Highlights
Methodology           

The findings set out in this report are based on our review and analysis of current measurement practice, which  the database makes readily available to the public.  Detailed findings appear in separate sections below, one for each major subject area of workers' rights, with additional sections for cross-cutting measurement issues such as the integrity of the monitoring process.   Each section addresses:

1. measurement issues that are unique or distinctive to that area;
2. strengths of current measurement practice in the area;
3. weaknesses;
4. a selection of measurement units reflecting current best practice;
5. suggestions for possible improvements, tailored to the needs in that subject area.

Key findings that apply to workplace measurement in general, rather than to specific subject areas, appear immediately below. Key findings for specific subject areas cannot be readily summarized, since they vary from subject to subject. 

Highlights

Searching for current best practices: 

1.      There is no single best model to be found in current use. Several monitoring organizations and companies have measurement guidance documents that cover all key areas of workers' rights in detail, showing up to hundreds of individual units of measurement apiece. ^[1] Each has its strengths in particular subject areas, but no one stands out overall. Our subject-by-subject analysis found best practices emerging from the efforts of a wide variety of practitioners, including those whose efforts focus only on one or a few subject areas, as well as those covering the full spectrum.

2.      Difficulties in measurement vary widely from subject to subject.  A subject like Wages, where the basic unit of measurement is objective and clear (an amount of money), presents obstacles to measurement that are very different from a subject like Abuse and Harassment; see detailed discussion in the sections below. Some subject areas (e.g., Freedom of Association) are also inherently more difficult to measure than others (e.g.,  Working Hours); again, details are discussed below.  These differences mean that overall solutions to measurement problems are less likely to be effective than piecemeal ones that are tailored to specific topics and questions.

3.      The quality of current measurement practice also varies widely from subject to subject, but not because of differences in measurability.  Current practice does not necessarily show better-developed units of measurement for easier-to-measure subjects, or vice-versa. As an example, monitoring currently does a much better job of measuring workers' ability to pursue complaints about Abuse and Harassment than complaints about Wages, even though Wages as a subject matter is inherently easier to measure.  In another example, the tracking of measurement results over time, to see how measurement results change, is a technique that is used for the subject of Non-Discrimination but not for most other subjects, even though tracking changes over time would be just as easy to do with those other subjects. Differences in the quality of measurement practice seem to have more to do with how directly units of measurement can be borrowed from long-standing legal and regulatory regimes (for example, Title IX of the U.S. Civil Rights Act of 1964, which addresses discrimination) than with any differences in inherent measurability from one subject to the next.

4.       There is high potential for cross-learning from one subject to another.  The disparities in measurement practice from subject to subject, noted above, offer many opportunities for easy improvements,  simply by borrowing successful practices from one subject area and extending or adapting them to others. Contrary to what might be assumed, inherent difficulties of measurement are often not the obstacle.  Many of the potential improvements for specific subject areas that are suggested in the analysis below are straightforwardly derived from what already exists in another subject area. 

Recognizing key problems: 

5.      Units of measurement that duck key interpretive issues are a chronic weakness. Current practice commonly sidesteps issues of definition and interpretation by framing measurement units in vague or evasive terms.  A simple example is the definition of a unit of measurement for factory temperature as whether temperatures are "adequately controlled" (record 454) or "comfortable" (record 1145) rather than whether they "stay between 50 and 96 degrees F" (record 740). The latter is clear and precise. Not all vagueness can be cured by substituting numbers for adjectives, however. A more complicated example, in the area of freedom of association, is defining the unit of measurement for the adequacy of factory policy. Framing the unit of measurement as whether the factory has "effective policies to ensure freedom of association" (record 1229) simply begs the question of what "effective" policies might be.  The monitor who puts down "yes" or "no" as the recorded result for that measurement unit may be using excellent  judgment,  but anyone else looking at that result has no way to tell what criteria for "effective" policies the monitor used - or whether completely different criteria might be used when the same measurement unit is taken in another factory. At least some greater degree of clarity, and some useful basis for comparison of one workplace with another, could be obtained with measurement units that cover the frequency and nature of elections, the mechanisms available to worker representatives to present demands to management, the availability of the minutes of meetings between management and worker representatives, and even whether workers' right to freedom of association is formally communicated to workers on a regular basis (although none of these is foolproof, either by itself or in combination).  Many measurement units in current use are cast in evasive terms, and often where definitions are most problematic or controversial. The desire to duck hard definitional questions is understandable, but the use of a measurement unit cast in evasive form virtually guarantees that the results of that measurement will be unreliable. ^[2]

6.      Qualitative units of measurement  are indispensable, especially for information gathered from workers themselves.   Quantitative measurement units can seem more reliable than qualitative ones, but several areas of workers' rights have key elements that need to be measured qualitatively. (The database separates qualitative and quantitative units of measurement for analytic purposes.)  Workers themselves are very often the most important source of information about violations of workers' rights, and on issues involving intimidation or voluntariness, ^[3] as well as issues like management responsiveness to grievances and retaliation for union organizing, the units in which their information is measured are likely be qualitative. Qualitative data can be gathered and managed with relatively rigorous methods, but current practice in this field tends to be haphazard and non-rigorous. ^[4]

7.      Worker inhibition needs to be tested for, and findings need to be adjusted or discounted for inhibition factors that are found, on a consistent basis.    Much of current practice relies on worker interviews to reveal problems. But relatively little attention is paid to measuring the likelihood that workers will be candid when interviewed.  As discussed below, ^[5] intimidation is the norm and may even be inherent in ultra-low-wage employment. Gauging how severe the risk of intimidation and other inhibiting factors is likely to be in any particular situation, and compensating accordingly, is a necessary part of using worker input in measurement results. Key indicators of intimidation and similar factors have not been consistently identified or defined in terms of measurement units, and measurement results related to intimidation have not been used to discount findings based on worker interviews when high intimidation potential is shown.

8.      Measurement units often have asymmetrical reliability.  Particularly for measurement units cast in yes/no form, as most are, ^[6] a "yes" answer may be much more informative than a "no" answer, or vice versa. In other words, the reliability of the unit of measurement depends on which result it produces. For example, if workers receive written pay slips with calculations and deductions clearly shown, and monitors are able to compare those with factory wage records, those two facts together make a reliable indicator that factory records are honest. But an absence of workers' pay slips does not necessarily indicate that factory records are false. At most, a measurement result that shows the absence of pay slips would raise a suspicion that would need to be pursued further. In evaluating the reliability of a particular unit of measurement, either by itself or in combination with others, it is important to consider reliability separately for each of the different measurement results that might be obtained. 

Focusing on highest potentials for improvement: 

9.      Measurement of incentives and deterrents is critical.  Even if a workplace offense like sexual harassment is hard to measure directly, it is possible to look for incentives in the workplace that increase or decrease its chances of occurring. Incentives mean focusing on self-interest:  is it in management's self-interest to prevent incidents of harassment? Is it in workers' self-interest to report them? For example, if a specific named manager has explicit responsibility for preventing harassment and meeting anti-harassment code standards, and that manager faces clear, serious penalties for failure and clear financial rewards for success, it makes compliance likelier than a situation in which management accountability is diffuse. Child labor is less likely when the employer must both compensate the child's family for loss of income, and hire an adult replacement, if caught; i.e., when violating the code standard is more expensive than complying. (This assumes, of course,  that there is a significant chance of getting caught).  Incentives are a recurring theme in the detailed discussions below. Measurement units that can show the level of incentive for and against particular kinds of worker abuses, or even measurement units that simply show whether specific types of incentives and deterrents are present, can provide a strong indicator of the likelihood that abuses are occurring. Equally, measuring a factory's incentive to communicate honestly with monitors can provide one helpful indicator of the likelihood that monitors are collecting accurate measurements in their interviews at that factory.

10.    There are several useful gauges for the overall integrity of the measurement process in any particular workplace.   They involve measuring the degree to which each of the following elements is present. Each is discussed in more detail in its own separate section:

Freedom of association - a right in itself, but also a strong indicator of the likelihood that workers are able to voice complaints internally to management and report them externally to monitors;
Education  about rights  -- an indicator of how likely both workers and managers will be to recognize violations of code standards when they occur (the first step toward being able to report them accurately to monitors);
Grievance procedures - how usable they are, and how effectively they are used in practice, as an important indicator of management's commitment to respecting workers' rights overall;
Transparency - how widely visible the results of the measurement process will be, as a general disincentive to poor measurement-taking;
Feedback and Revision - the likelihood of learning from experience, as a sign of whether the measurement process will be self-correcting over time.

To the extent each is present, each of these elements reinforces the others in supporting measurement reliability. See also the discussion of Monitoring below.

 Methodology 

This analysis is based on the Lawyers Committee's evaluation of the relevant set of measurement units that represent current practice in each subject area.  Using the database, anyone can easily locate the same sets of measurement units, second-guess our analysis and conclusions, and provide their own. We warmly invite readers to do so.  

Researchers for the Lawyers Committee assembled the units of measurement shown in the database and classified them for entry. Individual measurement units were classified not only by topic, ^[7] subtopic, ^[8] and source, ^[9] but also by the nature and form of the measurement unit, ^[10] and by whether there are any results for it (i.e., actual recorded data) that the source has reported to the public. ^[11] The database itself includes a fuller explanation. 

The database can be searched for any one classified category of measurement units, or for any combination of categories (for example, only the units of measurement for discrimination on the basis of religion that are qualitative in form and are based on surveys of worker opinion; or, only the units of measurement for child labor in quantitative form for which actual data results have been publicly reported by companies).  

The database can also be searched with a word-search feature (for example, all units of measurement that include the word "grievance," or "fire exit," or both  "worker" and "education" in the same measurement unit). The word-search feature can be limited to a single topic or subtopic, or it can be used independent of topic to find every measurement unit that includes the chosen word or words,  no matter how those units are classified. Searches can be narrowed by adding more words to a search; for example, "worker" yields 729 search results; "worker rep" yields 42; and "worker rep procedure" yields only four. ^[12]