i

Safety Science 49 (2011) 625–632

Contents lists available at ScienceDirect

Safety Science

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / s s c i

The relationship between the implementation of a Safety Management System and the attitudes of employees towards unsafe acts in aviation

Haytham Remawi a,⇑, Paul Bates b,1, Ian Dix c,2 a Griffith Aviation, Griffith University, Nathan, Queensland, Australia b Aviation at Griffith, Griffith University, Nathan Campus, Brisbane, Queensland 4111, Australia c Griffith University, Nathan Campus, Brisbane, Queensland 4111, Australia

a r t i c l e i n f o

Article history: Received 14 April 2010 Received in revised form 20 June 2010 Accepted 18 September 2010 Available online 12 February 2011

Keywords: Aviation Safety Airports SMS ICAO Sharjah Airport Griffith University

0925-7535/$ - see front matter Crown Copyright � 2 doi:10.1016/j.ssci.2010.09.014

⇑ Corresponding author at: Sharjah Department o Sharjah, United Arab Emirates. Tel.: +971 6 5081024 fax: +971 6 5581707.

E-mail addresses: [email protected] (H edu.au (P. Bates), [email protected] (I. Dix).

URL: http://www.griffith.edu.au (P. Bates). 1 Tel.: +61 (0)7 37355358; fax: +61 (0)7 37355204. 2 Tel.: +61 (0)7 3219 7317, mobile: +61 401 463 678

a b s t r a c t

Airports represent highly complex organisations, incorporating such interdependent operations as air- lines, ground transport, flight services, ground services, refueling, maintenance, customer services, cater- ing, administration and security. Airports, and especially International Airports, must ensure that their operations are conducted in a safe and efficient manner, as the consequence of any error or failure during operations has the potential for catastrophic outcomes. The international governing body for air transpor- tation, ICAO, requires that airports must implement a Safety Management System as a means of ensuring safe operations and eliminating or reducing the likelihood of low frequency/high consequence incidents. This research project sought to determine the extent to which the implementation of a Safety Manage- ment System (SMS) influenced the attitudes of airport employees toward unsafe acts. The hypothesis tested was that the implementation of an SMS into an airport will result in an improvement in attitudes toward safety. A Safety Culture Survey was utilized to measure these attitudes. Two International Air- ports were chosen to measure the extent to which the introduction of an SMS at one airport would influ- ence the safety attitude and culture of those employees. Sharjah International Airport, UAE, was used as the experimental group, as it did not have a formal SMS in operation, with another English-speaking International Airport (the Second Airport) being used as a control group, as it already had an SMS in oper- ation. A Safety Culture Survey was used as the pre and post-test measure over a 12-month period to determine the extent of influence of the introduction of the SMS at Sharjah Airport. The results of Phase 1 of the survey (June 2008) were compared against the results of Phase 2 of the survey (June 2009). The average score reported by participants at Sharjah Airport increased significantly from pre-test measure to post-test measure in relation to communication, safety rules, supportive environment, personal risk appreciation, work environment, and involvement. At the same time, the average score for personal pri- orities decreased significantly from Phases 1 to 2. Results indicate that participants at Sharjah Airport recorded a significant positive shift in attitude to the safety factors covered in the Safety Culture Survey, whilst at the same time responses from the Second Airport showed no such shift in attitude. The Second Airport showed neither decline nor improvement in responses. Whilst some methodological issues were identified, the results were of sufficient strength as to conclude that the outcomes reported support the hypothesis that the introduction of an SMS at Sharjah Airport has effected positive changes not observed at the Second Airport. Recommendations for ongoing research were made to further explore the strength of relationship between SMS and safety attitudes, as well as the relationship between safe attitude and safe behaviour.

Crown Copyright � 2010 Published by Elsevier Ltd. All rights reserved.

010 Published by Elsevier Ltd. All r

f Civil Aviation, P.O. Box 8, , mobile: +971 50 4827027;

. Remawi), p.bates@griffith.

.

1. Introduction

Left alone and poorly managed, most organisations will become less safe. Management neglect, worker apathy and an absence of analysis will all eventually create a less-safe operation (Stranks, 1994). On the other side of the equation, a successful Safety Management System can produce very positive safety outcomes. In this sense, a Safety Management System (SMS) is clearly good for business (Stranks, 1994). ICAO recognised the importance of

ights reserved.

Table 1 Multivariate effects on nine scale scores of occasion, airport, and their interaction.

Multivariate tests Effect Value F Hyp. df Error df Sig. Observed powera

Occasion Pillai’s Trace 0.163 11.553 9 534 *** 1.000 Wilks’ Lambda 0.837 11.553 9 534 *** 1.000 Hotelling’s Trace 0.195 11.553 9 534 *** 1.000 Roy’s Largest Root 0.195 11.553 9 534 *** 1.000

Airport Pillai’s Trace 0.427 44.223 9 534 *** 1.000 Wilks’ Lambda 0.573 44.223 9 534 *** 1.000 Hotell ing’s Trace 0.745 44.223 9 534 *** 1.000 Roy’s Largest Root 0.745 44.223 9 534 *** 1.000

Occasion* Airport Pillai’s Trace 0.130 8.894 9 534 *** 1.000 Wilks’ Lambda 0.870 8.894 9 534 *** 1.000 Hotell ing’s Trace 0.150 8.894 9 534 *** 1.000 Roy’s Largest Root 0.150 8.894 9 534 *** 1.000

��p < 0.01. * p > 0.05. *** p < 0.001.

a Ideally, the observed power of an effect should be in the 0.800–1.00 range.

Table 2 Multivariate effect of occasion and five covariates on nine scale scores (Sharjah Airport).

Effect Effect Value F Hyp. df Error df Sig. Observed power

Years in current position Pillai’s Trace 0.052 2.751 9 449 ** 0.956 Wilks’ Lambda 0.948 2.751 9 449 ** 0.956 Hotelling’s Trace 0.055 2.751 9 449 ** 0.956 Roy’s Largest Root 0.055 2.751 9 449 ** 0.956

Years working at airport Pillai’s Trace 0.045 2.377 9 449 * 0.919 Wilks’ Lambda 0.955 2.377 9 449 * 0.919 Hotelling’s Trace 0.048 2.377 9 449 * 0.919 Roy’s Largest Root 0.048 2.377 9 449 * 0.919

Number of courses completed Pillai’s Trace 0.073 3.941 9 449 *** 0.995 Wilks’ Lambda 0.927 3.941 9 449 *** 0.995 Hotelling’s Trace 0.079 3.941 9 449 *** 0.995 Roy’s Largest Root 0.079 3.941 9 449 *** 0.995

Exam assignment completed Pillai’s Trace 0.061 3.226 9 449 ** 0.981 Wilks’ Lambda 0.939 3.226 9 449 ** 0.981 Hotelling’s Trace 0.065 3.226 9 449 ** 0.981 Roy’s Largest Root 0.065 3.226 9 449 ** 0.981

Satisfactory amount of safety training Pillai’s Trace 0.050 2.632 9 449 ** 0.947 Wilks’ Lambda 0.950 2.632 9 449 ** 0.947 Hotelling’s Trace 0.053 2.632 9 449 ** 0.947 Roy’s Largest Root 0.053 2.632 9 449 ** 0.947

Occasion Pillai’s Trace 0.516 53.242 9 449 *** 1.000 Wilks’ Lambda 0.484 53.242 9 449 *** 1.000 Hotelling’s Trace 1.067 53.242 9 449 *** 1.000 Roy’s Largest Root 1.067 53.242 9 449 *** 1.000

* p > 0.05. ** p < 0.01. *** p < 0.001.

626 H. Remawi et al. / Safety Science 49 (2011) 625–632

SMS and recommended that all contracting states implement the requirement for SMS by November 2005. The GCAA adopted this recommendation in (via CAR Part X), requiring all Aerodrome li- cense holders to have a comprehensive SMS. An SMS is similar but more comprehensive than the Quality Assurance System (QAS) which is currently mandated through CAR Part IX. It differs in that whilst it has many of the same elements as an operational QAS, it is less driven by process alone, and also considers the hu- man element within the system. It relies on an examination of all operational aviation hazards and risks which may impact upon safe operation at the Airport. This includes unwritten risks such as human factors, cultural and environmental risks. The potential benefits of SMS are generally recognised throughout the world to the extent that many ICAO nations now require commercial operators to have documented safety management systems (ICAO Annex 6, Section 3.2). Within Australia, in 2005 CASR 139 (Aerodromes) introduced a requirement for SMS to be imple-

mented, and the more recent 2009 changes through CAO 82.3 (Low Capacity Regular Public Transport) and 82.5 (High Capacity Regular Public Transport) have had a similar requirement.

2. Safety culture

Whilst SMSs may be necessary to achieve consistent out- comes, they often do not, in isolation, prevent workplace safety incidents and accidents. There has been a growing awareness of the influence and importance of workplace safety culture in influ- encing safety outcomes. The current interest in the term ‘‘safety culture’’ can be traced directly back to the Chernobyl accident in 1986. Since then, numerous definitions of safety culture have abounded in the safety literature. Indeed, our recent review of the literature revealed several diverse definitions of the concept (Wiegmann et al., 2001). These various definitions of safety

Table 3 Univariate effect of occasion and three covariates on nine scale scores (Sharjah Airport).

Sharjah Airport Dependent variable Hyp. df Error df F Sig. Observed power

Years in current position Safety priority 1 457 2.775 NS 0.383 Communication 1 457 3.765 NS 0.491 Management commitment 1 457 0.060 NS 0.057 Safety rules 1 457 5.885 * 0.678 Personal priorities 1 457 0.025 NS 0.053 Supportive environment 1 457 0.502 NS 0.109 Personal risk appreciation 1 457 0.023 NS 0.053 Work environment 1 457 0.994 NS 0.169 Involvement 1 457 8.072 ** 0.809

Years working at airport Safety priority 1 457 0.232 NS 0.077 Communication 1 457 6.961 ** 0.749 Management commitment 1 457 2.043 NS 0.297 Safety rules 1 457 2.596 NS 0.363 Personal priorities 1 457 0.113 NS 0.063 Supportive environment 1 457 0.823 NS 0.148 Personal risk appreciation 1 457 0.023 NS 0.053 Work environment 1 457 1.176 NS 0.191 Involvement 1 457 1.027 NS 0.173

Number of courses completed Safety priority 1 457 0.081 NS 0.059 Communication 1 457 3.474 NS 0.460 Management commitment 1 457 1.921 NS 0.282 Safety rules 1 457 0.226 NS 0.076 Personal priorities 1 457 0.249 NS 0.079 Supportive environment 1 457 9.484 ** 0.867 Personal risk appreciation 1 457 0.338 NS 0.089 Work environment 1 457 1.811 NS 0.269 Involvement 1 457 1.006 NS 0.170

Exam assignment completed Safety priority 1 457 4.231 * 0.537 Communication 1 457 3.841 NS 0.498 Management commitment 1 457 19.202 *** 0.992 Safety rules 1 457 0.767 NS 0.141 Personal priorities 1 457 0.092 NS 0.061 Supportive environment 1 457 2.841 NS 0.391 Personal risk appreciation 1 457 0.014 NS 0.052 Work environment 1 457 6.511 * 0.721 Involvement 1 457 0.201 NS 0.073

Sat amount of tng completed Safety priority 1 457 14.394 *** 0.966 Communication 1 457 2.158 NS 0.311 Management commitment 1 457 5.736 * 0.666 Safety rules 1 457 0.647 NS 0.126 Personal priorities 1 457 0.005 NS 0.051 Supportive environment 1 [ 457 7.654 ** 0.788 Personal risk appreciation 1 457 0.034 NS 0.054 Work environment 1 457 0.550 NS 0.115 Involvement 1 457 8.304 ** 0.820

Occasion Safety priority 1 457 0.000 NS 0.050 Communication 1 457 93.299 *** 1.000 Management commitment 1 457 0.027 NS 0.053 Safety rules 1 457 51.182 *** 1.000 Personal priorities 1 457 26.058 *** 0.999 Supportive environment 1 457 51.470 *** 1.000 Personal risk appreciation 1 457 81.670 *** 1.000 Work environment 1 457 41.004 *** 1.000 Involvement 1 457 121.373 *** 1.000

* p > 0.05. ** p < 0.01. *** p < 0.001.

H. Remawi et al. / Safety Science 49 (2011) 625–632 627

culture are presented in Table 1. Most definitions originate from articles that have focused on safety culture in industries other than aviation as for example nuclear power, mining and manu- facturing (Reason, 1997). Nonetheless, there does appear to be several commonalities among these various definitions regardless of the particular industry being considered. These commonalities include:

1. Safety culture is a concept defined at the group level or higher, which refers to the shared values among all the group or orga- nisation members.

2. Safety culture is concerned with formal safety issues in an orga- nisation, and closely related to, but not restricted to, the man- agement and supervisory systems.

3. Safety culture emphasizes the contribution from everyone at every level of an organisation.

4. The safety culture of an organisation has an impact on its mem- bers’ behaviour at work.

5. Safety culture is usually reflected in the contingency between reward systems and safety performance.

6. Safety culture is reflected in an organisation’s willingness to develop and learn from errors, incidents, and accidents.

Table 4 Multivariate effect of occasion and five covariates on nine scale scores (the Second Airport).

Multivariate tests: Second Airport Effect Value F Hyp. df Error df Sig. Observed power

Years in current position Pillai’s Trace 0.261 2.596 9 66 * 0.913 Wilks’ Lambda 0.739 2.596 9 66 * 0.913 Hotelling’s Trace 0.354 2.596 9 66 * 0.913 Roy’s Largest Root 0.354 2.596 9 66 * 0.913

Years working at airport Pillai’s Trace 0.136 1.159 9 66 NS 0.523 Wilks’ Lambda 0.864 1.159 9 66 NS 0.523 Hotelling’s Trace 0.158 1.159 9 66 NS 0.523 Roy’s Largest Root 0.158 1.159 9 66 NS 0.523

Number of courses completed Pillai’s Trace 0.112 0.924 9 66 NS 0.417 Wilks’ Lambda 0.888 0.924 9 66 NS 0.417 Hotelling’s Trace 0.126 0.924 9 66 NS 0.417 Roy’s Largest Root 0.126 0.924 9 66 NS 0.417

Exam assignment completed Pillai’s Trace 0.080 0.642 9 66 NS 0.287 Wilks’ Lambda 0.920 0.642 9 66 NS 0.287 Hotelling’s Trace 0.088 0.642 9 66 NS 0.287 Roy’s Largest Root 0.088 0.642 9 66 NS 0.287

Satisfactory amount of safety training Pillai’s Trace 0.170 1.498 9 66 NS 0.658 Wilks’ Lambda 0.830 1.498 9 66 NS 0.658 Hotelling’s Trace 0.204 1.498 9 66 NS 0.658 Roy’s Largest Root 0.204 1.498 9 66 NS 0.658

Occasion Pillai’s Trace 0.178 1.583 9 66 NS 0.687 Wilks’ Lambda 0.822 1.583 9 66 NS 0.687 Hotelling’s Trace 0.216 1.583 9 66 NS 0.687 Roy’s Largest Rnnt 0.216 1.583 9 66 NS 0.687

��p < 0.01. ���p < 0.001. * p > 0.05.

Table 5 Multivariate effect of occasion and airport on six emergent scale scores.

Multivariate tests Tests Value F Hyp. df Error df Sig. Observed power

Occasion Pillai’s Trace 0.192 21.320 6 537 *** 1.000 Wilks’ Lambda 0.808 21.320 6 537 *** 1.000 Hotelling’s Trace 0.238 21.320 6 537 *** 1.000 Roy’s Largest Root 0.238 21.320 6 537 *** 1.000

Airport Pillai’s Trace 0.349 47.905 6 537 *** 1.000 Wilks’ Lambda 0.651 47.905 6 537 *** 1.000 Hotelling’s Trace 0.535 47.905 6 537 *** 1.000 Roy’s Largest Root 0.535 47.905 6 537 *** 1.000

Occasion* Airport Pillai’s Trace 0.197 21.900 6 537 *** 1.000 Wilks’ Lambda 0.803 21.900 6 537 *** 1.000 Hotelling’s Trace 0.245 21.900 6 537 *** 1.000 Roy’s Largest Root 0.245 21.900 6 537 *** 1.000

��p < 0.01. * p > 0.05. *** p < 0.001.

628 H. Remawi et al. / Safety Science 49 (2011) 625–632

7. Safety culture is relatively enduring, stable and resistant to change (Westrum and Adamski, 1999).

Given the numerous definitions of safety culture that have been proposed in the literature, it is not surprising that there is little consensus as to the exact number of indicators that reflect an orga- nisation’s safety culture. Indeed, numerous organisational indica- tors have been proposed, with some estimates ranging from as few as two to as many as 19 (Flin et al., 2000). Again, the numerous inconsistencies and often idiosyncratic labeling of these indicators creates difficulty in reconciling the variety of organisational indica- tors identified in previous reports (Global Aviation Information Network, 2002). Nonetheless, a closer inspection of these various reports suggests that there are at least five global components or indicators of safety culture. They include organisational commit- ment, management involvement, employee empowerment, reward systems, and reporting systems.

As reported by Stranks (1994), failure to institute systematic processes for the management of safety are not good for busi- ness, whereas the structure and content of SMS should be good for business, to the extent that some industries, such as the avi- ation industry, now mandate the development of SMS. Whilst the benefits to organisations of having SMS have been well docu- mented, it remains unclear as to how the implementation of these systems influence the attitudes of employees, and in partic- ular their attitudes toward unsafe acts. To this end the main purpose of this research project was to determine whether a rela- tionship exists between implementation of a Safety Management System (SMS) and consequent attitudinal change towards unsafe acts within an international Airport that had not had any formal SMS. The hypothesis being tested sought to determine whether the implementation of an SMP into an international airport would result in an improvement in the measure of safety culture at that airport.

Table 6 Multivariate effect of occasion and five covariates on six scale scores (Sharjah Airport).

Sharjah Airport Tests Value F Hyp. df Error df Sig. Observed power

Years in current position Pillai’s Trace 0.057 4.572 6 452 *** 0.987 Wilks’ Lambda 0.943 4.572 6 452 *** 0.987 Hotelling’s Trace 0.061 4.572 6 452 *** 0.987 Roy’s Largest Root 0.061 4.572 6 452 *** 0.987

Years working at airport Pillai’s Trace 0.061 4.865 6 452 *** 0.992 Wilks’ Lambda 0.939 4.865 6 452 *** 0.992 Hotelling’s Trace 0.065 4.865 6 452 *** 0.992 Roy’s Largest Root 0.065 4.865 6 452 *** 0.992

Number of courses completed Pillai’s Trace 0.016 1.235 6 452 NS 0.488 Wilks’ Lambda 0.984 1.235 6 452 NS 0.488 Hotelling’s Trace 0.016 1.235 6 452 NS 0.488 Roy’s Largest Root 0.016 1.235 6 452 NS 0.488

Exam assignment completed Pillai’s Trace 0.049 3.910 6 452 ** 0.969 Wilks’ Lambda 0.951 3.910 6 452 ** 0.969 Hotelling’s Trace 0.052 3.910 6 452 ** 0.969 Roy’s Largest Root 0.052 3.910 6 452 ** 0.969

Satisfactory amount of safety training Pillai’s Trace 0.059 4.752 6 452 *** 0.990 Wilks’ Lambda 0.941 4.752 6 452 *** 0.990 Hotelling’s Trace 0.063 4.752 6 452 *** 0.990 Roy’s Largest Root 0.063 4.752 6 452 *** 0.990

Occasion Pillai’s Trace 0.615 120.281 6 452 *** 1.000 Wilks’ Lambda 0.385 120.281 6 452 *** 1.000 Hotelling’s Trace 1.597 120.281 6 452 *** 1.000 Roy’s Largest Root 1.597 120.281 6 452 *** 1.000

�p > 0.05. ** p < 0.01. *** p < 0.001.

H. Remawi et al. / Safety Science 49 (2011) 625–632 629

3. Safety Culture Survey construction

A Safety Culture Survey was used to determine pre and post SMS intervention measures of employee attitude toward unsafe behaviours. The Safety Culture Survey was constructed by the author’s Research Supervisors, and was largely based on the UK Workplace Health and Safety Culture Survey. The questionnaire devised was presented as statements of opinion in which the par- ticipants were asked to response using a 5-point Likert-type scale, ranging from 1 (strongly agree) to 5 (strongly disagree). Demo- graphic questions were included to identify participants’ age, gen- der, current occupation, number of years that have been involved in their current position, and number of years that they have been involved in aviation.

All participants completed an identical questionnaire that was designed to measure the safety culture within the organisation. The questionnaire took approximately 20 min to complete, and the results of the research were confidential and anonymous. The questionnaires were distributed in both Airports in paper and on- line versions (through an internet website facility). Initially one month was allowed for collection of the forms. After the initial col- lection was made, the numbers were totalled and a further appeal for returns sent out through the coordination managers.

The survey was expected to achieve a number of objectives. The first objective was to gain an overall picture of the safety culture at each organisation – how the staff perceives safety or safety behav- iour at the time of application within that organisation. The second objective was to use the survey results to establish a baseline of measures of attitudes toward unsafe behaviours, against which any reimplementation of the survey could identify the extent of changes in attitudes (positive or negative), after initiating an SMS.

In addition to these uses, the survey had the capacity to also identify specific instances where staff report safety problems in the organisation anonymously. This provided Airport Management with the opportunity to investigate and rectify verified problems.

3.1. Survey implementation

Implementation of the Safety Culture Survey was conducted in two phases:

� Phase 1 involved implementing the Survey at Sharjah and the Second Airport in June 2008, prior to the implementation of the SMS at Sharjah Airport. � At Sharjah, 265 surveys were distributed to staff. All 265 sur-

veys were returned completed. � At the Second Airport, 34 surveys were distributed to a sample

of staff, with all 34 surveys returned completed. All the staff were employees of the Second Airport working in aerodrome operational areas. � Phase 2 involved re-implementing the survey at both Airports

12 months after the initial survey implementation in June 2009 and after the implementation of the SMS at Sharjah Inter- national Airport. � At Sharjah, 200 surveys were distributed to staff. All 200 sur-

veys were returned completed. � At the Second Airport, 47 surveys were issued to a sample of

staff, with all 47 returned completed. At The Second Airport there were an increased number of staff surveyed, because of the change in employee numbers. At Sharjah, due to restructur- ing there has been a ‘push’ to outsource some of the operational functions. Therefore, this resulted in a reduced sampling size.

3.2. Sample size

The total population surveyed was 299 in Phase 1 and 247 in Phase 2 from both Sharjah and the Second Airports. The response rate was identified as being the critical uncontrolled variable in this study.

It was calculated that a proportion sample would provide suffi- cient precision for the study (Cooper and Schindler, 2001). This as- sumed hat from Phases 1 and 2 95% of the participants would

Table 7 Univariate effect of occasion and four covariates on six emergent scale scores (Sharjah).

Sharjah Airport Dependent variable Hyp. df Error df F Sig. Observed power

Years in current position Safety rules as core to organisation 1 457 11.149 ** 0.915 Safety concerns as of little interest 1 457 0.081 NS 0.059 Safety rules as impractical 1 457 9.165 ** 0.856 Importance of personal appreciation 1 457 1.445 NS 0.224 Importance of personal communication 1 457 0.376 NS 0.094 Safety culture of conditions 1 457 0.981 NS 0.167

Years working at airport Safety rules as core to organisation 1 457 15.476 *** 0.975 Safety concerns as of little interest 1 457 0.474 NS 0.106 Safety rules as impractical 1 457 3.237 NS 0.435 Importance of personal appreciation 1 457 1.004 NS 0.170 Importance of personal communication 1 457 1.505 NS 0.232 Safety culture of conditions 1 457 0.821 NS 0.148

Number of courses completed Safety rules as core to organisation 1 457 0.228 NS 0.076

Safety concerns as of little interest 1 457 1.834 NS 0.272 Safety rules as impractical 1 457 0.285 NS 0.083 Importance of personal appreciation 1 457 0.684 NS 0.131 Importance of personal communication 1 457 1.816 NS 0.270 Safety culture of conditions 1 457 1.884 NS 0.278

Exam assignment completed Safety rules as core to organisation 1 457 6.089 * 0.692 Safety concerns as of little interest 1 457 0.552 NS 0.115 Safety rules as impractical 1 457 5.369 * 0.638 Importance of personal appreciation 1 457 7.162 ** 0.761 Importance of personal communication 1 457 0.218 NS 0.075 Safety culture of conditions 1 457 6.256 * 0.704

Sat amount of safety tng Safety rules as core to organisation 1 457 7.913 ** 0.802 Safety concerns as of little interest 1 457 5.591 * 0.655 Safety rules as impractical 1 457 4.079 * 0.522 Importance of personal appreciation 1 457 6.900 ** 0.746 Importance of personal communication 1 457 10.179 ** 0.889 Safety culture of conditions 1 457 0.297 NS 0.085

Occasion Safety rules as core to organisation 1 457 2.777 NS 0.383 Safety concerns as of little interest 1 457 394.872 *** 1.000 Safety rules as impractical 1 457 40.012 *** 1.000 Importance of personal appreciation 1 457 85.130 *** 1.000 Importance of personal communication 1 457 13.368 *** 0.954

Safety culture of conditions 1 457 69.277 *** 1.000

* p > 0.05. ** p < 0.01. *** p < 0.001.

630 H. Remawi et al. / Safety Science 49 (2011) 625–632

respond definitively to the questionnaire, and that this result could be applied to the total population with accuracy.

3.3. Analysis of results

As the results of the Phase 1 survey (June 2008) were to be com- pared against the results of the follow-up survey (June 2009), the data from the questionnaires were subjected to inferential statisti- cal analysis using a mixed-method analysis of variance (Kurskal– Wallis). The Kurskal–Wallis analysis of variance is designed to determine whether there are any differences between the variables overall, and whether there are any differences between the levels of the variables, which might be seen as the most appropriate re- sults analysis tool for this proposed study (Westrum and Adamski, 1999).

The hypothesis was that the experimental group responded more positively towards safety after the SMS implementation and therefore, induced a more positive safety culture.

4. Results

This research study hypothesized that differing emphases in the focus on safety concerns at the two airports (Sharjah – a new inter- est; the Second Airport – no change in focus) should lead to differ-

ential shifts in safety concerns as expressed by systematic changes across the nine scale scores by occasion and by airport. With that in mind, a series of MANCOVAs were performed with scale scores as dependent variables (DVs) and with occasion and airport as inde- pendent variables (IVs). It seems likely that personal characteris- tics such as differences in the length of time worked at an airport or the number of courses undertaken might also influence the scale scores of interest indicated above. To control for the effect of these personal characteristics, they were entered into MANCOVAs as covariates. Entering these covariates has the effect of controlling for their effects. Significant covariates clearly are of interest, and the extent to which occasion and airport have significant effects under these conditions is of even greater interest.

Personal characteristic variables were on the basis of the ab- sence or near absence of missing responses. For this reason, age in years and years in aviation were not included in the list of per- sonal characteristic variables to be entered as covariates. All six covariates were transformed into dummy variables (0, 1). An initial MANCOVA with occasion and airport as IVs and with the nine stan- dard scale scores as DVs was performed. The multivariate results were statistically significant for occasion, airport, and the interac- tion between these two IVs (see Table 1, Appendices).

In order to examine the effect by airport more closely, the file was split by airport and a second MANCOVA performed to examine the effect of airport affiliation together with six personal

Table 8 Multivariate effects for six emergent scale scores (Second Airport).

Second Airport Tests Value F Hyp. df Error df Sig. Observed power

Years in current position Pillai’s Trace 0.075 0.929 6 69 NS 0.343 Wilks’ Lambda 0.925 0.929 6 69 NS 0.343 Hotelling’s Trace 0.081 0.929 6 69 NS 0.343 Roy’s Largest Root 0.081 0.929 6 69 NS 0.343

Years working at airport Pillai’s Trace 0.098 1.256 6 69 NS 0.461 Wilks’ Lambda 0.902 1.256 6 69 NS 0.461 Hotelling’s Trace 0.109 1.256 6 69 NS 0.461 Roy’s Largest Root 0.109 1.256 6 69 NS 0.461

Number of courses completed Pillai’s Trace 0.067 0.822 6 69 NS 0.304 Wilks’ Lambda 0.933 0.822 6 69 NS 0.304 Hotelling’s Trace 0.071 0.822 6 69 NS 0.304 Roy’s Largest Root 0.071 0.822 6 69 NS 0.304

Exam assignment completed Pillai’s Trace 0.016 0.183 6 69 NS 0.094 Wilks’ Lambda 0.984 0.183 6 69 NS 0.094 Hotelling’s Trace 0.016 0.183 6 69 NS 0.094 Roy’s Largest Root 0.016 0.183 6 69 NS 0.094

Sat amount of safety training Pillai’s Trace 0.074 0.913 6 69 NS 0.337 Wilks’ Lambda 0.926 0.913 6 69 NS 0.337 Hotelling’s Trace 0.079 0.913 6 69 NS 0.337 Roy’s Largest Root 0.079 0.913 6 69 NS 0.337

Occasion Pillai’s Trace 0.049 0.598 6 69 NS 0.223 Wilks’ Lambda 0.951 0.598 6 69 NS 0.223 Hotelling’s Trace 0.052 0.598 6 69 NS 0.223 Roy’s Largest Root 0.052 0.598 6 69 NS 0.223

�p > 0.05. ��p < 0.01. ���p < 0.001.

H. Remawi et al. / Safety Science 49 (2011) 625–632 631

characteristic variables entered as covariates. Initial inspection of the effect of covariates (years in current position, years at airport, number of courses attended, recency of last course attended, amount of training [not satisfactory, satisfactory], examination by assignment [No, Yes]) indicated that the recency of the Last Course Attended was not associated significantly with scale scores related to either Sharjah or the Second Airport. So, this variable was omit- ted and a third MANCOVA performed.

The multivariate effects of occasion, and also of all five of the covariates were statistically significant in relation to the nine standard scale scores (see Table 2, Appendices). In the case of Sharjah Airport, not only occasion but also all of the five personal characteristics entered as covariates had statistically significant uni-variate effects with one or more of the nine standard scale scores (see Table 3, Appendices). It is worth reiterating that the important point isn’t so much that the covariates achieved statisti- cal significance but that the main effect for occasion remained significant for seven of the nine standard scales. The following fig- ure illustrates the effect of occasion in relation to those scales in more detail.

The average score reported by participants at Sharjah Airport increased significantly from Phases 1 to 2 in relation to communi- cation, safety rules, supportive environment, personal risk appreci- ation, work environment, and involvement. At the same time, the average score for personal priorities decreased significantly from Phases 1 to 2 (see Table 4, Appendices). The multivariate effects of occasion, and also of four of the five covariates were statistically non-significant in relation to the nine standard scale scores. Given the non-significant main effect for occasion, further analyses were not attempted (see Table 5, Appendices).

5. Analyses based on six emergent scales

An initial MANCOVA with occasion and airport as IVs and with the six emergent scale scores as DVs produced multivariate results indicative of occasion, airport, and the interaction between these

two IVs all being significant at the multivariate level (see Table 6, Appendices). The multivariate effect of occasion, airport, and the interaction of occasion by airport, were all highly statistically significant.

In order to examine the effect by airport more closely, the file was split by airport and a second MANCOVA performed to examine the effect of airport affiliation together with five (number of years at airport excluded as per previous parallel analyses) personal characteristic variables entered as covariates. Multivariate effects were statistically significant for occasion, and also for four of the five covariates (see Table 7, Appendices). As well as four of the five covariates having an effect on a selection of the emergent scale scores, occasion had a significant effect on five of the six attitudinal scale scores.

The average score reported by participants at Sharjah Airport increased significantly from Phases 1 to 2 in relation to safety rules as impractical, and the importance of personal appreciation. At the same time, the average score for safety concerns as of little inter- est, importance of personal communication, and agreement that safety culture focused on conditions (as opposed to blaming peo- ple) decreased significantly from Phases 1 to 2. The multivariate ef- fects of occasion, and also of all five of the covariates were statistically non-significant in relation to the six emergent scale scores. Given the non-significant main effect for occasion, further analyses were not attempted (see Table 8, Appendices).

The outcomes noted above support the hypothesis that the intervention at Sharjah Airport has produced significant attitudinal changes. The related reports of significant effects for variables such as the number of years at an airport, the recency with which courses had been attended, etc., underlines the complexity of the range of background variables that can impact on such attitudinal scores. The contrasting lack of statistically significant effects for either occasion or any of the five personal characteristic variables on attitudinal scale scores at the Second Airport further points to the significant influence of personal characteristic variables at Sharjah being most likely related to the vivifying effect of the

632 H. Remawi et al. / Safety Science 49 (2011) 625–632

intervention rather than manifesting as a generic influence regard- less of that intervention.

A methodological issue that provides an alternative explanation for the relative strength of effects obtained at Sharjah Airport vs. the Second Airport is the imbalance in the number of participants at those two sites. That is, whereas 465 participants completed these surveys at Sharjah Airport, only 81 did so at the Second Air- port. Note that each of the outcome tables includes the power of that effect as a reminder of the effect of sample size on such out- comes. Nonetheless, regardless of the difference in sample size, the extreme difference in outcomes does support the hypothesis that an active safety program has significant effects across a range of measures. Based on the patterns observed, these effects were mostly positive except for the observed significant tendency across occasions for participants at Sharjah Airport to be increasingly in- clined to view safety rules and procedures as not entirely practical during an emergency.

6. Summary and conclusion

The hypothesis being tested for this research project sought to determine whether the implementation of an SMP into an interna- tional airport would result in an improvement in the measure of safety culture at that airport. Overall the results of this research support the hypothesis that the introduction of an SMS will influ- ence the attitude of employees. The average score reported by par- ticipants at Sharjah Airport increased significantly from Phases 1 to 2 in relation to safety rules, supportive environment, personal risk appreciation, work environment, and involvement. At the same time, there was no corresponding increase in attitude for employ- ees at the Second Airport at the multivariate level or for any of the personal characteristic variables with the exception of years in cur- rent position.

A major limitation for the study is that any observed differences in the emphasis on safety are accompanied by systematic differ- ences across a range of other variables, including the difference in the number of persons recruited for this study from the two air- ports (almost 90% from Sharjah Airport). The presence of such sta- tistically significant differences in the composition of participants sampled from Sharjah vs. the Second Airport supports the conclu- sion that any statistically significant airport related differences could be explained not only by a differing emphasis on safety but also by age, Years Involved Current Position, Years Working For An Airport, Years Involved Aviation, Number Of Courses Attended, Exam Assignment, Last Course Attended. Which been analysed through the development of Sharjah International Airport Safety Management System.

However it seems clear from the results of this research that participants at Sharjah Airport experienced and expressed signifi- cant changes in levels of safety concern that extended across the nine scales more generally at the multivariate level and across six of the nine scales at the univariate level, and that this sense of changed attitude was not reported from participants at the Sec- ond Airport. That is, outcomes reported are consistent with the hypothesis that the introduction of an SMS at Sharjah Airport has

effected changes not observed at the Second Airport. Importantly, the Second Airport here represents a baseline or control condition in which one would not expect such changes given the maturity of the safety program already implemented. The good news is that even though a number of personal characteristic variables also pre- dict scores at Sharjah Airport at statistically significant levels, the effect of occasion is significant.

In order to further investigate the influence of safety manage- ment systems on employee attitudes toward and behaviour in un- safe acts, a number of recommendations have been identified.

1. Continue to measure attitudinal changes of participants at Shar- jah Airport toward unsafe acts as the SMS matures to determine if the changed attitudes observed in this research project are sustainable.

2. Continue to explore the relationship between the implementa- tion of SMS and employee attitude toward unsafe acts by repli- cating this research project, but this time ensuring that comparable numbers of employees are surveyed in each organisation.

3. Explore the relationship between employee attitudes toward unsafe acts and employee behaviour in unsafe acts. Of interest is whether an increase in employee positive attitude to safe acts is accompanied by an increase in safe behaviour.

4. Undertake an evaluation of the key components of an SMS implementation to determine which are critical to its successful implementation.

5. There would be value in undertaking an evaluation of how organisations engage and maintain commitment to SMS goals and objectives.

It is proposed that each of these research projects would further develop the understanding of the effectiveness of SMS in achieving positive safety cultures, and hence positive safety outcomes.

Appendix A

See Tables 1–8.

References

Cooper, D.R., Schindler, P.S., 2001. Business Research Method. McGraw-Hill, New York.

Flin, R., Mearns, K., O’Connor, P., Bryden, R., 2000. Measuring safety climate: identifying the common features. Safety Science 34, 177–192.

Global Aviation Information Network, 2002. Operators Flight Safety Handbook (Issue 2). GAIN, USA. <http://www.gain.org> (accessed 05.02).

Reason, J., 1997. Managing the Risks of Organizational Accidents. Ashgate, Brookfield, VT.

Stranks, J., 1994. Management Systems for Safety. Financial Times Pittman Publishing, London.

Westrum, R., Adamski, A.I., 1999. Organisation factors associated with safety and mission success in aviation human factors. In: Garland, D.J., Wise, J.A., Hopkin, V.D. (Eds.), Handbook of Aviation Human Factors. Lawrence Erlbaum, Mahwah, NJ, pp. 67–104.

Wiegmann, D.A., Zhang, H., Von Thaden, T., 2001. Defining and Assessing safety Culture in High Reliability Systems: An Annotated Bibliography. University of Illinois Institute of Aviation Technical Report (ARL-01-12/FAA-01-4). Aviation Res. Lab, Savoy, IL.