**2. Methodology**

Firstly, based on the problem statement, a conscious decision was made to analyse secondary data related to accidents and serious incidents related to aircraft maintenance.

Secondly, an 'Interpretivist' philosophical approach formed the basis of this research.

" ... ... .. interpretative and constructionist research does not only focus on the content of empirical data, but also on how the content is produced through language practices. Furthermore, research done from these philosophical positions does not predefine dependent and independent variables, but focuses on the full complexity of human sense making as the situations emerge. It is also assumed that there are many possible interpretations of the same data, all of which are potentially meaningful." Eriksson and Kovalainen [10] (p. 21)

Thirdly, an inductive approach was used during the design of this study. According to Eriksson and Kovalainen [10] (p. 24], "the research process develops, starting from empirical materials, not from theoretical propositions".

Considering these chosen philosophical positions and methods in relation to research, the data was not analysed by using existing taxonomies such as ICAO ADREP, ECCAIRS, HFACS, etc., but coded by using thematic analysis. In order to apply rigour to the research process, primary data was also collected from subject matter experts (SME's) to receive feedback not only about the outcome of the analysis but also about the methods used and the taxonomy developed. The overall research design and key steps followed, can be seen in Figure 2 below.

**Figure 2.** Research design and key steps.

### *2.1. Secondary Data: Accident and Serious Incidents*

In order to discern the maintenance-related accidents and serious incidents, two sources were consulted: Aviation Safety Network's (ASN) Accident Database supported by Flight Safety Foundation; and SKYbrary's Accidents and Incidents database. Generating the data set for the analysis involved the review of all accidents and serious incidents to identify the aircraft maintenance-related events for CAT category aeroplanes occurred between 2003 and 2017. These events, once identified, were then compiled within a singular dataset, which can be provided on request.

The ASN database contains data on worldwide accidents and hijackings involving airliners (of 12 or more passengers), military transport aircraft and corporate jets since 1919. In order to refine this data to appropriately match the scope of the study, it was first filtered for the date range of relevance and then further filtered for accidents only relating to CAT aeroplanes. The remaining data was then reviewed to ensure that only maintenance-related accidents were contained within the dataset.

The second source, SKYbrary's Accidents and Incidents Database, was also refined for the time period of interest, and then filtered for airworthiness events related to maintenance. Where accidents were identified in both the ASN and SKYbrary databases, the relevant information was merged within the dataset.

The official investigation reports for these events were then sourced and consulted to ensure the validity of the data provided for each of the events within the dataset. Some events listed on the ASN database did not have traceable official investigation reports. The majority of these events were omitted from the dataset as it was not possible to assure their validity. However, a small selection of these events were allowed into the dataset, when there was significant indication within the narrative of maintenance-related contributory factors.

### *2.2. Primary Data Collection: Subject-Matter Experts (SME)*

In order to scrutinise the results of the data analysis, SME feedback was sought. The method for this data collection was an online questionnaire, which was delivered through Qualtrics survey software.

The questionnaire had four open-ended questions and was distributed to five participants, who are from International Federation of Airworthiness and had extensive experience in design, production, operation, and maintenance domains including regulatory oversight and safety data analysis. The topics covered by the questions were:


The questionnaire was accompanied by a PowerPoint presentation which detailed the aim and objectives of the study alongside the project methodology. The participants were also provided with an Excel spreadsheet which gave an overview of the accidents/serious incident data, as coded by the developed taxonomy at the time.
