Introduction
Measurability issues
Current strengths
Current weaknesses
Best current practices (selected)
Possible improvements suggested by analysis

Introduction

The International Labor Organization does not address worker health and safety in either its fundamental ^[1] or its priority ^[2] conventions for international labor standards, although it has drafted one convention on workplace health and safety generally ^[3] and has nearly 60 specific standards, "the highest number in any single field." ^[4] However, nearly every model code of conduct and individual company code includes strong principles of health and safety protection, for example a requirement that employers"provide a safe and healthy working environment to prevent accidents and injury to health arising out of, linked with, or occurring in the course of work or as a result of the operation of employer facilities." ^[5] Such provisions reflect long-standing law in the United States and other developed countries. 

The area of health and safety has by far the best-developed units of measurement of any area of code standards,even though the subject includes many elements of different kinds and involves a good deal of technical complexity.Issues can be as simple as locked fire exits and as complex as chronic exposures to multiple chemicals with unknown long-term health effects. The stakes can also be extremely high: in one catastrophic case, a garment factory in Bangladesh caught fire in a building where over 900 workers were on duty.Locked gates blocked workers from escaping from the building until after fire-fighters managed to break them open, and more than 50 people died, including eight girls under the age of 14. ^[6] One study estimates that there were more than 100 factory fires in Bangladesh in the decade 1990-1999, resulting in 246 workers killed and over 5,000 workers injured. ^[7] Locked fire exits illustrate a typical measurement difficulty, since they can be unlocked when monitors visit and then relocked after monitors leave.

Measurability issues 

Because it is so relatively well developed, current measurement practice in this area offers a number of useful models for measurement units in other areas, especially in terms of educating workers about their rights, worker input into the measurement process, and systematic tracking of performance over time. 

The advanced development of measurement units in the health and safety area might be explained by the fact that many health and safety issues involve physical dimensions (e.g., temperature and noise levels) and physical objects (e.g., unlocked exits, sewing machine needle guards). Results of violations also tend to manifest themselves in physical terms: death, injury and illness. In other words, health and safety issues can be thought of as inherently more measurable than, say, issues of sexual harassment. 

However, the relatively greater success of developing measurement units for health and safety compliance seems to have as much to do with experience as with inherent measurability. Health and safety regulation has a long history in the Occupational Safety and Health Administration of the U.S. Department of Labor (OSHA), for example, and in the course of it many potentially ambiguous criteria (e.g., fire safety) have been defined in terms of standard units of measurement for government inspectors to use (e.g., location of fire extinguishers within fixed distance from every employee ^[8] ). This subject area is simply farther along the measurement learning curve than most others. 

The sheer number of measurement units in current use (one-third of the total number of all measurement units in all areas combined) suggests how well developed the measurement process is in this area. It also reflects the many dimensions of health and safety issues.This area shows by far the largest number of measurement units with actual results reported to the public, both from all sources and specifically from companies about their own supply chains. Public familiarity with health and safety measurements seems to breed company comfort in reporting such measurement results to the public, even when the results show less than perfect performance. 

Current strengths

Because current measurement practice is so extensive and so relatively successful, no brief discussion can cover all of its strengths. A quick look at the range and depth of measurement units within any of the identified subtopics ^[9] will provide many examples.  

This portion of the analysis concentrates on measurement issues that are common to a number of areas of workers' rights but that are most advanced in health and safety, with model practices that other areas could readily imitate. 

Worker education.How well workers understand what their rights are, and how to protect them, is critical in every area of workers' rights. Units of measurement cover both effort - i.e., the amount of education, training, and assistance workers and managers receive - and results - i.e., what they actually learn.  

Useful measurement units for education effort include observations about the frequency, ^[10] timing, ^[11] and length ^[12] of training sessions, as well as the number ^[13] and percentage ^[14] of relevant workers trained. Measurement units also address the specific subject matter covered by training. ^[15]  

Results of education are harder to measure than the pieces of the educational process, but current practice tries to gauge how successful training in health and safety has been by checking on whether both workers ^[16] and managers ^[17] can display their knowledge. It also tries to correlate training with tangible outcomes, in the form of reduced accident rates. ^[18]  

Worker input.In the health and safety area, workers are polled not only to identify problems, but also to get their subjective impressions ^[19] and record their satisfaction levels. ^[20] Most important, and unique to the area of health and safety, is checking on how regularly factory management is soliciting workers' views. ^[21] This is potentially a strong indicator of the extent to which workers' experience and needs are being incorporated into health and safety procedures and policies at the individual factory level. 

Statistical analysis. More than in any other subject area, key statistics (such as death ^[22] and lost-time accident rates ^[23] ) are tracked over time, ^[24] with changes explicitly measured in both absolute ^[25] and percentage ^[26] terms, over periods as long as 10 years. ^[27] Rates are compared to industry norms, ^[28] and accidents are statistically analyzed for severity ^[29] and potential relationship to specific causal factors ^[30] as well as frequency. 

Corrective practices. Whether or not accident causation is analyzed at the factory level ^[31] is itself a significant indicator.So is the presence or absence of specific follow-up actions taken in response to accidents. ^[32]  

Current weaknesses 

"Snapshot" distortion. Although current measurement practice covers quantifiable physical conditions in the area of health and safety, it does not differentiate well between conditions that are relatively permanent and those that can easily be altered when a monitor visits. Reliability of measurement results is obviously greater for items of physical infrastructure like handrails ^[33] or marked exits ^[34] than for conditions like whether exit doors are unlocked ^[35] or what the air temperature in the factory is, ^[36] even though the conditions themselves are equally tangible and measurable. For example, of more than a dozen different units of measurement about the location and number of fire extinguishers, ^[37] none addresses the possibility that fire extinguishers appear only at inspection time. ^[38]

This "snapshot" weakness is not limited to the health and safety area. But the larger number of concrete physical measurements in health and safety can create a false sense of accuracy, unless physical dimensions that are more or less permanent are treated separately from those that can easily change when monitors leave.  

Dependence on high-quality data. The statistical analysis described in Strengths depends on there being relatively accurate and complete data on deaths, injuries and illnesses in the particular workplace. But the paper trail may be as unreliable in this area as it is in others (for example, Wages or Working Hours).This is not a weakness for health and safety measurement overall, since there are many non-statistical measurement units in current practice as well, but it can undermine the statistical measurement results and analysis that are health and safety's greatest comparative strength. 

Little convergence on objective units of measurement.Although current practice in health and safety includes a number of clearly quantified units of measurement for physical conditions, such as factory temperature (e.g., between 50 and 96 degrees F.), vague and question-begging ^[39] measurement units for the same conditions are also commonplace (e.g., air temperature in the workplace "adequately controlled" ^[40] or "comfortable" ^[41] ). 

Best current practices (selected) 

Because current practice is so extensive and so relatively successful, a short list of selected best practices would be misleading. See Strengths above. 

In general, best current practices in health and safety are those that take advantage of well-developed statistical units of measurement for health and safety outcomes (e.g., lost-time incident rates; ^[42] number of accidents requiring sick leave per total number of work hours ^[43] ) and apply consistent forms of analysis to the results. Tracking changes in accident rates over time in the same workplace is the most basic example. Having standardized statistics makes it possible to track changes in the severity ^[44] as well as the incidence of accidents. 

Measuring current performance in a defined workplace (such as a factory) against the same workplace's previous performance is a valuable benchmarking exercise, as is measuring it against external standards. Current best practice includes comparison to industry-wide statistics. ^[45] It also includes comparison of a company's health and safety performance in its entire worldwide operations to performance in the company's U.S. facilities only. ^[46]  

Statistical analysis depends on the availability, accuracy, and completeness of accident reports from the workplace in question. Best practices therefore include careful scrutiny of whether such reports are available ^[47] and can be cross-checked for accuracy. ^[48]  

Possible improvements suggested by analysis 

1. More focus on causation. There is one current unit of measurement aimed at identifying a specific cause of accidents, ^[49] but most of the attention paid to the cause ofaccidentsfocuses only on whether accident analysis procedures ^[50] are in place, without probing further. Much more pointed analysis of causation could be pursued, using data that current practice already produces .For example, if injury documentation shows what part of the body was injured, those facts could be combined with data on the presence and use of different types of personal protective equipment (e.g., eye protection, ^[51] hand protection, ^[52] foot protection, ^[53] etc.), as well as with data on training in the use of such equipment. ^[54] Correlations could help to identify trouble spots and potential cures.
   
   
2. More cross-workplace comparisons. The existence of standardized statistical information in the health and safety area, in at least some cases, makes it possible to do more than simply compare a workplace's overall accident rates to industry norms ^[55] or country norms. Comparisons can also be made on the basis of specific contributing factors, both negative (e.g., too little sleep ^[56] ) and positive (e.g., training; protective equipment).
   
   
3. Identifyincentives and deterrents. Current practice touches on measurement units for a few incentive-related factors. Whether workers must pay for personal protective equipment ^[57] is a measurement unit that helps to show whether workers face a deterrent to having such equipment. Whether the factory pays hospital costs for workers injured on the job ^[58] is one factor in showing whether workers are deterred from getting hospital treatment when injured.But measurement for both negative and positive incentives could be much more systematic. Do managers personally benefit if health and safety improves? Do they lose if it declines? How, and to what extent? Are workers' own incentives consistent with improved health and safety, as opposed to a financial incentive to run greater health and safety risk? ^[59] The incentive framework in a factory is a strong indicator not only of management commitment, but also of the likelihood of success.
   
   
4. Measure worker satisfaction over time. Current practice looks to worker satisfaction, ^[60] but not to changes in the degree of worker satisfaction over time that could signal a pattern of improvement or decay.
   
   
5. Measure effectiveness of compliance policy over time. As in the area of Non-Discrimination, the availability of statistical results such as accident rates, and the ability to track those result over time, makes it possible to study the correlation between improvements and different factors such as management education and training; worker education and training; specific equipment; specific new policies and practices; a change in incentives; and even changes in monitoring practice. No other subject area has as strong a potential for measurable feedback on separate elements of compliance policy.The more workplaces in which such correlations are studied, the clearer the conclusions are likely to be.