14
Perceptions of organizational safety: implications for the development of safety culture SHARON CLARKE * Manchester School of Management, UMIST, Manchester, M60 1QD, U.K. Summary Regulatory bodies in the U.K. have emphasized the importance of ‘a positive health and safety culture’ to the safety performance of companies. A key feature of a company’s safety culture is shared perceptions amongst managers and sta concerning the importance of safety. This study recorded the perceptions of 312 British Rail train drivers, supervisors and senior managers concerning the relative importance of 25 railway factors. Each level also gave their estimates of the ratings of the other levels. The study found that whilst there was a shared perception of the importance of safety, intergroup perceptions were not realistic. It is argued that accurate intergroup perceptions are essential to the development of mutual trust and understanding between levels, which forms the basis for a positive safety culture. Suggestions for establishing the foundations for a positive safety culture are discussed. Copyright # 1999 John Wiley & Sons, Ltd. Introduction In the U.K., regulatory bodies, such as the Health and Safety Executive (HSE), have encouraged companies to develop a ‘positive safety culture’ as a means to successful health and safety management (HSE, 1991). Harrison (1972) defined organizational culture in terms of the beliefs and values of the organization, which act as prescriptions for the way in which organizational members should work. ‘Safety culture’ may be perceived as a subset of organizational culture, where the beliefs and values refer specifically to matters of health and safety. Schein (1992) suggests that the way in which senior managers instruct, reward, allocate their attention and behave under pressure, will be particularly salient in shaping organizational culture. Perceptions of senior managers’ attitudes and behaviours in relation to the safety and well-being of the workforce, therefore, will form the basis for the safety behaviour of workers, and therefore, the * Correspondence to: Sharon Clarke, Manchester School of Management, UMIST, P.O. Box 88, Manchester M60 1QD, U.K. Tel: 0161 200 3442. Fax: 0161 200 3505. email: [email protected] Contract grant sponsor: Science and Engineering Research Council (SERC). Contract grant sponsor: British Rail (BR). CCC 0894–3796/99/020185–14$17.50 Copyright # 1999 John Wiley & Sons, Ltd. Accepted 7 March 1997 Journal of Organizational Behavior J. Organiz. Behav. 20, 185–198 (1999)

Perceptions of organizational safety: implications for the development of safety culture

Embed Size (px)

Citation preview

Page 1: Perceptions of organizational safety: implications for the development of safety culture

Perceptions of organizational safety:implications for the developmentof safety culture

SHARON CLARKE*

Manchester School of Management, UMIST, Manchester, M60 1QD, U.K.

Summary Regulatory bodies in the U.K. have emphasized the importance of `a positive health andsafety culture' to the safety performance of companies. A key feature of a company'ssafety culture is shared perceptions amongst managers and sta� concerning theimportance of safety. This study recorded the perceptions of 312 British Rail traindrivers, supervisors and senior managers concerning the relative importance of25 railway factors. Each level also gave their estimates of the ratings of the otherlevels. The study found that whilst there was a shared perception of the importance ofsafety, intergroup perceptions were not realistic. It is argued that accurate intergroupperceptions are essential to the development of mutual trust and understanding betweenlevels, which forms the basis for a positive safety culture. Suggestions for establishing thefoundations for a positive safety culture are discussed. Copyright # 1999 John Wiley &Sons, Ltd.

Introduction

In the U.K., regulatory bodies, such as the Health and Safety Executive (HSE), have encouragedcompanies to develop a `positive safety culture' as a means to successful health and safetymanagement (HSE, 1991). Harrison (1972) de®ned organizational culture in terms of the beliefsand values of the organization, which act as prescriptions for the way in which organizationalmembers should work. `Safety culture' may be perceived as a subset of organizational culture,where the beliefs and values refer speci®cally to matters of health and safety. Schein (1992)suggests that the way in which senior managers instruct, reward, allocate their attention andbehave under pressure, will be particularly salient in shaping organizational culture. Perceptionsof senior managers' attitudes and behaviours in relation to the safety and well-being of theworkforce, therefore, will form the basis for the safety behaviour of workers, and therefore, the

* Correspondence to: Sharon Clarke, Manchester School of Management, UMIST, P.O. Box 88, Manchester M60 1QD,U.K. Tel: 0161 200 3442. Fax: 0161 200 3505. email: [email protected]

Contract grant sponsor: Science and Engineering Research Council (SERC).Contract grant sponsor: British Rail (BR).

CCC 0894±3796/99/020185±14$17.50Copyright # 1999 John Wiley & Sons, Ltd. Accepted 7 March 1997

Journal of Organizational BehaviorJ. Organiz. Behav. 20, 185±198 (1999)

Page 2: Perceptions of organizational safety: implications for the development of safety culture

safety performance of the company. The safety culture approach to accident reductionemphasizes the role played by social forces within an organization; whilst these are characteristicof the organization as a whole, social forces will act upon the individual operator's cognitions,perceptions and behaviour, with respect to health and safety at work.

Research has focused on de®ning the dimensions of safety culture (e.g. Zohar, 1980; Brownand Holmes, 1986; Dedobbeleer and Beland, 1991). The seminal work of Zohar (1980) factor-analysed a 40-item questionnaire to produce two primary dimensions: `perceived relevance ofsafety to job behaviour' and `perceived management attitude towards safety'. Studies havesupported the role of the ®rst dimension, e.g. `perceptions of personal threat from hazards' hasbeen found to in¯uence workers' willingness to participate in safety programmes (Goldberg,Dar-El and Rubin, 1991). However, whilst perceptions of personal risk are fundamental to safetybehaviour, Zohar argues that the cognitions that guide behaviour `are largely related to percep-tions of management attitudes about safety' (p. 101). More recent research, e.g. Cooper andPhillips (1994), rede®nes safety culture along the dimensions of direct and indirect in¯uences.Within this framework, perceptions of management attitudes and actions have a direct e�ect onworkers' behaviour; whilst indirect in¯uences include, status of safety o�cer, safety training ande�ect of safe conduct on promotion. Although there is little empirical evidence for the directe�ects of dimensions of safety culture on the actual behaviour of employees (Melia, Tomas,Oliver and Islas, 1992; Oliver, Tomas and Melia, 1993), there is at least correlational evidencethat positive management attitudes and actions have bene®ts, in terms of the safety performanceof the company (e.g. Gri�ths, 1985). Conversely, there is evidence that a perceived lack ofpositive attitudes and actions can have negative e�ects, e.g. Clarke (1996) demonstrates thatperceived managerial commitment to safety is predictive of incident reporting by sta�, with thosewho perceived managers as not committed to safety, less likely to make reports.

Whilst research evidence suggests that developing a positive safety culture will improve safetyperformance, there is less guidance on how companies might achieve such improvements throughcultural change. Weick (1987) suggests that highly reliable performance can be achieved throughthe development of an organizational culture that encourages `interpretation, improvization andunique action' (p. 124). For such a culture to exist, there has to be trust, openness and mutualunderstanding, on the part of both workers and managers. These qualities are re¯ected in theU.K. Health and Safety Commission's (1993) description of a positive safety culture, whichshould be characterized `by communications founded on mutual trust, by shared perceptions ofthe importance of safety, and by con®dence in the e�cacy of preventive measures. The viability ofdeveloping a positive safety culture will depend, therefore, on a number of factors, including: thequality of sta�±management communication; agreement at all hierarchical levels that safety isimportant; con®dence amongst the workforce that safety measures are adequate.

The Present Study

The present study will examine the perceptions of the importance of a number of safety issues atthree levels of an hierarchical transport organization, British Rail (BR). The three levels will be:sta� (train drivers); train drivers' supervisors and locally-based managers; senior managers. Ateach level, respondents will indicate their own views and their perceptions of the other two levels'views. The ®rst hypothesis concerns the nature of the respondents' own views. As a positive safetyculture depends on shared perceptions of the importance of safety issues, it is hypothesized that

Copyright # 1999 John Wiley & Sons, Ltd. J. Organiz. Behav. 20, 185±198 (1999)

186 S. CLARKE

Page 3: Perceptions of organizational safety: implications for the development of safety culture

own views at each hierarchical level will agree on the issues that are important to railway safety,demonstrated by shared high ratings.

Previous studies have used the technique of measuring perceptions of a `typical other' toexamine corporate culture and value congruence (e.g. Likert, 1961; Meglino, Ravlin and Adkins,1989) and attitudes about industrial relations (e.g. Walker, 1959, 1962; Stephenson, Brotherton,Skinner and Dela®eld, 1983). Stephenson and Allen (1987) emphasize the importance of mutualintergroup perceptions as each group will have perceptions of what the other will do. IntergroupUnderstanding (the extent to which there is a mismatch between one group's own view and theirperceptions of the other group) falls into three major categories: group assimilation (where twogroups wrongly perceive agreement between their own view and the view of the other); groupdi�erentiation (where two groups hold negative stereotypes about each other's view); realistic(where two groups have accurate perceptions of the other's view). The second hypothesis con-cerns the intergroup understanding demonstrated by each level's perceptions of the other levels'views. As a positive safety culture depends on trust and mutual understanding, it is hypothesizedthat perceived views will re¯ect accurate perceptions of the other levels' views (realistic intergroupunderstanding).

Biases in perceptions of other levels' views may arise from a number of sources. For example,whilst the literature has emphasized the importance of sta� perceptions of senior managers'attitudes to safety, sta� are likely to have little direct evidence of these managerial attitudes.Rather, ®rst-line supervisors will provide the sta�'s primary point of contact with the manage-ment structure, and therefore, supervisors' behaviour and expression of views will in¯uence thedevelopment of sta�'s opinions about management and their policies. These have been said torepresent what the organization `really wants' from its employees as `intentions communicated inthis way are not always consistent with those projected in o�cial company safety documents'(Leather, 1987, p. 171). This bias would be re¯ected in a close association between sta� percep-tions of supervisors/local managers and their perceptions of senior managers (hypothesis 3).

Similarly, senior managers are removed from the operational end of the company and mayhave little direct experience of sta�'s opinions. Stephenson et al. (1983) found that: `managers arethe only group who . . . permit their own views to determine their perceptions of how the `typicalother' sees things' (p. 38). Thus, in the absence of direct experience, managers may project theirown views onto their subordinates; this bias would be re¯ected in a close correspondence betweenmanagers' estimates of the ratings of the other levels and their own ratings (hypothesis 4).

Method

Sample

Questionnaires were distributed to a total of 1423 BR employees; of these, 312 completedquestionnaires were returned (a response rate of 22 per cent). Despite the low return rate, thesample represented a broad section of the driver population, drawn from ®ve di�erent areas ofthe BR network, across a wide range of age and driving experience. Supervisory sta� were drawnfrom the same ®ve areas as the driver sample. Senior managers were drawn from varyingdepartments, based throughout the U.K. The original sample comprised 186 train drivers,55 supervisors and local managers, and 71 senior managers; however, only complete question-naires were included, yielding useable data from 170 drivers, 53 supervisors and local managersand 66 senior managers. Drivers were aged between 21 and 64 years (M� 42.4 years), supervisors

Copyright # 1999 John Wiley & Sons, Ltd. J. Organiz. Behav. 20, 185±198 (1999)

PERCEPTIONS OF ORGANIZATIONAL SAFETY 187

Page 4: Perceptions of organizational safety: implications for the development of safety culture

and local managers between 34 and 64 years (M� 50.5 years) and senior managers between25 and 62 years (M� 42.6 years). Re¯ective of the industry average, 7 per cent of the seniormanagement sample were female (there were no female supervisory sta� or train drivers).

Questionnaire design and procedure

The questionnaire was developed on the basis of analyses of internal accident reports anddiscussions with BR managers. A total of 25 railway factors were included, representing a rangeof important safety issues, from top-level planning to drivers' skills (see Appendix 1).

Demographic details (sex, age, length of service, department/depot) were requested. The mainbody of the questionnaire comprised three sections. In the ®rst section respondents were asked torate a list of 25 items `in your own opinion' in response to the following question: `ConsideringBR at the present time, to what extent do the following factors have an adverse e�ect on the safetyof the railway?'

Respondents rated the factors by circling a ®gure on a 5-point scale (1� none at all; 5� verygreat). In the remaining two sections, respondents rated the same 25 items but were requested torespond `what you think is the opinion of local managers and supervisors', `what you think is theopinion of drivers' or `what you think is the opinion of top management' as appropriate. Itemswere listed in random order on the questionnaire, but in the same order for all three sections.

Questionnaires were distributed to drivers and supervisors at their depots and given to seniormanagers attending safety meetings. Data collection took place in the period October 1991±January 1992. Respondents were assured of con®dentiality and were not required to give theirnames. Questionnaires were returnable in envelopes provided via the internal mail system of thecompany.

Results

Descriptive statistics for the safety ratings of railway personnel are given in Table 1. This showsthat there is a high awareness of the safety implications across a range of issues, with all meanratings above 2.50.

Hypothesis 1, that there will be shared perceptions on the importance of safety issues acrosshierarchical levels, was tested using one-way MANOVA.1 Multivariate F was signi®cant(p5 0.001), indicating that there were signi®cant di�erences in the safety ratings depending onhierarchical level. Stepdown analyses were performed for each item. These revealed that safetyratings for the three hierarchical levels di�ered signi®cantly for: pay, management control,making pro®ts, reorganization (see Table 2).2

Actual di�erences in opinion seem to derive from traditional divides on pay/workingconditions, management control/supervision and the drive to make pro®ts. Adjusted marginal

1 As the data set showed evidence of multicollinearity, the four variables with the highest SMCs (all 40.60) wereexcluded from this analysis, leaving 21 dependent variables; items excluded were: sighting of signals, route knowledge,drivers' diagrams, drivers' skills (see Tabachnick and Fidell, 1989).2 Order of priority in stepdown analysis was determined by the size of univariate F. Items with signi®cant F-values forboth the univariate and the stepdown test are interpreted in this and following analyses (see Tabachnick and Fidell,1989).

Copyright # 1999 John Wiley & Sons, Ltd. J. Organiz. Behav. 20, 185±198 (1999)

188 S. CLARKE

Page 5: Perceptions of organizational safety: implications for the development of safety culture

means indicate that drivers (M� 3.83) and supervisors (M� 3.83) place a greater emphasison the importance of pay than managers (M� 2.57); managers (M� 3.75) and supervisors(M� 3.70) place a greater emphasis on management control than drivers (M� 3.04); makingpro®ts is viewed as more damaging by supervisors (M� 3.65) and drivers (M� 3.48) thanmanagers (M� 2.74).

It appears that there is some support for the ®rst hypothesis given the high ratings across thesafety issues by all respondents; however, the MANOVA indicates that signi®cant di�erences doexist between levels. Thus, whilst all levels share a concern for safety, di�erent priorities emergefor the di�erent hierarchical levels.

The perceptions that di�erent organizational levelshave of each other

Hypothesis 2 stated that there should be realistic intergroup understanding, i.e. accurateperceived views of the other levels. In order to examine this hypothesis: ®rst, a factor analysis wasconducted to reduce the complexity of the design (are there underlying dimensions in the safetyissues rated?); second, one-way MANOVAs were performed comparing own views andperceptions of the other levels (does each level perceive that there is a shared understanding ofthe importance of safety?); third, estimates were compared with each level's own ratings to assess

Table 1. Means and standard deviations for employees' safety ratings at three hierarchical levels

Railway item Drivers Supervisors ManagersMean S.D. Mean S.D. Mean S.D.

Sighting of signals 3.86 (1.31) 3.89 (1.40) 3.02 (1.27)Route knowledge 4.14 (1.23) 3.98 (1.32) 2.91 (1.24)Drivers' morale 4.20 (1.07) 4.15 (1.03) 3.68 (1.05)Top-level planning 3.35 (1.35) 3.74 (1.21) 3.50 (0.96)Quality of equipment 4.16 (1.16) 4.21 (1.08) 3.77 (0.89)Rostering 3.61 (1.33) 3.51 (1.20) 2.92 (1.06)Management control 3.04 (1.40) 3.70 (1.14) 3.74 (0.88)Track condition 3.98 (1.19) 3.79 (1.29) 3.27 (1.07)Signalling failures 4.08 (1.14) 4.06 (1.03) 3.80 (1.07)Drivers' diagrams 3.61 (1.23) 3.26 (1.15) 2.56 (0.98)Attitudes of the workforce 3.72 (1.23) 3.87 (1.13) 3.80 (0.96)Use of alcohol/drugs 3.92 (1.49) 4.02 (1.37) 3.52 (1.29)Training 3.95 (1.24) 3.94 (1.23) 3.58 (1.15)Cab comfort 3.72 (1.19) 3.38 (1.13) 2.77 (0.97)Communication 3.95 (1.19) 4.08 (1.03) 3.71 (0.89)Drivers' skills 3.89 (1.46) 3.89 (1.27) 3.24 (1.11)Reorganization 2.99 (1.39) 3.62 (1.32) 3.52 (1.14)Rule Book 3.40 (1.37) 3.53 (1.46) 2.83 (1.14)Making pro®ts 3.48 (1.56) 3.70 (1.29) 2.73 (1.16)Pay 3.83 (1.38) 3.85 (1.31) 2.56 (1.04)Quality of supervision 3.12 (1.35) 3.72 (1.03) 3.74 (0.90)Discipline 3.25 (1.33) 3.89 (1.10) 3.11 (1.14)Variety of work 3.30 (1.34) 3.25 (1.27) 2.95 (1.04)Project budgeting 3.04 (1.46) 3.32 (1.27) 2.52 (0.95)Quality of maintenance 4.28 (1.07) 4.30 (0.89) 3.70 (1.02)

1� no adverse e�ect on safety, ÿ5� very great adverse e�ect on safety. Drivers N� 170; supervisory sta� N� 53;managers N� 66.

Copyright # 1999 John Wiley & Sons, Ltd. J. Organiz. Behav. 20, 185±198 (1999)

PERCEPTIONS OF ORGANIZATIONAL SAFETY 189

Page 6: Perceptions of organizational safety: implications for the development of safety culture

accuracy (does each level have an understanding of any di�erences in priorities that exist betweenlevels?).

Principal components analysis, with oblique (oblimin) rotation, was used to reduce the numberof dependent variables.3 Five factors were retained, accounting for 63.4 per cent of the variance(see Table 3).

The items loading onto the ®rst factor might, if degraded or defective in some way, lead directlyto unsafe acts by sta�; this factor might be called `unsafe conditions'. The items loading onto thesecond factor are all associated with the decisional activities of senior managers (managerialdecisions). The third factor represents the working conditions of train drivers. The ®nal twofactors relate to line management: the fourth factor at a local, supervisory level and the ®nalfactor to line functions. There is little cross-loading above 0.45; however, quality of equipment(factor 1) cross-loads with factor 3, as equipment is related to working conditions; quality ofsupervision (factor 4) cross-loads with factor 2, as supervision is related to management control.Cronbach's alpha was satisfactory for each of the unweighted factors: unsafe conditions (0.877),managerial decisions (0.842), working conditions (0.818), local management (0.757), linefunctions (0.734). The unweighted factor scores were employed in further analyses.4

Three one-way MANOVAs were performed, with one independent variable (type of rating,with three levels: own rating and estimates of the other two levels' ratings) and the ®ve

Table 2. Univariate and stepdown F values for employees' safety ratings at three hierarchical levels

Railway item Univariate F* Stepdown F

Pay 25.37{ 25.37{Cab comfort 16.72{ 5.90Management control 11.18{ 26.55{Quality of supervision 8.65{ 5.99{Making pro®ts 8.17{ 6.39{Quality of maintenance 7.72{ 0.54Rostering 7.03{ 3.87Track condition 6.94{ 1.49{Reorganization 6.02 7.56{Discipline 5.72 3.65Project budgeting 5.64 1.85Drivers' morale 5.63 0.27Rule Book 4.99 1.18Quality of equipment 2.80 2.41Training 2.04 0.18Top-level planning 2.00 0.34Use of alcohol/drugs 1.92 1.40Variety of work 1.82 0.41Communication 1.40 1.23Signalling failures 1.02 0.98Attitudes of workforce 0.38 3.72

* (2,290) df.{Signi®cant at p5 0.002 in univariate context.{ p4 0.002.

3 As the inclusion of the original 25 items in the factor analysis did not lead to problems of multicollinearity, those itemspreviously excluded from the MANOVA were included in this analysis.4 Simple unweighted factor scores were preferred to weighted factor scores as the unweighted factor scores retain ameaningful representation of the mean ratings. Hence, where the term `factor' is used it denotes the unweighted ratherthan the weighted factor score.

Copyright # 1999 John Wiley & Sons, Ltd. J. Organiz. Behav. 20, 185±198 (1999)

190 S. CLARKE

Page 7: Perceptions of organizational safety: implications for the development of safety culture

unweighted factor scores as dependent variables. In each case, the multivariate F value wassigni®cant (p5 0.001), indicating that none of the levels perceive a shared understanding of theimportance of safety. Univariate and stepdown analyses were conducted for each factor.

DriversUnivariate tests show that the di�erences are greatest on issues of working conditions, unsafeconditions and line functions. However, stepdown analysis indicates that there is shared varianceamong these three factors. If working conditions is given priority, this factor is the only onewhich is signi®cant at p5 0.01 (see Table 4).

In their own view, drivers (M� 3.68) rate working conditions as having a much greater adversee�ect on safety than either supervisors (M� 2.28) or managers (M� 1.84). Using the Sche�eÂprocedure, di�erences between drivers' own and their estimates of supervisors' ratings andbetween the pair of estimates were both signi®cant (p5 0.001). It appears, therefore, that driversperceive that safety centres around the drivers' own immediate working environment, but theirconcerns, particularly in this area, are not shared by either managers or supervisors.

There is a close correspondence between the estimates for supervisors and managers, bothsigni®cantly lower than drivers' own ratings; although signi®cantly di�erent from each other, this

Table 3. Factor analysis for drivers' safety ratings (N� 169)

Railway item Factor loadings 40.45Unsafe

conditionsManagerialdecisions

Workingconditions

Localmanagement

Linefunctions

Track condition 76Signal sighting 72Route knowledge 71Signalling failures 69Use of alcohol/drugs 64Drivers' skills 61Quality of equipment 51 46Training 48Project budgeting 86Making pro®ts 79Management control 76Top-level planning 74Reorganization 59Pay 72Cab comfort 69Attitudes of the workforce 68Drivers' diagrams 67Drivers' morale 66Variety of work 46Rostering 45Rule book 70Discipline 68Quality of supervision 51 57Quality of maintenance 57Communication 47

Eigenvalue 8.20 3.67 1.54 1.37 1.08% variance 32.8 14.7 6.2 5.5 4.3

Loadings to two decimal places; decimal points omitted.

Copyright # 1999 John Wiley & Sons, Ltd. J. Organiz. Behav. 20, 185±198 (1999)

PERCEPTIONS OF ORGANIZATIONAL SAFETY 191

Page 8: Perceptions of organizational safety: implications for the development of safety culture

correspondence may re¯ect that drivers' perceptions may be biased by their lack of contact withsenior managers (partial support for hypothesis 3).

SupervisorsUnivariate tests show that there are di�erences for all of the factors, but that, due to sharedvariance, when given priority in stepdown analysis, only working conditions is signi®cant atp5 0.01 (see Table 5).

Supervisors perceive that managers will consider working conditions as less important(M� 2.71) than either themselves (M� 3.61) or drivers (M� 4.00). Using the Sche�e procedure,supervisors' own ratings and their estimates for drivers were signi®cantly higher than theirestimates for managers (p5 0.001); the estimates for drivers were signi®cantly higher thansupervisors' own ratings, but at the 5 per cent level. Thus, supervisors perceive that there aredi�erences between themselves and managers concerning the importance of safety, especially inrelation to working conditions.

ManagersUnivariate F tests indicate di�erences for each factor except managerial decisions. Stepdownanalysis indicates that there is shared variance between working conditions (signi®cant atp5 0.01) and unsafe conditions, when working conditions is given priority. Also, there is sharedvariance between local management (signi®cant at p5 0.01) and line functions, when localmanagement is given priority (see Table 6).

Managers estimate that drivers place a greater emphasis on working conditions (M� 3.65)than either themselves (M� 3.05) or supervisors (M� 3.04); signi®cant at p5 0.001. There wasalso a signi®cant univariate and stepdown e�ect for local management issues (p5 0.001). In thiscase, managers estimated that supervisors (M� 2.86) and drivers (M� 2.88) would place less

Table 4. Means, standard deviations and univariate and stepdown F tests for drivers' safety ratings(N� 143)

Railway factor Own ratings Estimates of supervisors'ratings

Estimates of managers'ratings

Mean S.D. Mean S.D. Mean S.D.

Unsafe conditions 3.97 (0.91) 2.93 (0.95) 2.61 (0.94)Managerial decisions 3.17 (1.10) 2.92 (1.20) 2.86 (1.42)Working conditions 3.68 (0.83) 2.28 (1.04) 1.84 (0.93)Local management 3.18 (1.09) 3.10 (1.30) 3.00 (1.33)Line functions 4.10 (1.01) 2.68 (1.15) 2.37 (1.16)

Railway factor Univariate F* Stepdown F (df)

3. Working conditions 263.67{ 263.67{ (2,284)1. Unsafe conditions 163.17{ 3.95 (2,283)5. Line functions 156.52{ 2.87 (2,282)2. Managerial decisions 3.17 3.00 (2,281)4. Local management 1.62 1.34 (2,280)

* (2,284) df.{Signi®cant at p5 0.01.

Copyright # 1999 John Wiley & Sons, Ltd. J. Organiz. Behav. 20, 185±198 (1999)

192 S. CLARKE

Page 9: Perceptions of organizational safety: implications for the development of safety culture

emphasis on local management compared to themselves (M� 3.21). Thus, managers perceive alack of shared understanding between themselves and their subordinates; with managers giving alower priority to working conditions, and a higher priority to local management.

Comparing managers' own ratings to their estimates of the other levels, it is evident that thereis not a close association, indicating that managers are not biased by their own views (rejectinghypothesis 4).

Table 6. Means, standard deviations and univariate and stepdown F tests for managers' safety ratings(N� 63)

Railway factor Own ratings Estimates of supervisors'ratings

Estimates of drivers'ratings

Mean S.D. Mean S.D. Mean S.D.

Unsafe conditions 3.38 (0.78) 3.39 (0.62) 3.67 (0.71)Managerial decisions 3.22 (0.68) 3.36 (0.76) 3.50 (0.94)Working conditions 3.05 (0.71) 3.04 (0.67) 3.65 (0.63)Local management 3.21 (0.82) 2.86 (0.84) 2.88 (0.74)Line functions 3.71 (0.79) 3.48 (0.75) 3.82 (0.70)

Railway factor Univariate F* Stepdown F (df )

3. Working conditions 40.81{ 40.81{ (2,124)1. Unsafe conditions 16.72{ 1.77 (2,123)4. Local management 8.54{ 11.36{ (2,122)5. Line functions 5.09{ 0.69 (2,121)2. Managerial decisions 2.73 6.06{ (2,120)

* (2,124) df.{Signi®cant at p5 0.01.

Table 5. Means, standard deviations and univariate and stepdown F tests for supervisors' safety ratings(N� 47)

Railway factor Own ratings Estimates of drivers'ratings

Estimates of managers'ratings

Mean S.D. Mean S.D. Mean S.D.

Unsafe conditions 3.97 (1.03) 3.97 (0.93) 3.37 (1.03)Managerial decisions 3.60 (0.89) 3.66 (0.98) 3.03 (1.37)Working conditions 3.62 (0.89) 4.00 (0.66) 2.71 (1.15)Local management 3.75 (0.94) 3.43 (0.88) 3.37 (1.21)Line functions 4.20 (0.82) 4.19 (0.84) 3.39 (1.27)

Railway factor Univariate F* Stepdown F (df)

3. Working conditions 45.94{ 45.94{ (2,92)1. Unsafe conditions 23.33{ 3.69 (2,91)5. Line functions 17.08{ 1.97 (2,90)2. Managerial decisions 6.63{ 0.37 (2,89)4. Local management 5.36{ 4.22 (2,88)

* (2,92) df.{Signi®cant at p5 0.01.

Copyright # 1999 John Wiley & Sons, Ltd. J. Organiz. Behav. 20, 185±198 (1999)

PERCEPTIONS OF ORGANIZATIONAL SAFETY 193

Page 10: Perceptions of organizational safety: implications for the development of safety culture

Comparing the perceptions of drivers, supervisorsand managers

Comparing managers' estimates to drivers' own ratings (see Tables 4 and 6), managers accuratelyperceive that drivers have a high awareness of safety issues, giving particular emphasis to linefunctions (3.82 compared to 4.10), unsafe conditions (3.67 compared to 3.97) and workingconditions (3.65 compared to 3.68). A similar ®nding is re¯ected in supervisors' estimates (seeTables 4 and 5). However, comparing drivers' estimates to managers' own ratings (see Tables 4and 6), drivers underestimate managers' concern over safety issues. This is particularly evident inrelation to working conditions, where drivers see managers as less likely to recognize theimportance of working conditions for safety (1.84 compared to 3.05). Drivers also under-estimated supervisors' concerns over working conditions (2.28 compared to 3.62).

Comparing supervisors' estimates to managers' own ratings (see Tables 5 and 6), supervisorshave fairly accurate perceptions of managers' views, giving greater emphasis to line functions(3.39 compared to 3.71) and unsafe conditions (3.37 compared to 3.38), and less emphasis toworking conditions (2.71 compared to 3.05). Managers' estimates of supervisors reveal slightunderestimates, but most notably, a misperception concerning local management issues: whilstmanagers think that supervisors lack concern over local management issues, this is not re¯ectedin supervisors' own ratings (2.86 compared to 3.75).

Discussion

From the results of this study, it is apparent that drivers, their supervisors and senior managersshare a perception that safety is important (supporting hypothesis 1). However, signi®cantdi�erences emerged in the relative in¯uence assigned to the various safety issues.

Di�erences in safety priorities may be less of a threat to the company safety culture, if thosedi�erences are acknowledged by other levels. Hypothesis 2 investigated the basis for intergroupunderstanding by examining the accuracy of each level's perceptions of the others. It was foundthat each level was aware that a shared understanding of safety did not exist, with each levelperceiving signi®cant di�erences between levels. Whilst this would initially indicate thatemployees had an understanding that di�ering perspectives would exist within the company,closer examination revealed that intergroup perceptions were not always accurate and showedsome degree of bias.

Drivers estimated that their supervisors and managers would have much less awareness of theimportance of safety compared to themselves. This general trend was particularly marked for theissues which drivers saw as most in¯uential: working conditions (e.g. pay, cab comfort), unsafeconditions (e.g. track condition, signal sighting) and line functions (quality of maintenance andcommunication). This may be related to hypothesis 3 which stated that drivers base theirperceptions of senior managers, with whom they have little direct contact, on their perceptions oflocal managers and supervisors. There is partial support for this hypothesis as drivers' ownratings di�er signi®cantly from both levels of managers; however, local managers and super-visors are perceived to be signi®cantly more concerned about safety than senior managers(2.28 compared to 1.84). These perceptions are actually inaccurate as both supervisors andmanagers share drivers' concerns about the impact of working conditions on safety (3.62 and3.05 compared to 3.68). Although managers' actual ratings are signi®cantly lower than the other

Copyright # 1999 John Wiley & Sons, Ltd. J. Organiz. Behav. 20, 185±198 (1999)

194 S. CLARKE

Page 11: Perceptions of organizational safety: implications for the development of safety culture

two levels, managers rate working conditions as more important than drivers imagine (3.05compared to 1.84); in the case of supervisors and local managers, there is no signi®cant di�erencebetween their ratings (3.62 compared to 3.68). This negative stereotyping of senior managers'views on safety is also re¯ected in supervisors' perceptions, where managers are viewed as lessconcerned about working conditions compared to supervisors' own ratings (2.71 compared to3.62). Whilst managers are less concerned than supervisors in reality, they have a greater degreeof concern than supervisors think. Although managers are aware of drivers' concerns overworking conditions (3.65 compared to 3.68), they are less aware that supervisors share drivers'concerns (3.04 compared to 3.62).

It was hypothesized that senior managers' perceptions may be biased due to their lack ofcontact with the operational end of the business (hypothesis 4). In general this hypothesis was notsupported, given that a close association was not found between managers' own ratings and theirestimates of other levels. However, it may be related to managers' misperceptions of supervisors'views; here, managers inaccurately estimated supervisors' views as very close to their ownconcerning working conditions (3.04 compared to 3.05) and unsafe conditions (3.39 compared to3.38). Thus, there is some evidence that managers wrongly perceive agreement between them-selves and supervisors, possibly connected with a tendency to colour their perceptions of otherswith their own opinions (Stephenson et al., 1983). However, the fact that this only occurs inrelation to estimates of supervisors and not in estimates of drivers, indicates that such a tendencydoes not extend to all other groups.

In addition to signi®cant di�erences on the issue of working conditions, senior managersperceive intergroup di�erences concerning the importance of local management (Rule Book,discipline and supervision). Managers perceive that both drivers and supervisors underestimatethe importance of local management compared to their own ratings (2.88 and 2.86 compared to3.21). In reality, drivers share managers' concerns over the impact of local management on safety(3.18 compared to 3.21), and supervisors, contrary to managers' beliefs, actually rate localmanagement issues as more important compared to senior managers' ratings (3.75 compared to3.21). Both drivers and supervisors have fairly accurate perceptions of managers' concerns overthe impact of local management on safety (3.00 and 3.37 compared to 3.21). In this case, thereis no support for hypothesis 4, as managers perceive their subordinates as lacking concerncompared to themselves, demonstrating wrongly perceived disagreement rather than agreement.

Implications of the study for the developmentof a positive safety culture

Cooper and Phillips (1994) state that a good safety climate is characterized by `a collectivecommitment of care and concern, whereby those in an organization share similar perceptions andpositive attitudes to safety'. This study has shown, in the company examined, that despite thepositive attitudes of organizational members, this collective concern was not recognized at anyhierarchical level. Intergroup perceptions were not realistic and revealed biased views of the safetyattitudes of other levels. It may be suggested that whilst shared perceptions of the importance ofsafety may exist, they will form an insu�cient basis for intergroup understanding and mutualtrust, as such an understanding will depend on accurate intergroup perceptions. The biases thatwere found to operate will in¯uence not only intergroup perceptions, but also sta�±managementcommunication, con®dence in management and commitment to safety; each of which is animportant element in the development of a positive safety culture.

Copyright # 1999 John Wiley & Sons, Ltd. J. Organiz. Behav. 20, 185±198 (1999)

PERCEPTIONS OF ORGANIZATIONAL SAFETY 195

Page 12: Perceptions of organizational safety: implications for the development of safety culture

Perceptions of senior management's attitudes and actions have been emphasized as part-icularly in¯uential in the development of a positive safety culture (e.g. Zohar, 1980; Brown andHolmes, 1986; Dedobbeleer and Beland, 1991; Cooper and Phillips, 1994). As noted earlier, theperformance of senior managers has an impact on the way that subordinates think and behave.Negative perceptions of senior managers may result from stereotyping due to a lack of directcontact; however, whilst direct contact is limited, the attitudes of senior managers will beevidenced through indirect media. For example, Cooper and Phillips (1994) noted a number ofindirect in¯uences on safety attitudes, including the status accorded to safety o�cers andcommittees. Other sources of information would include, communications with supervisors andexperience of operational problems. In the former case, impressions of senior managers will beformulated based on the way in which they are portrayed by supervisors. The perceived values ofhigher management may be more in¯uential on supervisors' communications than their ownpersonal values (Meglino et al., 1989). Thus, supervisors and local managers, who perceive theirown views as very close to those of drivers, may not be representing their own views but what theyperceive to be the `company views'. Negative stereotyping may be re¯ective not only of a lack ofdirect contact on the part of sta�, but of a more fundamental lack of con®dence in managers'commitment to putting safety before production. The intermediate role of supervisors in sta�±management communication needs more careful attention, given that although senior managersare aware of drivers' safety attitudes, they wrongly perceive agreement between themselves andthe supervisory level.

Although positive safety attitudes at senior management level are essential in developing apositive safety culture (Gri�ths, 1985), it cannot be assumed that such attitudes will cascadethrough the organization. This study has demonstrated senior management's attitudes andactions may be subject to negative stereotyping by sta� and supervisors. Thus, there is a need toregulate how these attitudes are transmitted to subordinates to ensure that managerial commit-ment to safety is accurately perceived. The perception of senior managers within a negativeinterpretative framework may be a product of mistrust that already exists within the company,e.g. trade union±management disputes (related to pay or working conditions). Therefore, thedevelopment of safety culture should not be viewed as a separate issue, but one that forms anintegrative part of the wider organizational culture (Waring, 1996).

Whilst implications for the development of a safety culture may be drawn from this study, ithas a number of limitations which should be considered. First, there was a relatively low returnrate of questionnaires, particularly from the driver population, which (although not atypical forthis population) raises the possibility of response bias and a non-representative sample; never-theless, the sample was fairly diverse, with responses drawn from a number of di�erent locations,and representative of a wide range of age and experience. Problems with distribution of thequestionnaires may also have contributed to the low return rate. Second, the sta� samplecomprised only train drivers, and not sta� drawn across the entire company, and so will onlyre¯ect the safety culture of this subgroup (Guest, Peccei and Thomas, 1994).

Acknowledgements

The research in this paper was supported by a studentship awarded by the Science andEngineering Research Council (SERC) and British Rail (BR). The helpful comments of twoanonymous reviewers are gratefully acknowledged.

Copyright # 1999 John Wiley & Sons, Ltd. J. Organiz. Behav. 20, 185±198 (1999)

196 S. CLARKE

Page 13: Perceptions of organizational safety: implications for the development of safety culture

References

Brown, R. L. and Holmes, H. (1986). `The use of a factor-analytic procedure for assessing the validity ofan employee safety climate model', Accident Analysis and Prevention, 18, 445±470.

Clarke, S. (1996). `The e�ect of habit as a behavioural response in risk reduction programmes', SafetyScience, 22(1±3), 163±175.

Cooper, M. D. and Phillips, R. A. (1994, January). `Validation of a safety climate measure'. Paperpresented at the BPS Occupational Psychology Conference, Birmingham, U.K.

Dedobbeleer, N. and Beland, F. (1991). `A safety climate measure for construction sites', Journal of SafetyResearch, 22, 97±103.

Goldberg, A. I., Dar-El, E. M. and Rubin, A. E. (1991). `Threat perception and the readiness to participatein safety programs', Journal of Organizational Behavior, 12(2), 109±122.

Gri�ths, D. K. (1985). `Safety attitudes of management', Ergonomics, 28(1), 61±67.Guest, D, E., Peccei, R. and Thomas, A. (1994, February). `Safety culture and safety performance:British Rail in the aftermath of the Clapham Junction disaster'. Paper presented at the Bolton BusinessSchool conference on Changing Perceptions of Risk, Bolton, U.K.

Harrison, R. (1972). `Understanding your organization's character', Harvard Business Review, May±June,119±128.

Health and Safety Commission (1993). Third Report: Organising for Safety, ACSNI Study Group onHuman Factors, HMSO, London.

Health and Safety Executive (1991). Successful Health and Safety Management, HMSO, London.Leather, P. J. (1987). `Safety and accidents in the construction industry: a work design perspective',Work and Stress, 1(2), 167±174.

Likert, R. (1961). New Patterns of Management, McGraw-Hill, New York.Meglino, B. M., Ravlin, E. C. and Adkins, C. L. (1989). `A work values approach to corporate culture: A®eld test of the value congruence process and its relationship to individual outcomes', Journal of AppliedPsychology, 74(3), 424±432.

Melia, J. L., Tomas, J. M., Oliver, A. and Islas, M. E. (1992, November). `Organizational and psycho-logical variables as antecedents of work safety: A causal model'. Paper presented at the Safety and Well-Being at Work Conference, Loughborough, U.K.

Oliver, A., Tomas, J. M. and Melia, J. L. (1993, November). `A psychometric study of a new measure ofsafety climate'. Paper presented at the Safety and Well-Being at Work Conference, Loughborough, U.K.

Schein, E. (1992). Organizational Culture and Leadership, Jossey Bass, San Francisco.Stephenson, G. M. and Allen, P. (1987). `Bargaining and Industrial Relations'. In: Warr, P. (Ed.)Psychology at Work, 3rd edn, Penguin, London.

Stephenson, G. M., Brotherton, C. J., Skinner, M. R. and Dela®eld, G. (1983). `Size of organization,attitudes to work and job satisfaction', Industrial Relations Journal, 14, 28±40.

Tabachnick, B. G. and Fidell, L. S. (1989). Using Multivariate Statistics, 2nd edn, Harper Collins,New York.

Walker, K. F. (1959). `Con¯ict and mutual misunderstanding: a survey of union leaders' and businessexecutives' attitudes to industrial relations', Journal of Industrial Relations, 1, 20±30.

Walker, K. F. (1962). `Executives' and union leaders' perception of each others attitudes to industrialrelations', Human Relations, 15, 183±196.

Waring, A. (1996). Safety Management Systems, Chapman and Hall, London.Weick, K. E. (1987). `Organizational culture as a source of high reliability', California Management Review,29(2), 112±127.

Zohar, D. (1980). `Safety climate in industrial organizations: Theoretical and applied implications',Journal of Applied Psychology, 65(1), 96±102.

Copyright # 1999 John Wiley & Sons, Ltd. J. Organiz. Behav. 20, 185±198 (1999)

PERCEPTIONS OF ORGANIZATIONAL SAFETY 197

Page 14: Perceptions of organizational safety: implications for the development of safety culture

Appendix 1: Items Used in the Safety Ratings Questionnaire

Glossary of Railway terms

1 Sighting of signals Ability of driver to accurately perceive what is being shown by thesignal (e.g. is a `danger' indication given)

2 Route knowledge Task-related knowledge, e.g. line speeds, position of signals3 Drivers' morale Group energy and enthusiasm, team spirit4 Top-level planning Long-term strategic activities of management5 Quality of equipment Standard of trains and related hardware6 Rostering Way in which shifts have been allocated to sta�7 Management control Managerial responsibility for controlling activities8 Condition of track E�ects of everyday wear and tear, e.g. alignment9 Signalling failures Signals may show the wrong message or fail to show any message

10 Drivers' diagrams Way in which turns of duty are arranged, e.g. a 2-hour journey,followed by a 2-hour break, another 2-hour journey, etc.

11 Attitudes of workforce Work-related attitudes, towards job, supervisors, etc.12 Use of alcohol/drugs As in violation of the company drug and alcohol policy13 Training In-house driver training, initial and refresher training14 Comfort of cab Ergonomics of seat, layout of instrumentation, etc.15 Communication Transmission of necessary information throughout company16 Drivers' skills Task-related skills, train-handling, braking into stations, etc.17 Reorganization Major organizational changes (BR having undergone several)18 The Rule Book Documentation issued to drivers and other sta� with prescriptive

and proscriptive rules for many activities19 Making pro®ts Managing productive activities that are pro®table20 Pay Monetary reward for the tasks undertaken21 Quality of supervision Standard of overseeing drivers' duties22 Discipline Maintenance of orderly conduct of sta�23 Variety of work Variety of routes, di�erent types of trains, etc.24 Project budgeting Allocation of resources for management projects25 Quality of maintenance Standard to which trains, track, etc. are maintained

Copyright # 1999 John Wiley & Sons, Ltd. J. Organiz. Behav. 20, 185±198 (1999)

198 S. CLARKE