Occupational safety and health inspection scores predict rates of workers' lost-time injuries
Introduction
One of the fundamental principles of personnel management is that monitoring and appraisal of performance are essential elements in ensuring that workers' activities are consistent with organizational goals. Most textbooks on management (Koontz and O'Donnell, 1972) personnel (Cascio, 1991) and industrial/organizational psychology (Muchinsky, 1993) assert that managers should explain to subordinates what is expected of them, work with them to set performance goals, provide guidance regarding methods by which the goals can be achieved, systematically monitor performance, and conduct periodic evaluations rewarding good performance, correcting errors, and punishing undesired behavior. The validity of these recommendations has been established in the course of extensive empirical research conducted by a wide variety of investigators in recent decades.
It seems logical that procedures found effective at the individual level would also be effective at the organizational level and would be just as appropriate for achieving safety goals as production goals. Thus, if they wish to gain compliance, units responsible for enforcing safety and health regulations should disseminate information about safety standards, provide guidance regarding the means by which the standards can be satisfied, and conduct regular inspections linked to enforcement actions. However, despite the apparent similarity between the appraisal of individual production performance and assessment of organizational safety performance, the value of workplace safety inspections remains a debated policy issue (Moore, 1993).
On the one hand, some studies (Mendeloff, 1976; New York Department of Labor, 1966; Wisconsin Department of Industry, 1971) have contended that safety inspections could prevent about one quarter of industrial accidents. Other studies, in contrast, have questioned the effectiveness of Occupational Safety and Health Administration (OSHA) inspections in reducing injuries on the job (Di Pietro, 1976; Mendeloff, 1979; Smith, 1976; Viscusi, 1979). Viscusi's (1979) study is of interest because it used nationally aggregated data to assess the joint effects on injury rates of the probability of an inspection and likely penalty for noncompliance. His analysis, which found no significant effects for inspection, did detect a significant reduction over time in injury rates which was uncorrelated with the occurrence of in-plant inspections. Other studies have found some evidence of the effectiveness of safety inspections. Smith (1979)used a longitudinal design which controlled for the effects of the industry to which a plant belonged, the plant's injury rate in the previous year, the size of the plant (number of employees), and the increase in employment during the inspection year. He found evidence of a significant effect of inspections on injury rates during the first of two study years, and moreover, reported that the reduction in injury rates was largest for the smallest plants. In his explanation of the finding that inspections had a significant effect in only the first of the two years, he suggested that there was likely to have been a decrease in the level of hazard (due to OSHA's policy of targeting the most hazardous plants first) and a dilution in the effectiveness of inspections (due to an influx of new inspectors from the first year to the second).
Robertson and Keeve (1983)noted that any research on the effects of inspections in reducing occupational injuries should also control for the effects of other variables (such as increases in workers' compensation payment rates) which could produce increases in injury reports, and thus mask the effect of inspections in reducing actual injuries. These investigators examined two data sets, the first of which consisted of longitudinal data from three metalworking plants. By controlling for increases in workers' compensation payment rates, they found that OSHA inspections had an effect on objectively verifiable injuries (e.g. lacerations and fractures), but not on `subjective' injuries (such as back pain and strain). An examination of aggregated data (by state and industry) also showed a significant effect of OSHA inspections when workers' compensation payment rates were statistically controlled.
Although these studies seem to establish that inspections do reduce occupational injuries, they have not identified mediating mechanisms by which this effect occurs. One explanation (Smith, 1979) focuses on the technical expertise inspections provide to those lacking specialized safety knowledge, while another emphasizes the motivating effect of a threatened fine for noncompliance (Viscusi, 1979). These developmental and motivational functions of inspections are quite consistent with the multiple functions served by individual performance appraisals (Schneider and Schmitt, 1986). That is, inspections can explain to managers what is expected of them, set goals for their performance, provide guidance regarding the methods by which the goals can be achieved, systematically monitor performance, and provide periodic evaluations which reward good performance, correct errors, and punish noncompliance.
The studies reviewed here suggest that inspections can lead to reductions in injury rates even when a comprehensive occupational safety and health (OSH) program has been established and access has been provided to safety expertise. In addition, they imply that there could be a continuing effect of inspections, even though such an effect was not found by others (Smith, 1979). However, the results of these studies raise a number of additional questions. For instance, is there a quantitative effect of inspection score over and above the qualitative effect of simply having an inspection? That is, among facilities which have been inspected, do those receiving higher scores on their safety inspections experience fewer injuries? And, do various performance indicators assessed during those inspections differ in their ability to predict those facilities' injury rates?
These questions, together with the results of previous research, lead to the following hypotheses. First, facility characteristics presumed to indicate the hazardousness of worker activities are expected to be positively correlated with employee injury rates. Second, inspection scores are hypothesized to be inversely correlated with injury rates. Finally, the inverse relation between inspection scores and injury rates is expected to be maintained when statistically controlling for differences in facility characteristics.
Section snippets
Method
Data on United States Navy shore facilities were classified into one of 11 types: shipyards, weapons facilities, aircraft repair depots, public works centers, naval air stations, hospitals, research and development (R and D) centers, bases, support facilities, training facilities and miscellaneous (mostly administrative) facilities. These facilities are principally staffed by civilian employees who are exposed to the full range of occupational hazards addressed by OSHA training regulations (
Results
The OWCP data tapes yielded a total of 3 843 accepted lost-time injuries, with 481 of these being severe lost-time injuries. The total work force for these 85 facilities was 185 688. Table 1 shows descriptive data, by facility type, as well as the ANOVA results for the comparisons among the facility types. All dependent variables were entered into separate ANOVAs. Concerns about non-normal distributions of dependent variables measuring rates and proportions (cf. Winer, 1962) led me to perform
Discussion
The analyses presented here strongly confirm the first hypothesis, i.e. that facility characteristics would have statistically significant positive correlations with employee injury rates. Moreover, the analyses also provide some support for the second and third hypotheses. Specifically, severe lost-time injury rates were inversely related to one type of inspection score (work place inspections), and this inverse relation was maintained even when statistical controls were introduced to partial
Acknowledgements
This article describes work supported by the Chief of Naval Operations, Environmental Protection, Safety, and Occupational Health Division (N-45), and by the Navy Medical Research and Development Command under work unit number 63706N-M0096.004-6202, Department of the Navy. This work was conducted while the author was in residence at the Naval Health Research Center. The views expressed in this paper are those of the author and do not reflect the official policy or position of the Navy,
References (32)
- et al.
The effectiveness of motor carrier safety audits
Accident Analysis and Prevention
(1992) - et al.
The effect of firm characteristics on truck accidents
Accident Analysis and Prevention
(1994) - et al.
The duration of speed reductions attributable to radar detectors
Accident Analysis and Prevention
(1993) - Bureau of Labor Statistics (1988) Occupational safety and health statistics (BLS Handbook of Methods, Bulletin 2285)....
- Bureau of Labor Statistics (1992) Occupational injuries and illnesses in the United States by industry, 1990 (Bulletin...
- Cascio, W.F. (1991) Applied Psychology in Personnel Management. Prentice-Hall, Englewood Cliffs,...
- Chief of Naval Operations (1991a) OPNAVINST 5100.23B: Navy Occupational Safety and Health (NAVOSH) Program Manual,...
- Chief of Naval Operations (1991b) OPNAVINST 5100.23C: Navy Occupational Safety and Health (NAVOSH) Program Manual,...
- Di Pietro, A. (1976) An Analysis of the OSHA Inspection Program in Manufacturing Industries, 1972–1973. U.S. Department...
- Doyle, E.J., Shepherd, S.L. and LaFleur, B.J. (1993) Costs for Department of the Navy civilians due to the federal...
Cited by (12)
Effects of disclosing inspection scores of health facilities
2022, Socio-Economic Planning SciencesCitation Excerpt :In addition to the impossibility of implementing diverse inspection strategies to test effectiveness in real life, the measurement of macro-level results of the activities performed by professional regulatory bodies is empirically challenging, due to the influence of numerous variables, especially in the area of healthcare [40]. On the other hand, despite the lack of evidence on the effectiveness of inspection activities of Brazilian professional regulatory bodies, the importance of regulation and inspection services is highlighted in the literature from different areas of knowledge [17–19,22–24]. In complex systems, it is deemed impracticable to perform agent-based simulations fully based in quantitative models; thus, researchers may benefit from integrating quantitative and qualitative methods, especially in the observation of phenomena in its broad sense, or its essential aspects [41].
An adaptive neuro-fuzzy inference system (anfis) model for assessing occupational risk in the shipbuilding industry
2014, Safety ScienceCitation Excerpt :Researchers have often preferred logistic regression when analyzing data from a case-control study since they can easily quantify the results in the form of odds ratio (OR) and associated confidence intervals. Roudsari and Ghodsi (2005), Lindell (1997) adopted logistic regression, while Waller et al. (1995), Fabiano et al. (2004), Salminen (2005), Siu et al. (2004), Smith et al. (2005) used the statistical correlation and significance study with the evaluation of P-value, odds ratio, likelihood ratio test to support the results. Poisson distribution is widely employed as an analytical tool in safety analysis, in reliability engineering and is also frequently used in epidemiological medical studies.
Occupational injury and accident research: A comprehensive review
2012, Safety ScienceWork environment risk factors for injuries in wood processing
2009, Journal of Safety ResearchIndustrial and occupational ergonomics in the petrochemical process industry: A regression trees approach
2008, Accident Analysis and PreventionCitation Excerpt :Researchers have often preferred logistic regression when analyzing data from a case–control study since they can easily quantify the results in the form of odds ratio (OR) and associated confidence intervals. Lindell (1997) and Roudsari and Ghodsi (2005) adopted Logistic Regression. On the other hand, Waller et al. (1995), Siu et al. (2004), Fabiano et al. (2004), Simo (2005) and Smith et al. (2005) used the Statistical Correlation and Significance Study using evaluation of P-value, Odds Ratio, Likelihood Ratio Test to support the results.