, as well as in others', should be
taken on context-relevant grounds.
Proponents of the Standard
View might agree with this as a matter of general fairness.
10 Because discrimination applies only where
fairness applies, it is possible to argue against regulation on the
libertarian grounds that private companies are not constrained by
considerations of fairness. If so, one would, of course, have to argue
against all discrimination laws that apply outside of the narrow sector
that libertarianism recognizes as public.
However, this point
seems to collide with the intuition about fairness upon which their view
of discrimination relies. They can try and deal with this by offering an
order of priority between these two intuitions. If they believe that the
intuition feeding the Standard View overrides the intuition explaining the
regulation of the insurance market, then they are committed to objecting
to regulating the insurance market. If they accept that the intuition that
supports regulation overrides the intuition feeding the Standard View in
this instance, then they are at least committed to arguing that
“genetic discrimination” is the wrong label. An unregulated
state of affairs may well be unfair, but it is not discrimination. In the
first option, the Standard View allows unfairness in the name of fairness
without having the resources within itself to explain why we should find
this order of priority compelling. What the second option reveals is that
the criterion for discrimination advocated by the Standard
View—context relevance—can seem arbitrary from the point of
view of the victims of decisionmaking. When the disadvantage is real, the
relevance criterion seems, well, irrelevant.
Item e: Resilience against Unfair Background
Factors
The Standard View satisfies the requirement of context sensitivity
specified in item b. I have, however, indicated that the merit of that is
countered by an inability to provide resilience against unfair background
factors.
A personal characteristic can be context relevant—and unfit to
ground a claim of discrimination—for reasons that are unfair.
Powerful groups shape positions to suit their interests and the work they
do is more highly valued.11
See Radcliffe
Richards J. The Sceptical Feminist. A Philosophical Enquiry, new
ed. London: Penguin; 1994:152. See also the Swedish court case Midwife
v. Örebro County Council (Labour Court, AD 1996 no. 41). The
court argued that a wage difference between a female midwife and a male
hospital engineer was not discrimination because the man's
qualifications had wider labor-market appeal.
Inequality and
hierarchy, on the labor market and elsewhere, have an effect on how
positions are formed and valued and on who does what where.
12 See Bradley H. Gender & Power in the
Workplace. Analysing the Impact of Economic Change. Basingstoke:
MacMillan; 1999:ch. 5, for how inequalities are attributed to
“natural” features, like female domesticity, and therefore not
discriminatory. On background inequalities making differences relevant:
“the assumption [is] that where such differences [in
productivity enhancing characteristics] do exist and,
howsoever derived (for example, as a consequence of unequal access to
educational or training opportunities, or the gendered allocation of
labour in the home) they are relevant to decision making,
regardless of the gendered consequences which may flow from them.”
See note 3, Conaghan 1999:31–2.
The Standard View struggles
to deal with that.
Let's return to Sara, who is too short for the assembly line. The
requirement works to the disadvantage of most women but not most men.
Earlier, I said that if the height requirement is motivated by technical
factors only, then passing Sara over for employment is not discrimination.
But let's pause for a moment. If the “typical worker”
were shorter than 170 cm, how likely is it that an assembly line would be
constructed in such a way that only people taller than that can operate
it? It is likely that the technical factors that explain the height
requirement and make it relevant are, in turn, explained by the male norm
on the labor market. Indirect discrimination does not solve this problem.
What makes a rule or criterion—rather than an individual
decision—context relevant may also be unfair, as this example
shows.
The Standard View cannot reach behind relevance to assess why
something is relevant in a certain context. It might be argued that this
problem would be lessened, or even solved, if the Standard View relied on
moral relevance—as Narveson seems to think that it does—rather
than context relevance. But that would not help the Standard View overall.
The context sensitivity of the Standard View is a strength. I have
indicated that it purchases this strength with an inability to meet the
requirement currently under consideration. Anyone tempted to try and meet
this requirement by switching to moral relevance, abandoning context
relevance, would, as I have argued, do so at the cost of not meeting
requirement b and ending up with an unworkable account.
The inability of the Standard View to offer resilience against unfair
background factors does, I think, partly explain why discrimination law
has proved so comparatively powerless. There is always “something
else” to explain why, for example, women make less money than
men—less experience, more absences, less bargaining power—all
of which can be traced back to gender inequality. As long as there is a
context-relevant “something else,” the Standard View sees no
discrimination even if the “something else” is unfair.
Item f: A Nonarbitrary and Non-Question-Begging
Principle for Ground Selection
Decisions can be taken on all sorts of irrelevant grounds, but we tend
to believe that there is something special about those characteristics
that can be ground for discrimination (let's call them
D-characteristics). If so, we should be prepared to offer a principle for
why things like sex, religion, ethnicity, sexual orientation, and
disability are on the list whereas hairstyle or eating habits are not.
There are a number of possible principles. The question is whether any of
them works, both in general and specifically for the Standard View.
One suggestion is that a D-characteristic is such that it can be the
focus of group identification. The characteristic matters to the
collective identity of people who have it. But we are after a principle to
evaluate the intuition that religion and sex are D-characteristic whereas
hairstyle probably is not, and people can identify with others on the
basis of anything they want. We cannot decide in advance what really
matters to people. This principle is question begging.
Even if it did work in general, this principle would not work for the
Standard View. If we made identity a condition, the D-status of a
characteristic would depend on how people conceive of themselves. Take the
Deaf community. People who identify with the Deaf community—not all
deaf people do—regard themselves as a linguistic minority, not as
disabled. If discrimination requires self-identification, disadvantage
based on deafness is discrimination on the ground of disability only if
the affected person regards him- or herself as disabled. This does not
suit the Standard View, in which discrimination is a matter of external
description. The Standard View does not enter into anyone's mind.
Alternatively, a D-characteristic is immutable. The idea is that it is
particularly bad to be disadvantaged because of a characteristic one
cannot help having. The first problem with this suggestion is that it does
not fit “the list.” Converts and transsexuals may suffer
discrimination on the basis of religion and sexual orientation not because
of what they cannot help being but of what they have turned themselves
into. Neither is it obvious why adopted characteristics should be less
worthy of protection. Even if this principle were convincing on its own,
it would not work for the Standard View. We have already seen (under d)
that one intuition on which the Standard View relies—fairness
requires that decisions are taken on relevant grounds—may come into
conflict with the intuition that it is unfair to be disadvantaged because
of a factor one cannot control. Immutability as ground for selecting
D-characteristics would seem to require adherence to that second
intuition.
A final suggestion is that D-characteristics are particularly potent
sources of disadvantage. As we have seen, though, the Standard View does
not require a notion of disadvantage separate from the unfairness. Even if
it did, using it to determine beforehand which characteristics are
D-characteristics would again beg the question.
