An organization and employee level investigation
of the relationship between high performance work systems
and workplace safety
Anthea Zacharatos
A thesis submined to the School of Business
in conformity with the requirements for the degree of Doctor of Philosophy
Queen's University Kingston, Ontario, Canada December 2001
Copyright @ Anthea Zacharatos, 2001
Acquisitions and Bibliographie Services
Acquisitions et services bibliographiques
The author has granted a non-
exclusive licence allowing the National Library of Caoada to reproduce, loan, distn'bute or sen copies of this thesis in microform, paper or electronic fomats.
The author retains ownership of the copyright in this thesis. Neither the thesis nor substantial e-cts fiom it may be printed or otherwise reproduced without the author' s permission.
L'auteur a accordé une licence non exclusive permettant a la Bibliothèque nationale du Canada de reproduire, prêter, distribuer ou vendre des copies de cette thèse sous la forme de microfiche/film, de reproduction sur papier ou sur format électronique. L'auteur conserve la propriété du droit d'aqui protège cette thèse. Ni la thèse ni des extraits substantiels de celle-ci ne doivent être imprimés ou autrement reproduits sans son autorisation.
Abstract
Two studies were conducted investigating the relationship between high performance work systems and occupational safety. Study 1 reveaied that to the extent organizations had adopted high performance work systems comprising ten hurnan resource practices, namely, the use of selective hiring and transfomational leadership, the provision of job quality, employment security. training and contingent compensation, as well as information sharing, reduced status distinctions, self-managed tearns, and the measurement of variables critical for success, they experienced greater workplace safety measured in terms of fewer lost time injuries. As part of Study 2, a pilot study was conducted in order to develop and test a questio~airerneasuring employee perceptions of the extent to which their organizations had adopted a high performance work system.
In the main part of Study 2, this questionnaire was used to investigate the relationship between high performance work systems and occupationai safety at the employee level as well as the mechanisms hypothesized to mediate the relationship. Trust in management and safety climate were fond to mediate the positive relationship between high performance work systems and safety performance measured in terms of more positive persona1 safety orientation (i.e., safety knowledge, safety motivation, safety cornpliance, and safety initiative) and fewer safety incidents (Le., injuries requiring fint aid and near-
misses).
Acknowledgements
First and foremost, I am grateful to my supervisor, Julian Barling, for his endless support and guidance and, most importantly, for his fiendship. 1 would aiso like to th&
my cornmittee members, Dan Gallagher and John Gordon, for their constructive ideas about this research, and Kevin Kelloway without whom I never would have figured out
LISREL. 1extend thanks to Nick Tumer for his coniagious enthusiasm and endless patience, and to Gai1 Hepburn and Catherine Loughlin who both assured me that it was actually possible to complete this work. I appreciate the time taken by Kara Arnold, Kate Dupré and Catherine Connolly in helping me with my survey items, and appreciate the help of Colette Hoption and Alysha Williams with the data entry. 1 would like to thank Dean McKeown for creating the web-based s w e y s . 1 am also appreciative of the Social Sciences and Humanities Research Council of Canada, the Queen's School of Business. the Industriai Accident Prevention Association, and those organizations that agreed to
participate in this research. Last, 1 am grateful to my parents and my fiends (including those mentioned above) for encouraging me al1 those times when 1 needed it and even
when I didn't. Thank you.
Table of Contents
CHAPTER ONE: Introduction............................................................................ 1 CHAPTER TWO: The Ten High Performance Management Practices.............................19
CHAPTER THREE: The Mechanisms by Which High Performance Management Practices Impact Workplace Safety...........................................................60 CHAPTER FOUR: Further Refinements to the Conceptual Models ................................. 67 CHAPTER F M : Study One............................................................................. 81
CHAPTER SN: Study Two- Pilot Study............................................................ 104 CHAPTER SEVEN: Study Two- Main Smdy .........................................................113 CHAPTER EIGHT: Discussion.......................................................................... -134
REFERENCES............................................................................... 144 Appendix A . Items Measuring Hi@ Performance Management Practices at the Organizational Level .............................................. -168 Appendix B . Cover Letters Used in Study One....................................... 173 Appendix C. Survey Sent to Human Resource Directors............................ 176
Appendix D . Sw e y Sent to Health and Safety Managers ........................... 181 Appendix E. Onginai Pilot Study Items................................................184 Appendix F. Web-Based Pilot S w e y.................................................. 190 Appendix G. Final Items fkom Pilot Study.............................................195 iv
Table of Contents (cont'd) Appendix H. T w t in Management Scale............................................. 199 Appendix 1. Affective Cornmitment Scale........................................... 200 Appendk J . Safety Climate Scale...................................................... 201 Appendix K. Safety Cornpliance Scale................................................ 202 Appendk L . Safety Initiative Scde .................................................... 203 Appendix M . Safety Knowledge Scale................................................. 204 Appendix N . Safety Motivation Scale................................................. 205 Appendix O. Perceived Risk Scaie .....................................................206 Appendix P . Paper-and-Pencil Survey for Shidy Two .............................. 207 Appendix Q. Web-Based Survey for Study Two.................................... 215 Vita..........................................................................................
224
List of Tables Table 1:
Numbers of Articles Dealing with Employee Safety and Total Articles Published in Mainstrearn Organizational Behaviour Joumals, 1990-1999.. ......................................................... .3
Table 2:
Descriptive Statistics and Intercorrelations Between Measures of the Ten High Performance Management Practices in Study One ................................................................................
91
Table 3:
Factor Loadings of Ten High Performance Management Practices in S ~ d One. y ................................................................... 92
Table 4:
Descriptive Statistics and hterconelations Between Measures in Smdy One.. ..................................................................... 93
Table 5:
Regression Results Predicting Total Lost Time Injuries.. ...............95
Table 6:
Descriptive Statistics and Lnterconeiarions Between Measures of the Ten High Performance Management Practices in the Pilot Study.......................................................................... -108
Table 7:
Factor Loadings for Scales in Pilot Study.. .............................. .110
Table 8:
Descriptive Statistics and Intercorrelations Between Measures in Study Two.. ............................................................... .120
Table 9:
Standardized Parameter Estimates for the Measurement Model.. .....123
List of Figures Figure 1:
Proposed model: Effects of a high performance work system on occupational safety at the organization level ..................... 16
Figure 2:
Proposed model: Effects of a high performance work system on occupational safety at the employee level ........................ 17
Figure 3:
Effects of a high pedormance work system on occupationai safety at the organization level ................................. 83
Figure 4:
Fully mediated mode1......................................................... -124
Figure 5:
Standardized Parameter Estimates for the Hypothesized Mode1......... 126
Chapter One Lntroduction
Most worken in developed counrries generally assume the organization for which they work will take al1 necessary measures to ensure that they r e m home safely at the end of the worliday. Despite this assumption, work-related injuries and deaths continue to occur at an a l m i n g rate. In 1999 alone, there were 833 work-reiated fatalities in Canada
in the same year, 379.395 Canadian worken suffered injuries serious enough to be compensated for either wages lost due to time off work, or for a permanent disability (Association of Workers' Compensation Boards of Canada, 2000). In the United States, there were 6026 fatal work injuries and approximately 3.8 million nonfatal injuries in 1998. In ternis of cost. these injuries resulted in an estimated 80 million production days
lost for that year alone and almost 60 million days in future years (Bureau of Labor Statistics, 2000; United States Census Bureau, 2000). These data illustrate the enonnous
cost of occupational injuries and fatalities for organizations in tems of production, but more importantly, in tems of lives altered and loa by these work-related eventsi. Fortunately. the issue of occupational safety is gaining increased attention in both the media and the workplace (Adams-Roy, Knap & Barling, 1995). This is most likely the result of large-scale industrial accidents such as Chernobyl, Three Mile Island and
1
It is wonh noting at this stage that these &ta refer o d y to occupationai injuries and not to occupational iiinesses, which rf includcd m the coimts wodd significantly augmait these stathics. Occupationai dhesses include nich ailments as tcndonitis, carpal m e 1 syndrome and contact dmnatitis and much iike occupationai injuries have si@~cant effctts on both woricplaccs and the lives of workm. Notwithstanding the ùnponancc of occupational illntss, the purpose of the cumnt rcscarch is to focus on workplace safety and management practiccs that are associatcd with fewcr workpiacc mjuries, exampIcs of which are cuts, burns, tracnrrcs and amputations as weii as fewer occupational fatalities.
Bhopal. These highly publicized events have increased the general public's awareness of the threift these industries pose to employee safety. It is not surprising then that workers consider saiary, benefits and daycare to be less of a concem than their work-related safety (Waidman, de la Peiia, Springen, Howard & Smith, 1989). Dwyer (1991), however, goes so far as to argue that the current focus on work-related accidents2,especially industrial ones, is more a huiction of the threat they pose for the general population than their threat to worker safety. This may in part help explain why, despite the alaming number of injuries which continue to occur in the workplace and increasing interest in the issue, occupational safety remains an underdeveloped area in the management sciences (McLain, 1995).
in 1982, Campbell, Daft and Huiin suggested that less than one percent of management research focused on occupational safety. A more cuiient inspection of rnainstream organizational behaviour joumals published between 1990 and 1999 demonstrates that this situation has not changed in the wo decades since the initial observation was made. Articles that dealt with any aspect of employee safety were counted as a proportion of al1 articles published in joumals such as the Academy of Management Journal, Academy of Management Review, Journal of Applied Psychology and Journal of Organizational Behavior. Of the 2,364 articles examineci, only 11 focused on employee safety, or one half of one percent of articles (see Table 1). This is not to
It is wonb noMg that the tcrrn "accident"refers to an evcnt that is not foreseen or expecrcd However, an andysis of safety-rcfated incidents would fmd tbat the causes of these incidents arc in the vast majority of cases predictable and even pccventabIe. Accidents simply do not "happen" but rather are the foresceable result of the actions of empIoyees and managm and, thereforc, arc events over which these partics have control. Therefore, with respect to safety, 1refiain fiom using the tenn accident whcneva possible. Whilt 1 have used the tenn ^accident" in many instances in the text, 1 have only &ne so when citing the works of otticr authors in an attempt to rcmaiu consistent with the ori@ authors' intentions,
Table 1 Nwnbers of Articles Dealing with Emdovee Safetv and Total Articles Published in Mainstream ûrganizational Behaviour Joumals. 1990-I 999
Acadtrny of Management Journal
Journal of Orgtnintional Bebavior
Journal of Occupational and Organizrtioaal Psychology
1990 199 1 1992 1993 1994 1995 1996 1997 1998 1999
TOTAL
Safety
Total
Total
Safcty
Total
O
39 35 38 66 71
29 28 28 21
1
O O
32 39 35 39 36
O O O O 1
72
25 31 34 27
O O 1
44
38
O
35
2
538
64 54
O O O O
41
35
JI
O 1 1
41 42 77
302
3
417
Say that safety research is not being published in other, more specializedjournals such as
Safety Science or the Journal of Occupational Health Psychology, however this analysis demonstrates that, despite its relevance. safety research continues to lack a place in mainstrearn organizational behavior publications.
Control-Oriented Management Practices Traditionally, the most frequently adopted method for managing occupational safety has been to take a control-oriented approach to human resources marling & Hutchinson. 2000). The control orientation is based on the assumption that workers are only motivated to exen as much effort as is necessary for them to complete a certain task. -4s such, it is the responsibiiity of the managerial hierarchy to use authority to control
employee behaviour and to coerce efficient performance from worken (Walton, 1985).
In terms of occupational safety, a control-oriented approach to management emphasizes the use of rules to enforce behavioun and the use of punishment for increasing employee compliance with the d e s . The implicit intent of this orientation is compliance with govemment reguiations and/or provisions in a collective agreement. Techniques such as goal-setting, feedback and rewards are used to control individual behaviours and to increase occupational safety (Barling & Hutchinson, 2000).
The popularity of the control-oriented approach to safety management is reflected
in the fact that much of the existing occupational safety literanue is focused on examining the role that individual ciifferences play in the causation of workpiace injuries. This attention is most Likely the result of earlier work such as that cited by Heinrich (1959), which States that an estimateci 88% of the 75,000 industrial accidents analyzed as
part of a study of accident causation were "caused prùnarily by the unsafe acts of
persons" (p. 11). For instance, the concept of 'accident proneness' was a major area of study for over 50 yean beginning in the 1920's (Hoyos, 1992; Kay, 1971;Sheehy & Chapman, 1987). Accident proneness assumes that some individuais are inherently more likely to be involved in an accident than are othen and, therefore, the majority of accidents are caused by a srnall proponion of worken. The appeal of this concept, which in part explains its enormous popularity, was its potential use as a means of identi-g
those
few employees who were thought to be responsible for a large proportion of accidents (Kay, 1971). Despite extensive work in the axa, very little empirical support has ever been found for the concept of accident proneness, nor bas a satisfactory technique for its application ever been developed (Dwyer, 199 1;Kay, 1971). Penonality characteristics comprise another area of occupational safety research that has received a great deal of attention. Numemus studies have examined the infiuence
of penonality characteristics on individuals' abilities to deal with stress and, in turn, the role that stress plays in workplace injuries (Sheehy & Chapman, 1987). Other penonality factors that have been examined in the safety literahue include neuroticim, aggressiveness, intelligence and uimvenion/extroversion, to name but a few (Sheehy & Chapman, 1984, 1987; Simonds & Grimaldi, 1963; Sutherland & Cooper, 1991).
Much of the research in this area is problematic, however, shce the data collected are mainly retrospective in nature, and causai inférences are hazardous at best (Sheehy & Chapman, 1987). It is quite plausible that being injureci at work couid induce negative mood or affect, for example (Iverson & Envin, 1997; Sutherland & Cooper, 1991).
Similarly, reported stress, aggression or neuroticism following a safety-related incident might not have been causative. hstead, especially in those incidents in which signifiant injuries or even fatalities occur, it is possible that these attributes could be a function of the incident itself or charactenstics of the workplace. In support of the latter, Frone (1998) found in a snidy of injury causation in adolescents that while penonality variables
were correlated with injury rates, the relationship no longer held once charactenstics of the work environment were accounted for. This suggests that penonality variables and
injuries may be related but only indirectly by way of other facton. Overail, liale conclusive evidence has been found for the relationship between penonality factors and accident causation (Hoyos. 1992; Sheehy & Chaprnan, 1984, 1987). Research demonstrates that both accident proneness and personality facton are not satisfactory predicton of occupational injuries. Moreover, this approach is unappealing as the implicit assumption is that those individuals who are thought to 'cause' safety-related incidents can and should be excluded from the organization with the overall aim of reducing the number of injuries. The moral and legal issues involved in excluding individuais who are considered 'accident prone' or who are of a certain personality type from specific occupations are complex and draw into question the ethicaiity of such a method (Walker, 1998).
Other individuai factors and their relationship with injury causation have also been considered in the literature. The possibility of life events contributing to injury causation has received some focus. The idea is that stresshl life events can explain why
some individuais may be more Iikely to be involved in dety-related incidents at specific penods in theu lives. This is in contrast with the concept of accident proneness, which
assumes that accident involvement is a stable trait or characteristic of a worker (Sheehy & Chapman, 1987). Much of the research in this area is also inconclusive given that,
similar to most other studies of injury causation. these studies use retrospective data. Consequently. it is very difficult to corne up with any steadfast explanations of injury causation (Hoyos, 1992). However, work conducted on the relationship between stressful life events, substance abuse ancilor fatigue, and injury causation has s h o w more promise. (Shain, 1982; Sheehy & Chapman, 1987; see for instance, Arnold et al., 1997 and Hoicom, Lehman & Simpson, 1993). The use of behaviour modification techniques and goal-setting is also consistent with the notion that employees are responsible for most occupational injuries (see for instance, Reber, Wallin, & Chhokar, 1984; Reber, Wallin, & Duhon, 1993). Much like the research focusing on the individual characteristics described above, these approaches
also foster control-oriented safety management by targeting unsafe employee behavioun (Barling & Hutchinson, 2000; Walton, 1985). The underlying idea is to identify employee behavioun that have the potential to be factors in injury causation and to modi@ them in order to increase work-reiated safety. The specific means for altering these behaviours can include knowledge of results, the use of reward and punishment or any combination of more than one technique. For instance, in a typical study of this nature conducted in a f m machinery manufacnuing plant, injury rates were reduced by a combination of goal-setting and knowledge of results. In this particular study,a specific. difficult, yet attainable goal was set. The goal was achieved if 90% of the employees perfonned theu jobs in a compietely safe manner. Employees were given individual feedback on their petfiorniance in the form
of praise, reprimand or corrective suggestions. They were also made aware of the percentage of their department's rnembers who were perfoming their jobs in a completely safe manner. The resulu of this study indicate that this approach effectively increased work-related safety (Reber & Wallin, 1984). Similar results have been found in a number of comparable studies in various settings (Chhokar & Wallin, 1984; Reber et al., 1984; Sulzer-Azaroff, Loafman, Merante, & Hlavacek, 1990). Other studies have just as effcctively set goals in tems of injuries, rather than intermediate critena such as unsafe behavioun (see, for instance, K m & Kopelrnan, 1987). Regardless of the apparent success of behaviour modification approaches for increasing safety, their focus rernains on the individual as the cause of workplace injuries, while neglecting managerial issues that may also play a role. Hence. these approaches
support the coercive 'control' mode1 of human resource management. In Waiker's (1998) words the behaviour modification approach assumes 'Viat the nature of the workplace and the work itself is less important that people's behaviour" (p. 2), and thai "we see who the most accident prone people are in our society. They are the underground miners,
construction workers and loggers. Office workers must of coune be extremely careful and safe people because they get hurt f a . less o W (Walker, 1998, p.2).
In fact, this wumption about worker responsibility holds m e for the accident proneness, personality factors and organizationai behaviour modification approaches. They have in common the belief that workers are not motivated to work safely, rather that their individual behavioun need to be controlled in order for the workplace to be safe. Not surprkingiy, unions do not support these control-oriented approaches to increasing workplace safety. Unions argue that the act of controlling employees as a means of
preventing injuries demonstrates a deep lack of respect for employees and their abilities on the organization's part. while simultaneously negleciing management issues that play
a part in occupational safety (Waiker, 1998). Similarly, Reason, Parker and Lawton (1998) argue that the use of rules and procedures is an ineffective means of rnanaging safety given it is impossible to dictate al1 the appropriate and inappropriate behaviours required to assure workplace safety. An over-reliance on rules constrains the behavioun of employees to the extent that, in many situations, individuals are unable to work safely without violating rules or procedures. The control-orientation, the authon argue, only serves to fûrther endanger workers.
Cornmitment-Oriented Management Practices More recently, there has been a growing realization that human resources are better managed by adopting a cornmitment-oriented strategy, or high performance work system approach3.This approach to management assumes that workers are capable of continuous improvement and will perfom at a higher level if they are encouraged to do so. This is achieved by, arnong other things, encouraging employee participation in
'
The ternis 'high cornmitment' aud 'hi& perfof~~~~t[lce' ue ofken used interchangeable in the literanirr to mean those ways of rnanaging human rcsoutces that rnove away fiom control-oriented or Taylorist approaches. in the strictest sensc, high commitment management practices arc those practices that Icad to greater organizational conunitment on the part of employees. More commonly, the terni is used in a looser sense to descrile those practices that lead to a range of beneficial employee attitudes bcyond simply cornmitment ùicluding, for instance, trust m management or perceptions of justice. The term, 'hi@ perfoxmancc management practices' (practiccs that comprise a 'hi@ performance work system') has a somewhat d i f f m t focus, in that it is used to d e s c n i thosc practices that arc associated with positive performance effects. These practices an for the most part similar to those associatcd with high comrnitment-the distinction lies in the outcomc (Wood, 1999a). in his discussion of tmninology, Wood argues that until hi& cornmiunent pmctices have cstablishcd positive effccts on performance. the term 'hgh performance' should be avoided 1 disape and belicvc thme is suff?~cientcvidcncc for the relationsbip between high perform;ince work systems and high performance management practices and organizationai and employec performance. Foiiowing Pfcffkr (1998a). I maintain the tcrm 'hi& performance' can be used interchangcably with 'high cornmitment' and have done so in this paper. 9
decision-making processes, by providing employees with high levels of quality training
and by paying employees comparatively better wages. By treating worken with respect and as capable and intelligent individuals, they will be more committed to the organization and more rnisting of management. in tum, the organization will reap the
benefits in terms of improved performance (Walton, 1985). As Margaret Wheatley (1997) States, ''you can't direct people into perfection; you can only engage them enough
so that they want io do perfect w o r k (p. 25).
In ternis of occupational safety, a commitmentsriented approach emphasizes the role of management in promoting a safe workplace, while getting away fiom the notion that it is ultirnately the employees who are to blame for any injuries that may occur. It follows fiom Walton's (1985) description of the cornmitment strategy that management must, for instance, ernphasize employee participation in training, employment security for al1 workers. and extensive sharing of information within the organization, if managers intend to improve their organization's safety performance.
Many studies highlight the importance of high performance work systems for organizational performance. In a study of 30 steel rninimills, Arthur (1994) found that the quantity. in ternis of labour houn, and quality of performance, as measured by scrap rates, were significantly bener in minimills operating under a commitmentsriented system than in mills managed in a control-oriented fashion. Furthemore, turnover was twice as high in the control-oriented minimills. Huselid (1995) examined 958 publicly
traded companies, and admuiistered a questionnaire to a representative of each Company
in order to assess whether the organizations in question had in place hi& performance work systems. M e r controlling for critical factors such as the size of the organization,
the effects of unions and employee compensation, high performance work systems were associated with significantly lower tumover rates, greater employee productivity in tems of sales per employee and both market-based and accounting-based measures of corporate performance. in addition, Ichniowski, Shaw and Premushi (1997) examined the productivity of
36 steel production lines in 17 organizations. They found that steel output was better when organizaiions had in place practices such as incentive pay, flexible job design elaborate screening of new employees, and the provision of employment security. Employees were also given the opportunity to work in problem-solving teams and were given off-the-job training in these organizations. On the other hand, employees of organizations with more traditional approaches to human resource management produced less than their peen. MacDufie (1 995) similarly found that human resource practices such as working in tearns, job autonomy, selective hiring, contingent compensation, reduced status distinctions and training were associated with greater productivity in the automotive industry, providing M e r evidence for the link between innovative human resource management practices and employee performance. Likewise, in a longitudinal snidy of 80 manufacturing companies, Panerson. West and Wall (2001) found that job quality and training,two components of a hi&
performance work systern, were positively associated with increased productivity anà,in ~ mthe , profitability of organizations. The results of this snidy are especially valuable as
the study examined changes over time and included multiple observations in the form of interviews, Company documents and t o m of production facilities. In sum. 1argue that cornmitment-oriented management practices such as those examined in these studies
should have sirnilarly positive effects on organizational safety performance as they do on
the various measures of organizational performance described above. The adoption of a cornmitment-oriented shategy for improving occupational
safety should be more effective than a control-oriented strategy because it would serve to increase employee trust in management and affective commitment to the organization. In Wheatley's (1985) ternis, trust and cornmitment are the means for engaging employees to do perfect work. In fact. a study conducted by Barling and Hutchinson (2000) found that tmst and affective commitment in an organization were perceived to mediate the relationship between cornmitment-oriented safety practices and safety climate. According to Zohar (l980a), perceived safety climate is defined as the shared perceptions of employees with respect to safety in their work environment. Furthemore, employee behaviours are dependent on these perceptions ( H o h m & Stetzer, 1996). Therefore, it follows that commitment-oriented strategies should increase employee tnist in management and affective commitment to the organization which, in h m , should result in safer work practices and fewa occupational injuries and fatalities.
Perceptions of a positive safety climate may d m mediate the relationship between management practices and occupational safety, and dong with trust and affective commitment will be discussed in greater depth in Chapter Three.
Hkh-Cornmitment Manaeement Practices for Safe Oraanizations Research leading to a clearly defined set of practices managers cari adopt in order to increase their organization's safety performance is lacking fiom the literature (Bimonte, 1994). in line with the cornmitment-onented strategy described by Walton (1985), Jeffrey Pfeffer (1998a) has developed a set of seven practices that he argues
increase the economic performance of a finn by way of the effective management of
human resources. Pfeffer (1998a) labels this 'profits through people" (p. 96). I believe these practices can be extended in two ways: First, three additional practices can be included with Pfeffer's seven principles and, second, 1 will argue that these ten practices apply equally to safety performance as they do to economic performance. Bnefly, the seven practices P feffer (1W8a) descnbes are: organizations must ensure employment security for their employees, they must subject al1 new personnel to a selective hiring process, employees should be provided with extensive training, the design of organizations should emphasize decentralized decision-making
and self-managed teams, the organization must make an effort to reduce status symbols that separate employees into different hieranihicai levels, information sharing throughout the organization should be cncouraged, and employees should receive relatively high compensation dependent on the organization's performance.
To provide an example, it would make little sense to offer rigorous selection and extensive training without ensuring employment security (Pfeffer, lW8a). indeed, empincai data show that fims promoting long-terni relationships with their employees are likely to offer their empioyees more training (Frazis, Gittleman, Harrigan & Joyce, 1998). This idea of bundling will be discussed in greater detail in Chapter Four.
Pumose of the Current Research The overall aim of the curent research was to go beyond the control-oriented approach to the management of occupationai safety and to assess the effectiveness of
high performance work systems for managing safety in organizations. While there is a paucity of research in this are& a growing body of Literan~ecalls for workplace, rather
than individual level factors to be considered in the effective management of workplace safety (Hohann, Jacobs & Landy, 1995; lsla Diaz & Diaz Cabrem 1997; Sheehy & Chapman, 1987). The major purpose of this research was, therefore, to M e r our understanding of the relationship between human resource management practices and safety performance at both the organization and employee levels. At the organization level, a mode1 relating a high performance work system
cornprishg ten management practices and workplace safety measured in tems of lost time injuries and days lost was considered (see Figure 1). At the employee level, this research explored mechanimis that mediate the relationship between the high performance work system and safety performance (see Figure 2). Specifically, 1 considered the roies of t m t in management, affective cornmitment to the organization
and perceived safety cIimate as rnediating mechanisms. Furthexmore, d e t y performance
a
.
.
.
at the employee level was conceptualized more broadly, extending beyond lost time injuries and resulting days lost and including personal safety orientation (compnsed
safety knowledge, safety motivation, safety cornpliance and safety knowledge) as well as safety incidents measured in tems of injuries requiring h t - a i d and near-misses. The literature supporting the development of these models and the inclusion of specific variables will be discussed in greater detail in the subsequent chapters. The following chapter provides the reader with a detailed discussion of each of the ten high perfoxmance management practices and the role each plays in workplace safety. Chapten Three and Four are centered on discussing the models that were tested in greater detail. The former focuses on the role of mist in management, affective cornmitment and perceived safety climate in mediating the relationship between the high performance work system and the safety outcomes. The latter provides M e r refinements to the mode1 and describes in greater detail the safety-related outcomes and the bundling of practices. Chapters Five through Seven descnbe studies that were conducted to evaluate the role of high performance work systems in workplace safety, while Chapter Eight provides the finai discussion and concluding thoughts.
Chapter Two The Ten High Performance Management Practices
The aim of this chapter is to descnbe in detail each of the ten high performance management practices: employment security, selective hiring, training, tearn working and decentralized decision-making, reduced status distinctions. information sharing, contingent compensation, transformational leadership, job quality and the measurement of variables cntical to organizational success. In each case. existing iiterature examining the role of the given human resource practice in promoting occupational safety will be discussed.
Practice One-Ensure
Emaloment Securitv
Pfeffer (1998a) claims that one of the most basic means by which organizations cm improve their performance is by ensuring employment security. There are number of
reasons why this is the case. Fint, iayoffs are cost ineffective given that with each employee laid off investments made in chat person in tems of training and development are also lost. Second, employment security promoies feelings of t m t in management and, in tum, results in employee cornmitment to organizational goals, including improved performance. Third, employment security encourages a long-tenn perspective within organizations. while discouraghg the quick fixes and self-interested behavioun that typically fail to benefit organizations.
In t e m s of safety, employment security was found to be one of the important factors differentiating hi& accident rate companies from low accident rate companies in 19
the both Smith, Cohen, Cohen and Cleveland (1978) and Zohar (l980a) studies. Smith et
al. (1978) report that tumover was greater and workgroup stability was lower in plants with higher work injury rates. Similarly, Zohar's (1980a) review of the literature revealed
that a more stable workforce with a greater number of older employees is characteristic of low accident rate companies. In addition, Grunberg, Moore and Greenberg (1996) in their
study of the wood products industry report that feelings of greaterjob insecurity are positively associated with the actual number of injuries and days of work missed because of an injury. Likewise, it was found in a study of food processing plant employees that greater feelings of job insecurity were related to lower safety motivation and safety compliance and greater numbers of workplace injuries (Probst & Brubaker, 2001). Given this research evidence, it is arguably in an organization's best interest to ensure job or employrnent security4 for their employees. In dangerous work environments, employment security amounts to fewer injuries and a reduction in the costs associated with a lack of workplace safety. The reasons provided by Pfeffer (1998a) as to why employment security benefits
an organization's economic performance can just as easily explain why employment security should benefit its safety performance. To begin, increased turnover results in an inexpenenced and less trained workforce, one more prone to being involved in costly
11 is wonh clanjrulg the terms job &ty and qloyment security as thcy are oficn uxd interchangeably in the literature. but in fact reflcct diffcrent concepts. When employees fecl that the Iikelihood they wiil lose thcir job (defuicdas a -c set of tasks (Parker & Wall, 1998)) is Iow and feel that they have conml over whether or not hcy losc thcirjob, thcy arc exhibiting feelings of job security (Brockner, Grovcr, Retd & Dewitq 1992). According to Pfcffer (1995) employment sccurity, on the othcr hanci, rcfers to a practice in which every action is taken to en- employees rctiiiu a job w i t b the organizaaon in this papa, the nvo tmns arc uscd inurchangeably whto discussing previous rcscarch in order to rcmain consistent with the tcrms used by the ori@ authors. In the currmt study, howcvcr, the provision of crnptoymcnt sccurity is the high performance work practice of mtercst
safety-related incidents (Cohen, 1977; Kincaid, 1996). Most research looking at high turnover and occupational injuries has focused on the implications of the increasing number of contingent, or temporary, worken for safety. Generally, it has been found that contract employees are typically less trained, less experienced, and receive less supervision than more permanent employees (Collinson, 1999; H o h m et al., 1995;
Kochan, Smith, Wells & Rebitzer, 1994; Wright. 1986). This is not surprishg if one considen that there is little incentive for an organization to invest in training temporary worken (Hohann et al., 1995; Rebitzer, 1995). Therefore, increased employment security ensures that employees receive more training and s u p e ~ s i o nresulting , in a more experienced workforce and, in tum, greater occupational safety (Kochan et al., 1994; Root. 198 1).
The mist in management that results fiom empioyment security (Pfeffer, 1998a; Walton, 1985) is another reason why long-term relationships between employer and employee should result in fewer occupational injuries. T w t in management has been shown to increase employee performance (McAllister, 1995). as has organizational cornmitment (Barling, Weber & Kelloway, 1996; Meyer & Allen, 1997). Arguably. both trust in management and affective cornmimirnt should increase safety performance as well (Barling & Hutchinson, 2000).
It has been found, for instance, that nuclear power plant workers who perceive safety risks to be high are less commined to the organization and esrimate the probability of an accident to be higher than do employees who are more committed to the organization (Kivimaki & Kalimo, 1993). Similarly, employees' perceptions of nuclear risk increase as their trust in top management decreases (Kivimaki et al., 1995). By 21
extension, 1 predict that employment security will increase employee cornmitment and trust in management, and will lead to fewer occupational injuries. Lastly, employment security should result in fewer occupational injuries in that it encourages a long-terni perspective. When organizations are actively investing time and money in employees, it is in their best interest to protect their safety. What typically seems to happen when management is not commiaed for the long-term is that employees are seen as dispensable and short-term profits ovenide any concern with safety (Jackall, 1988). Similarly, Sells (1994), in his discussion of the asbestos industry, daims that
much of the threat to worker safety inherent to the asbestos industry was due to the shortterm perspective that plagued management.
To summarize, fint, employment security ensures that employees are better trained and more experienced in their fields which, in tum, will result in fewer occupational injuries. Second, employment securiry builds employees' trust in management and cornmitment to the organization, which leads to greater employee motivation to perfonn better, including increasing worker motivation to perfonn jobs
more safely. Third, the long-term perspective that is encouraged by a work system where employment security is incorporated ensures that organizations do not consider their workforce as expendable. The result is that management will not imperil their workforce by cutting corners in order to satisQ short-term goals. Consequently, employment
security will lead to better economic performance in ternis of profits, as argued by Pfeffer (1998a), while also leading to better safety performance in teims of fewer occupational
injuries.
Practice Twc+Selectivelv Hire New Personnel
To date. the way in which selective h i ~ has g been used to achieve occupational safety is invariably consistent with a control-orientation, with its reliance on the "seleciive exclusion" ofpotentially hi&-risk employees. Typically, personality screening questionnaires are used to differentiate potential employees based on cntena such as
curent or fonner dmg addiction or alcoholism. emotional maturity, tnistworthiness and cornmitment. The majority of practical recommendations and research on selective hiring
has been focused on its use within a control-oriented framework, thereby rendering any consideration of selective hiring within a high performance system somewhat speculative. It is. nevertheless. wonh examining some of the research consistent with this control-oriented approach to managing occupational safety.
One frequently referred to scale is the Personnel Selection Inventory- Form 3s which screens applicants for safety consciousness (Jones, 199 1; Jones & Wuebker. 1988).
ïhis scale is comprised of four subscales measuring: (i) attitudes towards theft and crime. for honesty has been found to correlate with safety attitudes (ie: greater dishonesty is associated with lower safety consciousness). (ii) on-the-job violence, (iii) dnig use, and (iv) how responsible individuals feel for accidents they have previously been involved h.
The use of this employee screening device in a miik processing and delivery Company over the course of four yean resulted in a significant reduction in the accident rate and decreased expenditures on worker compensation clairns (Jones & Weubker, 1988).
Similar results were reporteci in a micking fïrm which had also irnplemented the screening inventory (Jones, 1991), suggesting that using such a device in the hiring process may effectively help reduce accidents at work. 23
Another screening inventory that has ofien been used in occupational safety research is the Employee Reliability Inventory (Borofsky, Bielema & Hohan, 1993; Borofsky & Smith, 1993, Bon>fsky,Wagner & Tumer, 1995). The seven subscales of this inventory examine: (i) freedom from disruptive alcohol and substance use. (ii) courteous job performance, (iii) emotional maturity, (iv) conscientiousness, (v) trustworthiness, (vi) long-term job cornmitment, and (vii) safe job performance (Borofsky et al., 1993). Use of the scale has also produced consistent results-
a significant reduction in turnover and
accidents across settings (Borofsky, et al., 1993, 1995; Borofsky & Smith, 1993). Despite the apparent success of these approaches, several major flaws in these studies bring question to our ability to infer that prescreening procedures caused a decrease in safety-related incidents, thereby limiting the usehilness of this data. First, these studies lacked control groups, leaving the results uninterpretable (Cook & Campbell, 1983). Alternative, plausible hypotheses could have just as easily explained the increases in safety reported in these studies.
For instance, in the Jones & Weubker (1 988) study evaluating the effectiveness of the Personnel Selection hventory- Fom 3S, the authors argue that despite the lack of control group, the apparent effectiveness of the prescreening inventory cannot be attribuied to the implementation of another intervention focused on accident prevention given that there were no other pro-
implemented concurrently. This does not
however eliminate other alternative hypotheses. For instance, an accident prevention program could have been put into effect prior to the implementation of the prescremhg inventory, or perhaps accident reduction couid have been the result of changes in general economic activity or more specificaliy, the sample h ' s productivity. Without a control 24
group, it is impossible to infer that the prescreening inventory accounted for the reduction in accidents. Similarly, a lack of control group is also a problem in the Borofsky and Smith (1993) study. In this case. the authon themselves note that demographic differences between the group hired before and the goup hired after the implementation of the prescreening inventory might account for the reduction in turnover. accidents rates and absenteeism they repon. The authon failed to collect demographic data on the two groups and while the personnel director suggested that that ihe two groups were not demographically different. the fact remains that alternative hypotheses can only be rejected with data. It is dso possible that other organizational changes implemented concurrently with preemployment screening were responsible for changes in safety levels. It is important. for instance. to consider the factors that lead the organization to choose preemployment screening. For instance, was the implementation part of a larger organizational change?
The methodological flaws inherent to the studies descnbed above aside. and perhaps more importantly, this type of prescreening is plagued by the notion, consistent with the control-onented approach, that work-related injuries are pnmady the fault of employees. This type of victim-blaming is unappealing to the extent that it places the responsibility for particular behaviours and ultimately workplace safety on employees
rather than placing the focus on the changes management can make to the workplace in order to improve safety levels. in contrast to the methods just discussed, a high performance system explicitly endoaes management's role in ensuring enhanced safety. Any organization committed to the idea of ensuring their success through human 25
resources, m u t pay a great deal of attention on how they hire new personnel (Pfeffer, 1998a). If an organization practices selective hiring, then it can be argued they will not
only acquire the benefits of reduced costs and increased profits (Pfeffer, 1998a), but also increased workplace safety. In support of this argument, both Cohen (1977) and Smith et al. (1978) found lower injury rate companies had more elaborate selection procedures
than did those organizations with higher injury rates. Pfeffer (1998a) provides a number of ideas for how selection can be effectively accomplished, two of which may be most informative with respect to occupational safety. First, involving teams in selecting future memben could be of benefit and, second, requiring applicants to go through several rounds of interviews in which the organization's values are emphasized also holds promise for enhancing occupational safety. 1 argue that selective hinng, implemented in such a way, will serve to increase an organization's safety performance. Significant empincal research is necessary, however, for a further understanding of this issue.
Practice Three-Provide
Extensive Training
According to Pfeffer (1998a), training is one of the most important dishguishng attributes of a high performance work system. The reason being that the foundation of the cornmitment-oriented organization is the workforce, and training directly benefits employees and organizations. By extension, improving an organizatioo's safety performance, like impmving economic performance, should also be achieved by way of increased training.
The effect of training on occupational safety is likely the most researched and practiced area in safety management and. therefore, a review of the training literature is beyond the current scope. in general, it has been found that employees who have undergone safety training suffer fewer work-related injuries than their untrained counterparts (Hale, 1984). While this does not seem h e d i a t e l y obvious fiom the literature given some studies fmd suppon for training while othen do not, Hale argues that the positive effects of safety training are immediately obvious when one considers those training programs that are well designed. It is important to note, however, that within a hi& performance work system,
sufficient training is not enough. Organizations need to go beyond this by also ensuring that employees are cornmitted to their organization, are more satisfied with their jobs and
are also committed to their safety and that of their fellow workers. An organization operating under a commitment-oriented system can achieve this by being perceived as being committed to safety training (Zohar, 1980a) and by offering training which exceeds the standards required by government regulations or collective agreements. A study by Tannenbaurn, Mathieu, Salas, Cannon-Bowen (1991) provides some
insight into the broad effects of training on performance, and has several implications for safety training. The researchen found that the extent to which training was perceived to have met naval minees' initial expectations, how satisfied uainees were with the training, and the learning of academic content al1 predicted subsequent coIIlfnitment to
the organization. These results are important because organizational cornmitment predicts work performance in gmeral, and safe working in particdar (Meyer & Allen, 1997, Parker, 27
Axtell & Tumer, 2001). This suggests that to be maximally effective training needs to go beyond the provision of knowledge related to how to do one's job safely. Arguably, satisfaction with training and meeting initial expectations are important predicton of organizational cornmitment and safety performance. Furthemore, training alone is not sufficient to achieve optimal performance. Rather, employers must also ensure that employees are empowered to use the knowledge they obtain (Parker, Wall & Jackson, 1997). Organizations need to provide oppomuiities for employees to use new skills following training (Parker et al., 1997), while also not punishing employees for doing so (Hamper, 1991). The role of training becomes paiticularly salient when work is dangerous. in many other work situations, learning can occur by way of trial and error. in the case of inherently dangerous work, however, the cost of a mistake is far too high, making training an especially important aspect of any effon to improve occupational safety (Weick. 1987). Both general and safety-specific training can only serve to benefit employees. By educating worken and, thereby, making them more capable of perfomiing their jobs
safely, any organization will reap the benefits of fewer work-related injuries and fatalities. Organizations need to be cornmitted to training, and need to spend money on
training and developing their employees if they want to see Unprovements in their safety perfomance.
Practice Four-Self-Manaeed
Teams and Decentralized Decision Makinq
There are a number of benefits that cm be derived from organizing employees into teams. Fint, an employee's peers tend to be more effective at controlling his or her
behaviour than is someone else h m higher up in the organization. Second, teams have the acided benefit of making al1 employees feel more responsible for the success of the organization. Third, teams remove layers of the hierarchy which, in effect, gives more control to those individuals who are closest to and best understand the situation. Fourth,
teams allow employees to pool ideas resulting in more creative solutions (Pfeffer, 1998a). Each of these benefits and their role in promoting workplace safety will be discussed in turne
With respect to occupational safety management, there is a growing field of
research that supports the claim that decentralization of decision making and implernentation of teams can benefit an organization in ternis of safety. For example, Simard and Marchand's (1997) study of 1,O61 workgroups drawn nom 97 manufacninng plants in Quebec, focused on the factors that predict workgroup conformity with Company safety niles. Their analyses showed that the quality of supenisor-ernployee relationships, as well as cohesion with the workgroup were the best predictors of the propensity to comply with safety d e s .
Similarly, in a mine where tearnwork was implemented, it was found that employees working in autonornous tearns experienced fewer accidents than did employees who continued to work individually (Trist, Susman, & B m , 1977). There is
also some support for the idea that teams in which memben effectivety provide their peen with feedback conceming their safety behaviours are less likely to M e r injuries 29
(Hohann & Stetzer, 1996). Littauer's (1956) statistical analysis of the proximal causes of industrial and military accidents aiso speaks to the importance of the role of teams in safety. He noted that the introduction of a new employee into the work context was predictive of safetyrelated incidents. This fuiding is funher supported by Goodman and Garber (1988) in their study of safety in underground coal rnining. They found that as the familiarity between memben of a working dyad decreased, safety infraftions increased. These studies demonstrate that employees working in teams tend to monitor each othen' behaviours and provide each other with feedback that, consequently, helps each
employee work more safely. In other words, the propensity to work safely is very much related to the reactions fiom others to one's own behaviour (Andriessen, 1978). Therefore, if employees behave in a safe manner and their coworkers support them in this behaviour. they will continue to exhibit it. Of course, the opposite would also be mie for unsafe behaviom. For feedback to occur between employees and for it to be effective, however, the organization must create an environment in which it is acceptable for workers to openly discuss the behaviours of others (Anderson, Hardy & Wesî, 1990; Bimonte, 1994; Preston & Topf, 1994). Tearnwork in organizations should also effectively reduce injuries given that working in teams causes individuais to feel more responsible for safety and more accountable for each othen' safety. Geller, Roberts and Gilmore (1996) found that a sense of belongingnas to a group and a sense of personal control were both predicton of workers' propensity to actively care for coworker safety. Further evidence for the role of increased responsibility in reduchg occupational injuries cornes h m a study looking at 30
flight crews (Tjosvold, 1990). It was revealed that with more cooperative goals, al1 memben of a flight crew felt responsible for safety and that under these conditions critical incidences were handled more effectively. Essentially, al1 team members were responsible and each peson felt more cornfortable expressing theu opinions and considering those of othea which, in effect, lead to greater air safety. Teamwork should also be an effective means for improving occupational safety because it provides those people who are more familiar with the situation greater opportunities for control. A move towards a tearn design results in fewer hierarchical levels within an organization (Pfeffer, 1M a ) thereby bringing employees closer to the level at which responsibility for their own work takes place. According to Nelkin and Brown (1984), p a t e r connol over their work is exactly what worken in dangerous industries want as their involvement in safety issues is an effective way of reducing occupationai injuries (Pasmore & Friedlander, 1982; Simard & Marchand, 1997). For instance, Mobil Oil Corporation found that the implementation of joint safety cornmittees in which ernployees and management regularly met to discuss safety issues had a significant effect on increasing workplace safety (Smith, 1994). Sirnilarly, in a study of chernical company employees working in teams, Alampay
and Beehr (in press) found that the safest working teams were those with the most control over varied aspects of their work. Working in teams with greater authonty over one's work was also found to be associated with fewer work-related injuries in a study o f manufacturing employees (Kaminski, 200 1).
Last, safety performance will irnprove with the implementation of teams because tearns promote the sharing of ideas that wiU result in better solutions, as illustrated by 31
Tjosvold's (1990) flight crew study. As mentioned earlier, he found that when faced with dangerous situations, memben of a flight crew perfonn more effectively as a group than as a hierarchy with the captain at the top of the chah of command. Tjosvold applies a theory of goal-interdependence to explain this. Each team rnember is motivated to contribute his or her ideas by their belief that they m u t work together in achieving their goals. An integral part of this is the exchange of resources such as ideas and effective communication. In this case, the sharing of ideas meant that as a group, the flight crew could corne up with the best solution possible in the face of danger, thereby making the work environment safer. Similar results were found in the Tnst et al. (1977) study described earlier. These researchen found that miners working in autonomous workgroups attributed their improved safety performance to their common goals, increased communication between members, and increased sharing of ideas. While a great deal of research needs to be conducted in the area of teams and
safety, there are a number of reasons to believe that the implementation of teams may be
an effective way to increase occupational safety. To summarize, team dynamics encourage peers to provide each other with feedback and encouragement for perfoming their jobs safely. Teams also give employees an increased sense of responsibility that motivates them to take an active interest in theu own safety and that of others. Furthemore, the implementation of a team structure ensures that those worken who most clearly understand what rnakes their job safe can take action to improve safety in their workplace. Last, teams encourage the sharing of ideas and solutions to problems related to occupational safety.
Practice Five-Reduce
Stanis Distinctions
Status distinctions in organizations remain the nom, and create unwanted barriers between people that breed resenmient and h m motivation (Pfeffer, 1998a; see also, Hamper, 1991). In a hi@-performance worlc systern, each employee f?om the shop floor to top management should feel that they cm contribute to diverse aspects of the organization. However, when select employees have special privileges such as the use of executive dining rooms. reserved parking or greater access to information, the message being conveyed is that these organizational memben are more Unportant than members not given the same advantages. If occupational safety is dependent on commitment from top management, as will be discussed in a later section, then one way to increase this commitment is to increase
the level of familiarity between management and shop floor employees. Status distinctions reduce this familiarity and make it difficult for management to fully appreciate the risks and challenges encountered by employees. By the same token, s t a t u distinctions also result in employees failing to perceive the importance management does place on their safety. in fact, an empiricai snidy of British Rail employees supports the idea that management and workers tend not to understand or appreciate the value each other places on safety (Clarke, 1999). The divide between these two p u p s can be so deep that followùig an incident resulting in a number of employees being killed, one survivor poignantly observes, "They lost money, but we lost lives" (kath on the Job, 1991).
Further evidence supporthg the relevance of reduced statu differences for occupational safety can be f o n d in attribution theory. DeJoy (1994) argues that in the 33
case of occupational safety management, where placing blame for events is so inherent, there is a tendency for conclusions about injury causation to be biased. For instance, it is perceived to be in top management's best interest to blame front-line workers for safety infractions. in contrast, workers may be inclined to hold management responsible for workplace injuries. DeJoy (1994) suggests that one of the ways of getting around this is by reducing the distances between employers and employees by way of increased
exposure. Essentially, when managers and employees see these status barriers being dismantled, they are more likely to see their own safety as dependent one on the other, and to feel an increased responsibility for joint safety. Milanovich, Driskell. Stout and Salas (1998) repoit that in airiine cockpits, individuals of higher statu, narnely captains, miss opportunities to listen to their fint officers, while subordinates are often too compliant and obedient in the presence of captains. Milanovich et al. (1998) also set out to study the effects of status distinctions. They asked 30 recently qualified military aviatoa to rate their expectations for two aviatoe. Each was described identically in terms of age, experience, mental status and health, however one was descnbed as a pilot while the other was described as a CO-pilot. The researchers showed that this status difference was critical. The participants in their
study held higher general and specific expectations for pilots than CO-pilots.This is important because CO-pilotswho hold higher expectations of pilots may then change their behaviour accordingly by, for example, becoming more abservient and this could have serious consequences in terms of performance more g e n d l y as well as safety
performance more specifically.
The second reason why reduction in status differences should positively affect occupational safety is that without unnecessary barries to communication there should be a greater flow of ideas through the organization. This point is best illustrated by a chemical plant operator referred to in Nelkin and Brown's (1984) research who observai that 'Vie biggest problem this plant has is that anybody with a degree thinks they're above the men on the bottom m g . There is no communication whatsoever. They have no respect for their chemical operaton. They think we're a bunch of dummies because we don't have a degree" (p.54). An incalculable amount of information is lost as a m u l t of
this type of distinction and resulting lack of communication between employees. One way of achieving a reduction in statu distinctions is through managementby-walking-around (MBWA). MBWA encourages management to be less O ffice-bound,
and spend at least a part of each day walking around and talking with the people whom they s u p e ~ s (Peten e & Wateman. 1982). in the safety context, not only would management be able to better see some of the hazards experienced by their subordinates. but as they l e m more about the daily safety problems encountered, they will be in a better position to engage in preventive efforts. in addition, employees who interact with their managers more frequently may be more prepared to trust their managers and may be more able to see that management does in fact hold occupational safety as a pnonty. The problems of mutual distnist identified by Clarke (1999) might be reduced. Overall, there is a paucity of research examining the d e of statu distinctions in injury prevention. Nevenheless, 1 betieve this management practice may be critical to a high performance system in which occupational safety is a final ouicorne. Overall,
reduced mnis distinctions wili likely m a t e greater associations betweenmanagement 35
and other employees, and will encourage the two groups to cornrnunicate and to M e r
understand the challenges faced by the other group.
Practice Six-Share
Information
in order to work effectively, worken need access to infonnation. In fact, based on his observations of the practices enacted by exceptional organizations, Fitz-Enz (1 997) suggests that information is one of the organization's most valuable resources. By
providing employees with information. the organization enables hem to have a bener understanding of the operation and its goals which, in effect should increase overall organizational hinctioning (Pfeffer. 1998a). Similarly, it would not be possible to work safely without hi11 information about al1 aspects of one's job in particular, as well as the organization more generally. In fact,
Ontario's occupational health and safety legislation is based on this very assumption given that a worker cannot exercise his or her right to refuse unsafe work without full information about the safety of that work. Much as the sharing of infonnation across organizational levels is critical for high performance in general, it is equally critical for optimal safety perfomance. There are some data supporting the importance of information sharing for occupational safety. Zohar (1980a) reponed that factones rated by trained, independent observers as h a h g bener safety programs were characterized by open communication
between management and workers. As well, Cohen (1977) and Smith et al. (1978) found that companies with beaer safety records where characterized by greater contact and
more open discussion between management and shop floor employees. Similarly, 36
H o h a n n and Morgeson (1999) found that to the extent that employees felt cordortable discussing safety-related issues with their supervisors, they were more highly commined to following safety procedures and practices and that this resulted in the lower occurrence of workplace injuries. Suppon for the value of information sharing also cornes fkom
Griffiths*(1985) study in which he amibutes his company's nim-around in safety performance in part to the use of an accident reporting system. The system ensured that al1 information relevant to safety-incidents resulting in loss or injury or 'near-misses' was recorded and then published in an effort to make sure that mistakes were not repeated. The obvious direct benefit of information sharing, therefore, is to provide employees with the necessary information to be able to do their job efféctively and safely. information sharing also ensures that managers are fully aware of the conditions
under which their employees are working and the necessary measures that need to be adopted to ensure their safety. Furthemore. information sharing will exen indirect effects on safety performance by enhancing employees' role clarity. Role ambiguity, or being unsure of what is
expected of one's selt is one of the major work-related stressors (Beehr, 1995; Sauter, Murphy & Hurrell, 1990). When employees believe, on the other han& that they have suficient clarity concerning their role, research shows consistently that they are heaithier and more productive, satisfied with their jobs and committed to the organization (Jackson & Schuler, 1985; Mathieu & Zajac, 1990).Therefore, it is fair to predict that access to
necessary information would have positive effects on employee safety as well. Aside h m the obvious benefit of providing sufficient information to perform
one's job safely, full sharing of information has other important secondary benefits. In 37
those situations where managers do share information, employees will have greater trust in management (Hohann, 1999). As previously noted, Clarke (1999). in a study of British Rail employees. found that both management and employees do not know how important safety is to the other group. The author argues that with greater communication. employees could sense those instances when management took safety seriously and would, in turn,show more trust in management. Employees also feel they are a meaningful part of the organization when there is greater sharing of safety-related information (Hohann, 1999), and there are a number of positive consequences of this. Arnong employees in a nuclear power plant, a lack of knowledge of safety-related issues was associated with higher levels of perceived risk (Sjoberg & Drottz-Sjoberg, 1991). In tum, perceived risk is negatively correlated with organizational cornmitment (Kivimaki et al., 1995) and job satisfaction, and positively correlated with task distraction (McLain, 1995). These findings are relevant to occupational safety management as they suggest that the positive consequences of sharing safety-related information include diminished perceptions of risk, greater organizationai comrnitment and job satisfaction and less task distraction-al1
important
factors in workplace safety. With respect to information sharing it is again worth mentioning that not al1
industries will benefit directly from an approach that relies on learning h m mistakes. Weick (1987) notes that employees working in dangerous industries simply do noi have the luxury of being able to do so. Instead, prevention becomes critical in these contexts. Employees must depend on the "imagination, vicarious experiences, stories, simuiatious, and other symbolic representations of technology and its effects" (p. 113). The sharing of 38
information is critical to this type of learning.
In surn, the sharing of information between members of the organization at al1 levels is a cntical component of a high performance work system. Information sharing is essential to any effort to reduce injuries, as information is the resource necessary for workers to be able to do their jobs to the best of theu abilities. Furthemore, the sharing of information encourages worken to trust management and to feel they are meaningfùl rnemben of the organization. This will only serve to encourage worken to want to work safely, while also making managers more aware and better able to deal with the safetyrelated issues their workers face.
Practice Seven-Provide
Com~ensationContingent on Safe Performance
Pfeffer (1998%1998c) argues that generally paying employees well enhances organizational performance. "If you're getting paid a wage that you're happy with. then you're happy at your work so you're switched on and alert. You don? mind doing your bit" clairns a North Sea oil worker (Collinson, 1999, p. 591). Well-paid employees feel valued by the organization, which in tum enhances their cornmitment to it. In addition, paying high wages increases the size and the quality of applicant pool during hiring, d l of which strengthens organizational performance in the long m. The contingent nature of compensation is even more important for organizations
(Pfeffer, 1W8a, 1998~).By explicitly choosing which behavioun are to be rewarded, the organization signals unambiguously the behaviours it values. Furthemiore, employees are motivated to contribute more to the organization when their own interests are in line with those of the organization. Occupational safety should reflect one such issue. Arguably, if
safety is considered a key performance indicator by an organization, then employees should be rewarded for theu efforts to improve safety, much as they would be for improving the h ' s profits within a more conventional high performance work system. Leather (1987)explicitly states that an organization should compensate individuals for the time and effort they devote to enhancing occupationai safety. Furthemore, employees cannot be expected to follow occupational safety requirements if doing so results in lost wages or other lost rewards, or if they suffer any f o m of inequity for working safely as compared to their peers who do not work as safely. There are data showing that paying people contingent on their safety performance may be effective in reducing occupational injuries. Several studies have investigated the effects of tokens and other reinforcers on occupational safety. For instance, Haynes, Pine and Fitch (1982) examined the effects of rewards such as fiee gasoline. cash and free bus passes on urban transit operaton working in teams. They found that this system produced marked reductions in accident rates. Sirnilar results were found in a study of roofîng crews (Austin, Kessler, Riccobono & Bailey, 1996). In this case, the crew was provided with time off as a reinforcer for complying with certain safety behavîours. Safety
cornpliance Ievels increased dramatically h m 55% to 95% as a consequence of this intervention. In a review of 24 such studies, McAfee and W i (1989) f o n d that consistent across studies, incentives or feedback Iead to lower numbers of accidents or improved d e y conditions. Nonetheles, several pmblems emerge with such an organizationai behaviour modification approach. Fim, the behaviours under consideration are typicaily highiy specific, limiting the possibilities that what is learned will generalize amss situations. 40
Second, there is widespread concem that such approaches rely on control to such an extent that they constitute manipulation (Pinder, 1998; Walker, 1998). Third, the longterm effects of this approach are not fully undentood (McAfee & Winn, 1989). Most studies of behaviour modification are focused on encouraging or discouraging specific behavioun such as seat belt or ear plug use, or keeping floon dry and aisles clear (see, for instance, Geller, Davis, & Spencer, 1983; Zohar, 1980b;
Chhokar & Wallin. 1984). Yet. othen go beyond merely focusing on reducing injuries. For instance, in an open-pit mine, it was found that giving employees tokens for safe performance significantly reduced the number of injuries (Fox, Hopkins, Anger, 1987). At the end of the month, employees were given tokens redeemable for goods if neither themselves nor a member of their own workgroup had been involved in a safety-related incident in the preceding month. More importantly, tokens were also provided to individuals or groups for making suggestions of ways to improve safety and for taking
unusual actions to prevent an injury or an accident. This approach is far more consistent with a high performance work system than simply providing rewards for specifically targeted behavioun. Nevenheless, as discussed in Chapter One,it has been argueci that organizational behaviour modification approaches to improving safety constitute manipulation of employees (Pinder, 1998; Walker, 1998). This is especially the case when there is exclusive use of individually-based rewards in the application of this approach. Furthemore, individually-based rewards raise other important issues, such as the potentially negative effect they may have on team work, and the fact that they cm result
in employees cutthg corners thereby M e r endangering themselves (Walker, 1998). It 41
is therefore, not surpnsing that these methods fail to be endoned by unions (Waker, 1998).
Group-level rewards for safety-enhancing actions are more in line with Pfeffer's (1998a) proposition that employees should be compensated for organizational performance. According to P feffer ( 1998~).individual incentive pay only acts to impair performance by eliminating the incentive to work in tearns and by encouraging worken to focus on short-tenn rewards. Ideally, compensation should be tied to the performance of the work group or fimi as a whole. In this way, contingent pay is part of a rneaningful system of management practices. Contingent compensation as a part of a high performance work system would serve to align the interests of each and every employee with the organization's overall goals. including those related to improving occupational safety levels. Therefore, to be consistent with a high performance system, organizations need to provide compensation to workers contingent on higher level such as group or organizational performance. Furthexmore, compensation needs to be linked to a wide
range of performance measures, of which some are safety-specific performance measures such as proactive behaviours. This would encourage greater proactivity among managers and employees as part of a long-term view of safety.
Practice Eiaht-ADD~ Transformationai leaders hi^ The level of importance management places on occupational safety directly affects how motivated employees are to perfom their job safely (Ho finann et al., 1995).
For instance, in reviewing the literature, Cohen (1977) found strong management 42
commitment to safety is a defining characteristic of successful occupational safety progmms. Similarly, in a study cornparhg high and low accident rate companies, the management of low accident rate companies was more strongly committed and more actively involved with work-related safety than was the case for those companies with
higher accident rates (Smith et al.. 1978). Griffiths (1985) attributes the exceptional safety record of his Company almost entirely to strong management cornmitment to occupational safety. The Chief Executive Officer of DuPont, an active participant in occupational safety, displays this type of conunimient by signing his letten. Chief Safety, Health and Environmental Offcer. Griffiths (1985) goes m e r , highlighting the importance of having a strong commitment to worker safety at ail s u p e ~ s o r yievels within the organization and not simply at the top management level. While there is a growing understanding of the central role management plays in
an organization's safety performance, little research has exarnined precisely what it is that a 'strong commitment to occupational safety' really means in terms of managerial action. Zohar (1980a) claims that management cornmimient to occupational safety cm be expressed in a nwnber of different ways. For example, safety matters would typically receive hi& prionty at meetings, safety offices hold high statu positions. safety training is emphasized, open channels of communication exist between al1 employees for the discussion of safety issues and new ideas for enhancing safety, good housekeeping is emphasized and there is a stable workforce. Gnffiths (1985) adds to this list a comprehensive safety policy,clear safety-related objectives and, k t , employee involvement. The strategies identified by Zohar (1980a) and Griffiths (1985) closely follow the cornmitment-oriented approach (Barhg & Hutchùwn, 2000; Wdton, 1985) 43
which emphasizes treating worken as capable, intelligent and responsible individuals. There are some research data supporting the role of senior management commitment to occupational safety. Simard and Marchand (1 9 9 9 , for exarnple, investigated the effect of senior management cornmimient on employees' willingness to take initiatives regarding safety. Based on nsponses fiom approximately 23,000 employees, they found that though management commitment was not directly associated with employees' safety initiative, it was the strongest predictor of supervisors' use of a participative style for the management of occupational safety. In nim, this participative style was the most significant predictor of employees' safety initiative. There are a number of reasons why transformational leadership is an appropnate leadership mode1 for enhancing occupational safety. Transformational leadership has probably received the most empincal scrutiny of any current leadership theory (Bass, 1998) and its validity is supported in a number of contexts (e-g., Lowe, Kroeck & Sivasubramanian. 1996), including unions (Kelloway & Barling, 1993) where access to formal rewards by leaders is limited and the importance of persona1 infiuence is magnified accordingly. Most importmtly, each of the four factors comprising transformational leadership, namely, idealized influence, inspirational motivation, intellectual stimulation and individualized consideration, lend themselves to the task of enhancing safety performance. Leaders act in a manner consistent with idealized influence when they act as role models for their followen. They are respected because of their beIiefs and because they consistently do what is moral and nght, not what is easy, or penonally profitable or beneficial. in addition, idealized influence builds mist (Bass, 1990, 1998), which is 44
critical because trust is thought to be a prerequisite for the performance outcornes associated with high performance or high cornmitment work practices.
Leaders who manifest inspirational motivation inspire their followen to go beyond their individual needs for the collective good of the organization or group. They also convince their followen that they can achieve at levels previously not thought to be possible, and use symbols and stones to clariQ the mission (Bass, 1990, 1998).
Using intellectual stimulation, leaders challenge their followers to confiont the validity and utility of long-held assumptions and to think for themselves. They also encourage their followers to solve problems in new and innovative ways and to see difficulties not as obstacles that interfere with the job that needs to be done, but rather as challenges or problems for which they must h d a solution (Bass, 1990, 1998). Last, leaders show individualized consideration by being demonstrably interested in their followen' needs and interests. They acknowledge the differences between followen and consider the needs and abilities of each individual when assigning tasks
(Bass, 1990, 1998). Leaders high in idealized influence would be able to convey the value of safety through their pmonal behaviours, those high in inspirational motivation would convince their followen that they could attain levels of safety not previously considered possible. The potential benefits of intellectual stimulation for enhancing safety performance are considerable. Intellectually stimulahg followers to confiont safety issues would help develop knowledge of new ways to achieve high safety levels, while also enhancing information sharing. Individualized consideration would be evident through leaders' real concern about their foliowers' safety at work, far beyond what would normally be 45
required to saris@ the minimal requirements of govemment regdations or collective agreements. Research examining the effects of transfomational leadership on safety is limited to only a handful of studies. O'Dea and Flin (2000) report in a study of managers in the offshore oil and gas industry, that managers' transformational leadership behavioun predicted worken' willingness to take initiative in safety-related matten. Similarly, Barling. Loughlin and Kelloway (2001) found, in a sarnple of restaurant worken, that to the extent that the participants*supervison exhibited transfomational leadership, workers experienced fewer occupational injuries. Similar results were found by Williams, Turner and Parker (2000) and Zohar (1001) in team contexts. Ii was found that the extent to which employees working in
teams followed safety procedures (Le., demonstrated safety cornpliance) and participated in activities which were beyond what was sirnply required (i.e., demonstrated proactive safety behavioun) was predicted by the transfonnational leadership behaviours of their
team leaders (Williams et al, 2000). Transformational leadenhip was also found to influence the number of micro-accidents (i.e., minor injuries) in employees working in teams in metal processing plant (Zohar, 2001). Collectively, these studies nippon the idea that transformational leadership plays an important role in occupational safety. A m e r reason why transformational leadership is appropnate within a high
performance system is that controlled-outcome research shows that transformational leadership can be taught to managers. Barling et al. (1996) found that bank managers who
underwent transformational leadership training were more intellectually shulating and paid greater indîvidualized consideration to subordinates than managers in the control 46
g"0UPOverall, the role of transformational leadership in a high performance work system cannot be overlooked. Research evidence demonstrates that transformationai leadership can make a critical difference to the performance of an organization and can also exert critical effects on safety performance at work. Furthemore, as Pfeffer (1998b), notes, it is the oqanizational leader who chooses the nature of the system to be implemented, and who helps to ensure successful implementation. As most organizations do not make use of high performance work practices (Ichniowski, Kochan, Levine, Olson & Strauss, 1996; Osterman, 1994). deciding to do so means going against the tide, which
is not always an easy task and one requinng strong leadership. For the reasons descnbed. transfomational leadership has been included in the current conceptualization of a hi& performance work system.
Practice Nine-Provide
Ouaiitv Work
Despite job quality not being included as a component in any of the major concepnializations of a high performance work system (e.g., Fitz-Enz, 1997; Pfeffer, 1998a), high quaiity work should be a cntical component of a high perfomance system
for two major reasons. In the h t instance, a high quality job, one that is "more fulfilling and effative" (Parker & Wall. 1998, p. ix), will ensure that employees maintain their
focus and attention. Second, and again referring to Wheatley's (1997) comment that "you
can't direct people into perfection; you can only engage them enough so that they want to do perfect work" (p. 25), a hi& quality job will engage people emotionally. Overall, a welldesigned job will ensure that employees are engaged mentaily,intellecnially and 37
emotionally, and optimally equipped to work safely. The job characteristics model (Hackman & Oldhman,1980; Oldham, 1996) is one of the most endurhg theones of work design. It consists of five job characteristics that comprise quality work, specifically: ski11 variety, task identity, task significance, task autonomy, and task feedback. These characteristics. by way of mediating psychological States, result in outcornes such as greater motivation and perfomance, job satisfaction, and lower turnover and absenteeism. What is most noteworthy about this model is its continued popularity as a tool for improving job quality. This continues despite its nurnerous forms and acknowledged limitations (for a more detailed discussion, see Parker & Wall, 1998).
Karasek's demand/control model provides a M e r and equally prevalent conceptualization of job design, one in which quality work is considered a function of both the autonomy an individual has in doing their work and the psychologicai demands of that work (Karasek. 1979; Karasek & Theorell, 1990).A quality job. following the demand/control model is one in which there are both high demands and high autonomy, with the premise being that those employees provided with C O ~ D Oover ~ theu work, will
be able to cope with the added demands. Much support has been found for the model in
general making it an important b a i s for job design. While there remains some question
in the literature concerning the proposed interaction between autonomy and demands. both the main effects are founded (see Parker & Wall, 1998 for an ovewiew). Research exarnùiing the role of work design on performance in general is extensive, yet the same cannot be said where occupational safety is concernai. Nevertheless, researchers have certainly focusad on the link between job dissatisfaftion 48
and injuries. The fundamental assumption on which this research is based is that employees who are dissatisfied with their work are less likely to concentrate on the task at hand. There are now data showing that job dissatisfaction is associated with task distraction, which itself is associated with employees' perceived nsk and fiequency of exposw to physical danger (McLain, 1995). Probst and Brubaker (2001) found job satisfaction to be positively related with safety motivation, safety knowledge, safety cornpliance and workplace injuries. Several Other studies have M e r demonstrated a link between job dissatisfaction and workplace injuries (Cooper & Sutherland, 1987; Holcom et ai., 1993; Zwerling et al., 1996). Studies focusing on job dissatisfaction and safety, however, are limited by their inability to uncover the predisposing causal factor. As Frone (1998) notes, job dissatisfaction itself is best conceived as an outcome of other job characteristics or ernployment expenences. Frone contrasted the possible effects of several job characteristics on teenagers' work injuries demonstrating that both work overload and boredom on the job were positively associated with work-related injuries. Thus, high quality jobs are critical for optimal safety, and management certainly has control over the quality of employees' work. Therefore, high job quality has been included as one of the ten management practices in the high performance work system. Current conceptualizationsof work and job design (Hackman & Oldham, 1980; Karasek, 1979; Parker & Wall, 1998; Sauter et al., 1990; Warr, 1987) suggest factors that would be critical to a quality job in which
employee safety is of concm, namely, appropriate work load, ernployee control, and role clarity. Each wiii be discwed in tum.
Ap~ro~riate Workload
The effects of workload on well-being has been studied for some time. Sparks, Cooper, Fned and Shirom (1997) found in an analyses of 3 1 quantitative and qualitative studies that there is a significant negative relationship between long hours at work and both physical and psychological well-being. Furthemore, the effects of long hours at
work may be more pronounced for jobs requiring greater attention, or for those that rely on repetitive tasks. This finding is M e r supponed by Raggatt's (1991) study of 93 Auscralian long-distance coach driven. He demonstrated that driving long houn was associated with stimulant use and sleep problems, which in tum predicted physical symptoms and greater numben of visits to medical practitionea. Interestingly, Raggatt also found a link between long houn of work and speeding, funher supporting the notion that work overload cm affect occupational safety. Baugher and Roberts (1999), similady. found that petrochemical workers who k i t they had high job demands were more womed about nsk at work than their less nished and less ovenvorked counterpans. Overall, these findings may be explained by the fact that work overload is a role stressor, and role stressors have been found to interfere with the ability of individuals to cope with demands (MacEwen & Barling, 1991). This type of situation, in tum, arguably predisposes people toward unsafe work. The idea that work overload, or what is frequently refend to as a 'production push', might be linked to occupational safety is by no means new and a considerable amount of research investigatïng the production push and safety issues has been reponed. Eyssen, Hofban and Spengler (1980), for instance, found that when managers felt hindered by an unusuaily heavy workload safety was compromised. Similarly, a midy by 50
Hofmann and Stetzer (1996) shows that an implicit push for production within an organization is associated with dimiaished workplace safety. There is general agreement that safety infractions increase during periods of economic growth, presumably because of the increase in the Pace of work that wodd be consistent with the need for production to meet demand (Conway & Svenson, 1998). Worth noting are Sparks et aL's (1997)
findings that there are numerous indications that having the ability to exert job-related control, such as the ability to refuse overtime without fear of recnminations (Schmitt, Colligan & Fitzgeiald, 1980). and the fieedom to choose work schedules (Barton &
Folkard, 1991) are cntical in rnitigating the effects of work overload on well-being and safety.
Far less work has been published on the effect of work underload on occupational safety. However, arguably role stressors fonn a U-shaped hinction such that we can expect similar negative effects at either end of the continuum. in support of this view,
Frone ( 1998)-in a study of injury causation in adolescents, found job boredom to be associated with a worhplace injuries. Fisher (1993), in a conceptual piece examining the causes and consequences of boredom, argues that boring tasks lead worken to seek stimulation fiom other sources and suggests that workers may,for instance, engage in horseplay, sabotage or thefi, and may work in an unsafe manner as a consequence of being b o r d at work (see. for instance, Hamper, 1991).
Lack of attention was also implicated in safety-related emrs in a laboratory study in which 60 volunteers engaged in a cornputer simulation involving filling a trailer with a
required amount of petmleurn, delivering it to a gas station and safely unioading the contents. The analyses showed that a lack of cognitive attention was the best predictor of 51
two different types of erron, namely spills and the placing of petroleurn in the incorrect cornpartment. Second, the results of a study of 7 1 transport drivers demonstrated that attention problems alone predicted actual involvement in vehicular incidents (Arthur. Barrett & Doverspike, 1990).
Emplovee Control Providing employees with greater autonomy (Hachan and Oldham, 1980; Oldham. 1996) or job-decision latitude (Karasek, 1979; Karasek & Theorell, 1990) is a component of most major theones addressing the elernents of a high quality job. Research findings show that autonomy or control is associated consistently with higher levels of work performance, more positive employee attitudes and higher levels of physical and psychological well-being. Furthemore, Patterson et al. (2001) dernonstrated longitudinally that job enrichment, a variable comprised of job variety, ski11 fiexibility and, most relevant to the current discussion, job responsibility predicted labour
productivity and firm profitability. The research of Wall, Corbett, Clegg, Jackson and Martin (1990) is also instructive in demonstrating the importance of worker autonomy. At the onset, empioyees working in an advanced manufacnuing environment were fhctioning within a system
where their supervisor had to be summoned each t h e a problern emerged with the equipment. The Company then moved to an 'operator-oriented' system, in which employees were given 15 minutes to repair erron before another party was surnrnoned.
They found both short-terni productivity gains and mental health benefits for employees associated with the operator system. These data suggest ihat when employees are 52
involved in their work, they will leam to take action to identify and correct problems. In terms of safety, they would be able to identim safety hazards and correct problems that might be associated with imminent danger, as well.
More specifically, there are now suggestions that greater levels of autonomy at work are associated with enhanced safety. Parker et al. (2001) report on a snidy conducted in a glas manufacturing plant in which occupational safety had been identified as a crucial concem. Their results showed that job autonomy enhanced employees*cornmimient to the organization, which in tum affected the employees' safety cornpliance with procedures. Similarly, Simard and Marchand (1995) demonstrated that a participative management style was the single best predictor of the extent to which employees were proactively involved in their own safety, as opposed to merely complying with safety regulations.
In addition, the participative approach to occupational safety management
was also the most salient predictor of two other factors which themselves are cntical for safety, namely group cohesion and cooperation between members of the group. Shannon et al. (1 996) m e r nippon the idea that control over one's work has a significant effect on safety levels. They report, in a study of 718 Canadian workplaces, that lost-tirne accident rates were lower in those workplaces in which worken participated in decision-making and in which employees were not expected to simply follow the instructions of management.
Role Clarity When employees are sure of how to do their jobs and are certain of what is required of them they have role clarity. A lack of role clarity (i.e., role ambiguity) is a major role stressor (Beehr. 1995; Sauter et al. 1990) and has been showri be consistently associated with productivity and health problems (see Jackson & Schuler, 1985). There
are several reasons to believe that role clarity may be associated with reduced safety risks. Houston and Allt (1 997) studied a group of junior medical doctors, al1 of whom had just assumed employment and were therefore considered to be somewhat unsure of
the role requirements of their jobs. The authon showed that the psychological distress associated with assuming the new role was related to an increase in the number of medical erron they made, the results of which could of course have serious personal and public safety effects. Hemingway and Smith's (1999) study of 252 nurses employed in four different hospitals fkther supports the potential huiction of role ambiguity in occupational injuries. The researchen showed that the level of role ambiguity experienced by the nurses was significantly associated with the number of reported injuries they experienced. To surnmarite, quality jobs should provide worken with opportunities to exert conml over their work, should inaintain employees' cognitive attention by being both interesthg and challenging, should not overburden the worker, and should have clearly defined expectations. High quality jobs are important to organizations because they
engage employees both mmtally and physically. This, in tum, should inspire employees to work more productively as well as more safely. As such. quality work is a natural 51
extension to the high performance work system.
Practice Ten-Measure
Variables Cntical to Oreanizational Success
Recognition of the importance of measuring desired behaviours in organizations is not new. Lawler (1996) reminds us, appropriately, that "organirations get the kinds of behaviours they measure and reward" (p. 232). With respect to the management of safety, organizations have not been remiss with respect to measurement, however, an examination of organizational practices highlights several issues. The fint issue is the extent to which the behaviours being measured provide
usefbl information for interventions. Pfeffer (1998b)uses the context of financial reporting systems to illustrate this point. He notes most financial reporting systems focus
on events that have already occurred, with less information about current conditions that could lead to direct action being amassed. Similarly, with respect to occupational safety, focusing on past safety incidents will provide less relevant information for interventions
than a focus on curent conditions, such as criticai ernployee attitudes or proactive behaviours that would predict subsequent safety performance. Second, the very behavioun that organizations typically measure are consistent with, and perhaps foster a control-onented climate. Organizations are preoccupied with measunng their cornpliance with government regdations or with the provisions of collective agreements, factors that as WheatIey (1 997) argues, "can never spell out the route to perfect safkty" (p. 25).
Furthemore, this approach may encourage a leadership style consistent with management-by-exception (Bass,1998) because managers would be preoccupied with 55
identifymg and correcting saf'ety-related infractions. Opportunities to go beyond corrective transactions are missed and there are few chances for employees to enhance safety through, for example, the sharing of ideas. Perhaps not surprisingly, the research data show that management-by-exception is usually inversely related to desired outcornes (e-g., Howell & Avolio, 1993). Following this argument, to the extent that cornpliance with governent saf'ety regdations and collective agreements involves management-byexception, it is highly unlikely thai it will result in great irnprovements in occupational safety levels. Simply measuring safety levels in ternis of numben of incidents, types of injury, and days lost also encourages the under-reqorting of safety infractions (Conway &
Svenson, 1998; Eisenberg & MacDonald, 1988; Parker, Carl, French & Martin, 1994; Pransky, Snyder, Dembe & Himmelstein, 1999; Veazie et al., 1994). Govements frequently ievy fines for safety infractions, Iegislation across numerous juridictions allows worker's compensation boards to increase premiurns for poor safety and collective agreements provide workers and their unions with numerous avenues for redress of safety idtactions such as grievances, work stoppages and even strikes. As a consequence, numerous incentives exist for managers to inaccurate1 report safety Uifiactions. Furthemore, measuriag safety levels in such a manner does not provide any insight into the number of near-misses or the nurnber of injuries not requiring medical attention that may occur with much p a t e r nequency. The focus of safety measurement for many organizations is dictated by the demands of extemal agencies, with the focus of most firms remainuig on the number of incidents, the severity of the resulting injuries, and in the worse case, the n u m k of 56
fatalities. Aside from the temptation to underrepon noted above, and the unreliability of these data (Donald, 1995), most significant is the fact that the idormation typically collected is simply not usehl in helping organizations l e m fiom their experiences nor in enhancing lunue safety. What is needed to enhance safety in the long-tem is a measurement approach that focuses on the proximal causes of the incident. For example, if a worker is injured or killed in a tractor incident, the cause of injury or death is not so much the fact that the tractor rolled over. Rather, the cause or causes run much deeper. Was it a lack of training or boredorn that caused the tractor to roll? Was it stress-related cognitive overload and a lack of concentration (MacEwen & Barling, 1991)? Or, perhaps a child using equipment designed for adults (Castillo, 1999) resulted in the rollover? instead of focusing on distal outcornes such as injuries and fatalities, benefits may be derived fiom assessing those attitudes and behaviours that are a direct consequence of
management practices and which, in tum, predict injuries and fatalities. For instance, the propensity for employees to take initiative to enhance safety and their participation in safety matters may be critical factors in predicting safety performance (Andnessen, 1978; Howlen & Archer. 1984; Simard & Marchand, 1994). in fact, in a study of 100 worlcplaces, Simard and Marchand (1994) found the behaviours associated with employee safety initiative (e.g., reporthg a hazardous work situation, exerting pressure on a supervisor to take appmpnate corrective action) fotm a coherent set of behaviours that predict accident frequency rates. Along the same lines, Griflin and Neal (2000) found individuals' howledge of safety matten, motivation to comply with safety standards and motivation to participate in safety matters as fiuther 57
predicton of safety performance. The fïndings support that idea that widening the scope of how organizations measure safety performance to include employee attitudes and behavioun that predict actual injuries may provide them with more usehl information with which they cm enhance future safety.
Within a high performance work system approach, it is also necessary to measure the management practices thought to influence work-related safety in the long-term. As
Pfeffer (1998b)argues, measuring the inputs into a high performance system is critical because doing so provides insight into the functioning of the organization. Therefore, it is necessary to measure, for instance. the extent to which managers display behaviours consistent with transfomational leadership, perceived employment security, the amount of training employees receive. the effectiveness of selection practices, the number of employees working in teams, and the quality of jobs, for 'ivhat gets rneasured gets done" (Pfeffer, 1998b. p. 27). Moreover, in order to encourage these practices they should be linked to compensation. This is impossible without their measurement, thereby, providing
a M e r reason why measurement is cnticai in organizations. It may also be worthwhile to keep mck of the mechanisms that are thought to mediate the relationship between hi& performance management practices and safety behavioun. For instance, measuring employees' levels of trust in management, anèctive cornmitment to the organization, and perceived safety climate rnay provide valuable information on the present condition of the organization with respect to safety. [n sum, the m e a s m e n t of variables cntical to organizational success should be
an important part of any conceptuaiization of a high perfomance work system. In order to manage for workplace sdety efflectively, it is necessary to measure distal causes of 58
safety incidents as well as employee attitudes and behaviours that predict subsequent safe
performance. Measuring these variables will provide organizations with the information necessary to prevent f i e r safety incidents from occurruig.
Chapter Three The Mechanimis by Which High Performance Management Fractices Impact Workplace Safety
Assuming the hurnan resource practices discussed in the previous chapter result in higher levels of occupational safety, the question of how such effects emerge remains to be answered. Following Barling and Zacharatos ( 1999). I suggest that a high peifonnance
work system would have a positive influence on safety performance in as much as the human resource practices would serve to increase levels of trust in management, affective cornmitment to the organization and positive perceptions of safery clirnate on the part of employees. in nim. each of these factors would result in higher levels of organizational
safety. The purpose of this chapter is to discuss each of these mediating links.
Trust in Management Trust is the "willingness of a pariy to be wlnerable to the actions of another party based on the expectation that the other will perfonn a particdar action important to the trustor. irrespective of the ability to monitor or control that other party" (Mayer, Davis. Schoorman, 1995, p. 712). Furthemore, Pfeffer (1998d) argues that trust is the single most important factor when evaluating management practices. " D o they convey and create trust,or do they signi@ distrust and destroy trust and mpect arnong people?' (p. 62). As such, aust is an important mediating link in a high performance work systw.
There are several elements of the high performance work system that will result in trust in management. For example, research shows thai trust in management is a proximal 60
consequence of transformational leadership (Barling, Moutinho & Kelloway, 2001) and mediates the relationship between transformational leadership and follower performance
(Jung & Avoiio, 2000). Pillai, Schescheim and Williams (1999) M e r demonstrate that trust in supervisors mediates the relationship between transfomational leadership and subordinates' organizational citizenship behaviours. Similarly, management systems that promote employment security are likely to heighten employees' trust in management, especialiy during turbulent and unstable times in the organization (Cascio, 1993). Shin and Lee (1999) have shown that this is also mie for organizations in the early stages of a lean production system.
The case can also be made for the roie of sharing safety-related information in the development of employees' trust in management. Fitz-enz (1997) suggests that especially
during dinicult times, information sharhg is likely to result in higher levels of employee trust. Clarke's (1999) study is informative in this regard. She focused on the safety perceptions of senior managers empioyed by British Rail on the one haad, and train
driven on the other. She found that while both groups ernphasized the importance of safety, perceptions of the importance accorded to safety by the other group were inaccurate. In particular, driven underestimated the extent to which managers were concemed with theu safety. in such an instance, it is unreasonable to expect the drivers to mst management where safety-related issues are concemed. The most obvious way of
alleviating this problem and building trust would be for managers to share infornation with employees, specifically about the importance they place on safety. While there does not appear to be any research on the effects of trust on safety performance, more general evidence for the relationship exists. In a revicw of the trust 61
literature by Kramer (1999), he reports that trust results in spontaneous sociability
(Fukuyama, 1995), the behavioural results of which include: cooperation between individuals, acts that extend beyond employee roles, working toward common goals, sharing information and being economical with limited resources. If extended to the realm of safety, these results suggest that employees who are tnisting of management may be more inclined to work in a safe manner, to look out for the safety of fellow employees and to take greater initiative in safety-related matten.
Affective Commitment Employees can be motivated to remain with the organization for diverse reasons.
in some cases, they stay because the costs of leaving are too great or because they feel somehow obliged to stay (Meyer & Allen, 1997). More imponantly, and of greater benefit to the organization, is the situation encountered when employees choose to remain with the organization because they want to (Meyer & Allen, 1997). in this case, they manifest affective commitment to the organization.
Evidence supporting the notion that individuai elements of a high performance system will result in higher levels of affective cornmitment can be found in the literature. For instance, Mathieu and Zajac's (1990) meta-anal ysis shows that leadership predicts overall loyalty, while others have dernonsttated that employees treated with consideration exhibit greater af'fitive commitment (Bycio, Hackett & Allen, 1995). More recent studies bave demonstrated that transformational leadership predicts affiective cornmitment
(Barhg et al., 1996, Barling, Moutinho et al., 2001), as well. Furthemore, Mathieu and Zajac (1990) show that job characteristics predict afféctive commitment to the 62
organization, as does role clarity. A job in which employees exert control and are able to use their skills and in which they know what is expected of them results in employees feeling greater a k t i v e cornrnitment to the organization. Employment security has also been associated with greater affective commitment (Ashford, Lee & Bobko, 1989).
Ln tum,affective cornmimient is an important and significant precursor to work performance (Barling, Moutinho et al., 200 1; Meyer, Allen & Smith, 1993; Meyer, Paunonen, Gellatly, Goffm & Jackson, 1989; Tannenbaum et al., 1991). Presumably, therefore, management practices such as the use of transformational leadership or the provision of quality work would intensify subotdinates' affective commitment to the organization, and would result in enhanced safety performance, as they would other dimensions of employee performance. Kivimiiki and Kalimo (1 993) support this idea in part with data showing that employees who are more comrnitted to the organization see the likelihood of a safety incident as significantly lower than those who are less committed. A study by Hackett, Bycio, and Hausdorf'(1994) more cleariy demonstrates the relationship between affective commitment and occupational safety. The authon studied the perfomance of 80 bus operaton, and demonstrated that affective commitment was significantly associated with the number of accidents occuning over one year.
Additiondly, Parker et al. (200 1) found that jobs allowing employees to work autonornously and in which there was quality communication within the organization led to a greater seme of affective commitment on the part of employees and, in tum, predicted safer working. Cohen and Ledford (1994) provide funher support for the role of management practices in increasing alfective cornmitment and occupational safety. In 63
their study of 169 self-managed and traditionai teams, organizational cornmitment was found to be negatively associated with the numbers of days of work missed following a lost time accident. While organizational commitment was not f o n d to be associated with safety levels in this particular study, this was more iikely because of the range restriction
in the safety data.
ferceived Safetv Clirnate Employees' perceptions of safety climate has been of interest for some time (see Zohar, 1980a). Safety climate is a subset of the overall organizational climate (Coyle, Sleeman & Adams, 1995) and is a refiection of the underlying safety culture (Flin, Mearns. O'Connor & Btyden. 2000). It consists of the perceptions employees have of their work environments with respect to safety policies, procedures and rewards (Grifin & Neal, 2000). Different conceptualizations of safety climate contain different
dimensions or facton which are thought to comprise it. For instance, Zohar (1980a) described safety climate in terms of eight factors. These are the perceived: (i) importance of safety training, (ii) effects of work Pace on safety, (iii) effects of safe work on promotion and (iv) on social statu, (v) management attitudes toward safety, (vi) statu of the safety cornmittee and (vii) of the safety officer, and (viii) the level of risk in the workplace.
Many of the practices associated wi& high performance work systems are hypothesized to result in positive perceptions of safety climate. For example, when management is seen to offer extensive training because it is committed to employee safety, rather than to simply comply with e x t d standards, perceived safety climate is 64
enhanced. Similarly, perceptions of safety climate will be more positive when organizations attempt to provide jobs of high quality, inasmuch as the workloaâ is considered appropriate by employees (Zohar, 1980a). Perceived safety climate is also a critical proximal predictor of safety behaviours.
H o h m & Stetzer (1996) found that employees in a chernical processing plant working in teams with more positive perceptions of safety climate also engaged in unsafe behaviours l e s frequently than did theu counterparts in teams with l e s positive perceptions of safety climate. Neal. Gnfi and Hart (2000) repori that safety climate was related not only to employee safety cornpliance but also to employees' participation in safety-related activities. Further support for the relationship between safety climate and safety performance is provided by Zohar (200 1) who found that safety climate predicted actual injuries in a study of production workers, similar findings to those of Barling, LoughIin et al. (200 1) in a sarnple of restaurant workers.
Overall. there is some evidence in the literatwe that the relationship between
human resource management practices and safety performance could be mediated by trust in management,affective cornmitment to the organization and safety climate. There is linle research examining the roles of mediating mechanisms in the more general hi@ performance work system literature, let aione in the context of a high performance work system in which workplace safety is the performance outcome. The curent reseafch will, therefore, serve to bring some insight into the role of attitudes in such a system, thereby shedding some much needed light upon the manner in which high performance work
practices corne to affect performance outcornes (Doaaid, 1995; Gardner, Moynihan, Park 65
& Wright, 2000; Guest, Michie, Sheehan & Conway, nad.).
Chapter Four Further Refinements to the Conceptual Models
Thus far the ten management practices that comprise a high performance work system have each been described, as well as the impacts that these are thought to have on occupational safety. The mediating mechanisms linking the management practices with safety performance. the final outcome of the mode1 at the employee level. have also been
addressed. There are two issues that remain to be discussed, however-the
interrelateci
nature of the management practices and the specific nature of safety performance. The purpose of this chapter is to elaborate on these two points as well as to describe the manner in which the models will be tested.
Bundles of Practices As mentioned previously, instituting one management practice logicaliy drives the
need for the application of other practices (Pfeffer, 1998a) such that a "'system of HR practices may be more (or les) than the surn of the pans" (Becker & Gerhart, 1996). For instance. organizations that are less cornmitted to providing employment security, relying more on contingent and contract workers, may see little benefit in expending scarce resources on selective hiring or oa training individuals. They may also be less likely to
share valuable, confïdential information with individuals who may subsequently work for the competition. Furthemore, to train employees and then to fail to compensate them for
their newly acquired skills begs the competition to hire them away at higher wages and to
reap the bmefits of the training.
Tearnwork encourages the sharing of ideas and the solving of problems related to work performance. including occupational safety more specifically. To institute a teambased structure without aiso emphasizing information sharing throughout the organization makes little sense. Furthermore. in order for teams to work independently and exert autonomy in decision-making it is necessary for members to have access to the necessary information. An emphasis on information sharing must also go hand in hand with a reduction
in status distinctions, for status distinctions arguably discourage communication between an organizationTsmemben. One way to encourage a reduction in status distinctions
would be through the use of transfomationai leadership. Specifically, nansfonnational leaders have the ability to create a vision of an organization in which there are less disparate groups and one in which employees at al1 levels are treated with qua1 respect. Under these conditions, everyone's safety would take equal priority. Furthermore, training and communication are cntical to job quality. The best jobs will be those in which employees are provided with sufficient information to do their jobs well. Employees also require training in order to develop the variety in skills that are necessary to do quaiity work. The ability to work in teams may also hirther improve the quality of a job.
Lasr, a f i e r role of transformational leadership is especially worth rnentioning. As Pfeffer (1998b)notes, and as discussed in Chapter Two, the leader of the organization
is typicaily the one who chooses the nature of the systern to be implemented Deciding to adopt a high performance work system and to ensure that the venture is successful is a chailenghg proposition that most organkations do not undertake (Ichniowski et al., 68
1996;Ostennan, 1994; Roche. 1999;Wood & de Menzes, 1998). Strong leadership is required at al1 levels in the hierarchy for an organization to implement a high performance work system. thereby M e r illustrating its significance in partnenhip with the other human resource management practices.
The data support the idea that the ten hurnan resource practices being proposed are conceptually and practically interrelated. Wood (1999b)analped the data fkorn 871 establishments in the United States and found that as long as quality circles are excluded Erom the analysis, the data strongly suggest that the practices consistent with a high performance work system are highly interrelated. Analyses of hurnan resource management practices in the United Kingdom replicate this (Wood & Albanese. 1995).In other words, organizations that use some of these human resource practices tend to use
othen as well. Ming's (1995) study takes the analysis of high performance systems, or high cornmitment bundles a step M e r . In a review of @or research on high
performance work systems, he showed that human resource management practices are more likely to yield positive effects for the fimi when they are introduced as part of a coherent system, rather than as single %est practices".
The Nature of Safetv Performance
Thus far, 1 have argued that high performance work systems are associated with improved safety performance. The question remains, however, as to what is really meant by safety perfoxmance. This section is dedicated to discussing this final outcome of the
high performance system in greater detail.
Overall, a high performance system such as the one descnbed in this paper should serve to reduce the number of injuries and fatalities in the workplace. Therefore, 1 argue that a workplace with this type of system in place will have better safety records, in the traditional sense, than an organization without such a system. Traditionally, workplace safety is measured in ternis of the number of injuries and the number of workdays lost as result of these injuries. in the wone cases, the number of
fatalities is dso recorded. Futher data on charactenstics of the involved workers and the injuries or fatalities are also maintained. These data are of the type organizations are typically expected to report to compensation boards and other governing organizations. rtiere are a number of issues with this type of data collection that deserve attention, however. First, under- and over-reporting of the number of illnesses and injuries is quite common (Eisenberg & McDonald. 1988; Pransky et al., 1999). At the organizationai Ievel. there are a nwnber of incentives for falsely reporting injuries as discussed in
Chapter Two. These include avoiding both fines for safety infractions and increased insurance premiums, as well as grievances and work stoppages resulting fiom safety
inhctions. At the individual Ievel, employees also have reason to falsely report injuries. Consider a contingent employee, or in some cases a permanent employee, who may hesitate to report an injury for fear of reprimand of loss of employment (Collinson, 1999). Employees may also be inclined to underreport safety incidents given that only those incidents resulting in injuries are of interest to most organizations and, finthemore, it is
often that case that injuries o d y attract attention h m the organization when they d 70
t
in t h e away from work. While the false reporting of incidents within organizations is of
concern, the point needs to be made that employees have little incentive to faisely report accidents to independent researchen (Grunberg et al., 1996). Second, actual injuries are considered relatively rare events (Veazie et al., 1994) and limiting the measurement of safety performance to injuries aione fails to provide organization's with useful infoxmation for the prevention of future safety incidents. Third, the ability of employees to accurately recall and self-report injuries is questionable. Employees ofientimes fail to recall an incident when asked to report injuries sustained over extensive penods of time (Landen & Hendncks, 1995; Veazie et al., 1994). While far fiom exhaustive, this list of concems with traditional measures of safety performance iIlustrates that the measurement of safety-related incidents in organizations is a chailenge to researchen and that, as researchen, we should be cntical of snidies relying solely on measures such as number of injuries and consequent lost-time as indicators of safety performance. The alternative is to include additional measures of safety performance. Two such
measures are the fiequency of injuries requiring fmt aid and the frequency of nearmisses. Injuries requiring b t aid are those injuries less serious than those requiring t h e away from work. Near-misses are those incidents ui which an injury was narrowly avoided (Hemingway & Smith, 1999). Injuries requiring fint aid as well as near-misses
are panicularly important considerations when measuring safety performance for they occur with pater fiequently than do more serious injuries and injuries quiring tirne away from work. Furthermore injuries requiring tint aid and near-misses may also be predictive of more serious fiiture înjuries. 71
Safety performance can aiso be measured in terms of behaviours consistent with safety compliance and those consistent with safety initiative (Griffin & Neal, 2000; Neal et al, 2000; Williams et al., 2000). Safety compliance refen to instances when employees follow safety-related ales and generally work in r safe manner (Griffin & Neal, 2000;
Simard & Marchand, 1994; Thompson, Hilton & Witt, 1998; Williams et al., 2000). Safety initiative, on the other hand, refers to behavioun in which employees are not simply working within safety standards. but are going beyond compliance and acting proactively to improve safety levels in their environment. For instance, an employee who volunteen to participate in safety audits or who pushes his or her supervisor to take action to improve safety is one that is acting proactively (Andriessen, 1978; G r i f b & Neal. 2000; Simard & Marchand, 1994; Williams et al., 2000).
Safety knowledge as well as safety motivation are M e r factors which have been found to be important to safety performance. Specificaily. both variables have been found to mediate the relationship between safety climate and safety compliance and safety participation. Safety knowledge is a reflection of the extent to which an individual feels that he or she h o w s about safety practices and procedures. Safety motivation, on the other hand, reflects how inclinai or motivated an individual is to perfom safety-related activities and procedures (Griffin & Neal, 2000; Grifnh Neal & Burley, 2000; Neal et
al., 2000). ûverall, safety knowledge, safety motivation, safety compliance and safety participation together reflect an individual's personal safety orientation. Collectively, they form an important measure of occupational safety performance. Safety incidents
consisting of injuries requiring f h t aid and near-misses form a fiuther me72
of d e t y
performance. Ln hand with, and not in place of, lost tirne injuries, these measures comprise a more comprehensive mesure of safety performance. According to Reason et al. (1 998), such broader measures are particularly important because they provide the organization with continuous Feedback-injuries
requking tirne away frorn work and
fatalities are rare events while the persona1 safety orientations of employees and the less severe safety incidents they experience are always present to be measured. As such, they provide usefùl information about the actual state of safety in the organization.
The Models to be Tested Thus far, 1 have been working toward describing a high performance system that 1 have argued will positively impact occupational safety. At the organizational level, this system is comprised of ten practices for managing human resources: the use of selective
hiring and transfomational leadership, the provision of quality work, employment security and training, reduced status distinctions, self-managed teams. as well as an emphasis on information sharing, contingent compensation and the measunment of variables critical to success. It was hypothesized that these practices will be positively related to workplace safety in ternis of fewer lost tirne injuries and days lost at the organizational level. At the employee level, these tm interrelated practices were hypothesized to be positively related to the mediating variables in the model, namely, trust in management, affective commîtment to the organization and safety climate. in
turn, by way of these mediating mechanisms, the human resource management practices were proposed to exert positive effects on safety performance as rneasured in texms of personal safety orientation (i.e., safety cornpliance, safety initiative, safety knowledge 73
and safety motivation). The mediating mechanisms were also hypothesized to be negativeiy related to safety incidents (Le., injuries requiring first aid and near-misses) and lost time injuries (again, refer to Figures 1 and 2). As indicated, there is a paucity of research relating high performance management practices to safety performance. This is a critical factor limiting the likelihood that management will adopt such systems for the optimal management of safety. What is required, therefore, is confimatory data that will encourage management to adopt cornmiment-oriented approaches to rnanaging occupational safety, as well as the organization as a whole. Furthemore, there is a dire need for research examining the mechanisms that mediate the relationship between cornmitment-oriented work practices and occupational safety at the employee level. The overarching purpose of this research was, therefore, to shed some light on the role of high performance work systems in
managing workplace safety and to open the floor to funve research in the area. Two main studies were conducted to test the proposed models and each will be discussed in tum in the following chapters. The purpose of Study One was to examine the proposed model at the organization level. Specifically, the relationship between the high performance work system and lost-the injuries was evaluated. The centrai purpose of Study Two, compnsing both a pilot study and a main study, was to evduate the model at the employee level. The purpose of the pilot study was to develop a mesure of high performance work systems to be used with a sample of fiont-line employees. in the main part of Study Two, ernployees were asked to report on the use of high performance management practices in their organizations using the n w e y developed in the pilot
study, the mediating mechanisms (Le., trust in management, affective commitment to the 74
organization and perceived safety climate) and safety outcomes in tems of personal safety orientation, safety incidents and lost time injuries. It is important to clarify at this point that the high performance management practices proposed as part of both the organization and employee level models were assessed in general as opposed to safety-specific tems. Alternative approaches to evaluating these models exist, specifically, the three options are: to measure both safetyspecific and universal practices. measure safety-specific human resource practices alone, or measure generalized management practices and their effects on occupational safety.
The latter approach was adopted in this case. As an example. transfomational leadership in general was evaluated as p a t of the high performance work system model. as opposed to transformational leadership in
which there was an explicit vision of a safe workplace. Similarly, when evaluating the presence of training, it was the prevalence of training in general that was evaluated rather than safety-related training. There are a number of reasons, both conceptual and methodological, why this approach was adopted. Conceptually, this approach follows fiom Pfeffer's (1998b)argument that performance indicatoa need to be a reflection of the facton the organization considers to be of greatest importance for its success. Additionally, these indicaton can be unique and
can extend outside the realm of cornmonplace financial performance indicators such as r e m on equity or retum on assets. Thus, following this argument, occupational safety is
an appropnate perfomance measure in organizations in which safety is of great consequence, regardless of the fact that safeîy as a performance measure falls outside of the domain of conventional outcomes.
Furthemore, numerous previous studies have successfblly linked high performance management practices to outcomes other than financial performance. For instance, as mentioned previously, Arthur (1994) found that cornrnitment-onented human resource systems were positively associated with employee turnover. Similarly, Huselid (1995) f o n d turnover and employee sales to be associated with high performance work practices such as information sharing, training and compensation contingent on performance. In a sîudy of the hotel industry, Hoque (1 999) found that human resource management practices such as job design, selective hiring, and training were positively associated with service quality, as well as the quality and flexibility of staff. Last, Tsui, Pearce, Porter and Tripoli (1997) report that ernployees in organizations that provided both training and employment security were more likely to exhibit organizational citizenship behaviours. These studies illustrate that performance can be measured in a nwnber of different ways and that high performance management practices can have positive effects on a variety of outcornes, depending on the organization's priorities. Following h m these
examples, improved occupational safety is also a feasible outcome to be expected h m
the implementation of a high perfomance work system. It was the purpose of this research, therefore, to evaluate the effects of nonsafety-specific high perfomance management practices on this unique outcome. While pedormance indicators such as sales per employee, customer satisfaction, Nniover and occupational safety are arguably precurson to more conventional h c i a l measures of performance, following Pfeffer (1W8b) they also stand alone as vaiid
measures of an organization's performance. While traditional financiai measures of 76
performance provide information about what has already occurred, these unconventional performance measures provide information about current conditions within the organization. They also provide organizational members some insight into the reasons why an organization may be perfoming a certain way in terms of more traditional measures. Further evidence for the idea that a high performance work system compnsed of general human resource practices can lead to an unconventional and specific outcome such as occupational safety cornes from the fact that general as opposed to safety-specific high performance management practices have been associated with safety-related
outcomes in the literature. For instance, in a study of semi-autonomous workgroups, Williams et al. (2000) found that general transformational leadership on the part of group leaders predicted group members' safety compliance and safety initiative. Other studies M e r illustrate the value of general high perfomance management practices in improving occupational safety. Geller et al. (1996) found that workers' propensity to actively case for the safety of coworkers was related to workers' sense of belongingness to their team as well as their sense of personai control, the latter being a h c t i o n of job
design. Generalized training can also exert positive effects on occupational safety for training increases perceptions of self-efficacy as well as organizational cornmitment, both
of which are precursors to occupational safety OeJoy, 1996; Tannenbaum et al., 1991). These studies provide a sampling of research in which a relationship was found betwem high performance work practices more genedly and occupational safety as a more specific outcome. What remained was for these individual practices to be tested together in a high performance work system in which occupationai safay was the 77
measured outcome. This was the purpose of the current research. There are also a number of methodological reasons why this approach was adopted in the current research. First, by considering nonsafety-specific hurnan resource management practices and safety-specific outcornes. one reduces the risk of confounding variables and inflated relationships. Retuming to the transformationai leadership example, when a supervisor consistently presents a safe workplace as an important belief or value, a behaviour reflecting safety-specific transfomational leadership, it is neither telling nor altogether surprising to find his or her subordinates reporting they work more safely. Likewise. the same issue arises when considering the process of selective hiring of employees in which safety has been a major focus. An infiateà relationship between selective hiring in which workpiace safety is given a prominent position and empioyee reports of their own safe behaviours is likely to result. Hence, in order to reduce the risk
of confounding independent and dependent variables in this study, the human resource management practices were measured in general as opposed to safety-specific ternis. A M e r reason why the general venus safety-specific approach was adopted in
the curreni research is one inherent to the fact that causality between variables cannot be detemiined. In t e s ~ the g curent models cross-sectionally, it is not possible to determine whether safety perfomance is a fwiction of the hi& perfomance work practices or whether it is safety performance that pushes organizations to adopt these practices. For instance, is it the case that selective hiring results in greater occupational safety or does
poor safety performance trigger the adoption of this practice? While it is acicnowledged that this issue camot be resolved withui the cumnt study, c o n s i d e ~ ggeneral management practices as opposed to safety-specific ones may provide greater insight. 78
In order to illustrate this point consider a study of British workplaces conducted by Wood, Barling, Lasaosa, Turner and Parker (200 1). The researchers found, contrary to
their hypotheses, that health and safety training was positively associated with injury rates. This relationship was explained by the idea that high injury rates likely lead organizations to adopt health and safety training. Had the researchen considered training in general rather than safety-specific training as a predictor of injury rates, this alternative explmation would have been less plausible for it is unlikely that high injury rates would have resulted in organizations adopting more training in general. in this case, a more general predictor would have drawn into question the alternative hypothesis as a possibility, while admittedly still not answering the question of causality. Nevertheless, this evidence provides a funher reason why generalized high cornmitment management practices constitute the predictor variables in the current study-they
may be somewhat
informative. Inherent to the use of generalized management practices as predictoa in this model, is the assumption that good general management will have safety as a focus if safety is important in the organizational context. Following from this argument, it is highly likely that a mesure of a non-safety-specific management practice will be highly correlated with its safety-specific translation. Data from the Australian Workplace Industrial Relations Survey (Department of Workplace Relations and Small Business. 1997) support this notion. It was found that the provision of training in generd was
significantly correlated with the provision of occupational heaith and safety training. Furthemore, universal training was more strongly correlated with lower reported injuries
than was safety-specific training (J. Barling, personal communication, October 2000). 79
This is most likely the result of either multicollinearity between the wo highly correlated training variables or a fuRher example of the effect reported above in the Wood et ai. (2001) study in which injury rates may have predicted the adoption of safety training. in essence, these data demonstrate that generalized predictors may be highly correlated with their more specific counterparts, therefore acting as proxies for these more specific variables. A M e r reason why the high cornmitment management practices to be assessed
in this mode1 were measured in generalized terms stems from the fact that al1 ten predictor variables could be rneaningfully measured in general terms, while the same ten predictors could not al1 be measured in safety-specific terms. The provision of employment security and job quality, as well as reduced statu distinctions could not be logically reinterpreted to reflect a safety focus.Therefore. in order to remain consistent across variables, the general focus was adopted. As mentioned above, 1 recognize that there were altemate approaches to
conducting this research. However. for the concepnial and methodologicai reasons just described I chose to test this mode1 at the most general Ievel and leave the testing of its alternate fonns to future research.
Chapter Five Study One
Numerous prior studies have examined the relationship between high performance work systems and diverse measiues of organizational perfomance including sales per employee, turnover, productivity and financial performance. These relationships have been evaluated at the organizational level most typically by mailing s w e y s to a set of organizations inquiring about their adoption of hi& perfomance work practices and the outcomes of interest (see for instance, Arthur. 1992, 1994; Hoque. 1999; Huselid. 1995). When available, data on the outcomes of interest are denved nom preexisting databases in order to minimize the chance of monomethod bias.
Huselid (1995) represents a typical hi& performance work system study of this
nature. ïhe author mailed s w e y s to the senior human resources professional in 3,452 fims inquiring about their fimi's adoption of high performance work practices and the
fim's tunover rate, one of the performance variables he considered. The s w e y responses from each organization were then paired with indicaton of fim financial performance, which were denved from a hancial database of publicly held American companies. The current study followed this approach with the exception that ùiformation about both the high performance work system and the safety outcomes was collected tkom organizational representatives. This was necessary given none of the d e s i d idormation was publicly available. Following Huselid (1995), a number of variables w e n controlied for in the regression analyses. Whether or not the organization was a subsidiary of a larger 81
organization, whether it belonged to the public or pnvate sector, the age of the organization, the total number of employees m.d whether or not the front-line employees were unionized or not, were al1 factors controlled for in the analyses. In his study of high performance work systems, Huselid (1 995) also factor analyzed the high performance work practices he hvestigated in order to determine their underlying factor structure. This was done in the current study as well.
Purnose The purpose of this first study was to determine whether a relationship e x i s ~ between the ten human resource practices descnbed within the high performance work system and occupational safety at the organizational level. It was hypothesized that organizations having more extensively adopted these high performance management practices would expenence lower numben of lost time injuries when controlling for demographic variables (see Figure 3). Given the amount of conceptual overlap between each of the high performance work practices outlined, a m e r purpose of this study was to detemine the underlying factor structure of the ten human resource practices. Before proceeding with this study, it was fint necessary to design a survey instrument measuring the extent to which participating organizations had adopted the ten
high performance management practices. This was required given a satisfactory instrument measuring each of the practices to be evaluated in this research did not exist prior to this study.
U . .
..
.
m
.
.
.
.
Taking into consideration the various dimensions of each human resource practice, 1 developed 63 s w e y items coIlectiveiy measuring the ten high cornmitment practices. These items were pretested with a number of human resource and safety
professionals as well as colleagues. in the case of transfomational leadership, items were adopted from Bass and Avolio (1995). The fmal items categorized by human resource practice cm be found in Appendix A.
Partici~ants Surveys were mailed to both the Health and Safety Manager and the Director of
Human Resources of 1471 cornpanies in Ontario, al1 memben of the industrial Accident Prevention Association (IAPA) in February 2001. Surveys sent to Human Resource Directon included three things: (1) demographic questions, (2) the survey measuring
high performance management practices and (3) questions about the number of lost t h e injuries and days lost due to injuries the organization had expenenced in the previous calendar year (Le., January 1.2000 thmugh December 3 1,2000). Surveys were sent to
Health and Safety Managers in order to check the reliabiiity of the safety data provided by the Human Resource Directon. As such the s w e y sent to Health and Safety
Managers consisted of two things: (1) demographic questions and (2) questions regarding lost Ume injuries and days lost due to injuries for the previous calendar year. This survey did not include questions investigating high cornmitment work practices. Completed s w e y s were received fkom 147 Directors of Human Resources (response rate = 10.19%) and 247 Health and Safety Managen (response rate = 17.13%). Respouse rates were calculated f i e r having accounted for the 29 sets of surveys that were retumed 84
undeliverable. Completed s w e y s from both the Human Resource Director and Health and Safety Manager were received from 33 organizations. It was possible based on these
paired surveys, of which 32 were usable, to compare the responses provided by the Health and Safety Managers and Human Resource Duecton on the safety-related questions. It was found that the responses provided by both parties correlated sufficiently
highly to conclude the safety data provided by the Human Resource Directon was reliable. The correlation between Human Resource Director and Health and Safety Manager reports of the total number of lost t h e injuries was 0.86 @ < .001, N = 32) and for the total number of days lost due to injuries it was 0.78 @ < .001, N = 32). The remaining analyses were conducted with the data provided by the Human Resource Directors. Of the 147 surveys retumed by Human Resource Directon, 138 were usable. in four instances, Human Resource Directon failed to provide safety data. in these cases, the safety data for the given organization were taken from the available Health and Safety Manager responses. The organizations in the final sample belonged to a wide range of industries, including chernical, automotive and construction. Private sector h
s constituted 87.69%
of the final sample. The organizatioas had been in existence for an average of 43 years (So = 3 1.88; range = 1.50 to 175 years) and had a mean number of 5 15 employees @Q =
828.75; range = 35 to 7465 employees). The vast majonty of the organizations' employees worked Full-time with 3.78% working part-tirne and 1.57% working on a contract basis. The rnajority of workplaccs had non-unionized front-line employees 85
(56.52%), another 34.06% had unionized front-line employees and the remaining 9.42%
of workplaces had both unionized and non-unionized front-line employees. Of the organizations in the final sample, 62.69% were subsidiaries of larger organizations. The
Human Resource professionals who responded to the survey had been in their positions
(m= 5.32; range = -17 to 30 yean) and with the organization an average of 8.03 yean (m= 8.53; range = 2 5 to 39.1 7 years).
an average of 5.05 years
Materials The matenals for this study consisteci of two sweys. The fint s w e y was sent to the organization's Director of Human Resources dong with two cover letten. one from rnyself one from the President and CE0 of the N A , both encouraging the respondent to participate in the snidy (see Appendix B for both cover letters). As mentioned earlier,
Human Resource Directors were asked to answer demographic questions and repon on the high performance management practices and injury rates in their organizations (see Appendix C for the actual questionnaire sent to Human Resource Directon). Health and Safety Managers were only required to respond to demographic questions and provide injury data (see Appendix D for the acnial questionnaire sent to Health and Safety Managers). Yiph oerformance manapemmt ~racticeswere measured with the 63 items I
developed for this purpose (again, refer to Appendix A). Fi@-five percent of the items were statements describing individual human resource practices to which participants provided the extent to which they felt this practice describeci their organization. An example of this form of item is '"This organization makes a point of keeping track of 86
factors that it considers critical for success". Responses to this type of question were on a five-point Likert scale ranging from 'strongly disagree' (1) to 'strongly agree' (5). A higher score reflected more extensive adoption of the high performance work practices at the organization level. For the majority of the remaining questions, respondents were asked to estimate
the percentage of employees in their organization to which a statement descnbing a high commitment management practice applied. This is a similar form of question to that employed previously in this type of research (see, for instance, Arthur, 1992; Becker & Huselid, 1998; Huselid, 1995). For instance, "What percentage of front-line employees received naining beyond that mandated by govement regulations in the last 12 months?" is a sample question. Yet other questions required specific answen such as the highest and lowest salaries and the number of hours of training received by fiont-line employees. In order to conduct analyses based on these varied forms of responses, al1 scores were standardized. Sample items for each of the ten high commitment work practices follow. Uniess otherwise indicated the responses were on the five-point Likert scale described above (Le., (1) = 'strongly disagree, (5) = 'strongly agree'). A sample item measuring employment security was, "Providing employment security to our employees is a pnonty
in this organization", while a sarnple item measuring selective hiring was, "Only the best people are hired to work in this organizationV*. Training was measured with items such as, "What percentage of your organization's fkont-he employees received training beyond that mandated by govemmcnt regulations in the last 12 months?', to which respondents
provided an estimated percentage of employees. Teams and decentralized decision87
making was measured with items such as, ''The development of teams is an important element of this organization's corporate strategy", while the following item, "Some memben of this organization have pnvileges that are mavailable to other members" (reverse scored) measured statu distinctions. information sharing was measured with an
item such as, "What percentage of the workforce is included in a formai Somation sharing program (e.g., a newsletter, emails)?" to which respondents again provided a percentage of employees. A sample item measuring contingent compensation was, ''The way in which employees in this organization are compensated encourages them to adopt a long-terni focus". Items measuring transformational leadership were adopted from Bass and Avolio (1 995) and a sample item was, "Leaders in this organization provide
employees with continuous encouragement". Job quaiity was measured with items such
as this: "Many employees in this organization perfonn simple and repetitive tasks as part of their work" (reverse coded). Last, the measurement of variables important to success was measured with items such as, " m a t percentage of the workforce completes attitude surveys on a regular basis?" to which respondents provided the percentage of employees.
Iniury rates. Respondents were asked to report the number of lost time injuries and total number of days lost due to each of eight specific types of injuries: (i) fractures, (ii) dislocations, sprains and strains. (iii) bruiskg and crushing, (iv) scratches and abrasions, (v) cuts, lacerations and puncnires, (vi) burns and scalds, (vii) eye injuries, and (viii) concussions and other head injuries. Respondents were also asked to report on the number of fatalities that occurred, if any. The questiomaire was sent out in Febniary of 200 1, and respondents were asked to report safety statistics for the previous caiendar year, January 1 to Decernber 3 1,2000. 88
ResuIts
Reliability coefficients for the scales measuring each of the ten high cornmitment management practices were calculated and those items that most negatively impacted the intemal consistency of each scale were removed. The resulting scales had the following reliability coefficients and numben of items: employment security (a = .63, number of items = 5); selective hiring (a = $57.nurnber of items = 5): training (a = .72, number of items = 5); teams and decenaalized decision-making (a = .8 1. nurnber of items = 6 ) ; reduced stanis distinctions (a = 39. number of items = 3); information sharing (a = .78, nurnber of items = 6): contingent compensation (a = 52, nurnber of items = 3); transformationai leadership (a = 3 3 . nurnber of items = 6);job quaiity (a = -77,nurnber
of items = 6): and measurement (a = .66. number of items = 7). While some of the scale reliabilities are quite low, it is worth noting that the number of items compnsing the scale cm have a significant impact on the scale reliability. For instance. the scale rneasuring contingent compensation was found to have
a reliability of a = 5 2 with three items. Using the Spearman-Brown formula (DeVellis. 1991), it is possible to calculate the reliability for the contingent compensation scale had it been comprised of more items. With six items, the reliability of the contingent
compensation scale rises to .68;with nine items, it becomes .77. This example demonstrates the sensitivity of a scale's coefficient alpha to the number of items. While some of the scales rneasuring high performance work practices have arguably poor reliabilities, this may be accounted for in part by the mal1 item numben. The first analysis conducted on these data had as its purpose u n c o v e ~ gthe
underlying factor structure of the ?enhi& cornmitment management practices. 89
Descriptive statistics and intercorrelations between the ten high performance management practices are found in Table 2. For each scale, the sum of scores on each item was calculated and the factor analysis conducted on the sumrned scale scores. Following Comrey (1978), this approach was chosen in order ensure the factor analysis was conducted on the most nomally distributed, continuous and measurement error-fiee
data possible. Using principal components extraction with listwise deletion a single factor emerged fiom the data. The results are displayed in Table 3. This single factor comprising a high performance work system accounted for 52.21 % of the total variance (Eigenvalue = 5 .Z)'. As the ten practices consisted of a single factor, that is to Say, a high performance
work system, I combined al1 ten subscales to conshuct a 52-item scale measuring the high performance work system for use in the next step, the regression analysis. 1
calculated the intemal consistency of this scale using the ten subscales as items to be 39. Next, a hierarchical regression analysis was conducted on these data Descriptive statistics and intercorrelationsof al1 variables are presented in Table 4. There are a number of assumptions that must be satisfied in order to conduct regression analyses.
Fint,the relationships between al1 independent and dependent variables must be lineu.
5
A factor analysis was &O conducted on the 52 individual items measuring aU tcn high performance work practices and similar rcsults to thosc rcported wcre fouad Spccrfically, using principal components extraction with listwisc dcletion a single factor emcrgcd h m the data. This single factor accomtcd for 28.08%of thc total variance (Eigenvalue = 14.60).
Table 2
Descriniive Statistics and Intercorrelations Betwccn Measurcs of the Ten Hidi Performaiice Management Practices in Studv One
Variable
1, 2. 3. 4. 5, 6. 7. 8, 9 10,
Eniployment Security Seleciive Iliring Training l'cams Status Distinctions Information Sharing Contingent Compensation Transforniational Leadership Job Quality Mcasurcmenl
la
SD
.O5
3.O8 3.01 3.30 4.23
.17 .O 1 -.O2
-.O1 .O 1 -.O2 -.O2 .O 1 .O5
2.23
4.07 2.12 4.38 4.05 3.97
1
2
.3 1** .23*+ .2îi** -21'
.35*+
.28**
.44*+
.28++ .2S*
.42*+ .48** SB** .48+*
.28**
.30**
.55+*
.44**
3
.52*+ .2O* .47** .29**
.35+* .33*+ .49**
4
5
.56** .67** .52*+ .70** .77** .69**
.52*+ .41°* .SO*+ .!il*+ .45**
6
.46** .56++ 38.. .67**
7
8
9
.44*+ .47++
.43**
.58** .58**
34'"
10
Table 3 Factor Loadin~rsof Ten High Performance Management Practices in Studv One
Factor Teams and decentralized decision-making Job quality Information sharing Measurement Transfomational leadership Selecrive hiring Reduced status distinctions Contingent compensation Training Employment security
Factor Loading Component 1 .89
m m -
=07
From visual inspection of the scatterplots, it was concluded that the assurnption of linearity was adequately met. Further assumptions of linear regression are that the variables are normally distributed and have homogeneous variances. These assumptions were not adequately met in these analyses as two of the variables were positively skewed (the total number of employees and the total number of lost time injuries) and the variances between variables were not homogeneous. Nevertheless, analyses were carried out on these data given the robustness of the regression technique (i.e., regression results are not greatly affected by violations in assumptions) and, for reasons of ecological validity. More specifically, the data are a reflection of the phenornenon in question. In the case of lost time injuries in particular, it is important to note that this variable is not normally distributed in the general population-most
organizations report fewer than
more injuries. Transforming these data would change their inherent meaning and, therefore, was deemed inappropriate. In the case of nurnber of employees, the data are also skewed given the number of outliers: 125 of the 138 companies in the final sample had between 35 and 999 ernployees, the remaining 13 companies had between 1O0 1 and
7500 employees. The results of the regression analysis for the totai number of lost time injuries are presented in Table 5. In Step 1, the demographic variables were entered into the equation. These were: whether the organization was a subsidiary of a larger organization, whether it was belonged to the private or public sector, the age of the organization, whether the front-line workers were unionized, non-unionized or both and, last, the total number of employees. None of the demographic variables with the exception of the total nimiber of employees were related to the totai number of lost M i e injuries. It was foumi, not 94
Table 5 Reeression Results Predicting; Total Lost Tirne Injuries
Predictor Variable Step 1 Subsidiary " Pnvate/public Age of Organization Unionized Total number of employees Step 2 High performance work system Total
Step 1
Step 2
-.12
-.O8
-,O3 .O0 J8
-.O4
.3g+* -
.39**
.O 1 .17*
-.32** .13**
.23**
Note: Standardized regession coefficients are reported. 1 = subsidiary; 2 = not a subsidiary 1 = private; 2 = public 1 = unionized; 3 = non-unionized; 3 = both unionized and non-unionized
surprisingly, that Iarger organizations in ternis of employee number, experienced more lost tirne injuries than did smaller organizations.
In Step 2, the variable measuring high performance management practices was entered into the equation. As predicted, high performance work systems were found to be positively related to safety performance measured in ternis of fewer lost time injuries. in other words, organizations having more extensively adopted high performance work systems also expenenced fewer work-related injuries. As well, high performance work systems accounted for 10% of the variance in lost t h e injuries, after controiling for demographic variablesb. Regression analyses were not conducted the number of days Iost due to injuries reported by participating organizations, for the variable provided little additionally valuable information. This is on account of the fact that the number of lost time injuries and number of days lost due injuries were very highly correlated one with the other & = -73, E < .O01. N = 1371,most likely given the number of responses in the lower ranges.
Data were also collected on the number of occupational fatalities, of which three were reponed. Given the rarity of this event, analyses were not conducted on these data either.
Regession analyses were also conducted on lost time injury &ta that were transformed for normality. The resuits of these anaiyses indicated that the concrol variabIcs entered in Step 1 accountcd for 112%of the variance in Ln.Only the total numba of employcts accounted for a signifîcant proportion of the variance in L ï i (J$= 30. p < .01). In Step Z it was found that the control variables in conjunction with high performance work systems accounted for 30.9% of the variance in lost timc injuries. As in Step 1, ody the cotai number of employees provcd a significant prcdictor. Overail, the regesion analyses conducted on aonnally M b u t e d Iost thne injury data provide fintbcr support for the rciationship between hi& performance work systcms and safcry performance. Aftcr c o n m i h g for dcuxograpùic variables, it was found that hi& performance work systcms accountcd for 19.7% @ < -01)of the variancc in lost cime injuries. 96
Discussion The aim of this study was to examine the underlying factor structure of the ten human resource management practices that comprise a hi@ performance work system. A M e r purpose of this study was to uncover whether organizations adopting a high performance work system approach to managing human resources experienced fewer lost time injuries than did organizations applying a more traditional approach to managing human resources. It was found that the ten hi& performance management practices, namely, the use
of selective hinng and transfomational leadership, the provision of quality work, employment security and training, reduced status distinctions, self-managed tearns, an emphasis on information sharing, contingent compensation and the measurement of variables critical for success, were reflective of a single underlying construct-the
high
performance work systern. This finding validates the idea presented in Chapter Four Fat the practices are highly interrelated one with the other forming a 'bundle' of practices
that are commonly applied together. This finding also supports the research of Wood and Albanese (1995), who found 14 human resource practices represented a single latent variable-high
cornmitment management. in the tenns used in the current research, these
14 variables amount to naking, selective hiring, measurement, job quality, information
sharing and employment security. Similarly, Ichniowski et al. (1997) found that practices reflecting, temwork, employment security, selective hiring, idonnation sharing and job quaiity were highly correlated. As the authors did not factor analyze these practices it is not possible to conclude what the underlying factor structure of these practices was.
Nevenheless, the results of this study support the findings of lchniowski et al. (1997) to the extent that the correlations between the ten practices considered in the current study were also high. The results of the factor analysis do not support the findings of Huselid (1995) who found the high performance work practices he examined comprised two
separate factors-those
that serve to develop employee skills and those that impact
employee motivation, but do support a later study by Huselid and Becker (1996) in which a single factor comprising a high performance management system was found.
When drawing cornparisons between the findings of this study and those of past research, the recommendation of Becker and Huselid (1998) m u t be considered. They argue that the results of those few studies in which a factor analysis was conducted on
human resource management practices should be interpreted with caution. The reason given is that the measures ernployed in these studies were not broad enough to capture the complexity of these practices with many of the measures used typically comprising a single item and, in some cases. two items per practice. The authoa suggest that a meaningful factor analysis cm only be conducted on data derived fkom more comprehensive measutes of human resource practices. The measures used in the current research satisb this criterion and therefore it can be argued the analyses conducted here provide more meaninghil results.
h was also found, at the organization level, that the adoption of a hi& performance work system is associated with greater workplace safety. In other words, to the exteni that firms ensure employment security, selective hiring, training, and job quality, provide their employees with the opportunity to work in teams, access to information and contingent compensation, and emphaske transfomational leadership, 98
measurement and reduced status distinctions, they expenence fewer workplace injuries.
Advances
This study, therefore, makes a number of significant contributions to the research.
Fim, in conducting the study, 1 have created an organization level measure of high performance work practices that more broadly captures the intricacies of each of the ten practices than have previous instruments. Furthemore. given this measure of the adoption of a high performance work system follows the recommendations of Becker and Huselid (1998) in that it includes multi-item measures for each practice, the factor analysis conducted as part of this study should provide more meaningful results than those of previous studies. The results of this study also provide strong support for the proposition that the adoption of high performance work practices at the organization level is associated with greater workplace safety. This fuiding supports earlier work on high performance work systems. which argues that these systems of work practices are associated with greater employee and firm performance. What is unique about the hdings of this study is that 1 have demonstrated that what constitutes organizational performance extends beyond W t i o n a l measures such as productivity and profitability. in fact, workplace safety cm be conceptuallled as a further indicator of firm perfomance. An implication of this
finding is that adopting a high performance work system with the intent of improving hancial performance or sales per employee. for instance, may additionally prove to have
a positive effect on workplace safety. The same rnay hold true in reverse. An organization adopting such a systern with the intent of impmving workplace safety may derive other 99
positive benefits, as well. More importantly, these findings provide much needed evidence for the belief that management practices play a cntical role in worlcplace safety and the magnitude of this relationship is particularly noteworthy. High performance management practices were found to account for 10% of the variance in lost t h e injuries expenenced at the organizational level. This finding suggests that managers can no longer look to employees as the source of workplace injuries, rather their own actions play a significant part in how safely employees work.
In fact, managers seeking to maxirnize workplace safety may denve the greatest benefits h m adopting a commitmentsnented approach to managing the workforce, one based on respect for the worker and his or her abilities. Essentially, treating employees as a valuable resource and ensuring they receive adequate aauiing, full information and quality work, as well as providing them with employment security, reduced status distinctions, the opportunity to work in teams and contingent compensation, among other things, may have a significant impact on their ability and desire to work safety. On the other hand. a control-orientated approach to managing human resources relying on the use of d e s and reguiations, may not be the most desirable and effective means of promoting workplace safety.
Limitations
Several methodological limitations of this study should be considered. First, the data analyzed in this study were cross-sectional in nature, thereby m a h g it impossible interpret causality between the variables being examineci. The ability to make causal 100
inferences is a fùnction of study design rather than data analysis. Despite the manner in which the relationships between variables were conceptualized. longitudinal research is necessary if one is to conclude that high performance work systems result in fewer losttime injuries. Second, a single organizational representative provided the data regarding the existence of high cornitrnent management practices in their respective Company as well
as the firm's overall safety performance. This is potentially problematic for three reasons: (1) there is no way to know whether the Human Resource Directors who filled in the survey were biased in their responses in order to portray their organizations in the most
positive light (2) the perceptions of the human resource management practices were those of the Human Resource Director alone, and (3) this approach increases the probability of mono-method bias. While al1 valid concerns. there are reasons why the impact of these limitations on the study results may be minimal. First, there is no reason to believe the human resource directon were biased in their responses as they were assured the anonymity and confidentiality of their responses. Furthexmore, responses to questions regarding human resource practices covered the full range of possibilities, h m most negative to most positive. With respect to the second point, I acknowledge that the rathgs of human resource practices corne into question given they are based on the perceptions of a single individual, however, this methodology follows fkom other studies in the area (see for instance, Huselid, 1995). In ternis of mono-method bias, one aspect of the research design may quel1 these concerns. Respondenu were asked whether the safety data came
kom actual records. It is presumed that such data would be free of respome bias. In 101
93.44% of the cases, participants claimed the data did in fact come fkom actual records.
Tkis of course. says nothing of the quality of the records the organizations keep. Last, the nsk of mono-method bias is limited by the fact that responses to questions were both
positively and negativeiy worded and were on different scaies (i.e., Likert-type scales and percentages) (Cook & Campbell, 1979). A M e r limitation of this study is the fact that safety performance was only
measured in terms of total lost time injuries. As mentioned earlier, these data come into question to the extent that they underestimate the nurnber of actual injuries (Eisenberg & McDonald, 1988; Pransky et al.. 1999). A better indicator of safety performance would have also considered micro-accidents-those
injuries that require a visit to the uifimiary
but do not require time off of work (see, for instance, Zohar, 2001). Such a measure would have provided a much more accurate picture of an organization's safety performance. Unfortunately, companies are not required to keep such data and therefore records of these more minor safety incidents do not typicdly exist. The smail sample site is a funher limitation of this study is. The final data set was based on responses nom 9.57% of the organizatious in the initial sample. in a review of the literature, Becker and Huselid (1998) report response rates for similar studies to be between six and 28%. The authoa report h m their ovm experiences that response rates
are dwindling as h
s receive more and more surveys and employees have less Mie to
respond. While the curent nsponse rate is not out of line with those reported in similar studies, it does lend some question to the generalizabilityof the results. 1 am unable to
detemiine whether the firms that responded somehow differ h m those in the general population.
A h a 1 limitation of this study was the fact that in the final 52 item scale may
have over-represented sorne practices that comprise a hi@ performance work system, while under-representing others. For instance, in the final scale, three items each measured reduced status distinctions and contingent compensation while the practice of measuring variables critical to organizational success was evaluated using seven items. It is possible that the factor analysis and even regression results could have been affected by this imbalance in item numbers.
Conciuding Thouehts
Despite these limitations. this study notably extends our understanding of the manner in which hi& performance work practices are 'bundled* together, providing
support for a single high performance work system factor. This study also provides evidence for a significant relationship between high performance work practices and occupational safety, thereby extending our understanding of the role these systems play
in workplace safety. The implications of these fmdings and avenues for fiinire research will br discussed in greater detail in Chapter Eight.
Chapter Six Study Two- Pilot Study
Having found a relationship between the hi& performance work system and occupational safety at the organization level in Study One, the question rernained: How do these practices exert positive effects on workplace safety? In order to answer this question it was necessary to consider this relationship at the employee as opposed to organization level. Specifically, it was necessary to measure employees' perceptions of
the ten work practices comprising the high performance work system in their organizations. It was dso necessary to investigate the mediating rnechanisms. trust in management, affective commimient to the organization and perceived safety climate, that
I argued relate these hi& commitment practices to safety performance. Before being able to investigate these relationships at the employee level however, it was necessary to create a s w e y instrument measuring employee perceptions of the ten high performance management practices as no such survey was available fiom the l i t e r a ~ eDoing . so was the aim of this pilot study.
Pumose n i e purpose of the pilot snidy was the development of s w e y items measuring the ten high commitment management practices to be useâ at the employee level. As was the case when evaluating the practices at the organizational level, a satisfactory measure
did not h d y exist.
Particiaants Surveys were distributed to 349 individuals falling into two categories: they were either recent graduates of the University's MBA program and their peen (255 individuals), or employees of the University's Physical Plant Services Department (94 individuals). It was required that participants be employed full or part-tirne, but not selfemployed in order to participate.
Of the initial sample, 142 voluntarily participated in this pilot study. The participants in the pilot study were 70 female and 72 male employed individuals. The participants had a rnean age of 32.39 years
= 9.33; range = 19 to 59 yean) and had
been working in their current organization for a mean of 5.5 1 years
= 8.14;
range =
.O8 to 33 yean). The mean length of t h e participants were in their current positions was
3.29 years (Se = 5.77; range = .O8 to 33 years). The response rate for this study was 40.69%.
Materials Materials consisted of a single questionnaire that was completed by the participants. The questionnaire consisted of limited demographic questions as well as items developed to measure the participant's perceptions of the extent to which their organization had adopted the ten hi& perfomance management practices being investigated. Between eight and ten items addressed each of the ten practices for a total of 82 items. See Appendix E for a listing of pilot s w e y items categorized by management
practice and Appendix F for a copy of the web-based pilot swey.
Items measuring employment security were adopted fiom Kuhnert and Vance (1992) with, "If 1 were to lose my current position, my organization would by very hard
to place me in another position elsewhere in the organization" being a sample item. A sarnple item measuring selective hiring was, "When new employees are hired, they must go through an extensive hiring process in which they are h t e ~ e w e da number of times".
Items such as, "The Company provides enough training for me to leam new ways to do my job" were used to measure training. Tearns and decentralized decision-making was measured with items such as, "My organization places a great deal of importance on team development**and items such as, "1 have the opportunity to interact with top management in my organization" measured statu distinctions. Monnation sharing was measured with items such as the following: "1 am given encugh information to understand my role in this organization". Items such as. "Part of my compensation is based on how well my workgroup or department performs" measured contingent compensation. Transformaiional leadership was measured with items adapted corn Bass and Avolio (1995). an example of which is, "My supervisor gets me to look at problems h m many
different angles". Job qudity and meanuement of variables critical to success were measured with the following items: "1 know what is expected of me at work" and "My organization makes a point of keeping track of factors that it considen critical for success", respectively. Ail responses were on a 5-point Likert-type scale ranging fiom 'stmngly disagree' (1) to 'strongly agree' (5). A higher score reflected the perception that the organization had more extensively adopted the ten human resource practices.
Procedure The pilot survey was conducted electronically. Potential participants were sent a letter by electronic mail encouraging them to participate in the study and directing them to the Uniform Resource Locator (URL) where the survey was posted.
Results
Following the procedure outiined by Churchill (1979), the items measuring the ten high performance management practices were refined in order to create the best possible measure of the construct. Items with iow correlations to the whole scale were removed in order to improve the internai consistency of each scale. Items with distinctly non-nomal responses were also removed from each scale. The final 5 1 items m e a s d g each of the ten practices can be f o n d in Appendix G. The resulting scales had the
following reliability coefficients: employment security (a = .70); selective hiring (a = .78); training (a = 3 6 ) ; teams and decentraiized decision-making (a = .78); statu
distinctions (a = .79); information sharing (a = .75); contingent compensation (a = .68); transfomational leadership (a = 39);job quality (a = .76);and rneasurement of variables
critical to success (a = .82). All scales had five items with the exception of information sharing that was comprised of six items. The data h m this pilot study were factor anaiyzed in order to uncover the under1ying structure of the variables. The descriptive statisacs and intercorrelations for
the ten high performance management practices are found in Table 6. Again, following Cornrey (1978), for each of the ten high performance management practices, the scores 107
Table 6 Descriptive Statistics and Intercorrelations Between Measures of' the Ten Iiigh Perfonnaiicc Manarenient Practices in the Pilot Study
Variable
1.
2. 3,
4. 5. 6,
7. 8. 9.
10.
Employment Security Selective Hiring Training Teams Siatus llistinclions Information Sharing Contingent Compensaiion Transformational Leadership Job Quality Measurcment
hl 29.15 29.4l 31.66
SI2
26.28
4.85 5.48 6.35 4.79 5.31
27.23
5 ,O4
21.53 28.74 28.20 34.23
3.91 5.82 4.56
26.M
6.70
1
2
3
.31 ** .49** .He+ .43** .46** .Sa*+ .38** .54** .50e* .SI** .49** .67** .19* SI** .43** .36** .46** .51+* .39** .44** 3 4 ' . .42*+ .47** .56**
4
5
6
7
8
9
1O
.69** .71**
.66*+
.29** .30**
.45** 6 .27** .74** .71** . X e + .61** .60** .54** .60** ,29*+ .60** .47** .65+* ,64+*
.64**
4
on items measuring the given practices were summed. These surnmed scores were then
used in the factor analysis. Using principal components extraction with listwise deletion, it was found that the ten high performance work practices reflected a single factor accounting for 55.99% of the variance (Eigenvalue = 5.60)~.See Table 7 for the results of this analysis. Given the ten practices comprised a single factor, 1 combined al1 5 1 items to form a single measure of a high performance system. The intemal consistency of the scale calculated with the ten subscales as items was a = .9 1.
Discussion The purpose of this pilot study was to create a measure of the extent to which a high performance work system had been adopted in an organization from the employee perspective. A M e r purpose was to investigate the underlying factor structure of the hi& performance work system. The fint of these aims was achieved by distributing the pilot items to a sample of individuals. and then refining the items based on intemal consistency. The result was ten scales measuring the individual high performance management practices. These scales were then factor analyzed and it was found that they represented a single cornmon factor, namely, the high performance work system.
' A factor analysis was ako conducted on the 5 1 individua1 items measuring all tcn high performance work practices and similar results to those reportcd were found Specificdiy, uing principal cornponcnts extraction with listwise deletion a single factor emcrged h m the data. This single factor accounted for 28.4 1% of the total variance (EigcnvaIue = 14.49).
Table 7 Factor Loadings - for Scales in Pilot Studv -
Factor
Information sharing Tearns and decentralized decisionmaking Reduced status distinctions Job quality
Training Transformational leadership Measurement Selective hiring Employment security Contingent compensation
Factor Loading Component 1 .86 .83 .83 .79 .77 .77 .75 .71 .60 .53
Advances These results extend current research in two important ways. Fint, the development of a scale that can be used to measure the extent to which employees believe their organization has adopied a high perfocmance work system was wanting fiom the literature. Such a scale is necessary for this type of employee level research to continue. This is of course not to Say that the scale developed in this study is without flaw but it does form a basis upon which future scales can be developed. The current findings also extend research in that they provide us with m e r understanding of the complementarity or 'bundling' that occun between human resource management practices within a hi& performance work system. The results of this pilot study suggest that much like at the organizational level, the ten practices as perceived by employees reflect a single underlying factor. Overall, this provides M e r evidence that the ten practices are conceptually and practically interrelated, though this finding should be interpreted with caution for the reasons that follow.
One aspect of the study methodology is particularly noteworthy-the use of a web-based survey. This technique has the added advantages of being a less expensive and more efficient means of collecting data. The greatest efficiency gains came fiom almost effortless distribution of surveys and the fact that manual data entry was not necessary. A disadvantage of this technique is that substantial technical knowledge required in creating a web-based survey. Fortunately, there is little reason to question the data derived h m web-based surveying. in a review of recent conference presentations made on the topic of web-based sweying, Kraut (2001) concludes that there is no difference between responses on paper-and-pend and web-based sweys. The method used to collect the 111
data does not appear to affect the results.
Limitations There are some limitations that must be acknowledged regarding this pilot study, and primarily concem the sample used. The sample was only in part representative of the
sample with which the instrument was to be used in the final study. Unfominately, it was not possible to know how many respondents came from the University's Physical Plant Services Department and, therefore, the more representative sarnple and how many came fiom the sample of former MBA students and their peers. in retrospect more elaborate demographic data would have helped resolve some of the questions surrounding the sample. Nevertheless, the items resulting from this pilot study were deemed satisfactory and were used in the main employee-level study, described in the following chapter.
Results regarding the underlying factor structure of the ten practices should be interpreted with caution, however, given non-representative sarnple. It is necessary that the structure be reexamined in the main study.
Concludinn Thouehts
In sum. the results of this study achieved two aims, the fint of which was to develop a measure of high cornmitment management practices that can be applied at the organizationai level. Second, this study provides m e r evidence that the ten human resource management practices examined are highly intemlated and reflect a single
underlying factor, namely, a hi& performance work system.
Chapter Seven Study Two- Main Study
To date, little is known about the mechanisms that mediate the relationship between high performance work systems and workplace safety let alone more general performance rneasures. Therefore, and as described earlier, one of the central purposes of this work is to understand not only whether hi& performance work systems affect workplace safety at the oqanizational level, but also to M e r our understanding of the means by which these positive effects occur.
in order to investigate this relationship, it was necessary to take the investigation to a more micro level and examine the behavioun and attitudes of empioyees. It has been
argued throughout this thesis, and particularly in Chapter Three, that the adoption of hi& performance work practices at the organizational level will have positive effects on employees' trust in management, affective cornmitment to the organization and perceptions of safety ciimate. in tum, it is these mediating mechanisms that have been hypothesized to positively impact safety performance at the employee level. Safety performance has been concepnialized to extend beyond simply numbers of lost time injuries, but rather to include ernployees' personal safety orientation (comprised of safety knowledge, safety motivation, safety initiative and safety cornpliance) and involvement in safety incidents (such as those requiring first aid as well as near-misses) as discussed in Chapter Four. These safety outcomes were chosen dong with loa tirne injuries for they provide much richer information about employees' safety performance than do more
conventional measures.
Pumose
The purpose of Study Two was to evaluate employee perceptions of the extent to which their organization's had adopted high performance work systems as well as the their trust in management, affective cornmitment to the organization and perceptions of safety climate. Employees were also asked to report on their behavioun consistent with both safety cornpliance and safety initiative, their safety knowledge and safety motivation, collectively compnsing their persona1 safety orientation. Furthemore, the fiequency in which they have suffered a range of injuries requiring fint aid and nearmisses was exarnined. Overall, this study served to answer the question, what mechanisms link hi@ performance work systems with greater workplace safety?
Partici~ants
Participants in this study were 196 employees of two organizations fkom the petroleum and telecommunications industries, both located in Canada. Eight-three participants worked in the petroleum company as plant and field operaton while another 1 13 worked in the telecommunications company as field technicians. Of the latter, 80 worked in one site while 33 worked in another. For a number of reasons, it was not possible to calculate the overall response rate for in many instances managers were responsible for distributing the surveys. There is no way of knowing how many of them
did so nor to how many individuals they sent the survey and, therefore, how many surveys were distributed. The exception is for one of the telecommunications sites, in which s w e y s were distributed directly to potential respundents. In this case, the survey was distributed to 123 individuals, wiîh a response rate of 26.83 %. 114
Of the questionnaires that were completed, 191 Furnished usable data. Two of these respondents were female and their s w e y s were removed from the final data set in order to elhinate any effects that gender may have had on responses. Participants in the final sample were 189 males with a mean age of 39.12 yean (Se = 7.92; range 22 to 58 years). On average, they had completed 13.84 years of schooling (So = 2.26;range 9 to 21 years) which amounted to just under one year of univenity or college. Eight-four percent worked full-time for an average of 41-68hours per week
= 8.04; range =
7.50 to 84 hours). The average number of h o m of overtime worked was 3.22 per week (SD = 4.02; range = O to 30 houn). Overall, participants had worked in their respective
organizations a mean of 12.38 years positions for 7.22 years
= 9.18; range = .25 to
35.08 yean) and in their
(u= 7.25; range = 2 5 to 31 years). The vast majority were
permanent employees (89.60%).
Matenals Each participant in this study was asked to fil1 out a survey asking demopphxc questions and measuring: Hieh Performance Work Svstems. The 51 items used were those developed in the Pilot Snidy (refer to Appendix G). Responses were on a five-point Likert scale ranglng h m 'stmngly disagree' (1) through 'strongly agree' (5). A higher score reflected the
perception that the organization had more extensively adopted a high performance work system.
Trust in manaeement. Six items h m Cook and Wall's (1980) measure of interpersonal trust at work were used to measure trust in management. A sample item Il5
was, "Management can be tnisted to make sensible decisions for the fïrm's friture*'.
Responses were be measured on a 7-point Likert-type scale with options ranging from 'strongly disagree' (1) to 'strongly agree' (7). A funher, general item was also added to the scale. namely, "1 have confidence and trust in management at rny company". Higher scores reflected greater trust in management (see Appendix H). Affective Cornmitment. Affective cornmitment to the organization was measured using seven items developed by Meyer and Allen (1997). "1 would be very happy to spend the rest of my career at this organization" and "1 do not feel 'emotionally attached' to this organization" (reverse scored) are examples of items. Responses were measured on a 7-point Likert-type scale and range fiom 'strongly disagree' (1) to 'strongly agree' (7). Higher scores indicated greater affective cornitment to the organization (see
Appendix 1). Safety climate. Sixteen items were used to measure four dimensions of safety clhate: management values (4 items), communication (5 items). training (4 items), and systems and pmcedures (3 items) (Neal et al., 2000). Example items were: "management is concemed for the safety of empioyees" and "The safety procedures and practices in this organization are usefûl and effective". Ali responses were on a 5-point Likert-type scale ranging from 'strongly disagree' (1) to 'strongly agree' (5). Higher scores reflected a more positive safety climate (see Appendix J).
Safetv cornpliance. Seven items derived fkom both Neai et al. (2000) and Williams et al. (2000) measured safety cornpliance. "1 use the correct safety procedures for carrying out my job" was a sample item. Al1 respooses were on a five-point Likerttype scale ranging h m 'strongly disagree' (1) to 'smngiy agree' (5). Higher scores 116
reflected greater safety cornpliance (see Appendix K). Safetv initiative. Safety initiative was measured with eight items from Turner and
Parker (2000). Sample items included: "1 am involved in improving safety policy and practices" and "1 oflen try new approaches to improving workplace safety". Responses to these items ranged h m 'strongly disagree' (1) to 'strongly agree' (5) on a five-point Likert-type scde. Higher scores reflected greater safety initiative (see Appendix L). Safetv knowledee. The extent to which employees feel knowledgeable about safety was measured with four items fkom Neal et al. (2000),including "1 know how to reduce the nsk of accidents and incidents". Responses to items were on a 5-point Likerttype scale where ' 1 ' reflected 'strongly disagree' and '5' reflected 'strongly agree'. Higher scores were associated with greater safety knowledge (see Appendix M). Safetv motivation. Four items measured safety motivation (Neal et al., 2000). A sarnple item was: "1 feel that it is worthwhile to put in effort to maintain or improve my penonal safety". Responses were on a 5-point Likert-type scale and ranged from 'strongly disagree' (1) to 'strongly agree' (5). Higher scores reflected greater safety motivation (see Appendix N). Iniuries. Workplace injuries were assessed in thne ways: the fnquency of injuries
requiring f h t aid, the fkquency of injuries almost sustauied (i.e., near-misses) and the number of lost t h e injuries and rwulting days lost. The incidence of workplace i njuria requiring fint-aid was assessed by asking
individuals to think back over the last six months and to report how often, fiom 'never' (1) to 'fkequently' (5). they had sustained injuries requiring h taid in each of nine injury
categones. The nine categories were: (i) fractures, (ü) dislocations, sprains and mains, Ili
(iii) bruishg and crushing, (iv) scratches and abrasions (superficial wounds), (v) cuts, lacerations and punctures (open wounds), (vi) burns and scalds, (vii) eye injuries, (viii) concussions and head injuries. and (ix)other. Six months is the recornmended maximum length of time over which employees can recail injuries they have sustained with any accuracy (Veazie et ai., 1994). Following Hemingway and Smith (1999). the incidence of near-misses was also measured using the same nine categones of injuries and the sarne response scale. in ihis case, employees were asked to report how frequently they had almosf sustained the particular type of injury over the last six months. Last, employees were also asked the number of lost-time injuries they had experienced and the nurnber of days off work that resulted in the last six months. Perceived Risk. In order to contml for the risk inherent to each workplace, employees' perceived risk was also measured using three items h m Neal et al. (2000). "There are significant dangers inherent in this workplôce" was a simple item. Responses ranged fiom 'strongly disagree' (1 ) to 'strongly agree' (5). Higher scores reflected greater perceived nsk (Appendix O).
Procedure Questionnaires were distributed to &ont-lineemployees of the petroleum Company in a paper-and-pend format (see Appendix P). They were distributed and the completed questionnaires collected in sealed mvelopes by management. In the case of the telecommunication Company, sweying was conducted electronically (see Appendk Q). Letters sent by electronic mail were delivered to participants either directly or via II8
their managers encouraging employees to participate in the study. The letters contained a
URL leading to the website on which the survey was posted.
Results Descriptive statistics and intercorrelations of al1 variables in this study are presented in Table 8. Al1 models were based on the covariance matrix and estimated
using maximum likelihood estimation as implemented in LISREL Vm (Joreskog &
Sorbom, 1992). Andenon and Gerbing's (1988) two-stage modeling approach was followed for these analyses. This method requires that one first atablish the fit of the measurement mode1 at which point one can test the stnictrual relations of interest. Latent variables were created to reflect two types of safety outcomes. In the first case, safety cornpliance, safety initiative, safety knowledge and safety motivation were made indicaton of a single latent variable-persona1 safety orientation Second, the incidence of acnial injuries and nw-misses were made indicaton of a single latent variable, narnely, safety incidents. Data collected on lost time injuries and days lost as a consequence were not included in the analyses given that the events were too hfkequent.
Of the employees sampled, 2.79% had expenenced a single iost time injury in the last six months. None of the respondents had more than a single lost tirne injury. Latent variables were also created to reflect the high performance work system and safety climate variables. This was done by conducting an odd-even split on the scales
(see for instance, Zacharatos, Barhg & Keiloway, 2000). In each case, the subscales
then acted as observed variables reflecting the latent facton-high performance
Descripiivc Stntistics and Iniercorrclütions Bctwcen Meiisures in Study l'wo (N = 189)
Variable HPWS
1. 2 3. 4. 5. 6,
h s t Safety Climatc Affective Cornmitment Compliance
7,
Knowledge
8, 9, 10,
Motivation First aid
Initiative
Near-miss
M
SD
1
153.18 32.42 64 .O4 34.64 29.34 29.57 16.70 17.78 10.38 1 1.68
21.12 9.16 9.27
.70** .54**
6.86
.hg**
3.59 4.04
.35** .28** .37** .14
1.90 1.88
--
2.10 -.31* 3.53 -.33**
3
4
--
-
7
8
9
-34'' ,27** .63** .27** .22** .42** . 2 P * .48** .27** .S9* .31 ** .72** .47** .lu* .37** .16* .55** .38** .48** -.30** 2 -.OS -.l 1 -.13 .O8 -.28** -.26* -.15 -.3Q** -.13 -.13
.I I .O3
.70**
2
5
6
.52** .53**
1O
work system, in one case, and safety climate in the other. This was required in order to increase the number of estïmated parameten and ensure the model was ovendentified, a necessary requirement for this type of analysis. These two scaies were chosen for the odd-even split arbitrarily-for
the simple fact that they had the greatest number of items.
This is acceptable, given that there is currently linle consensus on the best way to form indicator variables (Kelloway, 1996). The remahhg variables, trust in management and affective commitment. were treated as single indicaton of latent variables following Barling, Kelloway and Bremmerman (1 99 1). The pmposed measurement model, specikng six latent variables (Le.. hi& performance work system, mist in management, affective cornmitment. safety climate, persona1 safety orientation, and safety incidents) provided a reasonable, but not outstanding. fit to the data: X2 (4 1, N = 189) = 6 1.76. p < .05; goodness of fit index (GFI) = -95; adjusted goodness of
fit index (AGFI) = .90;nomed fit index (NFI) = -96;
comparative fit index (CFI)= .98;panimonious nonned fit index (PNFI) = -60. When considering potential modifications to the model, it is important to look to ihe literature rather than the data for insight (Anderson & Gerbing, 1988). 1 deerned the weakest aspect of the model to be the inclusion of affective cornmitment as a mediating mechanism. There exists evidence that affective commitment is associated with both
human resource practices and workplace safety. These hdings, however, are derived fiom snidies looking at limited factors such as the effect of transformational leadership on affective commiiment and, in a separate study, the role of affective cornmimient on workplace safety. None of these sniciies, with one exception, have looked at the relationship between human resource practices, affective commitment and workplace 121
safety more collectively.
The most similar study to the current one found that of a number of practices tested, affective commitment only mediated the relationship between job autonomy and communication quality and safe working (Parker et al.. 200 1). Surprisingly, the researchers did not find that affective comrnitment mediated the relationship between role overload and conflict (aspects of job quality), training adequacy or job security and workplace safety. Given this evidence, 1 decided to test a measurement model that did not include affective commitment. Because one variable was removed it was possible to make one of the split-scales latent variables a single indicator variable, in this case, safety climate. This is desirable with a srnail sample size as it reduces the number of parameten to be estimated. The results were more positive with the modified model demonstrating an acceptable fit to the
data:
(27, N = 189) = 35.16, ns; GFI = .96; AGFI = .93; MI = .97;CFI = -99; PNFI =
33.Standardized parameter estimates for the measurement model are presented in Table 9. Tfie measurement structure was then used to estimate the structural relations of interest. In order to demonstrate mediation, it is necessary to estimate three nested structural models (Kelloway, 1998). Fint, it is necessary to test the fully mediated mode1 (Figure 4) as well as a partiaily mediated model in which there are additional paths between the high performance work systern and the outcome variables, personai d e t y orientation and safety incidents. Furthemore, it is necessary to test a third non-mediateci model which consists of the partially mediated model with the paths from mist in
TabIe 9 Standardized Parameter Estimates for the Measurement Mode1 (N= 189)
High perfo&ance work system 1 . HPWS t 2. HPWS'3. Trust 4. Safety climate 5. Cornpliance 6. Initiative 7. Knowledge 8. Motivation 9. First-aid 1 0. Near-miss
Tmt
Safety
Penond
climate
safety
Safety incidents
orientation
.73 .90
management and safety climate removed. One must show that the fully mediated model provides a better fit to the data than does the nonmediated model and that the fully mediated model provides a more parsimonious fit to the data than does the partially mediated model in order to establish mediation (Kelloway, 1998). Before conducting the analyses,and given the high correlation between the mist and safety climate variables (1= S43,g < .01),the mode1 was m e r modified to allow
these two single indicator variables to covary. The proposed mediational model provided an excellent fit to the data. X2 (30, = 189)= 43.70,ns; GFI = .96; AGFI = .92; NFI = $96;CF1 = .99;P M I = .64. The partially mediated model also provided an excellent fit to
the data.
(28, = 189)= 36.84,-; GFI = .96; AGFI = .93;NF1 = .96;CF1 = .99; PNFI
= .60.The nonmediated model did not provide a good fit to the data,
(32,N = 189)=
103.53,g c .OOl;GFI = .90;AGFI = 33;MI = .89;CF1 = .92; PNFI = .64. The mediated mode1 clearly provided a better fit to the data than did the nonmediated model. in order to determine whether the fully or partially mediated model provided a better fit to the data a nurnber of factors were considered: the measures of comparative fit (NFI. CFI, and the 2 dircrmce test) and the overall panimony of each of the two models (Kelloway, 1998). In ternis of comparative fit, both models provided equally
acceptable fits to the data. The result of the $ din-ce
test, however, demonstrate that the
modeis were significantiy different, X2 diff-~e (2) = 6.86, c -05.By e x a m k g significant paths, however. it was found that the fully mediated model was the more parsimonious of the two, and therefore was retained for fùrther analyses. Standardized parameter estimates for the model are presented in Figure 5. As shown, safety incidents were predicted by both trust in management (f! = -.19; p < .OS) 125
and safety climate (p = -.35; p < .01). Persona1 safety orientation was predicted by safety climate (p = .76; < .01) but not by trust in management (P = -JO;
a.Trust in
management was predicted by high performance work practices (P = .77; E c .01) as was safety climate (B
= 62; p < .O 1).
Discussion Having found a relationship between high performance work systems and workplace safety at the organization level in Study One, the purpose of this study was to
further our understanding of the mechanisms which mediate the relationship. Narnely, it was hypothesized that high performance work systems serve to increase employee trust
in
management, affective commitment to the organiration and perceptions of safety climate which, in tum, result in greater safety initiative, safety compliance, safety knowledge and
safety motivation (Le., persona1 safety orientation) and fewer workplace injuries requiring fint aid and near-misses (Le., safety incidents). Support was generally found for this model. It was found that high performance work systems were positively associated with both tnist in management and perceived safety climate. In tum,both trust in management and perceived safety climate predicted injuries requiring f k t aid and near-misses (Le., safety incidents), while only safety climate predicted personal safety orientation-safety
knowledge, safety motivation,
safety compliance and safety initiative. An'tive commitment was not found to mediate the relationship between high performance work system and safety pedormance.
Advances
These results both replicate aspects of previous research and extend the literanire. The current findings support those of Neal et al. (2000) who found safety knowledge, safety initiative, safety motivation and safety cornpliance to be positively associated with more positive perceptions of safety climate. These findings also support those of Zohar (2000. 2001) and Barling, Loughlin et al. (2001) who both found safety climate to be related to the number of actual injuries expenenced by workers. These findings make several contributions to the current literature. These results extend our understanding of the mechanimis by which high performance work systems
may impact employee attitudes and behaviom and, in tum. desired outcomes. To date, there have been no studies that examine trust as the mediating mechanism between high performance work practices and any specified outcorne, including safety performance. The sarne holds true for the role of safety climate in the relationship between high performance work systems and safety-related outcomes. While studies have tested aspects of the current model, such as the effect of a single human resource practice on tmst or safety climate, or the effect of safety climate on safety outcomes, the current study is the first to test these relationships collectively in a comprehensive manner. Trust in management was found to play a significant role in mediating the relationship between high cornmitment work systerns and safety performance. Specifically, the more extensive adoption of a high performance work system lead to greater tnist in management which, in tum, resulted in fewer safety incidents experienceà by employees. This fmding points to the importance of trust in management as a factor in workplace safety. The fact that trust in management was found to mediate the 128
relationship between high commitment work systems and safety incidents is also of value to those interested in high performance work systems more generally. This finding suggests that trust in management is a possible outcome of such a system and precunor to performance outcomes, whether safety-related or otherwise, though this conclusion requires further investigation. Workers' trust in management was not found to mediate the relationship between high commitment work systems and persona1 safety orientation, however. This is a surprising result that may be attributable to the high correlations between study variables.
In addition and more specific to safety, the results of this study demonstrate that safety climate plays a mediating role linking high perfomance work systems to both fewer safety incidents and more positive personal safety orientations. This finding is important for it suggests mechanisms by which a positive safety climate is developed and numued. Recalling the fact that in the current study high performance work practices were measured in general venus safety-specific ternis, this finding becomes that much more interesting. It suggests that human resource practices not necessary geared toward safety, such as transformational leadership that does not promote a vision of a safe organization, and nonsafety-specific training, still exert positive effects on safety clhate, thereby demonstrating the value of generai 'good management' for pmmoting workplace safety.
In this snidy, afkctive commitment was not found to mediate the relationship between high performance work systems and workplace safety. There are a number of reasons why this rnight be the case. Firsf it is possible that afktive conmimient was too highly correlated with other mediating variables being investigated, namely trust in 129
management and safety climate. As such, affective commitment may have been forced out of the measurement mode1 as error variance.
Another possible explanation is sirnply that affective cornmitment does not play a critical role in mediating the relationship between high perfomance work systems and performance outcornes. This is somewhat consistent with the findings of Parker et al. (2001) who found that affective commitment only mediated the relationship between job
autonomy and communication quality and workplace safety. but did not mediate the relationship between the balance of high performance work practices examined and
safety performance. More generally, Gardner et al. (2000) reported similar findings. ïhey found that of the hurnan resource management practices they examined which in the terms used in this study amounted to selective hiring, job quality, uiformation sharing, training, measurement and contingent compensation, only job quality and information sharing were associated with p a t e r organizational cornmitment. Direct cornparisons between the current study results and those of these IWO other studies must be made with caution, however. Both the Parker et al. (2001) and Gardner et al. (2000) papers examined the roles of individual practices, while in the current study. a
set of ten human resource practices were examined collectively. Nevertheless, if only information sharing and job quality were found to be related to affective commitment in these studies, while other practices were not, it is not altogether surprising that in the current study affective commitment did not mediate the relationship between hi@ performance work systems, comprised of ten practices, and workplace safety. The results of the current study and the findings of both Parker et ai. (200 1) and Gardner et al. (2000) may raise some doubts about the d e of affective commitment as a mediathg mechanian 130
beween comprehensive high performance work systems and performance outcornes. Funher testing of these relationships is absolutely necessary, however, before any conclusions can be drawn with respect to the role of affective commiûnent in such a system. The results of this study aiso extend our understanding of the conceptualization of
a high performance work system. The measurement mode1 provided further support for the findings of the previous two studies that the ten hi& performance work practices are
highly interrelated and form a single underlying factor. This result is particularly noteworthy given the high performance work practices were measured by a number of items each, Following the advice of Becker and Huselid (1998) and are, therefore, more meaninghl than those of previous studies. The fact that the high performance work systems were measured h m the
employee perspective is a particular noteworthy aspect of this study. This diffen from similar studies in which data about the human resource practices corne directly fiom the
Hurnan Resource Director or equivalent. In this respect, the current study addresses one of the criticisms of earlier high perfomance work system studies found in the l i t e r a ~ e which is that researchen have made the fdse assumption that al1 employees in the organization experience the sarne human resource management practices (Martinez,
Pearce, Porter & Tsui, 2000). in asking employees to respond based on their individuai perceptions of human resource practices in their organizations, 1 have gone beyond this unitary view of human resource management.
Limitations
This study is not without limitations, however. First, this study relied exclusively on self-report data, which may have increased the probability of mono-method bias. Nevertheless, self-reported data were necessary for the variables considered, namely, the individual perceptions of the hurnan resource management practices, perceptions of safety climate, trust in management, affective commitment to the organization and the safety performance variables. One possibility may have been to use Company records of employee injuries. however, such a measure may have provided underestirnates of acnial injuries and is less nch than the cunent measures of safety performance to the extent that it focuses on a single aspect of safety performance. As well. company-held safety records
only account for those injuries that are most senous and occur less frequently. Overall, while monomethod bias is of some concem in this study, Crampton and Wagner's (1994) h d i n g that mono-method bias occun less frequently than most would p r e m e would suggest that the results are still valid. A second limitation of this study was that the nsk expenenced by participants in
their respective workplaces was not controlled for. Risk data were collected using a threeitem scale measuring perceived risk, however, they were not included in the h a I analyses given that the risk scale had an unacceptably low reliability (a = .46). Thirâ, the results of this study cannot be interpreted to mean that high performance work practices lead to greater safety performance by way of tnist in management and positive perceptions of safety climate despite the hypotheses being conceptuaiized in this manner. M y longitudinal data collection can provide us with the abitity to uncover causal relationships beh~eenthese variables. 132
Concludine- Thoughts
While this study is not without limitations, overall, it makes a significant contribution to the safety literature as well as the high performance work practice literature more generally. in one of the fint studies of this lcind, 1 demonstrated that there is a relationship at the employee level between high performance work systems and fewer safety incidents as well as more positive personal safety orientations. As well, it was found that trust in management and perceptions of safety climate mediate the relationship between high performance work systems and the number of safety incidents as well as employees' personal safety orientations, thereby shedding some light upon how high performance work systems exert their positive effects. The implications of these findings will be discussed in greater detail in the following chapter.
Chapter Eight Discussion
The results of the two studies that comprise this thesis make a number of significant contributions to the current literature. They extend our understanding of high performance work systems and their relationship with workplace safety at both the organization and employee levels. They also contribute to our understanding of the factors that may or may not mediate the relationship between high performance work systems and performance outcomes more generally and safety performance specifically. Last, these studies lend to our understanding of the interrelatedness of the human
resource management practices in a high performance work system.
F k t , it was found in a sample of 147 companies fkom varied industries that the extent to which the organization had adopted the high performance work system described in this thesis, comprising the use of selective hiring and transformational leadership. the provision of qudity work, ernployment security and training, reduced statu distinctions, self-rnanaged teams, information sharing, contingent compensation and measurement of variables critical for success, was negatively related to the number of work injuries experienced by workers in terxns of lost time injuries. This supports earlier findlngs that high performance work practices impact varieci aspects of organizational performance. It also extends our understanding of how best to manage for workplace safety.While the results of this snidy do not allow us to make causai inferences, there does appear to exist a m n g relationship between hi& cornmimient management practices and occupational safety at the 0rgani;Tational level. 134
The results of the current work also achieve the second aim which was to M e r our understanding of the facton that mediate the relationship between high performance work systems and employee performance. Support was found for the role of t m t in management as one such mediating mechanism. This is much in line with Pfeffer's assertion that trust plays a critical role in the success of a high performance work system
(Pfeffer, 19%.
1998b).in the current work, trust was found to mediate the relationship
between the high performance work system and the safery incidents experienced by worken. This finding suggests that trust may also play a similar role in high commitment work systems where the performance outcomes are substantially different than those currently investigated. Only hture research studies, however, can confimi this idea The current study results also m e r our understanding of the role of safety climate in such a system. It is well established in the literature that safety climate is related to outcomes such as safety knowledge, safety motivation, safety cornpliance and safety initiative as well as actual injury rates (Barling, Loughlin et al., 2001; Hohana & Stetzer, 1996; Neai et al., 2000; Zohar, 2000). What is less weli understood are the organizational facton that create positive perceptions of safety climate. This study provides strong evidence that high performance work systems are related to perceptions of positive safety climate. Again, given the cross-sectional nature of the data, it is not possible to conclude that high commitment work practices positively impact safety climate, however, this prelimuiary research fumishes us with some insight into this relationship and provides a starting point for funue research in this area. These studies also contribute to o u .understanding of the manner in which the
hi& performance work practices considered are interrelateci. It was found that the ten 135
human resource practices were reflective of a single underlying constnict, namely, the high performance work system . This suggesu that the practices are highly interrelated and organizations that, for instance, emphasize employee training and information
sharing between memben also provide their employees with employment security, as discussed in Chapter Four. The findings from this snidy are particularly meaningfbl in this respect given the rneasures of each of the ten practices, whether at the organization or employee level, comprised a number of items that more comprehensively measured the presence of each practices. This is an improvement over studies to date that have typically used û single item to measure each practice. The study results are also noteworthy for the following reasons. First, the relationship between hi& performance work practices and safety performance at both the organization and employee level was found using measures of high cornmitment management practices that had a more general venus safety-specific focus, as descnbed
in Chapter Four. For instance, the extent to which employees were provided with general as opposed to safety-specific training was measured. This approach was adopted for the methodological and theoretical reasons discussed. What this suggests, however, is that had the hi& performance work practices being measured been onented toward safety (Le., emphasizing the importance of worker safety in selection intemiews or creating a
vision of a safe organization as part of the transformational leadership dimension), the results of this study might have showa an even stronger relationship betweem the high performance work system and workplace safety. Therefore, it is possible that while organizations in which occupational safety is an important concern may benefit h m high pdonnance work systerns more generaüy, it may also be the case that they stand to 136
benefit even more so nom safety-specific high performance work systems. Second. in both Smdy One and Study Two, there is the distinct possibility that injury rates were underreported. In Study One respondents had no reason to underreport statistics given the anonymous nature of the survey and the vast majority did c l a h that results came from actual reports. Neveriheless, this does not eliminate the chance that recorded safety statistics were inaccurate as organizations have many reasons for keeping less than accurate records. Furthemore, the number of lost-time injuries does not provide us with any indication of the numerous injuries requiring first aid or near-misses that occurred. In Study Two, safety statistics may have also been underreported given the fact that employees are often unable to retrospectively recall injuries or near-misses. Therefore, if in fact the safety data collected underrepresented the number of lost-time injuries and workers experienced more minor injuries and near-misses than reported, the curent findings may underestimate the real strength of this phenornenon.
High cornmitment management practices have been the focus of increasing attention ever since Walton's seminal 1985 article describing the benefits to an organization of the cornmitment versus control-oriented approach to managing human resources. In his article, Walton argued that the employees would respond best to being neated as an invaluable organizational resource. Since then, a nurnber of others have expounded on this idea, most distinctly Jefiey Pfeffer (see, for instance, Pfeffer 1997, 1998a, 1998d). Pfeffer's work rnakes the distinct contribution of providing readers with a
set of practices for managing the workforce which he argues increase worker trust in management and cornmitment to the organization and that encourage people to take control over their work, to work smarter and more responsibly. The outcome of nich a 137
high performance work system, he claims, is improved employee performance and organizational profitability-what
Pfeffer calls, "achieving profits through peopleTT
(Pfeffer, 1998% p. 121). Many researchen have tested the proposition that cornmitment-oriented or high performance work system practices are associated with greater firm performance and profitability (see, for instance. Arthur, 1992. 1994; Hoque, 1999; Huselid, 1995; MacDuffie, 1995, Ichniowski et al. 1997), and support has been found for this nonnaditional mode1 for managing human resources. To date, however, the studies being reported have focused on employee-related performance measures including sales per employee, employee flexibility and tumover which are argued to be precunon of organizational performance and profitability more generally. This thesis makes a significant contribution to the literature by focusing instead on another critical individual and organizational outcome-worker
safety.
Cohen (1977) and his colleagues (see, as well, Smith et al. 1978) provided preliminary evidence that workers managed under a cornmitment-oriented approach would work more safely than those managed under a controlsriented system. They found that infornation sharing, training, employment security, selective hiring and a strong cornmitment on the part of management to safety issues were the hallmarks of low accident-rate plants. Cohen and his colleagues were remarkably ahead of their tirne for what they describeci was basically a high performance work system in which worker safety was the performance outcome. in the 24 years since this work was published, the discussion and mearcb of these
more humanistic management practices has flourished. Yet, to &te, there has not been a 138
m e r nor more comprehensive examination of the role of high performance work systems in increasing worker safety. Moreover, there has been a paucity of research examinhg the means by which high perfomance work practices corne to have positive effects on employee performance in gened, and worker safety, more specifically. This research was a preliminary attempt to draw together these two areas of snidy in order to understand whether high perfomance work systems are associated with greater workplace safety, and if so. by what means this effect occun.
Im~licationsof Findines The findings from this thesis have important implications for the management of
worker safety. Most importantly, they highlight the role of management in occupational safety. We can no longer assume that individual workers are primarily responsible for their own safety. Rather, this study demonstrates that organizational factors are significantly associated with safe work. The use of selective hiring and transformational leadership, the provision of quality work, employment security and training, reduced status distinctions, self-managed teams, information sharing, contingent compensation and measurement of those variables critical for success are al1 important factors in reducing injury rates. We now have preliminary evidence that the impact of these practices is due to their influence on employees' trust in management and perceptions of
safety climate. Fundarnentally, human resource practices that are based on respect for workers and which provide employees the means to work to their highest potential play a role in worker safety and should be encouraged.
The hdings aiso demonstrate that workplace safety can be managed, at least in 139
part, much as other aspects of empioyee performance. That is, this study provides preliminary evidence that organizations interested in improving employee safety performance should adopt high perfomance work practices that have been shown elsewhere in the literature to be associated with employee and corporate performance more generally. This work provides preliminary evidence that organizations will also reap the benefits of these systems in terms of employee safety, a desirable outcome when one considen the enonnous persona1 and financial costs associated with workplace injuries and fatalities.
Directions for Future Research Unfominately, many managers continue to hold firm financial performance as their foremost concem. They are iess concemed with health and safety issues and see the provision of occupational health and safety as a cost to be incurred (Goldenhar et al., 1999). This situation will not be remedied until M e r research demonstrates
undisputedly that workplace injuries exert financial cos& on organizations. This is a necessary direction for hture research. Once this missing link in the literature is remedied the findings of this study will be that much more rneaningfbl for they will provide a preliminary bais for concluding that high performance work systerns positively impact workplace safety which, in turn, positively anécts firm hancial performance. Only once this c o ~ e c t i o nis demonstrated empirically can we expect management to seriously consider the adoption of these practices for the management of workplace safety. Future research should look to overcome some of the limitations inherent to the 140
current work. Now that evidence exists for a relationship between high performance work practices and workplace safety, as well as the roles of trust in management and safety climate, it is necessary to conduct longitudinal research that will provide us with insight into the causal relationships. Furthermore, hiture studies should refrain frorn depending solely on self-report rneasures of the variables of interest. As well, hirther refinements to the rneasures of high performance work practices employed in the current studies are desirable. Also of interest, would be a further examination of other variables that may play a
part in workplace safety, both positive or negative. A replication of the current work in which control-oriented practices were also measured would allow us to make a direct comparison between the cornmitment-onented and control-oriented practices and their impact on workplace safety. As well, hiture research should re-examine the role of organizational cornmitment in mediating the relationship between high performance work practices and workplace safety. While affective commitment was not found to mediate the relationship between high performance work systems and workplace safety in the current study, it was found to mediate the relationship between some high performance work practices and performance outcomes in other studies (Parker et al, 2001, Gardner et al., 2000). Given these muted findings, it is not wise at this stage to eliminate it as a possible mediating mechanism and hiture research should focus on resolving the d e of this variable. The d e s ofjob satisfaction and perceptions of organizationd justice are m e r constructs that may be examined as mediators in the high performance work system-performance M.
The current research demonstrated that the ten human resource practices being considered were highly interrelated. Future research may be able to untangle these relationships and contribute some understanding to whether select practices play a greater part in workplace safety than do othen or whether al1 the practices are of equal importance.
Concluding:Thoughts
Research interest in the area of high performance work systems continues to develop as does interest in the important impact management can have on workplace safety. Nevertheless, very linle research has attempted to bring together these two distinct areas of study and none, to date, has done so as comprehensively as this work. The contribution ofthis thesis to our understanding in both these areas is significant. To the area of high performance work systems, this research provides evidence that the performance benefits accrued may be of a varied nature, thereby further illustrating the value of such systems. As well, this research has contributed significantly to our knowledge of how high cornmitment work systems impact employee and, in tum, organizational performance. To the realrn of safety management, this study provides much needed confirmation of the paramount role organizational rather than individual factors play in worker safety and this remains the most substantial contribution of this work. The hdings from this thesis suggest we need to take a much broader look at how we are
currently managing occupational safety if we have any intention of ensuring worken have the basic nght to return home safely to their families at the end of the workday.
References Adams-Roy, I. E., Knap, M. A., & Barling, J. (1995). Cornmitment to occupational health and safety in management training. In L. E. Tetrick & J. Barling
(Eds.), Chaneine e m ~ l o m e nrelations: t Behavioral and social ~ersiiectives@p. 133143). Washington, DC: Amencan Psychological Association.
Alarnpay. R. H., & Beehr, T. A. (in press). Empowerment. span of control and safety performance in work teams after workforce reduction. Journal of Occupational
Health Psvcholow. Anderson, J. C., dk Gerbing, D. W. (1988). Structural equation modeling in practice: A review and recomrnended two-step approach. Psvcholoeical Bulletin. 103, 41 1-423.
Anderson. N.. Hardy, G., & West, M. (1990). innovative teams at work. Personnel Management. 2S,48-5 1. Andriessen, J. H. T. H. (1978). Safe behaviour and safety motivation. Journal of Occu~ationalAccidents, 1,363-376. Arnold, P. K., Hartley, L. R., Corry, A., Hochstadt, D., Pema, F., & Feyer. A. M. (1997). H o m o f work and perceptions of fatigue among truck cirivers. Accident Analvsis
and Prevention, 29,47 1-477.
Arthur, J. B. (1993). The link between business mategy and industrial relations systems in Arnerican steel mlnimills. Industrial and Labor Relations Review. 45,488-
506.
Arthur, J. B. (1 994). Effects of human resource systems on manufacturing 144
perfomance and turnover. Academv of Manaeement Journal. 37,670-687. Arthur, W., Jr., Barrett, G. V., & Dovenpike, D. (1990). Validation of an
information-processing-based test battery for the prediction of handling accidents among petroleum-product transport dnven. Journal of Ao~iiedPsvcholoev. 75.62 1-628. Ashford, S. J., Lee, C.. & Bobko. P. (1989). Content, causes and consequences of job insecurity: A theory-based measure and substantive test. Academv of Management Journal. 32,803-829. Association of Worken' Compensation Boards of Canada. (2000). Work iniunes and diseases 11997-1 999)- National work iniuries statistics Dromam (Cat. No 01-102).
Mississauga, Canada: Author. Austin. J., Kessler, M. L., Riccobono, J. E., & Bailey. J. S. (1996). Using
feedback and reinforcement to improve the performance and safety of a roofing crew. Journal of Orginizational Behavior Manaeement. 16 (2). 49-75. Barling, J.. & Hutchinson, 1. (2000). Cornmitment vs. control-based safety practices, safety reputation, and perceived safety climate. Canadian Journal of Administrative Sciences. 17,76-84. Barling, J.. Kelloway, E. K.,& Bremermann, E. H. (1991). Reemployment predictors of union attitudes: The role of family socialization and work beliefs. Journal of ADDlied PsYc~oIoEY.76,725-73 1. Barling, J., Loughlin, C., & Kelloway, E. K. (2001). Develo~mentand test of a mode1 linkine transformational le ad en hi^ and occu~ationalsafety. Manuscript submitted for publication.
Barihg, J., Moutinho, S., & E. K. Kelloway (2001). Transformational leadership 145
and group performance: The mediating role of affective commitment. Revised manuscnpt submitted for publication. Barling, I., Weber, T., & Kelloway, E. K. (1996). Effects of transformational leadership training on attitudinal and financial outcornes: A field experiment. Journal of Amlied Psycholow. 8 1,827-832. Barling, I., & Zacharatos, A. (1999-August). Hieh oerformance work svstems: Management oractices for achievine ootimal safetv oerfomance. Paper presented a the Academy of Management Annual Meeting, Chicago. iL. Barroll, L. D. (1999). Consultinn ~roiect.Unpublished project, Queen's University, Kingston, Ontario.
Barton, J., & Folkard, S. (1991). The response of day and night nurses to their work schedules. Journal of Occu~ationalPsvcholorrv. 64,207-21 8. Bass, B. M. (1990). From transactional to transfomational leadership: Leaming to share the vision. Or~anizationalDvnarnics. I8,I9-3 1. Bass, B.M. (1 998). Transformational leadership: Industry, military and
educational impact. Mahway, NJ: Erlbaum. Bass, B. M.. & Avoiio, B. J. (1995). Multifactor leadenhia auestionnaire for research. Palo Alto, CA: Mind Garden. Baugher, I. E., & Roberts, J. T. (1999). Perceptions and w o q about hazards at work: Unions, contract maintenance, and job conml in the U.S.petrochemical industry. lndustrial Relations. 38,522-541. Becker, B., & Gerhart, B. (1996). The impact of human resource management on organkational performance: Pmgress and prospects. Academv of Management Journal. 146
39 779-801.
d
Becker, B. E., & Huselid, M. A. (1998). High performance work systems and firm performance: A synthesis of research and managerial implications. Research in Personnel and Human Resource Manaeement. 16,53401. Beehr, T. A. (1995). Psvcholoeical stress in the workdace. New York: Routledge. Bimonte, A. (1994, May). Win ernpowered workers= cornmitment to exceeding basic safety cornpliance. Occu~ationalHealth and Safetv. 63.36.
Borofsky, G. L., Bielema, M., & H o h a n , J. (1993). Accidents, turnover, and use of a preemployment screening inventory. Psvcholo~icalReoons. 73.1067-1076. Borofsky, G. L., & Smith, M. (1993). Reductions in turnover, accidents, and absenteeism: The contribution of a pre-employment screening inventory. Journal of Clinical PsvchoLom 49, 109-1 16. Borofsky, G. L., Wagner, J., & Turner, S. (1995). Sustained reductions in nunover and accidents associated with the ongoing use of a preemployment screening inventory: Results of a three-yea. longitudinal study. Psvcholo~calReoorts. 77.195-204. Brocher, J., Grover, S., Reed, T. F., & DeWitt, R. L. (1992). Layoffs, job insecurity, and survivon' work effon: Evidence of an inverted-U relationship. Academy of Management Journal. 35.41 3425.
Bureau of Labor Statistics. (2000). National census of fatal occu~ationaliniuries, 1999 (USDL Publication No. 00-236). Washington, DC: Author. -
Bycio, P., Hackett, R. D., & Allen, J. S. (1995). Further assessments of Bass's (1985) conceptualization of transactional and transformational leadership. Journal of
-gy.
80,468478.
Campbell, J. P., Daft, R., L., & Hulin, C. L. (1982). What to study: Generating and developing research questions. Beverly Hills, CA: Sage. Cascio, W. F. (1993). Downsizing: What do we know? What have we leamed? Academv of Management Executive. 7,954 04. Castille, D. (1999). Occupational safety and health in young people. In J. Barling & E. K. Kelloway (Eds.), Youne workers: Varieties of exnerience @p. 159-200).
Washington, DC : Amencan Psychologicai Association. Chhokar, J. S., & Wallin, J. A. (1984). Improving safety through applied behavior analysis. Journal of Safetv Research. 15, 14 1- 15 1. Churchill, G. A. (1979). A paradigm for developing better measures of marketing constructs. Journal of Marketing Research. 16,64-73. Clarke. S. (1999). Perceptions of organizational safety: implications for the development of safety culture. Journal of Oreanizational Behavior. 20.1 85- 198. Cohen, A. (1977). Factors in successhl occupational safety programs. Journal of Safetv Research. 9,168-1 78. Cohen, S. G., & Ledford, G. E.. Ir. (1994). The effectiveness of self-managing teams: A quasi-experiment. Human Relations. 47,13-43. Collinson, D. L. (1999). 'SuMving the ngs': Safety and surveillance on North Sea oil installations. Orpanization Studies. 20,579-600. Comrey, A. L. (1978). Common methodological problems in factor anaiytic studies. Journal of Consultine and Clinical P-holonv.
46.648659-
Conway, H., & Svenson, J. (1998-November). Occupational iujury and illness rates, 1992-96: Why they fell. Jnternational Labor Review, 3658. 148
Cook, J., & Wall, T. (1980). New work attitude measures of trust, organizational cornmitment and persona1 need non-hlfillment. Journal of Occuoational Psvcholow. 53.
39-52. Cook, T. D., & Campbell, D. T. (1979). Ouasi-exnerimentation: Desien and analvsis issues for field settines. Boston, MA: Houghton Mimin. Cook T. D., & Campbell, D. T. (1983). The design and conduct ofquasiexperirnents and tme expenments in field settings. In M. D. Dunnette (Ed.), Handbook of Industrial and Or~anizational Psvcholofzy (pp.723-326). New York: John Wiley. Cooper, C. L., & Sutherland, V. J. (1987). Job stress, mental health, and accidents arnong offshore worken in the oil and gas extraction industries. Joumal of Occuoational Medicine. 29,I 19-235. Coyle, 1. R., Sleeman, S. D., & Adams, N. (1995). Safety clirnate. Joumal of Safetv Research. 26,247-254. Crampton, S. M., & Wagner III, I. A. (1994). Percept-percept inflation of microorganizational research: An investigation of prevalence and effect. Joumal of 79.67-76. Apolied Psvcholoev. .
DeJoy, D. M. (1994). Managing safety in the workplace: An attribution theory analysis and model. Joumal of Safetv Resemh. 25,391 7. DeJoy, D. M. (1996). Theoretical models o f health behavior and workplace selfprotective behavior. Joumal of Safetv Research. 27 (2), 6 1-72.
Department of Workplace Relations and Small Business (1997). The 1995 Australian Workplace Industrial Relations Survey ( A m 95). Raw data.
DeVellis, R F. (1991). Scale develo~ment:Theorv and amlications. Newbury 149
Park, CA: Sage. DiPenio, V., & Guttentag, W. (Producers & Directon). (1991). Death on the job [Film]. (Available from Direct Cinema Lirnited, PO Box, 10003, Santa Monica, CA 904 1O)
Donald, 1. ( 1995). Psychologicai insights into managerial responsibility for public and employee safety. In R. Bull & D. Carson (Eds.), Handbook of ~svcholoevin leeal
contexts @p. 625-642). New York: Wiley. Dwyer, T. (199 1). Life and death at work: Industrial accidents as a case of social1v ~roducederror. New York: Plenum. Eisenberg, W. M., & McDonald, H. (1988-April). Evaluating workplace injury and illness records; testing a procedure. Monthlv Labor Review, 58-60.
Eyssen, G. M.,H o h a n , J. E., & Spengler, R. (1980). Managers' attitudes and the occurrence of accidents in a telephone Company. Journal of Occu~aiionalAccidents, 29 1-304.
Fisher, C. D. (1993). Boredom at work: A neglected concept. Human Relations, 46.395-41 7.
Fitz-Enz, 1. (1997). The 8 oractices of excemional comoanies: How ereat oreanizations make the rnost of their human assets. New York: AMACOM. Flin, R., Mearns, K., O'Connor, P., & R. Bryden. (2000). Measuring safety clirnate: Identifjmg the common features. Safetv Science. 34,177-1 92. Fox, D. K., Hopkins, B. L., & Anger, W. K. (1987). The long-term effects of a
token economy on safety performance in open-pit mining. Journal of Appiied Behavior Analvsis. 20,2 15-224. 150
Frazis, H., Gittleman, M.,Homgan, M., & Joyce, M. (1998-June). Results from the 1995 survey of employer-provided training. Monthlv Labor Review, 3- 13. Frone, M. R. (1998). Predicton of work injuries among employed adolescents. Journal of .4miied Psvcholoev. 83,565-576.
Fukuyama,F. (1995). Trust: The social virtues and the creation of ~romeritv. New York: Free Press. Gardner, T. M.. Moynihan, L. M., Park, H. J., & Wright, P. M. (2000-August). Unlockin~the black box: Examinine the processes throueh which human resource practices imrtact business wrformance. Paper presented at the Academy of Management Annual Meeting, Toronto, Canada. Geller, E. S., Davis, L., Spencer. K. (1983). uidustry-based incentives for promoting seat belt use: Di fferential impact on whiteîo llar versus blueîollar employees. Journal of Or~anizationalBehavior Manaaement, 5 , 17-29.
Geller. E. S., Roberts, D. S.. & Gilmore, M. R. (1996). Predicting propensity to actively care for occupational safety. Journal of Safetv Research. 27, 1-8. Goldenhar, L. M., Ruder, A. M., Ewers, L. M., Earnest, S., Haag, W. M., & Petenen M. R. (1999). Concerns of the drytleaning industry: A qualitative investigation
of Iabor and management. American Journal of Industrial Medicine. 35,112- 123. Goodman, P. S., & Garber, S. (1988). Absenteeim and accidents in a dangerous
environment: Empirical analysis of underground coal mines. Journal of Aaoiied Pwchotow. 73,81-86.
G r i h , M. A., & Neal, A. (2000).Perceptions of safety at work: A fhnework for linking safety climate to d e t y perfomance, knowledge, and motivation. Journal of 15 1
Occu~aiionalHealth Psvcholorrv. 5,347-358.
Griffin, M.A., Neal, A., & Burley, 1. (2000-August). The i m ~ a cof t su~wrtive leadership and conscientiousness on safetv behavior at work. Paper presented at the Academy of Management Annual Meeting, Toronto, Canada.
Griffiths, D. K. (1985). Safety attitudes of management. Ereonomics. 28.6 1-67,
Gnuiberg, L.. Moore, S., & Greenberg, E. (1996). The relationship of employee ownenhip and participation to workplace safety. Economic and Industrial Democracv, 17.221-241. Guest, D.. Michie, J., Sheehan, M., & Conway, N. (n.d.). Gettine inside the HRM-aerfomance relationshio. Unpublished manuscript.
Hackett, R. D.. Bycio, P.. & Hausdorf, P. A. (1994). Further assessments of Meyer and Allen's ( 1991) three-component mode1 of organizational conmitment. Journal of A ~ d i e dPsvcholow. 79,1523. Hachan, J. R.. & Oldham, G. R. (1980). Work redesimi. Reading, MA: Addison-Wesley.
Hale, A. R. (1984). 1s safety training worthwhile? Journal of Occu~ational Accidents. 6,17-33. Hamper, B. (1991). Rivethead: Tales fiom the assemblv line. New York: Wamer. Haynes, R. S., Pine, R. C., & Fitch, H. G. (1982). Reducing accident rates with organizational behavior modification. Acadernv of Managment Journal. 25,40741 6.
Heinrich, H.W. (1959). Industrial accident orevention: A scientific aaaroach. New York: McGraw-Hill. Hemingway, M.A., & Smith, C. S. (1999). Organizational climate and 1st
occupational stresson as predictors of withdrawal behaviours and injuries in nurses. Journal of Occuoational and Oreanizational Psvcholow. 72,285-299. Hohano, D. A. (1999). Safety communication: ûrganizational influences and consequences. Proceedines of 1999 Amencan Sociehr of Safetv Eneineers Best Practices
in Safety Management S-mwsium. 165-175. Hohann. D. A-, & Jacobs, R.. & Landy, F. (1995). High reliability process
indusmes: individual. micro, and macro organizational influence on safety performance. Journal of Safety Research .26,13 1- 149. H o f m a ~D.~A., & Morgeson, F. P. (1999). Safety-related behavior as a social exchange: The role of perceived organizational support and leader-member exchange. Journal of Amlied Psvcholoev. 84,286-296. Hofmann, D. A., & Stetzer, A. (1996). A cross-level investigation of factors influencing unsafe behavion and accidents. Personnel Psvcholow. 49,307-339. Holcom, M. L.. Lehman, W. E. K.. & Simpson, D. D. (1993). Employee accidents: Influences of penonal charactenstics, job characteristics, and substance use in jobs differing in accident potential. Journal of Safetv Research. 24,205-221 Hoque, K. (1999). Hurnan resource management and performance in the UK hotel industry. British Joumal of Industrial Relations. 37.41943. Houston, D. M., & Mt,S. K. ( 1997). Psychological distress and error making among junior house officen. British Joumal of Health Psvcholoev. 2,141-1 51. Howell, J. M.,& Avolio, B. J. (1993). Redicting consolidated unit performance: Leadership behavior, locus of control, and support for innovation. Journal of A ~ ~ i i e d Pmcholow. 78.891-902.
Howlen, M., & Archer, V. E. (1984-November). Worker involvement in occupational health and safety. Farnilv and Comunitv Health, 57-63. Hoyos, C. G. (1992). A change in perspective: Safety psychology replaces the traditional field of accident research. The German Journal of Pmcholow. 16,123. Huselid, M. A. (1995). The impact of human resource practices on turnover, productivity, and corporate financiai performance. Academv of Management Journal. 38. 645-672.
Huselid, M. A., & Becker, B. E. (1996). Methodological issues in cross-sectionai and panel eaimates of the human resource-firm performance link. Industrial Relations, 35.400-422.
Ichniowski, C., Kochan, T. A., Levine, D., Olson, C., & Strauss, G. (1996). What
works at work: Overview and assesment. Industrial Relations. 35,299-333. Ichniowski, C.. Shaw, K., & Prennushi, G. (1997). The effects of human resource management practices on productivity: A study of steel finishing lines. American Economic Review. 87.29 1-313. Isla Diaz,R.. & Diaz Cabrera, D. (1997). Safety climate and attitude evaluation measures of organizational safety. Accident Analvsis and Prevention. 29,643-650. Iveaon, R. D., & Erwin, P. J. (1997). Predicting occupational injury: The role of affectivity. Journal of Occuoational and Oreanizational Psvcholoev. 70,113-128. Jackall, R. (1988). Moral mazes: The world of comorate rnanaeers. New York: ûxford. Jackson, S. E., & Schuier, R. S. ( 1985). A meta-analysis and conceptuai critique
. .
of reseaich on role ambiguity and role conflict in work s e m . m t i o n a l Behavior 154
and Hurnan Decision Processes. 36,16-78.
Jones, J. W. (1991). A personnel selection approach to industrial safety. In I. W. Jones (Ed.), Preem~lo-menthonestv testinq @p. 185-194). New York: Quorum. Jones, J. W., & Wuebker, L. J. (1988). Accident prevention through personnel selection. Journal of Business and Psvcholoev. 3,187-198. Joreskog, K. G., & Sorbom, D. (1992). LISREL Vm:Analvsis of linear structural relations. MooresvilIe, Pl:Scientific Software.
Jung, D. L, & Avolio, B. J. (2000). Opening the black box: An experimental investigation of the mediating effects of trust and value congmence on transformationai and transactional leadership. Journal of Oreanizational Behaviour. 2 1,949-964. Kaminski, M. (2001). Unintended consequences: Organizationai practices and their impact on workplace safety and productivity. Journal of Occu~ationalHealth 6,127-1 38. PSVC~O~OW.
Karan, B. S., & Kopelman, R. E. (1987). The effects of objective feedback on vehicular and industrial accidents: A field experirnent using outcome feedback. Journal of Qr~anizationalBehavior Manament. 8,45056.
Karasek R. A. ( 1979). Job dernands, job decision latitude, and mental strain: Implications for job redesign. Administrative Science Ouarterlv. 24,285-308. Karasek, R. A., & Theorell, T.(1990). Healtfiv work: Stress. ~roductivitv.and the reconstruction of workine life. New York: Basic Books. Kay, H. (1971). Accidents: Some facts and theones. In P. B. Warr (Ed.), Psvcholow at work @p. 121- 145). Hamondsworth, England: Penguin.
Keiloway, E. K. (1996). Common practices in structurai equation modeiing. In C. 155
L. Cooper & 1. T. Robertson (Eds.), International review of industrial and orpanizational psvcholow: Vol. 11 @p. 141-180). Chichester, England: Wiley. Kelloway, E. K. (1998). Using LISREL for structural eauation modelinp.
Thousand Oaks, CA: Sage. Kelloway, E. K., & Barling, J. (1993). Mernben' participation in local union activities: Measurement, prediction, and replication. Journal of A~oliedPs~choloev.78, 262-279. Kincaid, W. H. (1996, August). Safety in the high-turnover environment. Occu~ationalHealth and Safetv, 65,22,24-25.
Kivimaki, M., & Kalimo, R. (1993). Risk perception among nuclear power plant personnel: A s w e y . Risk Analvsis. 13,42 1-434. Kivimaki, M., Kalimo, R., & Salrninen, S. (1995). Perceived nuclear nsk, organizational cornmitment, and appraisals of management: A study of nuclear power plant personnel. Risk Analvsis. 15,391-396. Kling, J. (1995-May). High performance work systems and f irm performance. Monthk Labor Review, 29-36. Kochan, T. A., Smith, M., Wells, J. C., & Rebitzer, J. B. (1994). Human resource strategies and contingent worken: The case of safety and health in the petrochemical industry. Human Resource Management. 33.55-77.
Krarner, R. M. (1999). Trust and distrust in organizations: Emerging perspectives, enduring questions. Annual Review of Psvchologv. 50,569-598. Kraut, A. 1. (2001). An e-mail letter to a fiiend. The Industrial-Oreanizational
Kuhnert, K. W., & Vance, R. J, (1992). Job insecurity and moderaton of the relation between job insecurity and employee adjustment. In J. C. Quick L. R. Murphy, & J. J. Hunell, Jr. (Eds.), Stress and well-beinn at work: Assessments and interventions
for occu~ationalmental health (pp. 48-63). Washington, M3: American Psychological Association. Landen, D. D., & Hendncks, S. (1995). Effect of recall on reporting of ai-work injuries. Public Health Rmorts. 1 10,350-354. Lawler, E. E., 1II. (1996). From the eround un: Six orincides for buildine the new lopic cornoration. San Francisco, CA: Jossey-Bass. Leather, P. J. (1987). Safety and accidents in the construction industry: A work design perspective. Work & Stress. 1, 167- 174. Littauer, S. B. (1956). The application of statistical control techniques to the study
of industrial and militas, accidents. Transactions of the New York Academv of Sciences, 18.272-277. Lowe, K. B., Kroeck, K. G., & Sivasubramaniam, N. (1996). Effectiveness correlates of ttïmsformationai leadership: A meta-analytic review of the MLQ literatwe.
leaders hi^ Ouarterlv. 7,385-425. MacDuffie, S. P. (1995). Hurnan resource bundles and manufacturing performance: Organizational logic and flexible production systerns in the world automotive indu-.
Industrial and Labor Relations Review. 48,197-221.
MacEwen, K. E., & Barling, J. (199 1). Effects of matemal ernployment experiences on children's behavior via mood, cognitive difficulties, and parenthg behavior. Journal of Marrigee and the Farnilv. 53,635-644. 157
Martinez. P. G., Pearce, J. L., Porter, L. W., & Tsui, A. S. (2000). Oreanizationlevel measures and their conseauences: A ~reliminarvstudv of strategc investment in
human resource management. Paper presented at the Academy of Management Annual Meeting, Toronto, Canada. Mathieu J. E., & Zajac, D. (1990). A review and meta-analysis of the antecedents, correlates, and consequences of organizational cornmitment. Psvcholoeical -
Bulletin. 1OS, 17 1- 194. Mayer. R. C., Davis, J. H., & Schoorman, F. D.(1995). An integrative mode1 of organizational trust. Academv of Management Review. 20,709-734. McAfee, R. B., & Winn, A. R. (1989). The use of incentivedfeedback to enhance work place safety: A critique of the Iiterature. Journal of Safetv Research. 20.7- 19. McAllister. D. J. (1995). Affect- and cognition-based trust as foundations for interpersonal cooperation in organizations. Academv of Management Journal. 38.24-59. McLain, D. L. (1995). Responses to health and safety nsk in the work environment. Acadernv of Management Journal. 38.1 726-1 742. Meyer, 1. P.. & .Allen, N. 1. (1997). Commitment in the wor research and a~dication.Thousand Oaks, CA: Sage. Meyer, J. P., Allen, N. J.. & Smith, C. A. (1993). Commitment to organizations and occupations: Extension and test of a three-component conceptualization. Journal of
&mlied Psvcholoev. 78,538-55 1. Meyer, J. P., Paunonen. S. V., Gellatly, [. H., G o f i , R. D., & Jackson. D. N. (1989). Organizational cornmitment and job performance: It's the nature of the
cornmimient that counts. Journal of Atmlied Psvchology. 74,152-1 56. 158
Milanovich, D.M., Driskell, J. E., Stout, R. J., & Salas, E. (1998). Staîus and cockpit dynamics: A review and empirical study. gr ou^ Dvnamics: Theorv. Research and Practice. 2,155-1 67.
Neal, A., Griffin, M. A., & Hart, P. M. (2000). The impact of organizational climate on safety climate and individual behavior. Safetv Science. 34.99-109.
Nelkin, D., & Brown, M. S. (1984). Workers at nsk: Voices from the workplace. Chicago. IL: University of Chicago.
O'Dea, A., & Flin, R. (2000).Safetv leaders hi^ in the offshore oil and eas industry. Paper presented ai the Academy of Management Annual Meeting, Toronto, Canada. Oldham. G.R. (1996). Job design. In C.L. Cooper & 1. T. Robertson (Eds.), International review of industrial and oreanizational ~svchology:Vol. 1 1 @p. 33-60). Chichester, England: Wiley. Osteman, P. (1994). How common is workplace transformation and who adopts it? Industrial and Labor Relations Review. 47,173-1 88.
Parker, D. L., Carl, W. R., French, L. R., & Martin, F. B. (1994). Characteristics
of adolescent work injuries reported to the Minnesota Department of Labor and Lndustry. Arnerican Journal of Public Health, 83,606-6 11. Parker, S. K., & Axtell, C., & Turner, N. (2001). Designing a safer workplace: Importance of job autonomy, communication quality, and supportive supervisors. Journal of Occu~ationalHealth Pmcholow. 6,2 11-228.
Parker, S. K.,& Wall, T. D. (1998). Job and work desipli: Oreanizine work to promote well-beinn and effectiveness. Thousand Oaks, CA: Sage. 159
Parker, S. K., Wall, T. D., & Jackson, P. R. (1997). "That's not my job": Developing flexible employee work orientations. Academv of Management Journal. 40,
899-929. Pasmore, W.. & Friedlander, F. (1 982).An action-research program for increasing employee involvernent in problem solving. Administrative Science Ouarterlv. 27,343-
362. Patterson, M. G.,West, M. A., & Wall,T.D.(201).Inte-mted manufacturine, empowement and comoay oerfomance. Manuscript submined for publication. Peten, T. J., & Waterman, R. H. (1982).In search of excellence: Lessons fiom America's best-run com~anies.New York: Harper & Row. Pfeffer, J. (1995).Producing sustainable competitive advantage through the effective management of people. Academv of Management Executive. 9.55-69. Pfeffer, J. (1997).New directions for or~anizationtheorv: Problems and promects. New York: Oxford University Press. Pfeffer, J. (1998a).Seven practices of successhil organizations. California Management Review. 40.96- 124
Pfeffer, J. (1998b,Spring). The real keys to high performance. Leader to Leader, 23-29. Pfeffer, J. (1998c,May-June). Six dangerous rnyths about pay. Harvard Business
Review, p. 109-119. Pfeffer, J. (1998d).The human eauarion: Buildin~ - arofits bv outtinn ~eoblefirst. Boston, MA: Harvard Business School Press. Pillai, R, Schriesheim, C. A., & Williams, E. S. (1999). Fairness perceptions and
trust as mediators for tramformational and transactional leadership: A two-sample study. Journal of Management. 25,897-933. Pinder, C. (1998). Work motivation in oreanizational behavior. Upper Saddle River, NJ: Prentice Hall.
Pransky, G.,Snyder, T., Dembe, A., & Himmelstein, J. (1999). Under-reporting oiwork-related disorders in the workplace: A case study and review of the literature. Ereonomics. 42,17 I - 182. Preston, R., & Topf, M. (1994, January). Transfomi fear io trust: Communicating about unsafe job incidents and settings. Occu~ationalHealth and Safetv. 63.60-6 1. Probst, T. M., & Bnibaker, T. L. (2001). The effects of job insecurity on employee safety outcornes: Cross-sectional and longitudinal explorations. Journal of Qccu~ationalHeaIth PsvchoIoev. 6,139-1 59. Raggatt, P. T. F. (199 1). Work stress among long-distance coach driven: A survey and correlational study. Journal of Oreanizationai Behavior. 12,565-579.
Reason, J., Parker, D., Lawton, R. (1998). Organizational controls and safety: The varieties of rule-related behaviour. Journal of Occuaational and Ooanizational Psvcholow. 71,289-304. Reber, R. A., & Wallin, I. A. (1984). The effects of training, goal setting, and knowledge of results on safe behavior: A component analysis. Academy of Management Journal. 27,544-560. Reber, R A., Wallin, J. A., & Chhokat, J. S. (1984). Reducing industrial accidents: A behavioral experiment. Industrial Relations. 23.1 19- 125.
Reber, R A., Wdlin, J. A., & Duhon, D. L. (1993). Preventing occupational 161
injuries through performance management. Public Personnel Management. 22.30 1-311. Rebitzer, J. B. (1995). Job safety and contract worken in the petrochemical
industry. Indusîrial Relations. 34,4047. Roche, W. K. (1999). in search of cornmitment-onented hurnan resource management practices and the conditions that sustain them. Journal of Management Studies. 36,653-678. Root, N. (1981, Mach). Injuries at work are fewer among older employees. Monthlv Labor Review, 30-34. Sauter, S. L.,Murphy, L. R., & Hurrell, J. J., Jr. (1990). Prevention ofworkrelated psychological disorders: A national strategy proposed by the National Institute for 45,1146-1 158. Occupational Safety and health (MOSH). American Ps~cholo~ist.
Schmitt, N., Colligan, M. J.. & Fitzgerald, M. (1980). Unexplained physical
symptoms in eight organizations. individual and organizational analyses. Journal of Occu~ationalPsvcholo~.53,305-3 17. Sells. B. (1994, March-April). What asbestos taught me about managing nsk. Harvard Business Review, 76-90. Shain, M. (1982). Alcohol, drugs and safety: An updated perspective on problems and their management in the workplace. Accident Analvsis and Prevention. 14,239-246. Shannon, H. S., Wdters, V., Lewchuk, W., Richardson, J., Moran, L. A., Haines,
T.,& Verma, D. (1996). Workplace organizational correlates of Iost-time accident rates in manufacruring. Amencan Journal of Industrial Medicine. 29,258-268. Sheehy, N. P., & Chapman, A. J. (1984). Accidents and safety. In A. Gale & A. J.
@p. 17 1- 190). Chichester, England: Wiley. Sheehy, N. P., & Chapman, A. J. (1987). Industrial accidents. tn C. L. Cooper & 1. T. Robinson (Eds.), International review of industrial and oreanizational ~ s v c h o l o ~ y
(pp. 20 1-227). Chichester, England: Wiley.
Shin, D.,& Lee, J. (1999-August). Tmst in lean ~roductionsvstems: Lean iob desig and worken' trust in management at Korean automobile ~lants.Paper presented a the Academy of Management Annual Meeting, Chicago, K.
Simard, M., & Marchand, A. (1994). The behaviour of first-line supeMsoa in accident prevention and effectiveness in occupational safety. Safetv Science. 17,169185. Simard, M., & Marchand, A. (1995). A multilevel analysis of organizational factors related to the taking of safety initiative by work groups. Safetv Science. 2 1. 113129.
Simard, M., & Marchand, A. (1 997). Workgroups' propensity to comply with safety niles: The influence of micro-macro organizational factors. Ereonomics. 40.172188.
Simonds, R. H., & Grimaldi, J. V. (1963). Safetv management: Accident cost and control. Homewood, IL: h i n . Sjoberg, L., & Drottz-Sjilberg, B. M. (1991). Knowledge and risk perception
among nuclear power plant ernployees. Bisk Analvsis. 11,6076 18.
Smith, M. J., Cohen, H.H., Cohen, A., & Cleveland, R. J. (1978). Characteristics of successfûl safety programs. Journal of Safetv Research. 10.5- 15. Smith, R B. (1994, June). Mobil refinery proudly waves flag as an OSHA 'Star' 163
VPP participant Qccu~ationalHealth and Safetv. 63,24-25,29. Sparks, K.,Cooper, C. L., Fried, Y., Shirom, A. (1997). The effects of hours of work on health: A meta-analytic review. Journal of Occu~ationaland Orpankational Psvcholow. 70,39 1308. Sulzer-Azaroff, B., Loafinan, B., Merante, R. J., & Hlavacek, A. C. (1990). Improving occupational safety in a large industrial plant: A systematic replication. 1 1,99- 120. Journal of Oreanizational Behavior Management,
Sutherland, V. J., Cooper, C.L. (1 99 1). Personality, stress and accident involvement in the offshore oil and gas industry. Personalitv and Individual Differences,
12, t 95-204. Tannenbaum, S. 1.. Mathieu, J. E., Salas, E., & Cannon-Bowers, J. A. ( 1991). Meeting trainees' expectations: The influence of training fulfillment on the development
of cornmitment, self-effkacy, and motivation. Journal of A ~ ~ l i Psvchdow. ed 76,759769. Thompson. R. C., Hilton, T.F., & Witt, L. A. (1998). Where the safety rubber meets the shop floor: A confirmatory mode1 of management infîuence on workplace safety. Journal of Safetv Research. 29.15-24. Tjosvold, D. (1990). FIight crew collaboration to manage safety risks. gr ou^ and Organization Studies. 1 5, 177- 191.
Trist, E. L.,Susman, G.L, & Brown, G. R. (1977). An experiment in autonomous working in an American underground coal mine. Human Relations. 30,201-236.
Tsui, A. S., Pearce, J. L., Porter, L. W., & Tripoli, A. M. (1997). Alternative approaches to the employee-organization relationship: Does investment in employees pay 164
off! Academv of Management Journal. 40,1089- 112 1. -
United States Census Bureau. (2000). Statistical abstract o f the United States: 2000 (120" ed.). Washington, DC: Author. -
Veazie, M. A., Landen, D. D., Bender, T. R., & Arnandus, H. E. (1994). Epiderniologic research on the etiology of injuries at work. Annual Review of Public Health. 15,203-22 1.
Waldman, S., de la Pefia, N.. Springen, K., Howard, J., & Srni& V. E. (1989, December 1 1). Danger on the job. Newweek, 4 2 4 , 1 8 . Walker, C. ( 1998July). Behaviour based safetv ~ r o m s or. "if it's rat r who are the rats?'Paper presented at the Association psvcholow. who is Pied P i ~ e and of Worken' Compensation Boards of Canada Congress, Winnipeg, Manitoba. Wall. T. D., Corbett, J. M., Manin, R.. CIegg, C. W., & Jackson, P. R. (1990). Advanced manufacturing technology, work design, and performance: A change study. Journal of Aooiied Psvcho~ow.75,691 -697. Wdton, R. E. (1985. March-April). Fmm control to cornmitment in the work place. Harvard Business Review, 77-84.
Warr, P. B. (1 987). Work. unem~lovment.and mental health. Oxford: Oxford University Press. Weick K. E. (1987). ûrganizational culture as a source of high reliability. California Mana~ementReview. 29,112-1 27. Wheatley, M. (1997, Summer). Goodbye, command and control. Leader to Leader, 2 1-28.
Williams, H.,Turner, N., & Parker, S. K. (2000). The cornmmsatorv role o f 165
transfomational leadenhio in oromotine safehr behaviors. Paper presented at the Academy of Management Annual Meeting, Toronto, Canada Wood, S. (1999a). Human resource management and performance. International Journal of Management Reviews. 1,367-4 13. Wood, S. (1 999b). Getting the measure of the transformed hi&-performance organization. British Joumal of Indushial Relations. 3 7 , 3 9 1 4 7. Wood, S., & Aibanese, M. T. (1995). Cm we speak of a hi@ cornmitment management on the shop floor? Joumal of Management Studies. 32,215-247. Wood, S.. Barling, J., Lasaosa, A., Tumer. N., & Parker, S. (2001). Qreanizational oraciices and safetv oerforrnance. Unpublished manuscnpt. Wood, S., & de Menezes. L. (1998). High cornmitment management in the U.K.: Evidence fiom the Worlcplace Industrial Relations Survey and Employers ' Manpower and Skills Ractices Survey. Human Relations, 5 1,185-5 15.
Wright C. (1986). Routine deaths: Fatal accidents in the oil industry. Socioloeical Review. 34,265-289. Zacharatos, A., Barling, J., & Kelloway, E. K. (2000). Development and effects of transformational leadership in adolescents. Leadershi0 Ouarterlv. 11.2 11-226. Zohar, D. (1980a). Safety climate in industrial orgaxiizations: Theoretical and applied implications. Joumal of A ~ ~ l i Pwcholow. ed 65.96-1 02. Zohar, D. (1980b). Promoting the use of persona1 protective equipment by behavior modification techniques. Journal of Safetv Research. 12.78-85. Zohar, D. (2000). A group-level mode1 of safay climate: Testing the effect of group climate on microaccidents in manufacturingjobs. Joumal o f Amlied P s v c h o l o ~ 166
85.587-596.
Zohar, D. (2001). The effect of safetv-climate and le ad en hi^ factors on microaccidents in work mouos. Manuscnpt submitted for publication. Zwerling, C., Sprince, N. L., Wallace, R. B., Davis, C. S., Whitten, P. S., &
Heeringa, S. G. (1996). Risk factors for occupational injuries among older worken: An analysis of the health and retirernent study. American Journal of Public Health. 86, 13061309.
Appendix A Items Measurhg High Performance Management Practices
at the Organizational Level
Unless otherwise indicated, responses to items were on a 5-point Likert scale with options ranging from 'strongly disagree' (1) to 'strongly agree' (5).
E mployment Security 1.
2. 3. 4.
5. 6.
7.
Roviding ernployment security to our employees is a priority in this organization. If an employee were to lose hisher job, this organization would ûy very hard to find himher another position elsewhere in the organization. What percentage of non-entry level jobs have been fiIled with intemal candidates in m e n t (Le., over the last three) yean? What percentage of the workforce could work in this organization as long as they want to? What percentage of UUs organization's workforce has been laid off in the 1 s t 2 years? (5) What percentage of this organization's work force has employment secwity? How many rounds of layoffs has your organization undergone in the 1 s t 2 years? (R)
The values and beliefs of this organization are discussed in interviews with potential employees. 2. Only the best people are hired to work in this organization. 3. Employees of this o r g d t i o n are involved in the hiring of their peers. 4. What percentage of job applicants does your fimi eventually hire? (RI 5. What percentage of job applicants undergo more than one interview before being hired? 6. What percentage of job applicants undergo more than two inteniews before being hired?
1 2 3 4 5
1 2 3 4 5
YO %
% %
rounds
1.
1 2 3 4 5 1 2 4 5 1 2 3 4 5 % % %
Training 1.
2. 3. 4.
5. 6.
Roviding employees with training beyond that mandated by govenunent regulations is a priority in this organization. 1 2 3 4 5 This organization subsidizes, assists or reimburses employees for training or courses taken outside of the workplace. 1 2 3 4 5 Employees in this organization receive additional compensation for training they get outside the organization. 1 2 3 4 5 What percentage of this organization's froibiine employees received training beyond that rnandated by govemment regulations in the last 12 months? Y0 What percentage of your organization's supervisory or managerial employees received training beyond that mandated by govemrnent regulations in the last 12 months? % What is the average number of houn of training beyond that mandated by govemment regulations received by a front-line employee in the last 12 months? hom
7. Whatistheaveragenumberofhoursoftrainingbeyondthat mandated by govemment regulations received by a supervisory or managerial employee in the last 12 months?
hours
Teams and Decentralized Decision-making The development of teams is an important element of this organization's corporate strategy. 2. This organization supports team development and training. 3. Employee suggestions are implemented in full or in part within this organization. 4. Decision-making by non-managerial ernployees is encouraged in this organization. 5. What percentage of employees in your organization work in semiautonomous groups (Le., groups with a high Ievel of responsibility)? 6. What percentage of employees in your organization work in problem-solvuig teams (i .e., teams with responsibility over a narrow range of issues)? 1.
1 2 3 4 5
1 2 3 4 5 1 2 3 4 5 1 2 3 4 5
%
%
Reduced Status Distinctions
The use of formal titles within this organization is enforced. (R) Employees at different levels within this organization are encouraged to interact. 3. Some memben of this organization have privileges that are unavailable to other members (e.g.: reserved parking and dining facilities). (R) 1. What percentage of the workforce is invited to company wide social events? 5 . What is the highest annual salary (including bonuses) paid to a full-time employee in your organization? 6. What is the lowest annual salary (including bonuses) paid to a fulltime employee in your organization? 1. 2.
1 2 3 4 5
1 2 3 4 5
1 2 3 4 5 %
S S
Ialormatioioii Sharing The findings fiom employee s w e y s are communicated tu members of this organization. 2. This organization keeps secrets from empioyees. (R) 3. Our cornpetiton know more about this organization than our employees themselves do. (R) 1. Information about how well the organization is performing financially is shared with employees. 5 . What percentage of the workforce is included in a formal information sharing program (e.g.. a newsletter, ernails)? 6 . What percentage of the workforce is formally notified about how the organization is perfomiing? 1.
1 2 3 4 5 1 2 3 4 5
1 2 3 4 5 1 2 3 4 5
% %
Contingent Compeusation 1. 2.
3. 4.
S.
This organization pays above market wages. The way in which employees in ihis organization are compensated encourages them to adopt a long-term focus. What percentage of the woMorce has access to company incentive plans, profit-sharing plans, a d o r gain-sharing plans? What percentage of the workforce is compensated (at least in part) based on their individual performance? (R) What percentage of the workforce is compnisated (at least in part) based on their perfomance within a group?
1 2 3 4 5 1 2 3 4 5 % %
%
Transformational Leadership 1.
2. 3. 4.
5.
6.
Leaders in this organization are focused on doing the right thing as well as on getting results. Leaders in this organization encourage employees to look at problems and come up with their own solutions and suggestions. Leaders in this organization listen to the concems of employees. Leaders in this organization express their confidence that the organization will achieve its goals. Leaders in this organization encourage employees to express their ideas and opinions. Leaden in this organization provide employees with continuous encouragement.
1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5
Job Quality
Fostering involvement in decision-mahg at al1 levels of this organization is an important element of the corporate strategy. 2. Many employees in this organization perform simple and repetitive tasks as part of their work. (R) 3. Providing ernployees with high quality jobs (i.e., jobs that are challenging, hilfilling, etc.) is a priority in this organization. 1. Ernployees of this organization are given lots of oppominity to decide how to do their work. 5 . What percentage of the workforce works overtime ho-? (R) 6 . What percentage of non-supe~soryemployees make decisions regarding how they do their work? 7. What percentage of front-line employees in your organization make decisions regarding how they do their work (e-g.: decisions about workflow. scheduling)? 1.
1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5
YO %
YO
Measu rement The measurernent of an employee's performance on the job is a pnority in this organization. 2. The measurement of occupational safety is a priority in this organization. 3. This organization makes a point of keeping track of factors that it considers criticai for success. 4. The measurement of turnover and absenteeism is a priority in this organization. 5. Wbat percentage of the workforce completes anitude surveys on a regular basis? 6 . What percentage of the workforce receives fomal performance appraisals? 7. What percentage of the workforce completes surveys measuring employee morale (Le., satisfaction. mut in management,loyalty)? 1.
(R) denotes a reverse-keyed item
1 2 3 4 5 1 2 3 4 5 1 2 3 4 5
1 2 3 4 5 Y0 % %
Appendix B Cover Letten Used in Study One
Dear Human Resources Director, Re: Human Resources and Occu~ationalSafety
We are h t i n g to invite you to panicipate in a study investigating the reiatiouship between human resource practices and occupational safety. Currently, very linle is known about the role of management pnctices in reducing occupational accidents and injuries. By completing the enclosed survey, you will contribute to a berter understanding of m;inagement's role in this very important issue. This particular survey asks you, as a senior Human Resource professional, to answer questions regarding the use of hurnan resource management practices in your organization and your organization's safety record. We ask that you please take ten minutes of your time to complete the cncloscd survey. A preaddressed stamped envelope bas been provided for you to r e m the survey. Be assurcd that your responses in this study are anonymous and will be aeated conf~dentially.As part of this study, the Health and Safety Manager in your organization has also k e n asked to complete a survey regarding yow organization's safety record. The code found on your survey will oniy be used to match yow survey with that of your organization's Safety Officer. Your organization ç w o t be identified based on ths number. Your participation in this study is voluntaty. Your consent to participate in this study will be assumed by the r e m of your completed suntey. If you are unable to complete this survey, please r e t m the blank survey in the envelope provided. Please answer each question to the best of your ability. If, for any reason, you have any conccm or cornplaints related to this study please contact Anthea Zacharatos at (6 13) 544-804 1, ~acharatosti3business.aueensu.caor Dr. Julian Barling at (613) 533-2477. If you would like a copy of the study results please contact Anthea Zacharatos and a summary of the results will be sent to you once they are available. In order CO maintain your anouymity pleue do not amch a request for smdy results to your m c y .
Thank you very much for taking the cime to complete and r e m this nuvty. Sincerely,
Anthea Zacharatos Doctorai Candidate
J u h Bariing, Ph-D. Professor of Organizatiod Behaviour
* Plcasc enjoy a cup of tea on us whiIe f&g
out the survty.
Dear Health and Safety Manager,
Re: iiuman Resources and Occu~ationalSafetv
We are writing to invite you to participate in a study investigating the relationship between human resource practices and occupational safety. Cmently, very little is known about the role of management practices in reducing occupationa1accidents and injuries. By completing the niclosed survey. you will contribute to a better understanding of management's role in this very important issue. This prncular survey asks you, as the Health and Safety Oficer, to answer questions regarding your organization's safety record. We ask that you please take ten minutes of your time to complete the enclosed survey. A pre-addressed stamped enveiope has been provided for you to r e m the survey. Be assured that your responses in this study are anonymous and will be treated confidentially. As part of this study, the Human Resources Director in your organization has aiso been asked to complete a survey regarding your organization's human resource management practices and safety record. The code found on your survey will only be used to match your survey with that of your organization's Human Resources Director. Your organization cannot be identified based on this number.
Your parûcipation in this study is voluntary. Your consent to paxticipate in ths study will be assumed by the return of your completed survey. if you are unable to complete this survey, please return the blank survey in the envelope provided. Please answer each question to the best of your ability. If, for any reason, you have any concems or cornplaints related to this study please contact Anthea Zacharatos at (613) 5448O4 1, azacharatos0business.aueensu.caor Dr. Julian Barling at (6 13) 533-2477, If you would like a copy of the study results please contact .Anthea Zacharatos and a summary of the resuits will be sent to you once they are available. In order to rnaintain your anonymity please do not attach a request for study results to your survey.
Thank you very much for taking the time to complete and return this survey.
Anthea Zacharatos Doctoral Candidate
Julian Barling, PhD. Professor of Organizational Behaviour
*
Please enjoy a cup of tea on us while filling out the survey.
INDUSIRIAL ACCIDENT PRMIUTION ASSOClffflON 250 Yonge Street. Toronto. Ontario MSB 2N4 Telephone (416) 506-8888 Fax (416 ) 506-8880
Dear Participant, Resuvchers fiom the School of Business at Queen's University are currentiy coaducting a study examinhg the relatioaship of management and organîzational practices and occupational d e y .
The contribution of this N y ro our understanding of how to manage accidents and injuries in the workplace is potentially sigmficant. This snidy differr fiom much exisMg safety research in that it does not focus on the engineering and monitoring of work, but rather on