What are the differences in management characteristics of heavy vehicle operators with high insurance claims versus low insurance claims?

Safety Science 70 (2014) 327–338

Contents lists available at ScienceDirect

Safety Science journal homepage: www.elsevier.com/locate/ssci

What are the differences in management characteristics of heavy vehicle operators with high insurance claims versus low insurance claims? Lori Mooren a,⇑, Ann Williamson a, Rena Friswell a, Jake Olivier b, Raphael Grzebieta a, Faisal Magableh a a b

Transport and Road Safety Research, University of New South Wales, Australia School of Mathematics and Statistics, University of New South Wales, Australia

a r t i c l e

i n f o

Article history: Received 25 September 2013 Received in revised form 10 July 2014 Accepted 10 July 2014

Keywords: Heavy vehicle safety Safety management Transport Trucks Driver pay

a b s t r a c t An exploratory survey of Australian organisations that operate fleets of heavy freight vehicles was undertaken to identify differences in management characteristics between those that have good safety records compared with those that have poorer safety records, using vehicle insurance claim rates as a proxy for safety. Fifty organisations that operate heavy vehicles and had either low or higher recent claim rates completed a questionnaire. These included various industry sectors, such as local government councils, utility companies, and freight transport companies. The questionnaire asked about the participants’ use of a wide range of safety management practices relevant to heavy vehicle drivers. The results showed that despite controlling for fleet size, companies with larger fleets had poorer claim rates. The results also suggested that higher claimers relied more on setting criteria and rules for vehicles and drivers, than low claimers. Low claimers seemed to focus more strongly on proactive risk assessment, and that drivers are paid for time worked and consulted on safety issues. A number of the findings were counterintuitive. For example, higher claimers more often than low claimers reported that they did more checking during recruitment, had more policies and some accreditation as well as doing more in-vehicle monitoring. The study showed that there are safety management characteristics that distinguish between good and poorer safety performers but that further research must assess both the use and quality of the safety management practices implemented. Ó 2014 Published by Elsevier Ltd.

1. Introduction This paper seeks to advance knowledge of what works best in reducing road trauma risk in heavy vehicle transport operations. In order to define the scope of this problem, comparative data from OECD countries is examined. These data can be used for comparison as rates, rather than raw numbers of fatalities. Typically, three types of rates are used. The rate for fatalities per population provides a public health rate that can be used to compare relative chances of road fatalities per person in a country. Two other rates are fatalities per number of registered motor vehicles and fatalities per number of kilometres travelled. These two rates are forms of exposure rates. OECD countries tend to keep the most reliable and consistent data to enable calculation of all three of these rates.

⇑ Corresponding author. Address: Transport and Road Safety Research (TARS), University of New South Wales, Level 1, Old Main Building West, Kensington, NSW 2052, Australia. Tel.: +61 2 9385 5666, mobile: +61 412 888 290; fax: +61 2 9385 6040. E-mail address: [email protected] (L. Mooren). http://dx.doi.org/10.1016/j.ssci.2014.07.007 0925-7535/Ó 2014 Published by Elsevier Ltd.

The types of motor vehicles registered can also enable a comparison between fatality rates of light versus heavy vehicles. Australian road safety efforts have succeeded in reducing the rates and numbers of road fatalities from being one of the highest rates of fatalities per 100,000 inhabitants from 25 in the early 1970s to 6.1 in 2010. By comparison road fatality rates in the United States of America (USA) are still relatively high at 11.1 per 100,000 people (World Health Organisation, 2013). However, looking at the fatal and serious injury crash rates for heavy trucks in Australia compared with the USA shows that Australia is not a better performer. Heavy trucks in the USA make up 3% of all registered vehicles, and account for 7% of vehicle miles driven but they are involved in 11% of all road fatalities (Bezwada, 2010). Similarly trucks and buses are only 3% of the total number of vehicles registered in Australian jurisdictions and represent only 8% of total vehicle kilometres travelled, but they are involved in 18% of fatal and serious injury crashes and hence have higher over-representation in road trauma statistics than in the USA (Australian Transport Council, 2011). Indeed, a study commissioned by the National Transport Commission (NTC) in 2002, benchmarking truck safety across Australia, Canada, France, Germany, New Zealand, Sweden,

328

L. Mooren et al. / Safety Science 70 (2014) 327–338

United Kingdom and USA found that the USA had the lowest rate of truck involved fatalities per exposure to 100 million kilometres (kms) travelled. The rate for Australian truck fatalities was 2.5 per 100 million kms travelled versus the American rate of 1.7 (Haworth et al., 2002). As the geographic area dimensions of the US and Australia are similar, the comparisons between them are worth noting. While there are similarities between the regulatory systems in the USA and Australia, Australian trucking allows considerably more liberal hours of service than in the USA which may account for some differences, but the US system is also more prescriptive and transparent than the Australian system (Mooren et al., 2012). For example, new entrants to the trucking industry in the US are audited within 18 months of operating against specific safety management criteria, whereas Australian companies do not go through this process. Also, safety ratings and compliance data on drivers and trucking companies can be accessed by the public in the USA. The Australian system requires compliance of trucks and drivers, but it is usually indirectly through Chain of Responsibility or duty of care legislation, that the authorities enforce safety management practices. From the viewpoint of workplace safety, too, the effective management of heavy vehicle (HV) driver safety is important because heavy vehicle drivers have one of the highest rates of serious occupational injury both on and off the road. Safe Work Australia reported that in the years 2003–2011 a cumulative total of 649 workers were killed in truck related incidents (Safe Work Australia, 2012). Of all Australian workers, people working in or around trucks have made up between one quarter and one third of all work related deaths in recent years. The trucking industry has been identified to be a high-risk industry for workplace injury in other countries, despite overall low road fatality rates. For example, in Japan the transport industry, including trucks, buses and taxis, has a crash fatality rate three times higher than that of private motor vehicles (Li and Itoh, 2013). Although a range of risk factors have been identified for HV driver injury (Department of Transportation U.S., 2006; Loeb and Clarke, 2007; Lueck and Murray, 2011; Parker et al., 1995; Williamson, 2005, 2007; Williamson and Friswell, 2013), organisational practices that may be used to manage the risks to drivers have received surprisingly little research attention. It has been recognised since the 1980s that workplace Health and Safety outcomes are determined, at least in part, by formal practices and policies in the workplace (Zohar, 1980). In particular, the role of the safety culture of the workplace has attracted a large literature (Zohar, 2010), but little of this research relates directly to the trucking industry. A recent literature search focussing specifically on heavy vehicle transport revealed little robust empirical research in the HV transport sector and little evidence for effective safety management characteristics that can reduce crashes and injuries (Mooren et al., 2014). This review identified some safety management practices and characteristics that have been found to have some effect on work related road safety (for both heavy and light vehicles). These include: safety characteristics of the fleet, driver recruitment practices, safety policies, safety training, driver remuneration, using in-vehicle monitoring devices, being accredited in a safety management program, communication and employee input into OHS, and employee discipline and incentives. Among the management practices that have been investigated in HV transport, strong evidence indicative of an effect has been reported only for payment practices (Belzer et al., 2002; Quinlan and Wright, 2008a; Rodriguez et al., 2006). In other work settings, such as in government agencies, the construction and manufacturing industries there is a number of safety management and organisational attributes that have been associated with safety outcomes including attitudinal, self-reported

behavioural, and incident rate changes (Fernandez-Muniz et al., 2007; Geldart et al., 2010), but it is difficult to gauge the relative importance of these management practices (Mooren et al., 2014). This is because the studies have often been constrained by the available data to focus on only small sets of practices or a collection of safety management practices, and the measures of safety performance have varied across studies. For example, Geldart et al. (2010) found that monitoring injury statistics, auditing, safety awards, and worker participation influenced lost time injury rates, but we do not know whether there are other more important safety management practices from this research alone or in combination with other evidence based factors. Fernandez-Muniz et al. (2007) found that a safety management system including policies, incentives, safety training, communications, preventative planning, emergency planning, enforcement, and incident reporting was associated with employee satisfaction with the number of personal injuries. Similarly in a light vehicle work related road safety study, Banks (2008) found that companies with comprehensive risk management strategies have fewer self-reported errors, fatigue and violations. But these studies did not quantify the contributions of each strategy. This paper reports the findings of a survey of companies that aimed to assess whether safety management practices and organisational characteristics differentiated heavy vehicle transport companies with good track records (low insurance claim rates) from those with poorer safety records (higher insurance claim rates). An extensive range of safety management characteristics relevant to organisations that operate heavy vehicles for transport tasks were examined in the survey on the basis that they have shown a potential to deliver safety benefits in previous research. While the main focus of this research is on occupational safety, as the work of heavy trucking necessarily involves public roads and other traffic, the results will also be relevant to road and public safety in general. Based on the review of literature (Mooren et al., 2014), it was expected that companies with low rates of insurance claims would exhibit: 1. Truck fleets that were well maintained and had a comprehensive set of safety features (Banks, 2008; Bruning, 1989; de Pont, 2005; Langwieder et al., 2001). 2. Rigorous and consistent journey and site risk assessment processes (Banks, 2008; Crum and Morrow, 2002; Oystein Saksvik et al., 2003). 3. Driver recruitment criteria that would endeavour to preclude high risk drivers(Banks, 2008; Vredenburgh, 2002). 4. Remuneration methods that would not encourage unsafe driving practices (pay for all hours worked) (Belzer et al., 2002; Corsi et al., 2002; Crum and Morrow, 2002; Monaco and Williams, 2000; Quinlan and Wright, 2008a; Williamson, 2007). 5. A comprehensive set of safety policies effectively communicated to drivers (Banks, 2008; Fernandez-Muniz et al., 2007). 6. Accreditation under an auditable safety management scheme (Baas and Taramoeroa, 2008; Naveh and Marcus, 2007). 7. Scheduling and rostering practices that minimise fatigue risk for drivers (Crum and Morrow, 2002; Feyer and Williamson, 1995; Golob and Hensher, 1994). 8. Comprehensive safety training of drivers (Arboleda et al., 2003; Huang et al., 2006; Wills et al., 2005). 9. Effective safety communications and driver input into safety decision-making (Geldart et al., 2010; Gregersen et al., 1996; Huang et al., 2006; Salminen, 2008). 10. Use of more in-vehicle safety monitoring devices (Wouters and Bos, 2000).

L. Mooren et al. / Safety Science 70 (2014) 327–338

11. Effective discipline for safety breaches and incentives for safety innovations (Banks, 2008; Fernandez-Muniz et al., 2007; Moses and Savage, 1994). 12. Lower incidence of injury and non-injury events, lower infringements and vehicle defect notices and fewer days off road due to mechanical failures (Blower et al., 2010; Cantor et al., 2010; de Pont, 2005).

2. Methods 2.1. Design The overall design of this study involved a comparison of the reported practices in good and poorer safety performing heavy trucking companies. A sample of trucking companies operating in Australia were categorised as good and poorer performing on the basis of the number of safety-related insurance claims made per truck. Managers of these companies were then surveyed in 2011–2012 to identify whether any of a range of safety management characteristics identified as potential influencers of safe performance in previous research discriminated between companies with good and poorer safety performance. Safety performance was indexed using vehicle damage insurance claim rates. Finalised heavy vehicle claims data for the period 2007–2009 were available from a major insurer of heavy vehicle fleets in Australia. The claims period was selected to (i) include the most recent years with a high percentage of finalised claims (mean per year = 98.4%), (ii) cover a sufficiently long period that the claim rate would have some resilience to atypical peaks in claims, and (iii) be as recent as possible to minimise the possibility of significant organisation changes since the claim period. Organisational claim rates per heavy truck for the period were calculated using claims where at least some fault was attributed to the insured (i.e., excluding claims for natural occurrences, 100% third party fault or unattributed fault). Companies were selected from this subset of claims to be invited to participate in this study on the basis of two criteria of interest: their claim rate and their heavy vehicle fleet size. Good performers were defined as those with the lowest 30% of claim rates (rate = 0 claims per truck) whereas poorer performers were defined as the highest 40% of claim rates (rate > 0.17 claims per truck). Fleet size may influence the extent and sophistication of the safety management practices that an organisation adopts (Eakin et al., 2010; Knipling et al., 2011; Mitchison and Papadakis, 1999) and so it was originally intended to also examine the effect of fleet size. Small and large fleet size groups within each claim rate group were created by selecting only from the largest 40% of companies and the smallest 40% of companies. This meant that large companies were empirically defined as those with 14 or more heavy trucks and small companies were defined as 8 or less heavy trucks. Companies with only 1 or 2 trucks were then excluded on the basis that they were so small they would be unlikely to adopt formal safety management practices. As the final sample size was smaller than anticipated, no comparisons were conducted in the final analysis between small and large companies. The selection of companies on size, however, controlled for the effect of this variable in the comparison of good and poorer performers as each group contained roughly similar proportions of small and large companies. Use of safety management practices was self-reported and included (i) vehicle acquisition, maintenance and disposal policies and procedures, (ii) driver recruitment, payment and training arrangements, (iii) other safety policies, (iv) participation in accreditation schemes, (v) journey and schedule planning practices, (vi) driver participation in Work Health and Safety man-

329

agement, (vii) driver monitoring and incentives, and (viii) safety data recording. 2.2. Participants Fifty organisations meeting the claim rate and size criteria participated in the study. They were drawn from across Australia and were recruited in two ways, either via their insurer or via their telephone listing. First, a major insurer extended an invitation via its insurance brokers to all clients that met the selection criteria to participate in the study. The brokers were asked to either forward the invitation or to agree (or not) to allow researchers to directly invite their client to take part. Unfortunately, this strategy resulted in very few companies volunteering (n = 17). To increase the sample, additional companies were recruited from a random sample of transport companies listed in the Australian Yellow Pages. Only those companies meeting the same claim rate and company size criteria that were applied to the insurance sample were included in the final sample. In total, there were 20 low-claiming companies (9 small and 11 larger), and 30 higherclaiming companies (12 small and 18 larger) in the final sample. 2.3. Materials The survey was developed to assess whether the participating companies used a range of safety management practices identified in the literature. Although not exhaustive, the characteristics that were included were judged to cover the main areas of safety management practice. There were 44 questions primarily in forced choice format. The survey took approximately one hour to complete as an interview and covered the following broad areas. (i) Descriptive information about the company – 10 questions including: the type of freight carried, the number of trucks of different size classes used, the number of employee, subcontract and freelance drivers1 employed and the lengths of trips and types of work assigned them, the number of employee drivers in different age groups (625, 26–55, 55– 65, >65) the number of management staff involved in the transport function, annual heavy vehicle fuel consumption, and the number of recorded driver injuries in the past year. Companies recruited through telephone listings were also asked the number of vehicle insurance claims made in 2007–2009 to allow classification of their claim rate as low or higher. (ii) Safety-related policies and management systems at the company – 5 questions including: whether (yes/no) policies exist on a range of individual safety-relevant issues at the company (e.g., fleet management, fitness for duty, driver training, etc.), how safety polices are applied to contract drivers, whether (yes/no) drivers formally agree to conduct or policy statements, whether (yes/no) safety is a formal performance criterion for managers, and whether (yes/no) the company participates in a range of individual safety accreditation schemes. (iii) Usual vehicle acquisition, maintenance and fleet turnover practices – 7 questions including the age of vehicles at purchase, the length of truck retention, the existence (yes/no) and nature of a truck disposal policy, the average heavy vehicle truck age, the existence (yes/no) of a vehicle purchasing policy, whether (yes/no) a range of individual safety features (e.g., anti-lock brakes, etc.) are included in

1 Note that we distinguished between subcontractors that are used regularly and freelance drivers used only occasionally.

330

(iv)

(v)

(vi)

(vii)

(viii)

(ix)

L. Mooren et al. / Safety Science 70 (2014) 327–338

the purchasing policy or considered in the absence of a policy, whether (yes/no) and how often trucks are scheduled for regular maintenance, and the total fleet days lost to mechanical problems in a year. Work scheduling, route selection, journey planning and risk assessment – 4 questions including: whether (yes/no) consideration is given to safety in route and journey planning, whether (yes/no) a range of individual safety features (e.g., rest area availability, etc.) are considered, whether (yes/no) risk assessments or safety audits are done at own depots, what proportion of delivery sites are assessed or audited for safety (all/most/some/very few/none), how driver schedules and rosters are determined (i.e., centrally, locally, using scheduling software), and whether (yes/no) a range of specific strategies are used to monitor schedules. Driver recruitment – 3 questions including: who undertakes recruitment, whether (yes/no) and which particular qualifications are required, and which particular history and performance checks (e.g., employment references, licence currency, etc.) are routinely conducted before hiring a driver. Driver pay arrangements – 2 questions including: the basis of pay for driving (e.g., hourly rate, rate per km, etc.) and whether (yes/no) non-driving activities such as loading/ unloading and queuing/waiting are paid. Safety training and education – 5 questions including: whether (yes/no) a range of individual types of training are provided to drivers, which of a range of communication strategies are used to inform drivers about safety matters, whether (yes/no) and how schedulers are trained in fatigue management, whether (yes/no) and what type of safety training is provided to managers, and whether (yes/no) safety managers have undertaken any Health and Safety training. Safety management participation arrangements – 2 questions including: whether (yes/no) drivers can participate in safety management and what specific mechanisms (e.g., Health and Safety committee, etc.) they can use, and what process is applied to deal with safety concerns raised by drivers. Driver behaviour management – 5 questions including: whether (yes/no) individual forms of in-vehicle performance monitoring are used, what actions are taken in response to working hours breaches and other unsafe behaviours, whether (yes/no) and what system is used to analyse safety incidents, and whether (yes/no) and what positive incentives are provided for safe performance.

2.4. Procedure In the first round of recruitment, the insurance policy numbers of companies meeting the claim rate and selection criteria were extracted by researchers from the de-identified claims data of a major insurer. The criteria were chosen to maximise the differences between company size groups and between claims groups while ensuring roughly equal numbers of candidates in the four categories. The insurer contacted the insurance broker for each of these companies, explained the purpose of the study and requested the broker seek permission from the company for researchers to contact them. When permission was received, a researcher contacted the company to invite a representative to take part in the survey. Unfortunately many insurance brokers were unwilling to invite their clients to participate, resulting in very few invitations and, consequently few acceptances. After persevering with this recruitment strategy for some time, this method was abandoned due to a lack of cooperation by brokers to agree to researchers contacting their clients.

Instead, companies were recruited to the study through invitations to a random selection of transport companies listed in the Yellow Pages. In the second round of recruitment, researchers telephoned transport companies with publicly listed phone numbers directly to explain the study and invite a representative to participate. Information about the study was sent to all interested companies and formal consent was obtained from willing participants. All participants were provided with a copy of the survey questions in advance. For both methods of recruitment, to maximise convenience and thus participation, volunteer companies could choose to complete the survey as a telephone interview at a mutually convenient time, an online survey or a paper survey returned via mail or scanned email. All participants were given a $75 gift voucher to compensate their time. The study was approved by a UNSW Human Research Ethics Advisory Panel (08/2011/52). 2.5. Response rate As indicated in Table 1, the overall response was 14% of those candidates who were contacted. A total of 84 managers volunteered to complete a survey with a higher response rate for those contacted directly from the randomised list of companies in the Yellow Pages than for those identified on the basis of their insurance policy listing. Information about insurance claim rates and company size could not be determined in advance for companies recruited through the telephone listing, so questions on fleet size and claims were added to the questionnaire. Applying the same recruitment eligibility criteria as applied to the insurance policyholder population meant that only 33 of the 67 companies recruited from the randomised Yellow Pages list met the eligibility criteria and were included in the final sample. These criteria were used as they were based on objective outcomes (safety-related claims) and a large population of companies. The study involved a tailored design method in which study participants were allowed to select their preferred mode of response (Dillman, 2009) – telephone interview, online survey or paper survey. Mode of response was fairly evenly spread across the study groups. Of the 50 participants in the final sample, 6 low-claiming and 15 higher-claiming companies completed an online survey, 3 low-claiming and 7 higher-claiming companies completed a telephone interview, and 7 low-claiming and 12 higher-claiming companies completed a written questionnaire. As the final sample size was smaller than anticipated, comparisons were not conducted in the final analysis between small and large companies. The selection of companies on size, however, controlled for the effect of this variable in the comparison of good and poorer performers as each group contained roughly equal proportions of small and large companies (small-low n = 9, large-low n = 11 and small-high n = 12, large-high n = 18). 2.6. Survey analysis Given the small sample size, the usual chi square statistical tests would not produce statistically significant findings for most

Table 1 Survey response rates and number eligible for the study using criteria established from insurance criteria for those who completed the survey questionnaire using both methods of recruitment. Sample source

N volunteered

N contacted

Response rate

N eligible

Insurer Yellow Pages

17 67

199 404

0.09 0.17

17 33

Total

84

603

0.14

50

L. Mooren et al. / Safety Science 70 (2014) 327–338

variables being compared. Because of the small sample size, we concentrated instead on identifying larger effect sizes between good and poor performing companies. The importance of associations was based on effect sizes of the odds ratios and risk ratios to compare the probability of a characteristic being present in organisations with good safety track records compared with those with poorer safety track records. Based on the recent work by Olivier and Bell (2013) on effect sizes for 2  2 contingency tables, the odds ratios were considered meaningful when they were 1.86 and above, or the reciprocal 0.54 or below. These are considered to be medium to large effect sizes based on Cohen’s (1992) effect sizes for / relative to the maximum attainable correlation /max. To guard against the possibility of over-interpreting results based on very low or high proportions of companies, medium effects were only considered important if the percentages of positive responses were not <10% nor >90% or if they were <10% or >90%, but the low and high claim group differed by more than 10% points. Thus, comparisons of binary variables for low and higher claiming companies and mean values of continuous variables were made. However, sometimes the spread of the data was too great for using this method for mean values (for example, larger fleet sizes ranged from 14 to 886). In these cases, log-linear regression was used. The negative binomial was used in lieu of the Poisson due to over-dispersed data. Rates were analysed similarly using the rate denominator as an offset variable. Where the distribution was too skewed to obtain adequate model fit, Mann–Whitney tests were used. 3. Results The 50 participating organisations came from various industry sectors, such as local government councils, utility companies, and freight transport companies. All of the companies had at least one employee driver and 25 companies (12 low-claiming and 13 higher-claiming) employed subcontractor drivers, either on a regular basis and/or as freelancers, as well as employees. The comparison between good and poorer performing companies revealed a number of differences. As a consequence, the results of the comparisons between organisations with low and higher claim rates have been organised into the following sections according to overarching management topics: freight and vehicle fleet, journey and risk management, staffing and driver recruitment, policies and safety accreditation, scheduling and training, communication and driver participation in OHS, work monitoring, driver discipline and safety incentives, and incidents and record keeping. For parsimony, only the data on practices and characteristics showing evidence of differences between low and higher claiming companies have been tabled and in a few cases differences have been described only in the text. 3.1. Freight and vehicle fleet The types of freight carried and fuel usage per truck did not differ between low and higher claiming companies. However, companies with higher claim rates tended to be larger with an average of 65 trucks compared to low claiming companies that had an average of 19 trucks in their fleets. This difference in absolute fleet size was significant (rate ratio = 0.29, p = 0.001), but it should be noted that the high mean number of heavy trucks for higher claimers was inflated by a minority of companies with fleets that were noticeably larger than the others (i.e., 119–886 trucks versus <60 trucks). To explore the relationship between claim rate and fleet size further, the claim rate per truck was examined for higher claiming, large companies (with >14 trucks). The higher claiming,

331

large companies with smaller fleets had a lower mean claim rate (0.38 claims per truck, SD = 0.16; n = 13) than the higher claiming, large companies with larger fleets (0.65 claims per truck, SD = 0.39; n = 5) but the difference was not statistically significant (Mann– Whitney U = 18, Standardised Test Statistic = 1.430, p = 0.173). The odds of higher claiming companies were less than half those of lower claimers in considering safety features when purchasing vehicles (Table 2). There was some indication in the data that consideration of Electronic Stability Programs (ESP) and front or rear underrun devices also had higher (doubled) odds among higher claimers, but the analysis was limited by the small number of companies considering each technology. None of the companies considered rear underrun devices alone, so when they considered them, they were also likely to consider other devices as well. For this reason, consideration of front underrun, rear underrun and Electronic Stability Program (ESP) technology were combined. The odds of companies considering any of the three was higher among higher claimers than low claimers. A similar percentage of larger (24%) and smaller (19%) companies considered any of these three devices. 3.2. Journey and risk assessment Companies were asked if they consider route safety factors in planning journeys, including grade separation, overtaking lanes, bridge capacity, road conditions, rest area availability, HAZMAT routes, over-dimensional vehicle access, speed limiting on poorer quality roads, traffic conditions, weigh stations, safety-cams, tunnels and low underpasses. There were few differences in responses by higher and low claimers, but the organisations that had low claim rates more often checked traffic conditions when planning transport journeys and considered speed limiting on poorer quality roads than did those with higher claim rates (Table 3). Companies were also asked if they audit their own worksites and the delivery sites. The odds ratios show that low claiming companies conduct risk or safety assessments of their own sites compared with higher claiming companies. 3.3. Staffing and driver recruitment There was not an important difference between higher and low claimers in their use of sub-contractor drivers nor their transport manager to driver ratios. The companies were asked about aspects of driver histories and performance that were checked prior to employment. Examples of recruitment checking criteria that were given in the questionnaire were: references from previous employers, licence currency, licence points, accident history, in-vehicle driving performance, selection test performance, health or other. Those with low claim rates more often checked accident histories of drivers before hiring them (Table 4). On the other hand, the odds for low claiming companies to check references, licence points or conduct performance tests when recruiting drivers were less than for higher claimers. In addition, companies were asked how many employee drivers were under 25 years of age, 26–55 years of age, 55–65 years of age and over 65 years of age. The higher claiming company odds ratio for employing any drivers over the age of 65 was more than four times that of low claiming companies, and higher claimers had nearly twice the odds of employing drivers aged 55–65, but showed no differences in the younger age groupings. Table 4 shows the results for staffing and recruitment practices with medium or greater effect sizes. Higher and low claimers did not differ on a number of aspects of the recruitment process including whether subcontractors were responsible for recruiting drivers. Higher claimers had higher odds

332

L. Mooren et al. / Safety Science 70 (2014) 327–338

Table 2 Considerations involved in truck purchasing decisions by companies with higher and low insurance claim rates where higher odds ratio indicates higher claimers have larger odds than low claimers.

a

Safety considerations applied to truck purchasing

Higher claim rate n (%) (N = 30)

Low claim rate n (%) (N = 20)

Odds ratioa

At least one vehicle safety feature considered in purchase ESP or front or rear underrun devices

16 (53) 9 (30)

14 (70) 3 (15)

0.490 2.429

Medium to large effect sizes were considered to be meaningful and were identified where odds ratios were >1.86 or <0.54 (Olivier and Bell, 2013).

Table 3 Journey and site risk assessment by companies with higher and low insurance claim rates where higher odds ratio indicates higher claimers have larger odds than low claimers.

a

Journey and site risk assessment practices

Higher claim rate n (%) (N = 30)

Low claim rate n (%) (N = 20)

Odds ratioa

Check traffic conditions prior to journeys Speed limiting on poorer quality roads Carry out safety audits at own worksites

6 (20) 3 (10) 24 (80)

8 (40) 4 (20) 19 (95)

0.375 0.444 0.211

Medium to large effect sizes were considered to be meaningful and were identified where odds ratios were >1.86 or <0.54 (Olivier and Bell, 2013).

Table 4 Driver staffing and recruitment practices for companies with higher and low insurance claim rates where higher odds ratio indicates higher claimers have larger odds than low claimers.

a

Driver recruitment and staffing practices

Higher claim rate n (%) (N = 30)

Low claim rate n (%) (N = 20)

Odds ratioa

Use recruitment firm Check accident history Check references Check licence points Conduct performance test Has employee drivers over age 65 Has employee drivers between ages 55 and 65

6 18 28 20 6 9 19

2 15 16 8 2 2 10

2.25 0.500 3.500 3.000 2.250 4.263 1.900

(20) (60) (93) (67) (20) (32) (66)

(10) (75) (80) (40) (10) (10) (50)

Medium to large effect sizes were considered to be meaningful and were identified where odds ratios were >1.86 or <0.54 (Olivier and Bell, 2013).

of using a recruitment firm for hiring, but this method of recruiting drivers was relatively uncommon. 3.4. Pay and conditions Companies were asked how drivers (employees, subcontractors and freelance drivers) were paid for driving. The options included: hourly rate, flat day rate, day rate with overtime, weekly rate, weekly rate with overtime, salary, flat rate per truckload, trip rate (based on kilometres travelled or tonnage carried) or other. Compared to low claiming companies, the odds of higher claimers paying employees by productivity (trip or truckload) methods were nearly five times higher (Table 5). All of the low claiming companies utilised time-based pay, i.e. hourly or salaries, for at least some of their employee drivers, compared to 87% of higher

claimers. It should be noted that 3 low claiming and 3 higher claiming companies used time-based and trip-based payment methods to pay their employee drivers, most probably reflecting different freight tasks assigned to different employee drivers within these companies. There was no difference between low and higher claiming companies in the ways that employee drivers were remunerated for loading and unloading. However, employee drivers in higher claiming companies had lower chances of being paid for hours spent queuing or waiting than employee drivers in low claiming companies. When subcontractor drivers were employed, the odds of higher claimers paying them trip rates were less than the odds of low claimers (Table 5) and the odds of paying for queuing and waiting time were higher. On the other hand, higher claimers were more

Table 5 Driver payment practices for companies with higher and low insurance claim rates where higher odds ratio indicates higher claimers have larger odds than low claimers.

a b

Driver payment methods

Higher claim rate n (%)

Low claim rate n (%)

Odds ratioa

For employees Employees paid for time worked Employee drivers paid either per truckload or trip Employees paid for all hours spent queuing or waiting

N = 30 26 (87) 14 (47) 24 (80)

N = 20 20 (100) 3 (15) 19 (95)

–b 4.958 0.211

For subcontractors (for companies that employ them) Subcontractor drivers paid a trip rate Subcontractor drivers paid for all hours spent queuing or waiting

N=8 3 (38) 7 (86)

N=4 3 (75) 2 (50)

0.200 7.000

For freelance drivers (for companies that employ them) Freelance drivers paid a trip rate Freelance drivers paid for all hours loading and unloading

N = 10 6 (60) 2 (20)

N=8 2 (25) 4 (50)

4.500 0.250

Medium to large effect sizes were considered to be meaningful and were identified where odds ratios were >1.86 or <0.54 (Olivier and Bell, 2013). Odds ratio could not be calculated as there was no variation in one group.

333

L. Mooren et al. / Safety Science 70 (2014) 327–338

Table 6 Safety policies and safety accreditation by companies with higher and low insurance claim rates where higher odds ratio indicates higher claimers have larger odds than low claimers.

a

Safety policies and accreditations

Higher claim rate n (%) (N = 30)

Low claim rate n (%) (N = 20)

Odds ratioa

Safety policies on fleet management Safety policies on fatigue management Safety policies on work monitoring NHVAS Mass Management NHVAS Basic Fatigue Management Have key performance indicators for safety management

20 25 14 11 11 10

8 10 5 3 4 3

3.000 5.000 2.625 3.281 2.316 2.982

(67) (83) (47) (79) (37) (35)

(40) (50) (25) (21) (20) (15)

Medium to large effect sizes were considered to be meaningful and were identified where odds ratios were >1.86 or <0.54 (Olivier and Bell, 2013).

Table 7 Scheduling and training by companies with higher and low insurance claim rates where higher odds ratio indicates higher claimers have larger odds than low claimers.

a

Scheduling and training practices

Higher claim rate n (%) (N = 30)

Low claim rate n (%) (N = 20)

Odds ratioa

Schedules are determined centrally Loads are scheduled centrally and rosters determined locally Local depots schedule and roster drivers Fatigue risk management training for employee drivers Driving skills (on road) training for employees Driving skills (classroom) training for employees Eco driving (fuel economy) training for employees Pre-trip vehicle inspection training for employees

12 7 10 21 17 7 5 14

13 (65) 1 (5) 4 (20) 10 (50) 5 (25) 1 (5) 1 (5) 6 (30)

0.359 5.783 2.000 2.333 3.923 5.783 3.800 2.042

(40) (23) (33) (70) (57) (23) (17) (47)

Medium to large effect sizes were considered to be meaningful and were identified where odds ratios were >1.86 or <0.54 (Olivier and Bell, 2013).

likely than low claimers to pay freelance drivers a trip rate and the odds of freelance drivers being paid for all hours loading or unloading were lower at higher claiming companies than low claiming companies. 3.5. Policies and safety accreditation The survey asked about a number of safety policies including: OHS risk assessment, OHS audits, OHS reporting systems, fleet management, driver recruitment and selection, driver training, fitness for duty, fatigue management, on-road behaviour, depot behaviour, work monitoring, work planning (e.g. journey planning), driver performance monitoring, seat belt use, vehicle selection, vehicle maintenance, depot conditions, accident prevention and response. The analysis showed that the odds of higher claimers having policies on fleet management, fatigue management and work monitoring were considerably greater than for low claimers (see Table 6). No differences were found between the claimer groups for any of the other policies. With regard to safety management accreditation programs, the survey asked if the company was accredited under the National Heavy Vehicle Accreditation Scheme (NHVAS) Mass Management, NHVAS Maintenance, NHVAS Basic Fatigue Management, NHVAS Advanced Fatigue Management, Trucksafe, ISO 9001 or other schemes. Higher claimers showed higher odds of being accredited under the NHVAS Mass Management and NHVAS Basic Fatigue Management compared with the low claimers. No other important differences were found for the claimer groups on membership of accreditation programs. Respondents were also asked if managers in their company had heavy vehicle safety management key performance indicators (KPIs). The odds for higher claimers having these KPI’s were nearly three times greater compared with those for low claimers. 3.6. Scheduling and training Respondents were asked whether drivers’ schedules are determined centrally, whether local depots schedule and roster drivers,

whether loads are scheduled centrally and rosters determined locally, whether rostering/scheduling software is used or if they managed scheduling and rostering in some other way. Companies with lower claim rates showed higher odds of scheduling work centrally, whereas higher claiming companies had greater odds of scheduling loads centrally while rostering drivers locally and/ or doing both locally (Table 7). Asked about whether schedulers were trained in fatigue risk management, a total of 20 higher claimers and 14 lower claimers answered this question with 15 higher claimers and only 6 low claimers saying that they are trained, so the odds of higher claimers doing this was 4 times greater than low claimers. The survey asked what type of safety training was provided for drivers, including the following types: OHS induction, Occupational Health and Safety, fatigue risk management, driving skills (on road), driving skills (classroom), pre-trip inspection, manual handling, loading/ unloading, or other. The odds of higher claiming companies reporting conducting various types of safety related training for their employee drivers, including fatigue risk management, driving skills (on road and in classroom), eco driving and pre-trip vehicle inspection, were higher compared to the low claiming companies. The survey asked whether managers responsible for OHS were provided specific safety training. No differences in this practice distinguished between higher and low claimers. 3.7. Communication and driver participation in Occupational Health and Safety The survey asked how safety information was communicated to drivers, in particular asking whether they used: newsletters, noticeboards, supervisors, union representatives, toolbox talks or staff meetings, special briefings, text messages, email, staff website or other means. There was little difference in methods or mean number of communication methods used by companies with low claim rates and by higher claimers, both averaging three modes of communication. The survey also asked whether and how drivers are involved in OHS decision-making – e.g. through an OHS committee or union

334

L. Mooren et al. / Safety Science 70 (2014) 327–338

Table 8 Communication and driver input for companies with higher and low insurance claim rates where higher odds ratio indicates higher claimers have larger odds than low claimers.

a b

Methods of safety communication and driver participation

Higher claim rate n (%) (N = 30)

Low claim rate n (%) (N = 20)

Odds ratioa

Drivers are involved in OHS decision-making Drivers are involved via an OHS committee There is a time limit for dealing with drivers’ safety concerns The time limit is monitored and enforced

24 14 11 9

19 (95) 6 (30) 11 (61) 11 (100)

0.211 2.042 0.389 –b

(80) (47) (38) (82)

Medium to large effect sizes were considered to be meaningful and were identified where odds ratios were >1.86 or <0.54 (Olivier and Bell, 2013). Odds ratio could not be calculated as there was no variation in one group.

representative, toolbox or staff meetings, suggestion box or other means. The odds of opportunities for driver input to safety decision-making were less in companies with higher claims than in companies with low claims (Table 8). Driver involvement via an OHS committee was more likely among higher than low claiming companies. When asked if there was a time limit for addressing drivers’ safety concerns, the odds of higher claimers were less than those of low claimers to set and monitor time limits for responding to drivers’ safety concerns. 3.8. Work monitoring Respondents were asked about monitoring driver hours and schedules. They were asked to indicate if hours and schedules are not monitored, in-truck logs or work records are reviewed after an incident, random spot checks of in-truck logs or work diaries are carried out, all in-truck recordings are reviewed, all diary records are reviewed or hours and schedules are monitored in another way. None of the work monitoring variables distinguished between higher claimers versus low claimers. The survey also asked whether the companies used other forms of in-vehicle monitoring, including: GPS tracking of the vehicle, fuel consumption, braking analysis gear change analysis, speed analysis, fatigue monitoring systems (e.g. Optalert) or other devices. Higher claiming companies were more likely to report GPS tracking, fuel consumption monitors, and speed analysis devices (see Table 9). Low claimers responded with a range of other types of monitoring practices not listed in the questionnaire choices, such as sending new drivers out with experienced drivers, documenting fatigue management records, and pre-start checks. 3.9. Driver discipline and incentives Higher and low claimers were found to differ on some questions about the company’s procedures for dealing with breaches of working hours and the actions taken on unsafe behaviour. Higher claimers had lower odds of using other methods for dealing with breaches of working hours, of doing formal investigations when people behave unsafely and of using other actions when people behave unsafely (such as issuing a letter to the driver and notes for safety meeting discussions, or the smaller companies would directly counsel the driver or dismiss them) compared to low claimers (see Table 10). In contrast, higher claimers had higher

odds of investigating the reasons for breaches of working hours and of applying discipline or penalties for breaches. Higher claimers had lower odds of using incentives to encourage safety innovations than did low claimers. 3.10. Incidents and record keeping No significant differences were found between higher and low claiming companies in terms of the records they keep on incidents, injuries, infringements or (truck) defect notices, nor to the infringement rates per employee drivers. The incidence of defect notices differed between higher and low claiming companies with low claiming companies having a mean rate of 0.10 and median of 0.03 defect notices per truck versus 0.34 and 0.15 defect notices per truck respectively for higher claiming companies (p = 0.02). While the data did not reveal significant differences in lost time injuries per driver between higher and low claimers, low claimers reported a trend for fewer average non-lost time injuries per driver (0.03, n = 18) than did higher claimers (0.07, n = 27; Mann–Whitney U = 173, Standardised Test Statistic = 1.86, p = 0.06). Importantly, this finding confirms that the variable used to index safety performance, a company’s 2007–2009 claim rate, is consistent with at least one important measure of safety outcome recorded at the time the survey was conducted in 2011–2012. 4. Discussion The results of this study showed some safety-related characteristics that distinguished companies with higher and lower safety performance outcomes. Some of these characteristics were as would be expected, but the survey also produced some surprising results. These results are summarised in Table 11. Overall, good performing (low claiming) companies were smaller, with fewer trucks. They also had fewer defect notices, did more safety-related checking and monitoring such as site safety audits, checking traffic conditions, speed limiting on poorer quality roads, and checking accident history at recruitment, were more likely to pay employee and freelance drivers for all time worked, actively monitored driver work and work load, and paid active attention to policy and compliance by having a formal approach to policy breaches, seeking driver input into OHS and responding quickly to safety concerns. Some of the characteristics that ‘‘first principles’’ as well as the literature would suggest should be more prevalent in HV

Table 9 In-vehicle monitoring by companies with higher and low insurance claim rates where higher odds ratio indicates higher claimers have larger odds than low claimers.

a

Type of in-vehicle monitoring used

Higher claim rate n (%) (N = 30)

Low claim rate n (%) (N = 20)

Odds ratioa

Use Use Use Use

19 (63) 13 (43) 9 (30) 1 (3)

7 4 3 4

3.200 3.059 2.429 0.138

GPS tracking device in-vehicle monitoring fuel consumption in-vehicle monitoring speed analysis in-vehicle monitoring other in-vehicle monitoring systems

(35) (20) (15) (20)

Medium to large effect sizes were considered to be meaningful and were identified where odds ratios were >1.86 or <0.54 (Olivier and Bell, 2013).

335

L. Mooren et al. / Safety Science 70 (2014) 327–338

Table 10 Driver discipline practices and safety incentives by companies with higher and low insurance claim rates where higher odds ratio indicates higher claimers have larger odds than low claimers.

a

Use of driver discipline practices and safety incentives

Higher claim rate n (%) (N = 30)

Low claim rate n (%) (N = 20)

Odds ratioa

Formally discipline and penalise drivers for breaches of working hours Investigate reasons for breaches of working hours Use other methods to deal with breaches of working hours Formal investigation when people behave unsafely Other actions when people behave unsafely Incentives are provided for safe work or management safety innovations

8 (27) 14 (47) 2 (7) 6 (20) 3 (10) 5 (17)

2 5 4 7 4 7

3.273 2.625 0.286 0.464 0.444 0.387

(10) (25) (20) (35) (20) (35)

Medium to large effect sizes were considered to be meaningful and were identified where odds ratios were >1.86 or <0.54 (Olivier and Bell, 2013).

Table 11 Summary of expected and unexpected findings comparing safety characteristics of low and higher insurance claimers where higher odds ratio indicates higher claimers have larger odds than low claimers. Low claimers report more frequently than higher claimers (expected)

Higher claimers report more frequently than low claimers (unexpected)

Fleet

Smaller fleets (averaging 15 trucks) More often consider safety features Fewer defect notices

Larger fleets (averaging 65 trucks) Consider underrun, ESP

Risk assessment

Check traffic conditions Speed limiting on poorer quality roads Safety audits at own sites

Recruitment and employment

Check accident history Fewer drivers >65 years and between 55 and 65

Check references, licence points, use selection tests

Pay and conditions

Pay employee and freelance subcontract drivers for time worked

Less likely to pay subcontractors a trip rate and more often pay for hours waiting or queuing

Policies

Have policies on fatigue management and work monitoring

Accreditation

NHVAS Basic Fatigue Management NHVAS Mass Management KPIs

Scheduling

Schedule and roster centrally

Communication and driver input

Encourage driver input into OHS Faster response to safety concerns

In-vehicle monitoring

Experienced supervise new drivers Document fatigue management Conduct pre-start checks

Driver discipline and incentives

Formal approach to policy breaches Offer incentives for safety innovations

companies with low insurance claim rates than higher claim rates, were not found. Counterintuitive findings, particularly in the areas of driver safety training, safety policies, and in-vehicle monitoring are difficult to explain. It appears that companies with higher claim rates conduct more safety related training than do the low claiming companies. This could mean that training is not one of the most important safety management practices, the type, or method, of training provided is not effective, or training was introduced in response to safety incidents and so has not yet had a positive effect on safety. It is possible that this effect may be at least partly related to smaller company size where training is less important as more direct and interactive control of work practices have more effect. Higher claiming companies also had a greater number of safety policies. The number of safety policies may not, however, reflect the level of the safety management in these companies. The existence of policies does not necessarily imply that they have been accepted by all of management nor been adopted into the ‘way we do things around here’ (Hopkins, 2006; Zohar, 1980). This issue, dubbed ‘‘paper compliance’’ (Quinlan et al., 2010), also applies to membership of accreditation systems where their simple existence cannot necessarily be an indicator that they have been adopted throughout the company. Further research and more in-depth examination is needed to determine whether these are just documents or they are actively promoted and enforced, and whether

Have in-vehicle monitoring devices: GPS, fuel, speed

these were introduced as a response to a problem with (claim) incidents. Low claiming companies seem to have a more formal and systematic approach to dealing with policy breaches. This may suggest that while rules exist in companies with higher claim rates, they may not be backed up as much with an active or formal safety management approach. Again, company size may also be influential as policies may have a more direct effect on middle managers and drivers in small companies with fewer layers and less dispersion of influence compared to large companies with many layers of authority and communication. Again, contrary to expectations, higher claimers used more methods of pre-employment checks when recruiting drivers, including reference and licence checks and (driving) performance tests. On the other hand, low claimers more often checked accident history prior to employing drivers and less often employed drivers over the age of 65. This may suggest that checking accident history is a more effective selection criterion than other pre-employment checks. The observed relationship between older employee drivers and claim rates also needs further investigation including whether there is an age-relationship with claims in these companies. In work monitoring, low claimers more often used less technological forms of monitoring such as sending new drivers out with experienced drivers and doing pre-start checks, but higher claimers used a greater number and range of in-vehicle monitoring

336

L. Mooren et al. / Safety Science 70 (2014) 327–338

systems than did low claimers. It may be that these companies have particular operational needs for installation of truck monitoring devices and meeting audit requirements for carrying greater mass and working longer hours, or for tracking delivery times. Whether or not the monitoring devices are checked in real time, or the audit requirements are maintained between audits is unknown. The higher claimers have greater odds of being accredited under the NHVAS Basic Fatigue Management scheme than low claimers. The literature contains mixed results on the value of accreditation schemes. One American study found that the safety performance of HV companies improved after certification to the ISO 9000 standard (Naveh and Marcus, 2007), but the reports of safety performance linked with NHVAS are inconclusive (Mooren and Grzebieta, 2011). Clearly, more research is needed to understand the role that accreditation systems play in encouraging safer performance in the heavy trucking industry. There also seems a much greater tendency for higher claiming companies to pay employee drivers by truckload or per trip, i.e., on a piece rate basis, compared with low claiming companies who tend to pay drivers on a time basis, whether on an hourly or weekly pay basis or by salary. The by-trip or truckload payment method has been found in other research to be linked with poor safety performance (Quinlan and Wright, 2008b; Rodriguez et al., 2006; Williamson, 2007; Williamson and Friswell, 2013). So, this finding is consistent with the expected results based on previous research. When low claimers employed subcontractors they paid them by trip but were more likely to include payment for queuing and waiting. Both high and low claimers used trip pay for some employee drivers which may reflect different freight tasks assigned to different employee drivers within these companies. On the other hand, freelance drivers were employed by both high and low claimers, but they treated them differently. While most higher claimers paid these drivers by trip and did not pay for any other aspects of work, most low claiming companies employing freelance drivers paid for all work. This pattern of findings reinforces the conclusion that paying drivers for work is associated with better safety outcomes. The results for subcontractor and freelance payments require replication in a larger sample of companies. Finally, while higher claiming companies reported the use of more types of communication modes with drivers, driver participation in safety management and encouragement of innovation in safety management seem to be more prevalent in companies with low claim rates compared with higher claim rates. This could imply that the quality and type of communication is more important that the number of types of communication used or that this is easier to achieve in smaller companies, or both. The general impression from the survey results is that the heavy vehicle transport companies with low insurance claim rates per truck tend to take a more active and substantive approach to managing safety in their organisations, whereas the higher claiming companies take a more passive or static method of managing safety. The higher claimers tend to rely more on setting criteria and rules for vehicles and drivers, than do the low claimers. Low claimers seem to more strongly focus on proactive risk assessment, ensuring that rules are agreed, and consulting drivers on safety issues.

drivers in Australia (Friswell, 2013 (Friswell, 2013; Williamson et al., 2009) and internationally (Peignier et al., 2011) have had similar response rates. Furthermore, two studies of response rates for surveys in the trucking industry confirmed similar response rates using similar approaches to that found in the current study (Lau, 1995; Lawson and Riis, 2001). Nevertheless, this study took a unique approach to understanding the predictors of safe performance in this industry by actively comparing the characteristics of companies with demonstrated good and poorer safety outcomes. The results therefore provide the foundation for further research to show whether the characteristics that distinguished good and poorer performing companies in this study will be confirmed in another study. We had hoped that by recruiting trucking companies through an insurer we would have validated safety outcome data and achieve better response rates. We encountered problems as the insurance brokers, who were the link with companies, were unwilling to allow us to contact companies directly despite strong assurances that any participation in this study would be anonymous. Consequently, our response rates almost certainly underestimated the actual response rates as many brokers (but unfortunately an unknown number) did not invite participation on our behalf. Given the low numbers of participants recruited for this study, performance differences between large and small companies were not pursued separately in the analysis. Despite controlling for fleet size by ensuring the two performance groups had similar representation of large and smaller companies, however, fleet size was associated with claim rates since higher claimers had on average a larger number of trucks in the fleet. This effect clearly needs to be pursued in further research. Also, as vehicle kilometres travelled was not taken into account, some differences between higher and low claiming companies may have been a function of more exposure to on-road risk. However, the finding that fuel usage per truck (an index of distance travelled) was similar in low and higher claiming companies argues against this possibility. Further research is needed to understand whether this has an effect on safe performance. Another limitation was that the claims status was established for a period before the survey period using claims data (or reported claims) from 2007 to 2009, and survey period in 2012. Therefore, the safety management features reported in 2012 may have been put in place as a response to claims in the prior 3-year period, or there may have been other significant changes such as company acquisitions. However, as the non-lost time injury comparisons showed that higher claimers tended to have higher rates than low claimers, it would seem that perhaps these safety management efforts have not resulted in changing their risk compared with the low claimers. Finally, the cross-sectional nature of the study may have masked possible explanations of seemingly counterintuitive findings. For example, companies with higher claim rates may have introduced more policies, training and monitoring systems as a response to their heightened insurance claims. However, this study was not able to determine when or if these changes occurred before, during or after the period selected for comparison. 5.2. Conclusions

5. Limitations and conclusions 5.1. Limitations Clearly, the company response rate is a limitation of this study. A considerably higher response rate was intended, however, low response rates are a continuing problem with this study population. A number of other studies involving light and heavy truck

The value of this study is that some safety management characteristics distinguished heavy vehicle transport companies with low insurance claim rates from with those with higher claim rates. While there may be important characteristics or combinations of characteristics that could not be tested in this study, this research provides some directions for further research by indicating where to look further to examine aspects of safety management that can

L. Mooren et al. / Safety Science 70 (2014) 327–338

predict safety performance. Clearly, pay methods are important considerations in safety management. All of the low claiming companies paid their drivers for the time worked, whereas the higher claimers had nearly five times the odds of paying drivers by truckload or per trip by comparison. The unexpected results, such as higher claiming companies having more safety policies, assist to refine the focus of further research into the content and implementation of these policies. Those with low insurance claims reported having more active management characteristics than higher claimers, such as checking traffic conditions, speed limiting on poorer quality roads, conducting safety audits, gaining OHS ideas from drivers, acting on drivers’ safety concerns, and taking a formal approach to addressing safety breaches. Conversely, the higher claimers had in place safety equipment and policies, but it was unclear, from the survey, how these were used in practice. It may be that applying specific criteria in the selection of drivers, trucks and monitoring equipment, together with having a range of safety policies is not as important as other practices such as risk assessments, enforcement of rules, and involving drivers in safety management. More research is needed to confirm the features of an effective safety management system that were found in this study. There are many questions raised by these findings that suggest the need for validating their importance for safety management in further studies that take into account company size. In particular, this will require observational studies including in-depth or site visits, discussions with managers and drivers, and observations of documents and the workplace and operations. This further research is needed to better define how elements of safety management are applied and received in organisational settings. Moreover, assessing the safety outcomes from combinations of safety management characteristics are likely to reveal a fuller picture of how a safety management system should be constructed. Then, the development of a holistic approach to safety management that can potentially improve occupational and road safety in organisations that operate heavy vehicles may be possible.

Acknowledgements The authors thank the participating company representatives. We acknowledge the financial and technical support provided by ARC Linkage Grant LP100100283 partners, the NSW Centre for Road Safety, Transport for NSW, Transport Certification Australia, National Transport Commission, Zurich Financial Services, and the Motor Accidents Authority of NSW to make this paper possible. Prof. Williamson is supported by an NHMRC Senior Research Fellowship. Prof. Grzebieta was supported by the NSW Centre for Road Safety at Transport for NSW. The authors also acknowledge the contributions and assistance provided by Mr. Peter Johansson, Mr. Roger Hancock, Dr. Soames Job and Dr. Charles Karl.

References Arboleda, A., Morrow, P.C., Crum, M.R., Shelley, L., Mack, C., 2003. Management practices as antecedents of safety culture within the trucking industry: similarities and differences by hierarchical level. J. Saf. Res. 34, 189–197. Australian Transport Council, 2011. Road Safety. Retrieved from: (13.02.12). Baas, P., Taramoeroa, N., 2008. Analysis of the Safety Benefits of Heavy Vehicle Accreditation Schemes. Austroads, Sydney. Banks, T.D., 2008. An Investigation into How Work-Related Road Safety can be Enhanced, Centre for Accident Research & Road Safety – Qld (CARRS-Q). Queensland University of Technology. Belzer, M., Rodriguez, D., Sedo, S., 2002. Paying for Safety: An Economic Analysis of the Effect of Compensation on Truck Driver Safety. Science Applications International Corporation, Federal Motor Carrier Safety Administration.

337

Bezwada, N., 2010. Characteristics and contributory causes associated with fatal large truck crashes. Department of Civil Engineering, Kansas State University, Manhattan. Blower, D., Green, P., Matteson, A., 2010. Condition of trucks and truck crash involvement. Transport. Res. Rec.: J. Transport. Res. Board 2194, 21–28. Bruning, E.R., 1989. The relationship between profitability and safety performance in trucking firms. Transport. J. 28, 40–49. Cantor, D., Corsi, T., Grimm, C., Ozpolat, K., 2010. A driver focused truck crash prediction model. Transport. Res. Part E: Logis. Transport. Rev. 46, 683–692. Cohen, J., 1992. A power primer. Psychol. Bull. 112, 155–159. Corsi, T., Barnard, R., Gibney, J., 2002. Motor carrier industry profile: linkages between financial and safety performance among carriers in major industry segments. Analysis Division U.S. Department of Transportation, Washington, DC, pp. 1–84. Crum, M., Morrow, P., 2002. The influence of carrier scheduling practices on truck driver fatigue. Transport. J. 42, 20–41. de Pont, J., 2005. Assessing Heavy Truck Safety in Tasmania. Department of Infrastructure, Energy and Resources, Hobart. Department of Transportation U.S., 2006. Report to Congress on the Large Truck Crash Causation Study. National Technical Information Service, Virginia. Dillman, D.A., 2009. Internet, Mail, and Mixed-Mode Surveys: The Tailored Design Method. Wiley, New York. Eakin, J., Champoux, D., MacEachen, E., 2010. Health and safety in small workplaces: refocusing upstream. Can. J. Pub. Health (March/April), S29–S33. Fernandez-Muniz, B., Montes-Peon, J., Vazquez-Ordas, C., 2007. Safety management system: development and validation of a multidimensional scale. J. Loss Prev. Process Ind. 20, 52–68. Feyer, A., Williamson, A., 1995. Work and rest in the long-distance road transport industry in Australia. Work Stress 9, 198–205. Friswell, R., 2013. Fatigue and occupational safety in short haul, light transport. PhD Thesis. School of Public Health and Community Medicine. University of New South Wales. Geldart, S., Smith, C.A., Shannon, H.S., Lohfeld, L., 2010. Organizational practices and workplace health and safety: a cross-sectional study in manufacturing companies. Saf. Sci. 48, 562–569. Golob, T., Hensher, D., 1994. Driver Behavior of Long Distance Truck Drivers: The Effects of Schedule Compliance on Drug Use and Speeding Citations. Institute of Transport Studies, University of California, Irvine. Gregersen, N.P., Brehmer, B., Moren, B., 1996. Road safety improvement in large companies. An experimental comparison of different measures. Accid. Anal. Prev. 28, 297–306. Haworth, N. et al., 2002. Truck safety benchmarking study. Melbourne, National Transport Commission, Australia. Hopkins, A., 2006. Studying organisational cultures and their effects on safety. Saf. Sci. 44, 875–889. Huang, Y.-H., Ho, M., Smith, G.S., Chen, P.Y., 2006. Safety climate and self-reported injury: assessing the mediating role of employee safety control. Accid. Anal. Prev. 38, 425–433. Knipling, R., Nelson, K., Bergoffen, G., Burks, S., 2011. Safety management in small motor carriers: a synthesis of safety practice, CTBSSP Synthesis 22, Washington, DC. Langwieder, K., Gwehenberger, J., Hummel, T., Bende, J., 2001. Benefit potential of ESP in real accident situations involving cars and trucks. In: 18th ESV Conference, Nagoya, Japan. Lau, S.W., 1995. Truck Travel Surveys: A Review of the Literature and State-of-theArt. Metropolitan Transport Commission, Oakland, California. Lawson, C., Riis, A.-E., 2001. We’re really asking for it: using surveys to engage the freight community. Transport. Res. Rec. 1763, 13–19. Li, Y., Itoh, K., 2013. Safety climate in the trucking industry and its effects on safety outcomes. Cognit. Techn. Work, 1–12. Loeb, P.D., Clarke, W.A., 2007. The determinants of truck accidents. Transport. Res. Part E: Logis. Transport. Rev. 43, 442–452. Lueck, M., Murray, D., 2011. Predicting Truck Crash Involvement: A 2011 Update. American Transportation Research Institute. Mitchison, N., Papadakis, G., 1999. Safety management systems under Seveso II: Implementation and assessment. J. Loss Prev. Process Ind. 12, 43–51. Monaco, K., Williams, E., 2000. Assessing the determinants of safety in the trucking industry. J. Transport. Stat. 3, 69–80. Mooren, L., Grzebieta, R., Williamson, A., Olivier, J., 2012. Comparing heavy vehicle safety management in Australia and the United States. In: ACRS National Conference – A Safe System: Expanding the Reach, Sydney. Mooren, L., Grzebieta, R., Williamson, A., Olivier, J., Friswell, R., 2014. Safety management for heavy vehicle transport: a review of the literature. Saf. Sci. 62, 79–89. Mooren, L., Grzebieta, R., 2011. Review of Australian alternative compliance schemes. In: TRB Annual Meetings. Transportation Research Board, Washington, DC. Moses, L., Savage, I., 1994. The effect of firm characteristics on truck accidents. Accid. Anal. Prev. 26, 173–179. Naveh, E., Marcus, A., 2007. Financial performance, ISO 9000 standard and safe driving practices effects on accident rate in the U.S. motor carrier industry. Accid. Anal. Prev. 39, 731–742. Olivier, J., Bell, M., 2013. Effect sizes for 2  2 contingency tables. PLoS One 8. Oystein Saksvik, P., Torvatn, H., Nytro, K., 2003. Systematic occupational health and safety work in Norway: a decade of implementation. Saf. Sci. 41, 721–738. Parker, D., West, R., Stradling, S., Manstead, A.S.R., 1995. Behavioural characteristics and involvement in different types of traffic accident. Accid. Anal. Prev. 27, 571–581.

338

L. Mooren et al. / Safety Science 70 (2014) 327–338

Peignier, I., Leroux, M.-H., de Marcellis-Warin, N., Trépanier, M., 2011. Organizational safety practices of hazardous materials carriers. Transport. Lett. 3, 149–159. Quinlan, M., Bohle, P., Lamm, F., 2010. Managing Occupational Health and Safety: A Multidisciplinary Approach, third ed. Palgrave MacMillan, Melbourne. Quinlan, M., Wright, L., 2008a. Remuneration and Safety in the Australian Heavy Vehicle Industry: A Review Undertaken for the National Transport Commission, Melbourne. Quinlan, M., Wright, L., 2008b. Safe Payments: Addressing the Underlying Causes of Unsafe Practices in the Road Transport Industry. National Transport Commission, Melbourne. Rodriguez, D., Targa, F., Belzer, M., 2006. Pay incentives and truck driver safety: a case study. Ind. Labor Relat. Rev. 59, 205–225. Safe Work Australia, 2012. Work-Related Traumatic Injury Fatalities, Australia 2010–11, Canberra. Salminen, S., 2008. Two interventions for the prevention of work-related road accidents. Saf. Sci. 46, 545–550. Vredenburgh, A.G., 2002. Organizational safety: which management practices are most effective in reducing employee injury rates? J. Saf. Res. 33, 259–276. Williamson, A., 2005. Fatigue and coping with driver distraction. J. Aust. College Road Saf., 19–21.

Williamson, A., 2007. Predictors of psychostimulant use by long-distance truck drivers. Am. J. Epidemiol. 166, 1320–1326. Williamson, A., Bohle, P., Quinlan, M., Kennedy, D., 2009. Short trips and long days: safety and health in short-haul trucking. Ind. Labor Relat. Rev. 62. Williamson, A., Friswell, R., 2013. The effect of external non-driving factors, payment type and waiting and queuing on fatigue in long distance trucking. Accid. Anal. Prev. 58, 26–34. Wills, A., Biggs, H., Watson, B., 2005. Analysis of a safety climate measure for occupational vehicle drivers and implications for safer workplaces. Aust. J. Rehabilit. Couns. 11, 8–21. World Health Organisation, 2013. Global Status Report on Road Safety 2013: Supporting a Decade of Action, Geneva. Wouters, P.I.J., Bos, J.M.J., 2000. Traffic accident reduction by monitoring driver behaviour with in-car data recorders. Accid. Anal. Prev. 32, 643–650. Zohar, D., 1980. Safety climate in industrial organizations: theoretical and applied implications. J. Appl. Psychol. 65, 96–102. Zohar, D., 2010. Thirty years of safety climate research: reflections and future directions. Accid. Anal. Prev. 42, 1517–1522.