Contribution of Risk Perception and Communication in Aviation Safety Events

Abstract

Although effective risk management during operations relies on risk perception and risk communication, the aviation industry has not systematically considered the contribution of these two constructs to safety events. This study analyzed a representative sample of safety investigation reports (1) to identify the degree to which risk perception and communication and their factors have been influential overall and across various flight operation stages of investigated events, and (2) to examine whether their contribution has changed with time. The analysis of 140 reports showed environmental factors affected risk communication and perception most frequently, whereas emotional and physiological factors were found in the sample with very low frequencies. Also, risk communication and perception and their factors did not appear with the same frequency across the various flight stages, and a few variations were observed over time. The aviation industry could consider the results of this study to steer its efforts toward mitigating the adverse effects of factors related to ineffective risk perception and communication. This could include the inclusion of respective factors in safety reporting schemes, investigation methods and analyses and, possibly, a tailored approach to the various flight stages and targeted risk literacy interventions.

Keywords

aviation security and emergency management aircraft safety emergency management pilot safety safety safety performance and analysis crash analysis crash data general hazard analysis

The consistently low rates of aviation incidents recorded in statistical studies ( 1 ) can be attributed to the ongoing attempts of all stakeholders to apply effective risk management, which constitutes a pillar of safety management ( 2 , 3 ). However, in addition to organizational methods and approaches usually applied with time allowances in the office environment, each human operator performs real-time risk management at local levels by understanding and processing constantly the unfolding reality, assessing and deciding, acting and—ultimately—determining the effectiveness of (in)actions by observing their results ( 4 ). Nevertheless, owing to continuous and cumulative changes in the operational environment, unexpected combinations and the nature of risks might render standardized recovery practices—those aimed at “normal” conditions that are not directly adjustable—ineffective ( 5 ).

Therefore, real-time risk management cannot be approached merely as a rationalized process of identification, evaluation, and prioritization ( 6 , 7 ). Risk perception is an essential aspect of risk management in the daily aviation experience ( 8 ). It represents the functional assessment of risks and the perceived outcomes from interactions with those risks ( 9 ). While performing a multitude of tasks, aviation personnel must consider the context in which they operate (e.g., environment, equipment, procedures, colleagues) and the effects of their responses ( 10 , 11 ).

Complementary to perception, risk communication entails disseminating perceived threats and relevant actions to control risks ( 12 ). Risk communication also targets erroneous assumptions and information differences across various mediums (i.e., oral communication, journals, and memos) ( 13 ). Enabling an early warning system allows shared understanding, as well as coordination of activities ( 14 – 17 ), and good communication influences the audience’s understanding of risk ( 18 ). Nonetheless, communication breakdowns in aviation continue to be a concern and contribute to incidents and accidents. A study that considered 151 investigation reports, mainly from commercial aviation, identified communication issues in 63% of the events, with the most frequent problems involving human–human interactions (i.e., mainly between pilots and between pilots and air traffic controllers), verbal and local communications as well as unfamiliarity of the receivers with the messages transmitted ( 19 ). In general aviation, the most common causes of communication breakdowns in air and ground operations include radio interference, mismanagement or misapplication of communication means, malfunction of communication equipment, workload, and language proficiency ( 20 ).

As risk perception and communication are inextricably linked within an effective risk management practice, the literature discusses various factors contributing to those two constructs. For instance, familiarity with a hazard primes the operators to be more tolerant of it, perceive the risk as low and take little action to communicate it ( 18 , 21 , 22 ). In addition, fatigue may impair mental decision-making abilities, particularly the attentional spectrum and memory capacity, impairing both risk perception and communication ( 23 – 25 ). Thus, familiarity and fatigue coupled with confirmation bias shape the perceived level of risk and the ability to control it ( 26 ).

Hansson and Aven advocated for integrative risk analysis to lay a firm basis for developing this field ( 27 ) and acknowledged the obstacles faced by contemporary social developments and technical reforms that introduce an increasingly challenging operating environment ( 28 ). However, while safety and accident investigation methods are intended to help detect, clarify, and mitigate hazards and structural deficiencies, risk perception has been mostly ignored owing to its inherent fragility caused by subjectivity and the impact of the latter on the risk management process ( 29 – 33 ).

Taxonomies widely used in aviation, such as the Human Factors and Accident Classification System (HFACS) ( 34 ) and the Systematic Human Error Reduction and Prediction Approach (SHERPA) ( 35 ) do not explicitly refer to risk communication and perception and their factors. Similarly, mandatory and voluntary reporting schemes lack fields dedicated to the specific aspects ( 36 , 37 ). Therefore, despite the acknowledged critical roles of risk communication and perception, currently, those constructs have not been subject to systematic consideration and analysis during investigations, safety reporting, or classification of causes of adverse events. Consequently, the aviation sector lacks information about the degree to which risk perception and communication and their factors have contributed to occurrences, (serious) incidents, and accidents. This, in turn, deprives the sector of the ability to indicate trends, identify weak areas, and plan for necessary interventions.

In light of the above-mentioned issues, this study analyzed safety investigation reports to (1) examine the frequency with which risk perception and communication factors are influential in aviation safety events, (2) detect differences across the various flight phases, and (3) explore variations over time. As a result, this research offers insights into how risk perception and communication historically have affected aviation operations resulting in unwanted events, and highlights possible avenues for interventions and further research through the collaboration of industry and academia.

Methodology

Study Sample

As this study focused on the individual’s capacity to perceive and communicate risks effectively regardless of operation type, we did not focus on a particular aviation sector. Instead, we considered military, general, and commercial aviation safety events in our sample. Under a nondeterministic approach suggested and adopted by several authors in the safety context ( 38 – 42 ), the term “safety event” used in this paper denotes any investigated case regardless of its outcome severity (e.g., incidents, serious incidents, or accidents). The authors acknowledge that military and commercial aviation staff may be well-conditioned to respond to adverse situations, owing to systematic training and education, continuous exposure to complex systems, and other factors. However, this does not exclude the conditioning of general aviation individuals based on their experiences relative to their operating context. Furthermore, cases in which commercial pilots also engage in general aviation activities (e.g., instructors, personal leisure) and the transition of military pilots to commercial aviation cannot be excluded.

The sample size was set to 140 investigation reports as this study is part of a larger project, and the researchers had only two months to record and analyze data. We analyzed cases from 2009 to 2018 to ensure the analysis results would sufficiently reflect the current situation. We consulted the aviation safety network (ASN) online platform (https://aviation-safety.net/) to retrieve investigation reports because of its public access. A systematic screening of a decade’s archive of individual repositories of different state authorities and agencies would need the investment of a considerable amount of time and resources. Nevertheless, ASN is consistent with the International Civil Aviation Organization (ICAO) Annex 13, is regularly updated with information and reports from all ICAO countries, and occasionally offers a direct link to English editions of reports not originally published in English.

When accessing the ASN platform, the lead author retrieved a list of 2,006 investigated safety events (i.e., [serious] incidents and accidents) in which the end-users were not incapacitated or fatally injured. As a means to verify the representativeness of our sample, we considered the number of reports found in ASN and issued by the U.S. National Safety Transportation Board (NTSB) (i.e., 1,094 reports). The lead author filed a request directly to the NTSB’s archive service to provide a list of events based on the criteria applied to ASN as explained above. The response from NTSB returned 1,093 reports, meaning an extremely low difference from the reports derived from ASN. Based on this result, we considered ASN to be an adequately representative source.

Next, the lead author and two interraters screened the reports to identify those including distinct references to the risk perception of involved individuals and/or their role as senders or receivers of risk communications during the events. This round of screening resulted in the inclusion of 504 reports. Krippendorff’s alpha tests were executed at the first round of screening with an acceptable agreement level of equal to or higher than 80% ( 43 – 45 ). In addition, the results showed high-intercoder reliability (Krippendorff’s alpha = .9711).

During the next step, the lead author and the two interraters screened reports, including information as per the NTSB’s taxonomy (Table 1), which classifies personnel actions during safety events ( 46 ). The researchers used the same taxonomy system for the reports published by ICAO countries other than the United States. We could not identify any other regional classification with distinctive codes for risk perception and communication. Owing to the high intercoder agreement in the previous step, in the second round, only the frequency of report inclusions was calculated (R1 = 99.6%, R2 = 99.8%, R3 = 98.41%). This step of the study resulted in eliminating 89 reports (n = 415).

Table 1.

Personnel Coding ( 46 )

	Categories	Description
Risk perception	Action/decision—information processing	Expectation/assumption
		Identification/recognition
		Understanding/comprehension
		Complacency
	Psychological—personality/attitude	Confidence/reliance on equipment
		Motivation/respond to pressure
		Personality
		Self-confidence
		Accuracy of communication
		Common phraseology
Risk communication	Personnel issues—task performance—communication	Crew/duty change-over
		CRM/MRM techniques
		Following instructions
		Interpretation/understanding
		Issuing instructions
		Lack of communication
		Language/accent
		Read back

Note: CRM = crew resource management; MRM = maintenance resource management.

During the final screening round, the sample was stratified per part of the flight when/where most of the findings were concentrated. This was performed by maintaining an equal number of reports per year (n = 14) in descending chronological order, eliminating 275 reports. We randomly chose 14 reports per year subject to not exceeding two reports referring to the same flight stage (i.e., flight planning, pushback, taxi, takeoff, cruise, descend, final approach, and landing). This specific criterion was adopted to explore whether or not risk perception and communication and their factors had been found more frequently in certain flight phases than in others. Figure 1 presents the screening process explained above. The final sample of 140 reports represented 22 countries and covered the 10-year period mentioned above.

Figure 1.

Safety reports screening process.

Collection and Analysis of Data

Following the works of Thomson et al. ( 47 ) and Chionis and Karanikas ( 33 ), this study focused on broad categories of factors for risk perception (emotional factors [P-EMF], cognitive factors [P-COF], physiological factors [P-PHF], and environmental factors [P-ENF]); and risk communication (emotional factors [C-EMF], cognitive factors [C-COF], physiological factors [C-PHF], environmental factors [C-ENF], timeliness-related factors [C-TIF], and channel-related factors [C-CHF]). The analysis of each event was based on the questions listed below.

If the event referred to risk perception issues:

To which part of the flight was risk perception linked?

Which category/categories of risk perception factors (P-EMF, P-COF, P-PHF, P-ENF) was/were addressed as contributing to the event?

If the event referred to risk communication problems:

To which part of the flight was risk communication linked?

Which category/categories of risk communication factors (C-EMF, C-COF, C-PHF, C-ENF, C-TIF, C-CHF) was/were addressed as contributing to the event?

We calculated the frequencies for each risk perception and communication constructs and their related factors to provide an overview of the attribution of safety events to risk perception and communication. Also, depending on the sample distribution, through chi-square and Fisher’s exact test, we examined any associations between risk perception and communication factors and flight stages. We used the chi-square test and, wherever more than 20% of cells had expected frequencies lower than 5, Fisher’s exact test ( 48 ). We hypothesized no statistically significant differences in the distribution of risk perception and communication factors during safety investigations across the various flight stages.

The tests above were also conducted to explore significant differences in the frequencies of risk perception and communication factors over time. In relation to the latter, we grouped the reports in three periods by maintaining balanced sample sizes (2009 to 2011: 42 reports, 2012 to 2014: 42 reports, and 2015 to 2018: 56 reports). Based on the premises that (1) risk perception and communication are of paramount importance for the whole aviation sector regardless of activity, which can be equally affected by the various factors, and (2) these constructs would have been increasingly attended to over time during staff training and education, our additional research hypothesis was that the frequencies of risk perception and communication factors that contribute to events are lower in the most recent period, 2015 to 2018.

The overall level of statistical significance for the statistical tests was set to 0.05 ( 49 ). To avoid the buildup of errors, we adjusted the level of significance for the two tests (factors versus flight stages and period) based on the Bonferroni correction: P_crit = α/k, where k is the number of comparisons ( 50 ), resulting in P_crit = 0.05/2 = 0.025 per test. We ran all tests using SPSS v. 25 ( 51 ) and selected the Bootstrap method with a 95% confidence interval and 1,000 samples to overcome the limitations of the sample distribution, as well as employing the Monte-Carlo simulation with a 99% confidence level and 10,000 samples to calculate the exact significances.

Results

The contributions of risk perception and communication factors to the investigated events are shown in Figures 2 and 3 respectively. Across the whole sample, environmental factors contributed the most to ineffective risk perception and communication (76.4% and 74.3%, respectively). The rest of the factors varied among average frequencies (38.6% to 65.7%), except for the physiological factors (1.4% and 2.9% for risk perception and risk communication respectively) and emotional factors (20% and 17.9% for risk perception and risk communication respectively). In risk communication, timeliness was a contributory factor in 40.7% of the reports and channel-related issues in 65.7% of the investigated events.

Figure 2.

Distribution of risk perception factors’ contribution as reported in the investigations.

Figure 3.

Distribution of risk communication factors’ contribution as reported in the investigations.

The categories of risk perception and risk communication factors found across the sample are presented in Table 2. Most of the investigation reports concerning risk perception identified one or two of the four categories of factors, with about 14% not referring to the influence of any factor. Conversely, one to four of the six risk communication factors were recorded with comparable frequencies. One-fifth of the investigation reports did not attribute the respective events to any risk communication factor.

Table 2.

Number of Cases per Number of Factors Found in Each Report, Regardless of Contribution to Events

Risk perception		Risk communication
Reported factor categories (P-EMF, P-COF, P-PHF, P-ENF)	Investigated events (N; %)	Reported factor categories (C-EMF, C-COF, C-PHF, C-ENF, C-TIF, C-CHF)	Investigated events (N; %)
0	19; 13.6	0	28; 20.0
1	42; 30.0	1	30; 21.4
2	64; 45.7	2	21; 15.0
3	13; 9.3	3	32; 23.0
4	2; 1.4	4	24; 17.1
		5	2; 1.4
		6	3; 2.1
Total	N = 140	Total	N = 140

Note: P-EMF=Perception Emotional Factors, P-COF=Perception Cognitive Factors, P-PHF=Perception Physiological Factors, P-ENF=Perception Environmental Factors, C-EMF=Communication Emotional Factors, C-COF=Communication Cognitive Factors, C-PHF=Communication Physiological Factors, C-CENF=Communication Environmental Factors, C-TIF= Communication Timeliness Factors, C-CHF=Communication Channel related Factors.

The contribution percentages for risk perception across the different flight phases varied between 10.71% and 61.42%, with flight planning, final approach, and landing presenting higher frequencies. The percentages for risk communication detected contributions ranging from 25.71% to 87.85% per flight phase, with cruise, final approach, and landing being the most frequently linked with this construct. Overall, the findings indicated that risk communication was more commonly found in the sample as a contributory factor than risk perception across all flight phases apart from flight planning (Table 3).

Table 3.

Contribution of Risk Perception and Risk Communication per Flight Stage

Flight operations stages	Code	Risk perception, N (%)	Risk communication, N (%)
Flight planning	FP	63 (45)	43 (30.71)
Push back	PB	15 (10.71)	36 (25.71)
Taxi	TA	26 (18.57)	56 (40)
Takeoff	TO	49 (35)	64 (45.71)
Cruise	CR	46 (32.85)	71 (50.71)
Descend	DE	32 (22.85)	46 (32.85)
Final approach	FA	53 (38.85)	78 (55.71)
Landing	LA	86 (61.42)	123 (87.85)

The tests on the distribution of the contributing risk perception and communication factors across the flight stages showed,

During the flight planning phase,

○ Cognitive factors (P-COF) of risk perception were more frequent: χ² (1, N = 140) = 10.847, p = .001.

○ In relation to risk communication, environmental (C-ENF) factors were found more frequently, whereas timeliness (C-TIF) and channel (C-CHF) factors were less frequently present: χ² (1, N = 140) = 8.900, p = .004; χ² (1, N = 140) = 9.083, p = .003; and χ² (1, N = 140) = 11.552, p = .001, respectively.

In the push back phase, only risk communication timeliness (C-TIF) was detected more frequently: χ² (1, N = 140) = 11.443, p = .001.

During the taxi phase, environmental (C-ENF), timeliness (C-TIF), and channel (C-CHF) factors relating to risk communication were more frequent: χ² (1, N = 140) = 8.077, p = .004; χ² (1, N = 140) = 11.363, p = .001; and χ² (1, N = 140) = 6.848, p = .012, respectively.

In the takeoff phase, cognitive (P-COF) and environmental (P-ENF) factors of risk perception were more frequent: χ² (1, N = 140) = 5.816, p = .018; and χ² (1, N = 140) = 7.329, p = .010, respectively.

In the cruise phase,

○ Cognitive (P-COF) factors of risk perception were more frequent: χ² (1, N = 140) = 6.493, p = .014.

○ Cognitive (C-COF) and timeliness (C-TIF) factors relating to risk communication were more frequent: χ² (1, N = 140) = 11.124, p = .001; and χ² (1, N = 140) = 6.837, p = .013.

In the landing phase, emotional (P-EMF, C-EMF) factors concerning both risk perception and risk communication were observed less frequently: χ² (1, N = 140) = 9.767, p = .002; and χ² (1, N = 140) = 8.306, p = .006.

The tests in relation to the frequencies across the three time periods (2009 to 2011, 2012 to 2014, and 2015 to 2018) revealed the following significant differences:

Cognitive (P-COF) factors related to risk perception were observed significantly less frequently in the latest period of 2015 to 2018: 16.4% (χ² [1, N = 140] = 9.892, p = .008).

Timeliness (C-TIF) and channel (C-CHF) factors of risk communication were found more frequently in the latest period 2015 to 2018 at 17.9% and 17.1% respectively: χ² (1, N = 140) = 7.699, p = .020; and χ² (1, N = 140) = 8.327, p = .017.

Discussion

The analysis of 140 safety investigation reports showed that risk communication factors contributed to safety events more frequently than risk perception factors across all flight phases apart from flight planning. It was outside the scope of this study to explain this difference and examine the possible increased vulnerability of risk communication compared with risk perception during aviation operations. However, we could not exclude the case that communication includes aspects that may be more accessible during investigations (e.g., data from voice recording equipment). In contrast, risk perception involves cognitive processes that are difficult to investigate and support with hard data.

Risk perception will influence what will be communicated, and, reciprocally, risk communication informs and influences risk perception ( 52 ). Ineffective or inadequate risk communication may lead to ambiguities in risk perception with reference to the clues to be considered by the involved personnel or examined during investigations ( 53 ). Furthermore, although the application of crew resource management (CRM) principles may facilitate an adequate flow of information by mandating explicit verbal communication, the question remains as to how many of the problems CRM solves are created from incomplete risk perception ( 54 ). Therefore, although the literature suggests these two constructs interrelate practically, risk communication issues might have been discovered more frequently had more immediate access to the respective information been possible.

Nevertheless, the results showed similar frequencies in the common categories of risk perception and communication factors (i.e., physical, cognitive, emotional, and environmental). More specifically, environmental factors were found to have the highest percentages (76.4% and 74.3% respectively for risk perception and communication), whereas the opposite was observed for emotional and physiological factors. The latter category recorded the lowest frequencies for both constructs at the levels of 1% to 3% respectively for risk perception and communication. This may be explained by operators in critical roles possessing medical certifications to work based on various clinical examinations (i.e., respiratory and cardiovascular systems) that are performed in multiple oversight cycles ( 55 ). This might also generate “fit-to-work” expectations that could lead investigators not to focus on physiological factors.

The high percentages of environmental factors, including the physical environment and organizational influences, can be attributed to the recordable and relatively visible physical agents affecting operations (e.g., weather conditions) and the emphasis on sociotechnical system complexities established determinants of several safety events, such the interactions among processes, culture, people, and infrastructure ( 56 , 57 ). However, the low percentages of emotional factors might indicate that residual psychological effects could be neglected (e.g., anxiety, depression, embarrassment) over tangible organizational and environmental inputs, along with difficulties in assessing their impact on risk perception and communication.

The high frequency of channel-related and temporal factors of risk communication in reports is concerning when considering the intensive addressing of communication means in aviation training, the CRM programs included ( 58 , 59 ). Risk literacy has been found to stabilize risk perception and communication and shape comprehension and utilization of the risk information environment ( 60 ). Therefore, although CRM programs aim to build competencies to understand and use risk information, among other objectives, this may not have been entirely achieved owing to an inadequate focus on risk communication capacity.

Additionally, the findings revealed that no categories of risk perception or risk communication factors were reported in 14% and 20% of the sample, respectively. Thus, the general attributions of events to risk perception and communication without further explanations about their underlying factors might reflect using constructs as legitimate direct causes of events without having examined the constituents of such constructs ( 61 ). On the positive side, the distribution of the frequencies with which one or more categories of factors were reported to have contributed to events (Table 2) indicates an awareness of risk perception and communication constructs within most of the investigation teams. Nonetheless, as it remains unknown whether the whole set of risk perception and communication factors were examined regardless of their contribution to each event, research on the degree to which investigators are aware of and consistently consider such factors is warranted.

Flight Phases

When examining the distribution of the two constructs across the various flight phases, the results suggest that flight planning, final approach, and landing were more frequently affected by issues related to risk perception. In addition, the cruise, final approach, and landing phases appeared more frequently influenced by risk communication problems. Therefore, although the cruise is the lengthiest flight stage, issues with both risk perception and communication were more frequently recorded in the phases of approach and landing. This is confirmed by literature suggesting that during those phases, human involvement prevails and workload varies, changes influence situational awareness in the operational status ( 62 ), and environmental factors affect an aircraft’s status during the final approach ( 63 ). In general, the final approach and landing have been tagged as the most complicated flight stages, where most accidents take place ( 64 – 66 ).

Moreover, the statistical tests suggested some differences in the prevalence of risk perception and communication factors. Cognitive factors mediate the end-user’s understanding of a mission and its conditions in the flight planning phase. At the same time, the risk-related information shared is usually standardized (i.e., briefing process, checklists, weather information). Also, flight planning takes place in a relatively controlled environment. It is a scheduled activity, a fact that might explain the statistically lower frequencies of timeliness- and channel-related factors of risk communication. Although this could also justify the lower frequencies of environmental factors of risk communication, it should be noted that environmental factors can negatively influence risk communication from complacency over routine precursors (e.g., expected turbulence) encountered in regular routes ( 67 ).

The takeoff phase is characterized by minimum communication requirements, and demands for increased attention on internal and external parameters to ensure the safe separation of the aircraft from the ground. This could explain the higher frequencies of cognitive and environmental factors when risk perception issues arise. With regard to safety events unfolding during the cruise phase, physical and psychological stressors may increase the workload during parallel activities when dealing with unexpected events, affecting both perception and communication ( 68 – 70 ). This could be intensified under complacency and unpreparedness as this phase is usually straightforward and typically requires minimum mental and physical effort from pilots. Multiple factors have been highlighted as leading to miscommunication in this phase, such as quality of audio signals, the accent of the pilot or controller, the English language proficiency of the operator, and failure to use standard phraseology ( 20 )—all linked to cognition. The above could also explain the more frequent appearance of timeliness factors related to communication for this particular flight phase.

Furthermore, the increased effects of timeliness factors in risk communication during the push back phase could be explained by the involvement of multiple operators and groups (e.g., ground crews, air traffic controllers, pilots) within the relatively short duration of this activity compared with the preceding and following flight phases. Communication during push back tasks might be more vulnerable to background noise, variance in operational language fluency, and visual signals, which can delay communication ( 71 ). Concurrent information and signals may increase mental workload and, consequently, decrease operators’ available window for action to ameliorate any hazardous situations ( 54 ).

The increased contribution of environmental, temporal, and channel factors relating to risk communication during the taxi phase can be similarly explained by the involvement of multiple operators and the execution of parallel tasks by the aircrews (e.g., pretakeoff checklists). Thus, communication during taxiing could be more vulnerable owing to external factors affecting the performance of concurrent activities and communications with other groups (e.g., air traffic controllers). Nonetheless, it seems cognition and emotions are not significantly affected. The latter could be attributed to the adherence to standard protocols during this phase, which possibly eases pilots’ workloads ( 72 ).

With reference to the landing phase, the lower involvement of emotional factors for both risk perception and communication could be attributed to the high environmental inputs that elevate the mental workload ( 70 ). Landing is highly demanding as pilots have to coordinate multiple tasks while keeping themselves on high alert for any spontaneous hazard cues that might arise ( 73 ).

Temporal Variations

Interestingly, the results of the statistical tests revealed that in the latest period, 2015 to 2018, cognitive (P-COF) factors related to risk perception were observed less frequently, whereas the opposite was true for timeliness- (C-TIF) and channel-related (C-CHF) factors of risk communication. The slightly increased contribution of C-TIF and C-CHF can be respectively linked to production pressures and saturation, or quality of communication media ( 74 , 75 ) despite technological advancements and end-user training on relevant communication aspects.

The cognitive factors concerned, the developments in human factors research and automation in the aviation industry could have provided enough barriers through the years, which should have been efficient to mitigate perception-related cognitive factors ( 76 ). It is claimed that improvements in human–machine interactions and automation have alleviated the mental workload on human operators, thus decreasing the influence of related cognitive factors ( 77 ), supporting information processing ( 78 ) and relieving operators from ordinary tasks ( 79 ). However, there are opposing views and equivocal findings on this topic. For example, contextual control loop automation has been suggested to counter the effects of automation surprise ( 80 ). At the same time, it has also been suggested as culpable for an operator’s cognitive restrictions and the disturbance of controlling capabilites ( 81 ). Moreover, inept human–automation interaction has been suggested as a precursor to increased task management for automation cross-checking, elevated workload, and potential errors ( 82 ).

Limitations

The current study is subject to various limitations. First, the safety investigation reports do not mention what the investigators checked in relation to risk communication and perception. Instead, the reports refer to the respective factors when they contributed to the events. This did not allow the researchers to examine the degree to which investigation teams consistently examine risk perception and communication, regardless of their contribution to the events. Second, although we employed random sampling in selecting the 140 reports out of the 415 identified as candidates for the analysis, we cannot exclude the possibility that the findings are not entirely representative of the whole aviation environment.

The limitations above are significant when considering that the various worker groups (e.g., technicians and other ground staff) are not always involved in the investigated aviation events, either as safety investigators or involved personnel or as persons of reference. In addition, there are variations in job demands and specialized training among the general aviation, military, and commercial aviation contexts. For example, CRM training programs expectedly focus with different weights on aspects deemed as more critical per type of operation. In general, sector-specific and organizational cultures and practices influence the degree to which risk perception and communication are considered in safety management initiatives and viewed as critical elements of operations. Furthermore, as the concept of the “average” flight crew does not apply in reality, even within each operation type, the findings of this study must be only seen as reflective of the overall picture in the aviation sector and not conclusive for each commercial, general, and military aviation domain separately.

The third limitation relates to the depth and quality of accident investigation processes and how the derivative reports can vary within and across aviation authorities and investigative teams and across time periods. For example, where accidents are typically investigated by diverse and highly experienced teams of experts, less severe events could be subject to more cursory investigations by small teams with relatively less experience, or even by a single investigator. Nonetheless, including reports from 22 countries and encompassing 10 years offers confidence that the results are sufficiently representative, especially when considering the relative differences revealed for risk perception and communication factors. We also note that although slightly more than half of the initial sample of reports regarded investigations conducted by NTSB, only about a third of the final dataset analyzed included NTSB reports. Subsequently, there was no relative overrepresentation of the NTSB’s region of responsibility.

In relation to the analysis process, we cannot exclude the effects of biases entirely. Nonetheless, the lead author has a specialty in aviation psychology and experience as a safety investigator. In addition, the interraters selected had specific expertise in aviation safety, and analysis of investigation reports from two distinct backgrounds: law and psychology. Finally, the fourth author has extensive industry and technical expertise in aviation safety, and the third author has an academic interest in aviation as well as extensive industry knowledge.

Conclusions and Recommendations

This study examined the factors related to risk perception and communication in aviation by analyzing 140 safety investigation reports. Although statistical reports suggest the last phases of flights are more critical and susceptible to the effects of several risk-related factors ( 75 , 83 , 84 ), the analysis of the investigation reports indicated that other flight phases might be increasingly affected by risk perception and communication factors. The risk perception concerned, cognitive factors were found to be more influential on flight planning, takeoff, and cruise phases; environmental factors were found to contribute more frequently during the takeoff phase. The risk communication concerned, environmental factors were more frequently found to be involved in the flight planning and taxiing phases; temporal factors were found to be more influential on push back, taxiing, and cruising phases. Channel-related factors were more influential during the taxiing phase.

Additionally, the reports revealed that cognitive and environmental factors have contributed to events less over time. Conversely, timeliness- and channel-related communication factors were found more often in recent investigations. Although the former finding indicates successful system interventions over time, the latter suggests a possible underestimation or underresearch of the respective factors. Therefore, although aviation organizations are expected to follow a rigid risk management framework focusing on numerical/parametric data ( 8 ), a pragmatic approach to real-time risk management at the operator level could be more effective than simulated forecasts ( 85 ).

Prospective analyses could focus on experimental studies on the evolution of attitudes toward risk perception and communication variables in commercial, general, and military aviation, perhaps separately. The relationships between personality traits and risk perception and communication in aviation are also areas that need further investigation. Moreover, it would be useful to research the influence of the various consecutive or parallel activities involved during different flight phases on risk perception and communication. Furthermore, communication training for aviation personnel at the operator level could enhance risk literacy when such training is fused effectively with current programs such as CRM.

Finally, an inclusive framework for evaluating risk perception and communication factors during investigations could support the aviation sector’s efforts to understand safety events better and improve future safety performance. This could be achieved by supplementing accident investigation methods, analysis taxonomies, and safety reporting tools with fields dedicated to risk perception and communication, possibly starting with the principal factors considered in this study and later enriching them depending on the degree of detail required. Also, as the analysis of large qualitative datasets demands considerable time and resources, machine learning (ML) and natural language processing (NLP) could be considered. Although ML may produce ineffective predictions ( 86 ), it has been used for classifying aviation safety reports ( 87 ), and NLP has been applied to content analysis of safety reports in the aviation and construction sectors ( 88 – 90 ).

Footnotes

Author Contributions

The authors confirm contribution to the paper as follows: study conception and design: D. Chionis; data collection: D. Chionis; analysis and interpretation of results: D. Chionis, N. Karanikas; draft manuscript preparation: D. Chionis, N. Karanikas, A.-R. Iordan, A. Svensson-Dianellou. All authors reviewed the results and approved the final version of the manuscript.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Dimitrios Chionis

Nektarios Karanikas

References

Airbus. A Statistical Analysis of Commercial Aviation Accidents 1958/2017. Airbus, 2018.

ICAO. Safety Management Manual. Montreal: International Civil Aviation Organization, 2018.

Boyle

Health and Safety: Risk Management, 5th ed. Routledge, London, 2019.

National Research Council. Aviation Safety and Pilot Control: Understanding and Preventing Unfavorable Pilot-Vehicle Interactions. The National Academies Press, Washington, D.C., 1997.

Fiksel

Resilient by Design: Creating Businesses That Adapt and Flourish in a Changing World. Island Press, Washington, D.C., 2015.

Hubbard

D. W.

The Failure of Risk Management: Why It’s Broken and How to Fix It. John Wiley & Sons, Inc., Hoboken, NJ, 2009.

ISO. ISO 31000:2018 Risk Management – Guidelines. ISO Copyright Office, Geneva, Switzerland, 2018.

Müller

M. H.

Risk Perception and Risk Management. OITAF Congress 2017, Bolzano, 2017.

Aven

Risk Assessment and Risk Management: Review of Recent Advances on Their Foundation. European Journal of Operational Research, Vol. 253, 2016, pp. 1–13.

10.

Stanton

N. A.

Harvey

Allison

C. K.

Systems Theoretic Accident Model and Process (STAMP) Applied to a Royal Navy Hawk Jet Missile Simulation Exercise. Safety Science, Vol. 113, 2019, pp. 461–471.

11.

Taisne

Perttu

Tailpied

Caudron

Simonini

Atmospheric Controls on Ground- and Space-Based Remote Detection of Volcanic Ash Injection into the Atmosphere, and Link to Early Warning Systems for Aviation Hazard Mitigation. In Infrasound Monitoring for Atmospheric Studies ( Le Pichon

Blanc

Hauchecorne

, eds.), Springer, Switzerland, 2019, pp. 1079–1105.

12.

Friedensen

Risk Communication: Communicating Risks as a Function of Good Risk Management. Proc., 4th International Energy Conversion Engineering Conference and Exhibit (IECEC), San Diego, CA, 2006.

13.

Morgan

M. G.

Fischhoff

Bostrom

Atman

C. J.

Risk Communication: A Mental Models Approach. Cambridge University Press, Cambridge, 2002.

14.

Fiske

Introduction to Communication Studies. Routledge, New York, 1990.

15.

Macrae

Analysing Near-Miss Events: Risk Management in Incident Reporting and Investigation Systems. The London School of Economics and Political Science, London, 2007.

16.

Yanmaz

Yahyanejad

Rinner

Hellwagner

Bettstetter

Drone Networks: Communications, Coordination, and Sensing, Ad Hoc Networks, Vol. 68, 2018, pp. 1–15.

17.

Downey

D. H.

Bedard

J. C.

Coordination and Communication Challenges in Global Group Audits. Auditing: A Journal of Practice & Theory, Vol. 38, No. 1, 2019, pp. 123–147.

18.

Walaski

P. F.

Risk and Crisis Communications. John Wiley & Sons, Inc., Hoboken, NJ, 2011.

19.

Karanikas

Passenier

The AVAC-COM Communication Model and Taxonomy: Results from Application to Aviation Safety Events. Proc., International Cross-Industry Safety Conference, Amsterdam, 2019.

20.

Molesworth

B. R.

Estival

Miscommunication in General Aviation: The Influence of External Factors on Communication Errors. Safety Science, Vol. 73, 2015, pp. 73–79.

21.

Song

Schwarz

If It’s Difficult to Pronounce, It Must Be Risky: Fluency, Familiarity, and Risk Perception. Psychological Science, Vol. 20, No. 2, 2009, pp. 135–135.

22.

Orr

T. J.

Bellanca

J. L.

Eiter

B. M.

Helfrich

Rubinstein

E. N.

The Effect of Hazard Clustering and Risk Perception on Hazard Recognition. In Advances in Human Factors in Simulation and Modeling ( Cassenti

D. N.

, ed.), Vol 780, Springer, Cham, 2019.

23.

Shorrock

S. T.

Errors of Perception in Air Traffic Control. Safety Science, Vol. 45, 2007, pp. 890–904.

24.

Smith

A. P.

A UK Survey of Driving Behaviour, Fatigue, Risk Taking and Road Traffic Accidents. BMJ Open, Vol. 6, No. 8, 2016, p. e011461.

25.

Fan

Smith

A. P.

The Impact of Workload and Fatigue on Performance. In Human Mental Workload: Models and Applications ( Longo

Leva

, eds.), Communications in Computer and Information Science, Vol 726, Springer, Cham, 2017, pp. 90–105.

26.

Nickerson

R. S.

Confirmation Bias: A Ubiquitous Phenomenon in Many Guises. Review of General Psychology, Vol. 2, No. 2, pp. 175–220, 1998.

27.

Hansson

S. O.

Aven

Is Risk Analysis Scientific?

Risk Analysis, Vol. 34, No. 7, 2014, pp. 1173–1183.

28.

Waterson

Jenkins

D. P.

Salmon

P. M.

Underwood

‘Remixing Rasmussen’: The Evolution of Accimaps within Systemic Accident Analysis. Applied Ergonomics, Vol. 59, 2017, pp. 483–503.

29.

Redmill

Subjectivity in Risk Analysis. Redmill Consultancy, 2001.

30.

Koradecka

Pośniak

Widerszal-Bazyl

Augustyńska

Radkiewicz

A Comparative Study of Objective and Subjective Assessment of Occupational Risk. International Journal of Occupational Safety and Ergonomics, Vol. 16, No. 1, 2010, pp. 3–22.

31.

Vasvári

Risk, Risk Perception, Risk Management – A Review of the Literature. Public Finance Quarterly, Vol. 1, 2015, pp. 29–48.

32.

Paek

H. -J.

Hove

Risk Perceptions and Risk Characteristics. 2017. https://oxfordre.com/communication/view/10.1093/acrefore/9780190228613.001.0001/acrefore-9780190228613-e-283?print=pdf. Accessed June 20, 2019.

33.

Chionis

Karanikas

Safety Methods in Aviation: To What Extent Do They Address the Effects of Risk Perception and Risk Communication?, Proc., 33rd EAAP Conference Connecting People, Organisations and Technology in Aviation, Dubrovnik, 2018.

34.

Wiegmann

D. A.

Shappell

S. A.

The Human Factors Analysis and Classification System (HFACS). Report No. DOT/FAA/AM-00/7. Federal Aviation Administration, 2000.

35.

Stanton

N. A.

Systematic Human Error Reduction and Prediction Approach (SHERPA). In Handbook of Human Factors and Ergonomics Methods ( Stanton

N. A.

Hedge

Brookhuis

Salas

Hendrick

H. W.

, eds.), CRC Press, Boca Raton, FL, 2004, Pp. 37-1–37-8.

36.

Speyer

J. -J.

Example Application of Aircrew Incident Reporting System (AIRS). AIRBUS, France, 2004.

37.

SKYbrary. EUROCONTROL Voluntary ATM Incident Reporting System (EVAIR), SKYbrary. 2021. https://www.skybrary.aero/index.php/EUROCONTROL_Voluntary_ATM_Incident_Reporting_System_(EVAIR). Accessed August 6, 2021.

38.

Hubbard

Evans

Problems with Scoring Methods and Ordinal Scales in Risk Assessment. IBM Journal of Research and Development, Vol. 54, No. 3, 2010, pp. 2:1–2:10.

39.

Byrne

Creating a Just Culture. The RoSPA Occupational Safety & Health Journal, Vol. 42, 2012, p. 27–29.

40.

Duijm

N. J.

Recommendations on the Use and Design of Risk Matrices. Safety Science, Vol. 76, 2015, pp. 21–31.

41.

Dekker

Just Culture: Restoring Trust and Accountability in Your Organisation, 3rd ed. CRC Press, Boca Raton, FL, 2016.

42.

Karanikas

Nederend

The Controllability Classification of Safety Events and Its Application to Aviation Investigation Reports. Safety Science, Vol. 108, 2018, pp. 89–103.

43.

Krippendorff

Content Analysis: An Introduction to its Methodology. SAGE, Thousand Oaks, CA, 2004.

44.

Hayes

A. F.

Krippendorff

Answering the Call for a Standard Reliability Measure for Coding Data. Communication Methods and Measures, Vol. 1, No. 1, 2007, pp. 77–89.

45.

Gwet

K. L.

Handbook of Inter-Rater Reliability, 4th ed. Advanced Analytics LLC, Gaithersburg, 2014.

46.

US NTSB. Aviation Coding Manual. USA National Transportation Safety Board, Washington, D.C., 1998.

47.

Thomson

M. E.

Onkal

Avcioglu

Goodwin

Aviation Risk Perception: A Comparison Between Experts and Novices. Risk Analysis, Vol. 24, No. 6, 2004, pp. 1585–1595.

48.

Kim

H. Y.

Statistical Notes for Clinical Researchers: Chi-Squared Test and Fisher’s Exact Test. Restorative Dentistry & Endodontics, Vol. 42, No. 2, 2017, pp. 152–155.

49.

Berman

Wang

Essential Statistics for Public Managers and Policy Analysts, 4th ed. SAGE, Los Angeles, CA, 2017.

50.

Armstrong

R. A.

When to Use the Bonferroni Correction. Ophthalmic & Physiological Optics, Vol. 34, No. 5, 2014, pp. 502–508.

51.

IBM Corp. IBM SPSS Statistics for Windows, Version 25.0. IBM Corp, Armonk, NY, 2017.

52.

Schmälzle

Renner

Schupp

H. T.

Health Risk Perception and Risk Communication. Policy Insights from the Behavioural and Brain Sciences, Vol. 4, No. 2, 2017, pp. 163–169.

53.

Kammerbauer

Minnery

Risk Communication and Risk Perception: Lessons from the 2011 Floods in Brisbane, Australia. Disasters, Vol. 43, No. 1, 2019, pp. 110–134.

54.

Hutchins

The Cognitive Consequences of Patterns of Information Flow. Intellectica, Vol. 1, No. 30, 2000, pp. 53–74.

55.

EASA. Easy Access Rules for Medical Requirements. European Aviation Safety Agency, Köln, 2019.

56.

Leveson

N. G.

A New Accident Model for Engineering Safer Systems. Safety Science, Vol. 42, No. 4, 2004, pp. 237–270.

57.

Nixon

Braithwaite

G. R.

What Do Aircraft Accident Investigators Do and What Makes Them Good at it? Developing a Competency Framework for Investigators Using Grounded Theory. Safety Science, Vol. 103, 2018, pp. 153–161.

58.

Jimenez

Kasper

Rivera

Talone

Jentsch

Crew Resource Management (CRM): What Aviation Can Learn From the Application of CRM in Other Domains. Proceedings of the Human Factors and Ergonomics Society 59th Annual Meeting, Vol. 59, 2015, pp. 946–950. https://doi.org/10.1177/1541931215591274.

59.

Chidester

T. R.

Anca

Crew Resource Management, Risk and Safety Management Systems. In Crew Resource Management ( Kanki

B. G.

Anca

Chidester

T. R.

, eds.), Elsevier Inc., 2019, pp. 227–247. https://doi.org/10.1016/B978-0-12-812995-1.00008-7.

60.

Kusumi

Hirayama

Kashima

Risk Perception and Risk Talk: The Case of the Fukushima Daiichi Nuclear Radiation Risk. Risk Analysis, Vol. 37, No. 12, 2017, pp. 2305–2320.

61.

Karanikas

Chionis

Plioutsias

“Old” and “New” Safety Thinking: Perspectives of Aviation Safety Investigators. Safety Science, Vol. 125, 2020, p. 104632.

62.

Lin

C. J.

Lin

P.-H.

Chen

H.-J.

Hsieh

M.-C.

H.-C.

Wang

E. M.-Y.

H. L. C.

Effects of Controller-Pilot Communication Medium, Flight Phase and the Role in the Cockpit on Pilots’ Workload and Situation Awareness. Safety Science, Vol. 50, 2012, pp. 1722–1731.

63.

Xie

Sun

Jiao

Association Rules Mining with QAR Data: An Analysis on Unstable Approaches. Proc., IEEE 1st International Conference on Civil Aviation Safety and Information Technology (ICCASIT), Kunming, China, IEEE, New York, 2019, pp. 603–606.

64.

Hinkelbein

Schwalbe

Neuhaus

Wetsch

W. A.

Genzwürker

H. V.

Incidents, Accidents and Fatalities in 40 Years of German Helicopter Emergency Medical System Operations. European Journal of Anaesthesiology, Vol. 28, No. 11, 2011, pp. 766–773.

65.

UK CAA. CAP1036: Global Fatal Accident Review 2002 to 2011. UK Civil Aviation Authority, Sussex, 2013.

66.

IATA. Loss of Control in-Flight Accident Analysis Report, 2010–2014. International Air Transport Association, Montreal, 2015.

67.

IATA. Environmental Factors Affecting Loss of Control In-Flight: Best Practice for Threat Recognition & Management. International Air Transport Association, Montreal, 2016.

68.

Wickens

C. D.

Gordon

S. E.

Liu

Stress and Workload. In An Introduction to Human Factors Engineering ( Wickens

C. D.

, ed.), Longman, New York, 2004, pp. 324–350.

69.

Novak

Mrazova

The Effect of Physiological Stressors on Pilot’S Decision Making During Unfavourable Simulated Conditions: An Explorative Study. Incas Bulletin, Vol. 7, No. 2, 2015, pp. 153–162.

70.

Saputra

A. D.

Priyanto

Muthohar

Pilot Metal Workload in Flight Operation: A Case Study of Indonesian Civilian Pilot. Aceh International Journal of Science and Technology, Vol. 6, No. 1, 2017, pp. 37–43.

71.

SKYbrary. Pushback. 2019. https://www.skybrary.aero/index.php/Pushback#Accidents_and_Incidents. Accessed January 14, 2020.

72.

FAA. Standard Operating Procedures and Pilot Monitoring Duties for Flight Deck Crewmembers. US Federal Aviation Administration, 2017.

73.

FAA. Chapter 8: Approaches and Landings. In Airplane Flying Handbook, US Federal Aviation Administration, 2016, pp. 8_1–8_36.

74.

Ligda

S. V.

Fischer

Mosier

Matessa

Battiste

Johnson

W. W.

Effectiveness of Advanced Collaboration Tools on Crew Communication in Reduced Crew Operations. In Engineering Psychology and Cognitive Ergonomics. EPCE 2015. Lecture Notes in Computer Science ( Don

Wen-Chin

eds.), Springer, Cham, 2015, pp. 416–427.

75.

Kelly

Efthymiou

An Analysis of Human Factors in Fifty Controlled Flight into Terrain Aviation Accidents from 2007 to 2017. Journal of Safety Research, Vol. 69, 2019, pp. 155–165.

76.

Mouloua

Hancock

P. A.

Human Performance in Automated and Autonomous Systems. CRC Press, Boca Raton, FL, 2019.

77.

Gawron

Automation in Aviation-Guidelines. The MITRE Corporation, McLean, 2019.

78.

Kaber

D. B.

Perry

C. M.

Segall

McClernon

C. K.

Prinzel

L. J.

Situation Awareness Implications of Adaptive Automation for Information Processing in an Air Traffic Control-Related Task. International Journal of Industrial Ergonomics, Vol. 36, No. 5, 2006, pp. 447–462.

79.

Naranji

Sarkani

Mazzuchi

Reducing Human/Pilot Errors in Aviation Using Augmented Cognition and Automation Systems in Aircraft Cockpit. AIS Transactions on Human-Computer Interactions, Vol. 7, No. 2, 2015, pp. 71–96.

80.

De Boer

Dekker

Models of Automation Surprise: Results of a Field Survey in Aviation. Safety, Vol. 3, No. 3, 2017, p. 20.

81.

Oliver

Calvard

Potočnik

Cognition, Technology, and Organisational Limits: Lessons from the Air France 447 Disaster. Organization Science, Vol. 28, No. 4, 2017, pp. 729–743.

82.

Mosier

K. L.

Fischer

Morrow

Feigh

K. M.

Durso

F. T.

Sullivan

Pop

Automation, Task, and Context Features: Impacts on Pilots’ Judgments of Human–Automation Interaction. Journal of Cognitive Engineering and Decision Making, Vol. 7, No. 4, 2013, pp. 377–399.

83.

BOEING. Statistical Summary of Commercial Jet Airplane Accidents. 2019. http://www.boeing.com/resources/boeingdotcom/company/about_bca/pdf/statsum.pdf. Accessed January 5, 2020.

84.

EASA. Annual Safety Review 2019. European Union Aviation Safety Agency, 2019.

85.

Zhong

Liu

Lin

Luo

The Method of Oilfield Development Risk Forecasting and Early Warning Using Revised Bayesian Network. Mathematical Problems in Engineering, 2016, pp. 1–10.

86.

Faria

J. M.

Machine Learning Safety: An Overview. 2018. https://www.researchgate.net/profile/Jose-Faria-3/publication/320567319_Machine_Learning_Safety_An_Overview/links/59f37efda6fdcc075ec34986/Machine-Learning-Safety-An-Overview.pdf. Accessed August 4, 2021.

87.

de Vries

. Classification of Aviation Safety Reports Using Machine Learning. Proc., International Conference on Artificial Intelligence and Data Analytics for Air Transportation (AIDA-AT), Singapore, 2020.

88.

Tangui

Tulechki

Urieli

Hermann

Raynal

Natural Language Processing for Aviation Safety Reports: From Classification to Interactive Analysis. Computers in Industry, Vol. 78, 2016, pp. 80–95.

89.

Pimm

Raynal

Tulechki

Hermann

Caudy

Tangui

Natural Language Processing (NLP) Tools for the Analysis of Incident and Accident Reports. Proc., Conference on Human-Computer Interaction in Aerospace (HCI-Aero), Brussels, 2012.

90.

Tixier

A. J.-P.

Hallowel

M. R.

Rajagopalan

Bowman

Automated Content Analysis for Construction Safety: A Natural Language Processing System to Extract Precursors and Outcomes from Unstructured Injury Reports. Automation in Construction, Vol. 62, No. 7, 2016, pp. 45–56.