The aviation system consists of various interactive and interdependent components including manufacturing and engineering, the operation of aircraft, infrastructure, personnel licencing and organisational certification that enables organisations to provide a service to society (The Department of Transport 2017). This industry enables increased accessibility to fast and efficient transportation of goods and people on a personal and industrial level (Njoya 2013). The Department of Transport (2017) noted the symbiotic relationship between the aviation industry and society.

Continual growth in the international air transport system requires developing and implementing policies and practices to increase mutually beneficial value for all stakeholders (Njoya 2016). Consistent with this view, Chialastri and Pozzi (2008) argued that optimal performance and sustainable credibility amongst stakeholders (both internally and externally) originate from coherence between principled organisational management and the lower-level outcome-based functions.

The aviation industry remains a diverse environment where various components play a vital role in the overarching realisation of an efficient and reliable transportation solution (Lazur et al. 2014). Resilience in civil aviation maintains sustainable credibility amongst all stakeholders and emanates from unremitting coherence between an organisational and procedural structure focussing on error-avoidance within the extensive aviation system (Chialastri & Pozzi 2008).

Organisational resilience, as a desired capacity of an organisation’s sustainability, supports the discourse to dynamically reinvent business models and organisational strategies in response to circumstantial changes before such adaptation becomes critical (Hamel & Välikangas 2003) and thus facilitates fluid adaptation to the high velocity of changes in the industry (Patriarca, Di Gravio & Constantino 2016).

The South African aviation industry provides scheduled and non-scheduled air transportation through the utilisation of all the components of the aviation system (eNCA 2019; Mabotja et al. 2019; The Department of Transport 2017). Despite the significant size of the South African aviation industry, commercial challenges such as high fixed costs, low-profit margins and significant impact from volatility and fluctuations in the external or macro-environment are omnipresent (Mhlanga 2018). Many organisations in the industry fail to operate sustainably because of these challenges and a misconstrued realisation of the stringent operational requirements (Wynbrandt 2017). Sustainability in a complex socio-technical system, such as the aviation industry, requires rapid adaptation to external fluctuations that may stimulate an imbalance between the demand of the organisation’s service and the organisation’s operating cost (Alexandra, Limnios & Schilizzi 2014; Mhlanga 2018; Mhlanga & Steyn 2016).

Organisational resilience and competitive advantage in the marketplace are directly related to each other, and the relationship between organisational resilience and competitive advantage is embedded in an organisation’s ability to ‘maintain above-average returns even after absorbing the shocks of the competitive environment’ (Teixeira & Werther 2013:335). To achieve this, the configuration of intellectual, financial, technological, human and other resources needs to be renewed and developed continuously (Teixeira & Werther 2013). Although organisational resilience is desirable, resistance to change is omnipresent in organisations (Edgeman 2015).

The characteristics of a resilient organisation include the ability to manage internal and external complexities and changes through holistic thinking, acknowledgement of multiple perspectives and reframing the complexities and changes as moving targets (Teixeira & Werther 2013). Leem, Vargo and Seville (2013) maintain that organisational resilience is a multidimensional approach to manage uncertainty facing organisations during business activities. Many possibilities exist to respond to uncertainties. Still, the centralisation of organisational controls (such as quality assurance and corporate governance) to manage such uncertainty yields increased organisational knowledge, creative problem solving and sustainable adaptation to better future responses to similar uncertainties (Lee et al. 2013).

Although various definitions of corporate governance have been coined in the past three decades, its essence remains ‘the system by which companies are directed and controlled’ to ensure all-inclusive sustainability and value (Marx & Mohammadali-Haji 2014:234). The International Organization of Standardization (2015:vi) describes a quality management system as a multi-level organisational and strategic system ‘to improve its overall performance and a sound basis for sustainable development’. Various theorists (Burnard, Bhamra & Tsinopoulos 2018; Lee et al. 2013) argue a strong theoretical resemblance between socio-technical resilience of an operational system and congruent application of corporate governance and quality assurance.

Dutta, Lanvin and Wunsch-Vincent (2017) propose that internal organisational controls associated with corporate governance and quality assurance increase the viability of innovation through strategic direction and leadership, knowledge and resource management, process control, conformance and customer satisfaction. The advantages of integrating internal controls within an organisation include profound value for stakeholders across geographic boundaries (Marx & Mohammadali-Haji 2014). Because of the complexity of the aviation system, the necessity for an inclusive and stringent governance framework that holistically addresses an organisation’s technical performance and commercial integrity through improved accountability remains omnipresent in the aviation industry specifically (Tjørhom 2010).

Consistent with this view, Chialastri and Pozzi (2008) argued that coherence between principled organisational management and the lower-level outcome-based functions stimulates optimal performance and sustainable credibility. Despite the focus on holistic coherence between principled organisational management and the outcome-based functions within an organisation, various researchers (Amankwah-Amoah 2018; Mhlanga & Steyn 2016; Njoya 2013) identified detachment between outcome-based procedural supervision (or quality assurance) and the principled organisational management (or corporate governance) in the aviation industry.

The risks that emerge from detachment between quality assurance and corporate governance include reduced stakeholder reputation, reduced competitiveness and the inability to adapt to the rapidly changing macro-environment (Amankwah-Amoah 2018). This inability to rapidly adjust internal controls in a sustainable and reliable manner to enable rapid and autonomous metamorphosis, in response to the changing environment, points to a decline in an organisation’s resilience capacity (Lazur et al. 2014; Lee et al. 2013; Patriarca et al. 2016; Teixeira & Werther 2013).

Previous research suggests the need to further interrogate the relationship between management control processes that address commercial management and technical outcomes (Lazur et al. 2014; Njoya 2016). Furthermore, Alexandra et al. (2014) argued that a lack of knowledge on the connection between resilience and organisational control systems contributes to the inability to maintain agility.

This study aimed to explore the relationship between organisational control processes, such as quality assurance and corporate governance, and organisational resilience in the South African aviation industry through stakeholders’ perception.

The qualitative phase of the larger sequential mixed-method study by Serfontein and Govender (2020) introduced a four-pronged framework of resilience properties embedded in organisational control frameworks such as quality assurance and corporate governance in the South African aviation industry. These components are Strategic management and company culture, Monitoring and awareness, Exposure management and Responsive adaptation as depicted in Figure 1.

Through the application of quantitative data analysis, a variable, or individual identifiable attribute of a subject, can be examined and explained (Agresti & Finlay 2009; Creswell 2014). Organisational resilience and its respective properties, namely Strategic management and company culture, Monitoring and awareness, Exposure management and Responsive adaptation, thus manifest as variables to organisational control systems (quality assurance and corporate governance).

Variables are structured according to temporal order and observatory measurement (Creswell 2014) and can display a consequential relationship to another variable (Kothari 2004). In such a cause-and-effect relationship, the variable that stimulated the consequential change in the other variable is denoted as the dependent variable (DV) (Beckett et al. 2014). The independent variable (IDV) measures a behavioural variation in response to the DV (Saunders, Lewis & Thornhill 2009). Organisational resilience in a complex sociotechnical system, and its respective indicators were identified as the IDV(s).

As such, organisational control systems are denoted as the IDV and organisational resilience as the DV. The DV consisted of four subcategories, namely Strategic management and company culture (DV1), Monitoring and awareness (DV2), Exposure management (DV3) and Responsive adaptation (DV4).

Yin (2018) asserts that studies concerned with a business environment should consider contextual elements such as the industry at large, policy and regulatory conditions to set the contextual boundaries. This study was embedded in the South African aviation industry within a contemporary timeframe and considered the Department of Transport’s (2017) policy to categorise stakeholders to the study’s setting.

The shared characteristic of the target population was based on The Department of Transport’s (2017) definition and categorisation of stakeholders to the South African civil aviation system, as indicated in Table 1.

Considering the accessibility of possible participants from the population (Martínez-Mesa et al. 2014), the target population consisted of an estimated 11 697 stakeholders to the South African aviation industry. As the Department of Transport (2017) classified this homogenous group of stakeholders into the subcategories defined in Table 1, the homogenous group of stakeholders to the South African civil aviation system presents a stratified population with different strata of shared characteristics (Creswell 2014).

Because Alvi (2016) recommends that a quantitative study generally needs 100–150 respondents to reach data saturation indicative of the population’s perception, the target sample of 200 participants was selected from the target population, by using non-probability convenience sampling (Rahi 2017).

To maintain the applicability of the survey in relation to the phenomenon of interest, the questionnaire design and content mirrored the embedded organisational resilience properties that the authors (Serfontein & Govender 2020) identified in organisational control systems in the South African aviation industry through prior qualitative research.

The questionnaire embodied a Likert-type rating scale to realise numeric values associated with the level of agreement to each survey item (Greener 2008). The scale of agreement ranged from 1 = strongly disagree to 5 = strongly agree.

A preliminary pilot study is advisable to validate the instrument in support of criterion validation and benchmarking (Brough 2009; Kothari 2004). The pilot study sample should be representative of the study sample (Saunders et al. 2009), and a sample size of 12 participants is appropriate (Junyong 2017). Twenty-two purposively selected participants from the study sample were approached to participate in the pilot study, and 18 individuals participated. Following the completion of the survey, informal engagement with each participant took place to identify and correct weaknesses in the survey. The survey was condensed, and various criteria reworded to avoid ambiguity.

The layout of the survey used for data collection purposes consisted of eight sections. The first section included ethical considerations and obtained informed consent from the participants. The second section included general instructions on the completion of the questionnaire. The third section included a screening question not only to ensure that participants represent the study sample, but also to classify the participant into the stratified sample described in Table 1. Thereafter, a section on organisational controls (IDV) in the South African aviation industry followed, and a section for each of the organisational resilience categories (DV1, DV2, DV3 and DV4) was included. The survey was designed on an online platform that could only be accessed through a unique identifier code available to purposively selected participants. The design of the survey included electronic controls to ensure that informed consent is obtained before the remainder of the survey could be accessed. An electronic control was also activated to ensure that each question is answered before a participant could proceed to a subsequent section.

Because the respondents were purposively selected from the study population, participants were knowledgeable on the phenomenon of interest and committed to the voluntary completion of the survey. For increased accessibility, this study used the online platform SurveyMonkey to host the survey questionnaire between 28 January and 11 February 2020. After 14 days, the survey’s unique identifier provided to voluntary research participants was deactivated to cease data collection.

Data were extracted from the online platform, retained in an electronic database and reworked to remove redundant information by isolating participant responses to the survey questions. This sanitised version of the database was provided to a statistical consultant for analysis. Data collected from 203 respondents were analysed by using Microsoft Excel 2016 and STATA 15 (software for statistical and data science). This data set exceeded the intended sample size.

The measurement and analysis of each item in a survey questionnaire are not advisable if there is not a specific methodological rationale for the detailed and cumbersome approach (Boone & Boone 2012). Because the survey used for this study grouped items associated to each variable together, it is advisable to rather measure the scale score for each variable (or category) in the survey (Boone & Boone 2012; Willits & Theodori 2016). The practice to report categorically on the scale score for the different variables, instead of each item constituting such categories, is known as composite scoring (Joshi et al. 2015).

The Cronbach alpha was applied to determine internal consistency through value measurement (Saunders et al. 2009). In exploratory research, a Cronbach alpha above 0.6 indicates satisfactory internal consistency (Beckett et al. 2017). The factor analysis method was employed to indicate the validity of the questionnaire’s constructs (Bolarinwa 2016). With specific reference to a sample size of above 200, as is the case with this study, a factor loading of 0.4 is an acceptable indication of validity (Taherdoost 2016).

Data analysis followed a two-dimensional approach to address descriptive and inferential statistics. Whilst descriptive statistics allow numeric description and comparison of quantitative data (Saunders et al. 2009), the inclusion of inferential statistics in the analytical approach permits identifying correlations to predict relationships between variables (Hahn 2015).

The introductory page of the survey questionnaire included pertinent information about anonymity and voluntary participation. Participants could only proceed to the subsequent sections of the survey once informed consent was acknowledged (Yin 2018). In addition, ethical clearance was obtained from the Da Vinci Institute of Technology Management as part of the doctoral research.

Data were collected from 203 participants representing the different strata of the homogenous study population defined in Table 1. The frequency of the data collected between the different strata of the study population is reflected in Table 2.

This study followed a composite approach to report aggregated descriptors for each variable in the relationship structure. Table 3 indicates the descriptive outcomes for the IDV organisational control systems (IDV) as a portmanteau for the nature of the aviation industry (IDV1) and quality assurance and corporate governance (IDV2).

The factor loading results for the IDV yielded an acceptable level of construct validity (Beckett et al. 2017) and indicate that the measurement tool employed for this study examined the construct as intended for a sample size exceeding 200 participants (Taherdoost 2016). As a Cronbach alpha above 0.6 in exploratory research indicates satisfactorily reliability (Beckett et al. 2017), the survey addressing the IDV is an accurate and reliable reflection of the phenomenon of interest within its context.

The valid and reliable research instrument indicates a scale score of 8.55 out of 10 for the statements that constituted the IDV. The scale score indicates the degree of agreeability of the participants to the questionnaire statements (Kothari 2004). Therefore, research participants indicated an 85.50% agreement level to the statements presented to them during quantitative data collection.

Table 4 reflects the descriptive outcomes for the components of the composite DV organisational resilience (DV). This variable consists of Strategic management and company culture (DV1), Monitoring and awareness (DV2), Exposure management (DV3) and Responsive adaptation (DV4).

The descriptive results for the DV yielded an acceptable construct validity as indicated by the respective factor loading results (Beckett et al. 2017) and suggest that this section of the data collection instrument is an accurate measurement of the subject matter within its setting (Taherdoost 2016). The Cronbach alpha measurement indicates admissible reliability of the data collection instrument (Beckett et al. 2017).

Therefore, the survey content associated with the IDV is an accurate and reliable reflection of the topic analysed. The valid and reliable data collection tool yielded a scale score of 17 out of a maximum of 20 for the IDV and indicates an 86.55% of respondent agreement with the statements included in this section of the survey.

The above-mentioned results form the basis on which the relationships between the variables were analysed through inferential statistics. The inferential statistics methods used are correlation and regression analysis.

As a prerequisite to correlation and regression analysis, statistical assumptions such as normality, multicollinearity, heteroskedasticity and autocorrelation must be tested (Zaid 2015). To determine the stability of the data set, the Shapiro-Wilk W test for normal data was performed. As the p-value for this test returned a result of 0.23244 below the 0.5 threshold, the data set presents a normal distribution (Agesti & Findlay 2009).

This study employed the variance inflation factor (VIF) to assess collinearity, and the VIF results were 3.55 (DV1), 5.07 (DV2), 3.85 (DV3) and 4.80 (DV4), respectively. As none of the variables indicated a VIF above 10, the possible adverse effects of significant multicollinearity can be eliminated (Agresti & Findlay 2009), and the different effects of individual variables can be determined from the data set (Saunders et al. 2009).

Heteroskedasticity, or the ‘extent to which data values for the dependent and independent variable have unequal variance’ (Saunders et al. 2009:592), can be diagnosed by the White test (Coenders & Saez 2000). Where the White test indicated a p-value above the 0.05 threshold, the variance of the residuals is homogenous (Agresti & Findaly 2009). The p-value indicated a measurement of 0.3669, and no further diagnostic tests to correct heteroskedasticity were required.

Reliability in regression analysis results can also be impacted by autocorrelation, which refers to the degree of similarity between the different values of variables (Saunders et al. 2009). The Durbin-Watson tests to detect autocorrelation yield a result between zero and four, where a value of two indicates no autocorrelation, a value of zero indicates a positive result and towards four a negative autocorrelation. (Saunders et al. 2009). As results between 1.5 and 2.5 do not require further investigation, the Durbin-Watson statistic of 1.951636 indicates the acceptability of the data set (Agresti & Findlay 2017).

The above-mentioned analytical outcomes indicate the appropriateness of the data set for correlation and regression analysis (Zaid 2015). The Pearson correlation method was used to examine the relationship between the variables. This method yields a coefficient (r-value) indicating the strength of the linear relationship between the IDV and DV (Boone & Boone 2012). The level of certainty through which deductions on these relationships can be made is indicated by the statistical significance, or the p-value, of the data set (Beckett et al. 2017).

To indicate the intensity of the relationship between variables, Cohen (1988) postulates that a correlation coefficient (r-value) between 0.1 and 0.3 indicates small correlations, a value between 0.3 and 0.49 implies a moderate correlation and a value of 0.5 or above indicates a strong correlation. Agresti and Finlay (2009) emphasise that the r-value (coefficient) and the p-value (significance level) must be compared to determine if a statistically significant relationship exists. Where the data return a correlation coefficient (r) > significance level (p-value), a statistically significant relationship exists between the variables (Agresti & Finlay 2009).

The relationship between the IDV of organisational control systems (IDV) and the DV of organisational resilience in a socio-technical system (DV) was tested. Also, the relationship between organisational control systems (IDV) and each of the four composite categories of the DV, namely organisational resilience (DV1, DV2, DV3 and DV4), were subjected to the same examination. Calculations from Pearson’s correlation method revealed the results reflected in Table 5.

The results indicate a significance (p-value) below 0.001 for each of the combinations of variables subjected to examination. As this indicates the statistical significance of the deductions made (Beckett et al. 2017), this study carries a probability in excess of 99% to make an accurate deduction. Where the significance of the relationship between the IDV and compounded DV was tested, the results indicate a >99% certainty that a statistically significant relationship between organisational control systems (IDV) and organisational resilience (DV) exists.

Through the same criteria, the data also indicate a >99% certainty that a statistically significant relationship exists between the IDV (organisational control systems) and each of the individual DVs, namely Strategic management and company culture (DV1), Monitoring and awareness (DV2), Exposure management (DV3) and Responsive adaptation (DV4). This not only indicates that a statistically significant relationship was identified between organisational control systems (quality assurance and corporate governance) and organisational resilience in the South African aviation industry, but it also indicates a relationship between organisational control systems (quality assurance and corporate governance) and each of the components of organisational resilience.

Following the determination that a statistically significant relationship exists between the IDV and each of the DVs, the collective data set was further examined to determine the nature of these relationships by conducting regression analysis (Zaid 2015). This method indicates a straight regression line as an indicator of the behavioural trend observed in the interrelationship of two variables (Brough 2019). The regression line can intuitively predict future behaviour of the DV through adjustment of the IDV’s value (Lane 2017) and is of value in this study to further understand the nature of the relationships identified.

The linear relationship between a combination of a DV and IDVs can be positive or negative (Joshi et al. 2015). A negative relationship implies a decline in the behavioural measurement of the DV in response to an increase in the behavioural indicator value of the IDV (Harpe 2015). The rate of change is predicted by the regression coefficient (r) that presents a gradient indication to simulate predicted behaviour of the DV (Bonett & Wright 2015).

Table 6 indicates the regression coefficient results for the relationship between the IDV and each of the DVs.

Through the application of analytical rules, the data revealed a positive relationship between the IDV and DV in all the variations examined (Harpe 2015; Joshi et al. 2015; Lane 2017). Therefore, the results shown in Table 6 indicate a definitive expected increase in the behavioural measurement value of organisational resilience (DV) when the measurement for organisational control systems (IDV) strengthens.

Similarly, an increase in the behavioural measurement of organisational control systems is expected to stimulate an elevation of the measurement values for each of the components of organisational resilience (Strategic management and company culture – DV1, Monitoring and awareness – DV2, Exposure management – DV3 and Responsive adaptation – DV4). The degree of the expected change is informed by the value of the coefficient (Bonett & Wright 2015), and the highest anticipated increase was identified in the relationship between the IDV and the DV of Strategic management and company culture (DV1).

The predictive value of a regression model is dependent on its confidence and significance levels (Beckett et al. 2017). As a measurement of central tendency, the standard error informs the confidence level of the predictive power embedded in a regression model (Beckett et al. 2017). The confidence level (t-value) indicates the degree of certainty that a regression relationship exists (Zaid 2015). To predict the probable range of a variable’s measurement in a regression model, the confidence interval informs the upper and lower limits of the expected range (Zaid 2015). Table 7 shows confidence and significance indications for the data set.

The t-value of < 0.001 indicates >99% confidence that the regression model has predictive power (Beckett et al. 2017). The deduction can be made that the application of this regression model will forecast the behaviour of the causal relationships examined with accuracy. The confidence interval’s upper and lower limits indicate, to a degree of confidence, the range in which the predicted regressed value of the DV will present itself (Zaid 2015). The standard error value for all regression combinations does not exceed 1. As such, it indicates a marginal expected deviation from the predicted data observations and increases the significance of the regression model (Beckett et al. 2017).