About
the database
Background
Definitions
Need for reliable measurements
Difficulties of measurement
What analysis can show
The global sweatshop issue is evolving
from promises to proof – from getting major multinational corporations
to pledge that their products will not be made under sweatshop conditions,
to finding out whether those pledges are actually being kept.
As
this shift occurs, it brings up a very practical question.
What pieces of information are needed to show what the working
conditions actually are, in any particular workplace located anywhere
in the world? In other words, what should count as proof that companies
are meeting their promises?
A
multinational company selling consumer goods like clothing can use hundreds
or thousands of factories around the world to make its products. Up
to now, the problem of getting any information at all, from so many
different workplaces, has tended to overshadow the problem of whether
the right kinds of information are being sought. But figuring out the
right measurements, measurements that can reliably show whether working
conditions in a specific workplace do or do not meet pre-agreed standards,
is critical.
Several
efforts are underway to specify some or all of measurements that should
be taken. Companies trying to
watch over their own contractors and suppliers, sweatshop critics trying
to document abuses, and a variety of monitoring organizations offering
their services are each using their own sets of measurements, sometimes
taking the shape of questionnaires
that monitors visiting a factory are asked to fill out and procedures
they are asked to follow. But there is no widely accepted form and no
consensus. The state of the art is experimental at best.
The
purpose of this analysis is to advance the state of the art, by looking
at every unit of measurement currently being used by any participant
in this field; identifying particular strengths, weaknesses, and best
practices from that collective experience; and offering potential improvements.
The analysis uses a database
we have constructed, and put on the internet for public use, to make
the many scattered pieces of this collective experience easy to find,
sort, and compare.
Our
findings are summarized in Highlights
and detailed in sections that separately cover each of the major topics
of workers rights. The findings are based on our analysis of current
measurement practice, but we invite readers to use the database for
their own purposes and to draw their own conclusions. We hope that a
clear view of experience to date will help every participant in this
process to improve on current practice, to recognize key measurement
limitations and compensate for them, and to move toward
more uniform measurements that all participants on all sides
can use. Without standardized yardsticks,
the process of proving that global companies are keeping their
no-sweatshop promises, or breaking them, will not be able to advance
far.
About
the database
Our
researchers compiled 2,900 individual units of measurement being used
in this field, from 100 different sources ranging from individual companies
and industry groups, to monitoring organizations, to non-governmental
watchdogs, to the International Labor Organization. We then organized
them in a database with many different searching and sorting capabilities,
to make it as easy as possible to find and analyze current measurement
practice on any particular subject and from any particular angle. The
database itself includes
a full description of its purpose,
content, and methodology.
Background
In
the last decade, public outrage over extremely abusive working conditions
revealed to be occurring in the production of such familiar products
as sport shoes, soccer balls, and clothing in general has turned into
a strong demand for multinational corporations to take responsibility
for the treatment of workers who make their products.
Looking
for the quickest, cheapest ways to produce labor-intensive goods in
a global economy, those companies have been moving much of their manufacturing
to low-wage countries where even basic legal protections for workers
are non-existent and where union organizing is strongly discouraged
or prohibited. The result is that millions of the workers the global
economy uses– typically, women and children – are subjected to conditions
far below what international labor rights standards call for.
As the public learned about these
conditions through the media, attention focused on the brand-name companies
involved and created pressure on them to be accountable. Recognizing
the importance of labor rights as basic human rights, the Lawyers Committee
for Human Rights devoted itself early on to pursuing mechanisms of corporate
accountability for labor rights.
A
number of companies have responded to public pressure by adopting codes
of conduct, which set out basic
workers’ rights and workplace standards for their suppliers and contractors
to follow worldwide. Some have also required their suppliers and contractors
to submit to monitoring to determine their compliance with the new workplace
codes. The Lawyers Committee for Human Rights played a major role in
helping establish and develop the Fair
Labor Association, a non-profit, multi-stakeholder organization
made up of industry, non-governmental organization (NGO), and college
and university members, operating an independent monitoring system to
hold participating companies accountable for compliance with a uniform
workplace code of conduct.
Definitions
Monitoring
involves the collection of information about particular workplaces,
and to be credible it needs a system of measurement.
[1]
The measurement process asks questions, gets answers, and
holds those answers up against a defined code standard. It can be broken
down into separate subjects, such as Freedom
of Association or Health and
Safety, but in each case the three key elements are the same. To describe these three elements, this analysis
uses the terms units of measurement,
measurement results, and code
standards respectively.
Units
of measurement. These are the many different yardsticks that
an observer could apply, the defined pieces of information that the
monitoring process is supposed to collect. For example, to help determine
the degree of fire safety, one unit of measurement might be the width
of the aisles on the factory floor. Another might be whether fire exits
are unlocked. In effect, units of measurement are the questions that the process is meant to
answer (“How wide are factory aisles?” or “Are all exits unlocked during
factory hours?”). Units of measurement are meant to be the same from one case to the
next: asking the same questions each time is part of a consistent process
of evaluation. This analysis
refers to each separate piece of desired information as a separate unit of measurement. A 62-item set of questions for Fair Labor Association
monitors to use during their factory visits
[2]
is thus a collection of 62 different measurement
units. Each of the 2,900
items in our database,
collected from a wide range of different sources,
[3]
is a unit of measurement.
Measurement
results. These are the answers, in specific
cases, to the questions posed by units of measurement.
If the measurement unit is “width of factory
aisles,” the measurement result
in a specific factory might be “44 inches wide.”
Measurement units are
the yardsticks; measurement results
are the readings taken with the yardsticks. Results
will vary from workplace to workplace; units
stay the same.
Units
of measurement are much more likely to be disclosed to the public than
measurement results, particularly
when the source is a company or industry. In general, companies find
it much easier to disclose the list of items they are looking for, as
evidence of workplace conditions, than to disclose what they or others
have actually found: “we want to know the width of the aisles,”
but not “the aisles were 44 inches wide in factory A; 31 inches
wide in factory B”; etc. Independent observers like non-governmental
organizations do not feel the same constraint, and to date they have
been the public’s main source of measurement results for individual
workplaces. But NGOs’ measurement activities tend to be limited, understandably,
to only one or a handful of workplaces. Only one-third of the units
of measurement we identified in our research were accompanied by actual
measurement results in the sources we derived them from, and most of
those results were for a company’s operations only in the aggregate,
or for conditions in an entire country, rather than for an individual
factory or other specific workplace. The database
shows the details.
Code
standards. These are the criteria for compliance with workplace
codes of conduct that are set out in the codes themselves. In effect,
they are the promises that companies are making – what they say their
workplaces will be held to. Code standards are easy to locate, but they
nearly always come in general terms such as “Employers shall provide
a safe and healthful working environment to prevent accidents and injury
to health . . . .”
[4]
How to apply them case by case is the key practical
question, requiring much more detail in the form of units of measurement,
monitoring procedures, rules of interpretation, etc. The ambiguity of code standards can create problems in determining whether they are
being met, even when reliable measurement results are available. See
Difficulties of measurement below.
Indicators.
The term indicators
is often used in monitoring to describe characteristics of a workplace
that are particularly significant in determining whether code violations
are occurring. In effect, indicators
are an intermediate step between individual units of measurement and
overall code standards. Indicators
tend to be derived from a number of individual measurement units taken
together. For example, analyzing the proportion of women to men in the
lowest-wage positions in a factory, the comparative success of women
and men in being promoted and in earning higher actual wages over time,
and other such statistics can produce a strong indicator
of possible discrimination on the basis of gender.
Defining
indicators in terms of their measurable components is a step that is
often neglected, even though it is a necessary part of generating indicators
that are reliable. The definitional problem cannot be resolved by simply
defining the indicator itself as a unit of measurement (e.g.,
“Is there discrimination?” (record
805), since all that does is to shift the problem onto the shoulders
of the monitor.
[5]
Reliability
is defined and discussed in Elements
of Reliability.
Need
for reliable measurements
To
be effective, monitoring needs to produce information that can be dependably
used to gauge the degree of compliance in any particular workplace with
defined workplace standards. The information itself needs to be reliable,
of course. It also needs to be in consistent form, gathered using standardized
methods from workplace to workplace and from country to country so that
compliance determinations are consistent and comparisons between workplaces
are fair. These needs put a premium
on defining the units of measurement that the monitoring process will
use.
Findings
based on consistent measurement of workplace conditions against fixed
standards were not needed to begin the international movement against
sweatshops. The early exposes that launched major public concern were
based on conditions so outrageous that they would have failed any imaginable
set of measurement criteria. Abuses were so blatant that measurement
was not necessary.
Once
code standards were adopted, however, attention had to shift to the
question of whether those standards were being complied with, case by
case. On the line between compliance and non-compliance, inevitably,
there are finer distinctions and more subtle offenses than the ones
that first made headlines. In answering compliance questions, obviousness
is no longer a sufficient guide.
One
imperfect way of answering compliance questions is simply to trust the
judgment of on-site investigators. Select the individuals carefully,
send them out to specific factories or other workplaces, and then rely
on them to combine what they see, what they hear from talking to workers
and others, and their own “know it when I see it” intuition into an
informed personal conclusion as to whether standards are or are not
being met. Conceptually, this is much easier than trying to figure out
each of the pieces of information that would be needed by a third party
to reach such conclusions independently. “Trust the on-site observer’s
judgment” has therefore been a mainstay of workplace evaluation so far,
both for companies and for their critics.
Faith
in a monitor’s judgment can only go so far, however. As the number of
workplaces to be evaluated grows, from a handful at first to many hundreds
for even one multinational corporation, more and more different monitors
need to be employed. Inevitably their skills, experience, and judgment
will vary, to the point at which findings of compliance or non-compliance
may depend more on the choice of monitor than on the facts of the case.
An approach that puts a high degree of faith in the monitor’s intangible
abilities forces the debate to be about the monitor’s credibility, instead
of about the facts in the particular case. But endless debates about
credibility are what the fact-based approach to monitoring was meant
to overcome.
For
these reasons, monitoring programs have increasingly tried to specify
the information that they are gathering to support their conclusions,
using checklists of facts that monitors are required to establish during
a factory visit, questions they are required to ask of workers or managers
or others, procedures they are required to follow in selecting people
to interview, and so forth. The
use of consistent measurements and systematic procedures gives onlookers
more confidence in the integrity of monitoring programs. It also allows
a monitor’s findings to be spot-checked by someone else, when those
findings are called into question in a particular case. In other words,
the need to define units of measurement has been recognized, and
a number of attempts are underway.
Difficulties
of measurement
Using
a measurement process to determine code compliance is both complicated
and controversial. Early failures
have fed the controversy.
The
practical difficulties are several. Some code standards, and the workers’
rights in international law on which they are based, are inherently
difficult to measure against. They are framed in broad conceptual terms,
which need to be converted into measurable components before they can
lend themselves to pass/fail assessments, much less to the application
of anything like a 1-to-10 scale of compliance. Even when standards
are articulated sharply, measurement can be difficult because in many
situations, perception plays as significant a role as concrete fact.
And even when standards and measurement units are precise and concrete,
weaknesses in the monitoring process may undermine the ability to collect
reliable measurement results.
Framing
of standards. The code standards for workers’ rights now
being applied, and now requiring measurement for compliance, are grounded
in the fundamental rights articulated in the Universal Declaration of
Human Rights
[6]
in 1948, beginning with the right not to be discriminated
against (art. 2) and the right not to be held in servitude (art. 4).
The rights of workers, set out in intentionally
broad terms, include
- the right of everyone “to just and favourable conditions
of work” (art. 23(1));
- the right to “equal pay for equal work” (art. 23(2));
- the right of “[e]veryone who works to just and favourable
remuneration ensuring for himself and his family an existence worthy
of human right dignity…” (art. 23(3);
- and the right “to form and to join trade unions for
the protection of [the worker’s] interests” (art. 23(4)).
The
sections that follow, analyzing measurement progress in each specific
area of workers’ rights, offer many illustrations of this problem. The
problem tends to be most acute where there is most resistance to interpretation
of workers’ rights in line with international legal standards. Some
code-of-conduct standards, for example, call for “freedom of association” in general terms without
specifying the right to form and join trade unions,
[8]
even though the right to form and join trade
unions is explicitly at the heart of the international standard.
As
well as failing to acknowledge the explicit contents of established
international standards, resistance can also take the form of failing
to deal with key issues posed by an established standard. Again using
freedom of association as the example, the international standard is
cast primarily in terms of the right of workers to form and join trade
unions. In weighing compliance with the right to freedom of association
where unions are barred by the state, should any form of organization
other than an independent trade union be taken into account; and if
so, how? Disagreement
on approaching this critical issue, discussed at length below,
[9]
has stood in the way of articulating clear code standards
for freedom of association – much less translating them into measurable
terms -- for a majority of the workplaces that codes are intended to
cover.
Perception.
In some areas of workers rights, such as Abuse
and Harassment, perceptions are especially important and in
practice are central to defining a violation of code standards. For example, if a gesture or word directed toward
a female worker is perceived by her as having a sexual connotation that
creates shame or fear, that perception is a key element in determining
if sexual harassment has occurred. Context, cultural and otherwise,
also needs to be taken into account—for example, a gesture or word that
would be innocuous in one context could constitute harassment in another
[10]
--and context varies widely from place to place. Following
the Health and Safety model and
generating a set of fixed criteria for sexual harassment (e.g., gesture
A in country Z constitutes harassment under conditions 1, 2, and 3)
could offer some guidance but would be very unlikely to capture the
reality of sexual harassment violations.
Instead, measurement needs to address the workers’ subjective experience directly. Measuring
what workers perceive can be done, but making those measurements of
subjective information reliable takes special techniques, both in data
collection and in correlation with information from other sources, which
are time- and resource-intensive and which most monitoring schemes have
not systematically developed or incorporated.
[11]
Measuring
perception is also essential whenever worker intimidation and fear are
involved (see Monitoring), as they very
frequently are, no matter which area of workers rights is being addressed.
Freedom of Association is a particularly
sensitive area, since violations of free association rights and intimidation
of workers so often go hand in hand.
Weaknesses
in monitoring process. The collection of measurement results
by a monitor can also be unreliable for a number of other reasons, discussed
in Elements of reliability.
Experience is full of examples of monitors being lied to, mistrusted,
misled on purpose or by accident, or simply misunderstood. Monitors
can also miss key facts due to their own lack of training, experience,
diligence, motivation, access, or time. Figuring out which units of
measurement are more and less reliable when applied in the monitoring
process, and for which purposes, is no easy task. For example, when
monitoring for potential violations of the right to freedom of association,
a key source of information is most often unavailable. Union organizers
and representatives tend to be the most sensitive detectors of violations;
but where unions are barred and access by organizers to the unorganized
labor force is denied, there will seldom be any such witnesses to consult.
Since this is the most common situation in the global workplace, it
makes it especially difficult to monitor successfully in the area of
freedom of association, even apart from definitional uncertainties.
Finally, measurement is also made
more difficult by skepticism about the monitoring process in general,
which in a few conspicuous cases has had its weaknesses dramatically
demonstrated.
[12]
Some monitoring skeptics fear that measurement creates an
illusion of rigor in a process that is inherently flawed. Others, who
support monitoring, believe that units of measurement cannot capture
the important subtleties in violations of workers rights, and that pursuing
questions of independently reliable measurements will distract attention
from the importance of choosing the right monitor.
What
analysis can show
Taking an overview of current measurement
practice will not make the difficulties of measurement in this field
disappear. It can, however, point to the most effective ways to reduce
those difficulties that have been tried. It can also suggest how much
farther those successful techniques could be taken. Indirect indicators
can sometimes be effective where direct measurement is not, for example.
Seeing whether the right incentives for reliable measurement are in
place can often be as important as the particular measurements themselves.
Dividing the compliance question
into its component parts and looking at each systematically can make
it easier to identify measurement problems and potential solutions.
Many of measurement’s difficulties are specific to only one topic or
only one kind of measurement unit. Approaching problems topic by topic
avoids the fallacy that sweeping, across-the-board solutions are the
only ones that will work. Specific solutions targeted to specific problems can go far. For
monitoring skeptics, the step-by-step approach can also help focus attention
on where the most serious residual problems may lie and what kinds of
solutions would be likeliest to help. Having a clear sense of the different
components of reliability can also help in recognizing whether a potential
solution is addressing the real nature of the problem.
We suggest possible improvements
in each topic area, but not with any sense of having identified the
optimal or ultimate set of measurements in any case. Our purpose instead
is to illustrate the possibilities, drawing heavily on current experience
and offering relatively simple extrapolations from it. If this analysis
stimulates others to go farther and improve on our findings and suggestions,
moving toward wider understanding and agreement on measurement’s potential
and the nuts and bolts of its implementation, our goal will have been
achieved.
