The Role of Organizational and Individual Variables in Aircraft Maintenance Performance

Aviation maintenance has been identified by the FAA as an area where better efficiency is needed to cope with ever increasing workloads. However, aviation maintenance has also been identified as one of the major causes of accidents. Consequently, if further efficiencies are to be achieved, they cannot come at the cost of reduced safety margins. The present study employed a safety climate approach to assist in the development of a model that can help to explain morale, psychological health, turnover intentions, and error in the aviation maintenance environment. An instrument called the Maintenance Environment Survey was developed and administered to 240 personnel responsible for maintenance of a large military helicopter fleet. Data collected through the survey were used to develop a structural model that predicted 45% of the variance in psychological health, 67% of the variance in morale, 27% of the variance in turnover intentions, and 44% of the variance in self-reported maintenance errors. The model shows the pathways through which organizational level and individual level variables can influence work outcomes and leads to suggestions for interventions that can help to improve maintenance efficiency. Explaining Aircraft Maintenance Performance 3 The role of Organizational and Individual Variables in Aircraft Maintenance Performance The importance of the maintenance function was captured by Weick and colleagues when they observed that: “Maintenance people come into contact with the largest number of failures, at earlier stages of development, and have an ongoing sense of the vulnerabilities in the technology, sloppiness in the operations, gaps in the procedures, and sequences by which one error triggers another” (Weick, Sutcliffe, & Obstfeld, 1999, p. 93). A significant proportion of these errors come at the hands of the maintainers themselves as the ever-increasing complexity of aviation places greater demands on those responsible for their maintenance. Figures emerging from the United Kingdom Civil Aviation Authority (CAA) show a steady rise in the number of maintenance error mandatory occurrence reports over the period 1990 to 2000 (Courteney, 2001). A recent Boeing study of worldwide commercial jet aircraft accidents over that same period shows a significant increase in the rate of accidents where maintenance and inspection were primary factors (cited in ICAO, 2003). The FAA, in its strategic plan for human factors in aviation maintenance through to 2003, cited statistics from the Air Transport Association of America (ATA) showing that the number of passenger miles flown by the largest US airlines increased 187% from 1983 through to 1995. Over that same period, the number of aircraft operated by those airlines increased 70% but the number of aviation maintenance technicians increased only 27%. The FAA concluded that the only way the maintenance program could cope with the increased workload was by increased efficiency at the worker level (cited in McKenna, 2002). Despite the awareness of the importance of maintenance to the aviation industry and the growing problems confronting maintenance, until recently, empirical research into the nature of maintenance work and related human factors has been negligible. The development of descriptive models of human error and accident causation (Reason, 1990; Senders & Moray, 1991) and the recent adaptation of Reason’s model to aviation maintenance (Reason & Hobbs, 2003) are major steps in the right direction. Research on error classification schemes (e.g., Patankar, 2002; Shappell & Weigmann, 1997) and, more recently, safety culture (Taylor & Thomas, 2003; Patankar, 2003) represent other bright spots in a surprisingly sparse research literature. However, what are needed in addition to the descriptive accident causation models, classification schemes, and culture surveys are empirically validated models that capture the major influences on maintenance work and provide a means of assessing these influences. Models of this kind can provide the basis for predicting unsafe organizational states and designing interventions that will lead to reductions in maintenance errors. The present study set out to develop such a model within the context of aviation maintenance using a multivariate methodology that has its roots in what has become known as the safety climate approach. This approach is described in the following paragraphs. Safety Climate and Safety Culture Over the years, the concepts of safety culture and safety climate have developed almost in parallel through the safety literature. Safety climate is operationalised in the current study as the individual’s perceptions of the organizational policies, procedures, and rewards relevant to safety in the organization (Guldenmund, 2000; Griffin & Neal, 2000). This definition sets it apart from safety culture, which is usually regarded as a stable, deep-seated aspect of an organization Explaining Aircraft Maintenance Performance 4 that expresses itself through climate (Guldenmund, 2000, p.221). Whereas the assessment of safety culture requires tangible means of measurement such as in-depth interviews and analysis of stated safety goals and polices (Guldenmund, 2000; Mearns & Flin, 1999), safety climate is assessed through self-report questionnaires. Constructing a Measure of Safety Climate Attempts have been made to define a core set of constructs for safety climate (see Flin, Mearns, O’Connor, & Bryden, 2000). Although not entirely successful in establishing core dimensions, this research is useful in suggesting constructs that should be considered for inclusion in research on maintenance errors. Recent publications relating to the assessment of safety climate in aviation maintenance also provide guidance. Taylor and Thomas (2003), for example, used a self-report questionnaire called the Maintenance Resource Management/Technical Operations Questionnaire (MRM/TOQ) to measure what they regarded as two fundamental parameters in aviation maintenance: professionalism and trust. The dimension of professionalism is defined in their questionnaire in terms of reactions to work stressors and personal assertiveness. Trust is defined in terms of relations with coworkers and supervisors. Questions relating to these areas also appear in the questionnaire to be used in the current research. Patankar (2003) constructed a questionnaire called the Organizational Safety Culture Questionnaire which included questions from the MRM/TOQ along with items from questionnaires developed outside the maintenance environment. Following the application of exploratory factor analytic routines to a dataset generated from respondents that included 124 maintenance engineers, Patankar identified four factors as having particular relevance to the safety goals of aviation organizations: emphasis on compliance with standard operating procedures, collective commitment to safety, individual sense of responsibility toward safety, and a high level of employee-management trust. Turning to the general safety literature, there are now a host of questionnaires that purport to measure either safety culture or safety climate. Wiegmann and his colleagues (Wiegmann, von Thaden, Mitchell, Sharma, & Zhang, 2003) drew upon 13 such measures to construct their Commercial Aviation Safety Survey (CASS), an instrument designed for use with pilots. Most of these questionnaires are multidimensional, covering a range of factors that the authors consider to be of relevance to safety performance. The availability of so many questionnaires tapping an array of safety-related constructs presents a challenge to researchers interested in constructing a safety climate survey for use in specific settings such as maintenance. That challenge was addressed in the present study by using the principle of triangulation to isolate the constructs relevant to a maintenance environment. Drawing upon the distinction between culture and climate made earlier, this methodology entailed a close examination of the safety culture in an organization in order to derive questions for inclusion in a safety climate survey. The first step in the triangulation process involved a search of the safety literature to identify potential constructs for inclusion in the questionnaire. As already mentioned, there is no shortage of surveys in the literature and some researchers have attempted to identify core safety climate constructs (e.g., Flin et al., 2000). The second step involved the analysis of a maintenance incident database and the associated incident investigation reports. The database and incident reports highlighted the relevance of factors such as inadequate training, poor supervision, and individual factors such as stress and fatigue as causes of maintenance-related Explaining Aircraft Maintenance Performance 5 incidents. The third method involved a series of focus group interviews with maintenance personnel and their supervisors to ascertain their perceptions of factors that impact on maintenance work. Content analyses of these interviews highlighted organizational concerns such as scheduling and resources. Information collected in these three phases was then used as the basis for the construction of a questionnaire to measure organizational and individual factors considered likely to impact on maintenance performance. The resulting questionnaire, called the Maintenance Environment Survey (MES), was broader in scope than many of the existing climate or culture surveys. It contained questions intended to define the following constructs: a) safety climate, b) morale, c) psychological health, d) job turnover intentions, and e) maintenance errors. The construction and validation of the MES was a necessary first step towards the development and validation of a structural model showing how the various factors captured by the survey interact to influence maintenance errors. Despite the proliferation of studies reporting new safety climate questionnaires, there are few studies in the safety literature that have taken the extra step of constructing models to illustrate the interactions among the psychological factors captured by the questionnaires. Using climate surveys in combination with the techniques of multivariate analysis, especially path analysis and structural equation modeling, it is possible to capture elements of the accident causation process and to test different models of how the components of the system work. These models can then be used to direct interventions aimed at improving safety performance in the maintenance environment. The rationale for the model to be tested in the present study is set out in the following paragraphs. Developing a Model to Predict Maintenance Errors Regarding the relations between safety climate and maintenance errors, there is now a substantial body of empirical evidence from the general safety literature to support the contention that measures of climate are related to safety outcomes. This relationship has been demonstrated in cross-sectional surveys where scores on safety climate scales have been linked with accidents (Donald & Canter, 1994; Zohar, 1980), in longitudinal studies (Neal & Griffin, 2002), in intervention studies (Donald & Young, 1996), in individual as well as group-level studies (Hofmann & Stetzer, 1996; Zohar, 2000), and across a very wide range of industrial settings. These settings include hospitals (Neal, Griffin, & Hart, 2000), the offshore oil industry (Mearns, Flin, Gordon, & Fleming, 2001), the power industry (Donald & Young, 1996), and chemical processing plants (Hofmann & Stetzer, 1996). Most of these studies used regression and bivariate correlations to demonstrate the existence of a relationship between safety climate and safety performance. However, a small group of studies have used path analysis or structural equation modelling (SEM) to explain the observed relationships (e.g., Hofmann & Morgeson, 1999, Neal et al., 2000, Tomás, Melia, & Oliver, 1999; Oliver, Cheyne, Tomas, & Cox, 2002). Together, the two groups of studies provided the basis for a hypothesized SEM model that was expected to capture variance in selfreported maintenance errors. The model is shown in Figure 1. Explaining Aircraft Maintenance Performance 6