The link between Attitude, Risk Perception, Experience

and Behaviour in Australian General Aviation

Justin L. Drinkwater

B Aviation (Hons)

A thesis to satisfy the requirements of the degree of Doctorate in Philosophy, at the University of New South Wales

May, 2014

i PLEASE TYPE THE UNIVERSITY OF NEW SOUTH WALES Thesis/Dissertation Sheet

Surname or Family name: Drinkwater

First name: Justin Other name/s: Lee

Abbreviation for degree as given in the University calendar: A VIA 1900

School: Science Faculty: Aviation

Title: The link between Attitude, Risk Perception, Experience and Behaviour in Australian General Aviation

Abstract 350 words maximum: (PLEASE TYPE) In Australia, the General Aviation (GA) sector exhibits a much poorer safety record than the airline industry; it is responsible for 93% of fatal accidents, whilst representing only 55% of the total flight hours. Leading factors said to account for the high fatality rate in GA include human error and poor decision-making; as a result, a concerted effort has been made to improve pilot' decision-making in GA. However, these efforts have not clearly addressed the role of attitudes in effective and safe behaviour (Thomas, 2004). In a similar vein, poor (or inferior) risk perception has been identified as a possible 'hole' in the defence of accidents in the literature (Reason, 1990). Therefore, the main aim of present study was to investigate the link between pilot attitude, risk perception, flight experience and risky flight behaviour. In order to do this, three experiments were undertaken. The first experiment involved the use of a questionnaire designed to investigate the link between pilots' attitude, risk perception and experiential data. The second experiment was designed to examine self-reported risk-taking behaviour of pilots. It therefore involved the use of a questionnaire to gather self-reports of intended behaviour. The third experiment was designed to investigate flight behaviour in a simulated high-risk situation and augment this behavioural data with an interview to gain further insight into the causal factors of pilots' risk management behaviour. The results revealed that in Australian GA, older pilots, those with more flight hours, or those that fly regularly did not exhibit superior perception of the risks in a given situation than their younger or less experienced counterparts. They did however exhibit lower levels of self-confidence (attitude). This finding differs from aviation research undertaken in the United States (Hunter, 2005), where risk perception and attitude were both related to experience. In terms of behaviour, risk perception was found to be the only predictor of effective risk management behaviour. Therefore, holding a 'good' attitude, or having a higher level of experience appeared to have no influence on the decisions by pilots as to their risk management behaviour. It is important to highlight that the sample in this research is different than that in the US research; the results obtained in this research are based upon a relatively small cohort of pilots that are limited in terms' of their experience, age and vocation. Specifically, no airline pilots were included in this study, only General Aviation and trainee pilots were utilised. The generalisability of these results within the wider aviation industry and other domains remains untested.

Declaration relating to disposition of project thesis/dissertation

I hereby grant to the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or in part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all property rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.

I also authorise UniversitY, Microfilms to use the 350 word abstract of my thesis in i s rtation Abstracts International (this is applicable to doctoral

...97. /.f?.~./.l.~...... 'h' ~w.~~;; ·. . li a& Date The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years must be made in writing. Requests for a longer period of restriction may be considered in exceptional circumstances and require the a roval of the Dean of Graduate Research.

FOR OFFICE USE ONLY Date of completion of requirements for Award:

THIS SHEET IS TO BE GLUED TO THE INSIDE FRONT COVER OF THE THESIS

ii COPYRIGHT STATEMENT

'I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation. I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstract International (this is applicable to doctoral theses only). I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of ::~::siso;;ition 'al/c.

Date . ~l/11 . /ltr ......

AUTHENTICITY STATEMENT

'I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations ·n formatting, they are the result of the :~~:::sian t";l.~~: ...... ······························

Date .... 7.:/u .. ./.(tt ...... ORIGINALITY STATEMENT

'I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or :i:::· Pl!i"7[J!f'ession . isackno~ledged.' oate 7/U/!Lf: ...... Abstract

In Australia, the General Aviation (GA) sector exhibits a much poorer safety record than the airline industry; it is responsible for 93% of fatal accidents, whilst representing only 55% of the total flight hours. Leading factors said to account for the high fatality rate in GA include human error and poor decision‐making; as a result, a concerted effort has been made to improve pilot’ decision‐making in GA. However, these efforts have not clearly addressed the role of attitudes in effective and safe behaviour (Thomas, 2004). In a similar vein, poor (or inferior) risk perception has been identified as a possible ‘hole’ in the defence of accidents in the literature (Reason, 1990). Therefore, the main aim of present study was to investigate the link between pilot attitude, risk perception, flight experience and risky flight behaviour. In order to do this, three experiments were undertaken. The first experiment involved the use of a questionnaire designed to investigate the link between pilots’ attitude, risk perception and experiential data. The second experiment was designed to examine self‐reported risk‐taking behaviour of pilots. It therefore involved the use of a questionnaire to gather self‐reports of intended behaviour. The third experiment was designed to investigate flight behaviour in a simulated high‐risk situation and augment this behavioural data with an interview to gain further insight into the causal factors of pilots’ risk management behaviour. The results revealed that in Australian GA, older pilots, those with more flight hours, or those that fly regularly did not exhibit superior perception of the risks in a given situation than their younger or less experienced counterparts. They did however exhibit lower levels of self‐confidence (attitude). This finding differs from aviation research undertaken in the United States (Hunter, 2005), where risk perception and attitude were both related to experience. In terms of behaviour, risk perception was found to be the only predictor of effective risk management behaviour. Therefore, holding a ‘good’ attitude, or having a higher level of experience appeared to have no influence on the decisions by pilots as to their risk management behaviour. It is important to highlight that the sample in this research is different than that in the US research; the results obtained in this research are based upon a relatively small cohort of pilots that are limited in terms of their experience, age and vocation. Specifically, no airline pilots were included in this study, only General Aviation and trainee pilots were utilised. The generalisability of these results within the wider aviation industry and other domains remains untested.

iv Acknowledgements

First, I must express my sincerest thanks to Dr. Brett Molesworth, who persisted with me, guided me and pushed me during this thesis. Brett was a perfect model of dedication and professionalism in his work, and an inspiration during this project. Credit goes to Brett for the excellent flight simulator that the department now has, which was a joy to use. Thanks go to both Dr. Steven Shorrock and Dr. Boyd Falconer who were both excellent supervisors in the early stages of this work. Steve’s ability to provoke thought and to provide an alternate point of view was invaluable. Boyd’s encouragement, passion and enthusiasm were similarly indispensible. To ‘the boys’ in the research office, in no particular order, Dr. Tae Ryang Koo, Dr. Phillipe Estrade, Dr. Ryan McCabe and Dr. Alan Griffiths ‐ a big thank‐you for the great environment in which to work. I’m not sure that I will ever be blessed to be in such a welcoming and fun (and caffeinated) workplace again. I’d like to thank Professor Jason Middleton and Professor Ann Williamson for their input into the formative stages of the work, and valuable feedback on the thesis more recently. Thanks also to the administrative staff at the aviation department, without whom the office would not run (or be half as nice to work in). I’d also like to thank the staff at the Bankstown campus of UNSW for accommodating research within their busy schedules. Special thanks go to Captain Brian Horton for being so enthusiastic and welcoming of research in general. Of course, I must thank all the pilots that took part in this research ‐ your time is greatly appreciated. Finally, my biggest and best thanks go to my four girls, my young bloke, and the rest of the Drinkwater clan. I couldn’t have done this without the total support you’ve given me. Claire – I really have no idea why you put up with this, or how you did it for that matter, but I thank you from the bottom of my heart.

v Table of Contents

ABSTRACT­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ IV ACKNOWLEDGEMENTS ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ V LIST OF TABLES­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ IX LIST OF FIGURES ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ X 1 INTR ODU CTIO N ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 1 1.1 THE AUSTRALI AN AVIAT ION INDUSTRY ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐3 1.2 LINKS BETWEEN PSYCHOLOGICAL, EXPERIENTIAL, DEMOGRAPHIC FACTORS AND BEHAVIOUR ‐‐‐‐‐6 1.3 AIMS OF THE PRESENT RESEARCH‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐9 1.4 SUMMARY ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 10 2 RISK AND RKIS PERCEPTION ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­11 2.1 INTROD UCTION ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 11 2.2 RISK ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 11 2.3 RISK PERCEPTION ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 19 2.3.1 Hazard perception. ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­21 2.3.2 The theories arising from societal risk perception. ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­23 2.3.2.1 The psychometric model.‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐27 2.3.2.2 The cultural theory of risk perception. ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐30 2.3.2.3 Other theories of risk perception.‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐35 2.4 CONCLUSION ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 39 3 ATTI TUD ES­­­­­­­­­­­ ­­­­­­­­­­­­­­­­­­­­ ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­41 3.1 INTRODUCTION ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 41 3.2 ATTITUDE: DEFINITIONS AND COMPONENTS ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 41 3.2.1 Classic definitions and components of attitudes. ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­42 3.2.2 Recent theories of the nature of attitude/s.­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 44 3.2.2.1 Single attitude theories.‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐45 3.2.2.2 Dual attitude model. ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐46 3.2.2.3 Constructionist models.‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐48 3.2.2.4 Other models. ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐50 3.2.3 Function of attitudes. ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­54 3.3 ATTITUDE CHANGE ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 57 3.3.1 Dual process models of attitude change. ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 58 3.3.1.1 Systematic or central route to persuasion.‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐59 3.3.1.2 Heuristic or peripheral route to persuasion. ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐61 3.3.2 The unimodel of persuasion. ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­62 3.4 CONCLUSION ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 63 4 THE RELATIONSHIP BETWEEN ATTITUDE, RISK PERCEPTION AND BEHAVIOUR ­­­­­­­­­­­64 4.1 INTRODUCT ION ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 64 4.2 AH ISTORY OF THE RESEARCH ON THE ATTITUDE – BEHAVIOUR RELATIONSHIP ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 65 4.3 FACTORS THAT INFLUENCE THE ATTITUDE‐BEHAVIOUR RELATIONSHIP ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 68 4.3.1 Attitude aggregation and specificity ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 68 4.3.2 Attitude accessibility. ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­70 4.3.3 Attitude stability and contextual factors ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 72 4.4 ATTITUDE‐BEHAVIO UR MODELS ‐‐‐‐‐‐‐‐‐‐‐‐ ‐‐‐‐‐‐‐‐ ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 74 4.5 CRITIQUES OF AND ADDITIONS TO THE TPB ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 80 4.6 SUMMARY OF THE ATTITUDE BEHAVIOUR MODELS ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 83 4.7 RISK PERCEPTION AND BEHAVIOUR ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 83 4.7.1 Summary of risk perception ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­91 4.8 SUMMARY ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 91 vi 5 METHODOLOGICAL OVERVIEW­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­93 5.1 INTRODUCTION ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 93 5.2 BACKGROUND TO THE METHODOLOGY ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐94 5.3 SUMMARY ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐100 6 EXPE RIM ENT ONE ­ RISK PERCEPTIONS AND ATTITUDES OF PILOTS ­­­­­­­­­­­­­­­­­­­­­­­­ 101 6.1 INTRODUCTION ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐101 6.2 METHOD ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐105 6.2.1 Measures­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 105 6.2.2 Participants. ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 110 6.2.3 Procedure. ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 112 6.2.4 Data analysis.­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 113 6.3 RESULTS‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐114 6.4 DISCUSSION ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐119 6.5 CONCLUSION ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐125 7 EXPERIMENT TWO – SELF­REPORTED BEHAVIOUR OF PILOTS IN HIGH­RISK SITUATIONS 127 7.1 INTRODUCTION ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐127 7.2 BACKGROUND ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐129 7.3 EXPERIMENTAL DESIGN ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐131 7.3.1 Experiment Two A – Pilot Study to Determine Flight Scenarios. ­­­­­­­­­­­­­­­­­­­­­­­ 133 7.3.2 Experiment Two B – The Full Self­Report Study. ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 136 7.4 METHOD ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐139 7.4.1 Experiment Two (A)­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 139 7.4.1.1 Participants. ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐139 7.4.1.2 Procedure. ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐140 7.4.1.3 Results. ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐141 7.4.2 Experiment Two (B). ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 144 7.4.2.1 Participants. ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐144 7.4.2.2 Design.‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐145 7.4.2.3 Materials. ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐147 7.4.2.4 Procedure. ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐148 7.5 RESULTS‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐148 7.5.1 Data analysis.­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 148 7.5.1.1 Participant risk ratings of the flight (and what they relate to). ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐153 7.6 DISCUSSION ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐155 8 EXPERIMENT THREE ­ REVEALED BEHAVIOUR OF PILOTS IN A SIMULATION OF A HIGH­ RISK SITU ATIO N­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 160 8.1 INTRODUCTION ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐160 8.2 BACKGRO UND ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐162 8.3 EXPERIMENTAL DESIGN ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐163 8.4 METHOD ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐170 8.4.1 Participants. ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 170 8.4.2 Materials. ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 171 8.4.3 Procedure. ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 173 8.5 RESULTS‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐176 8.5.1 Dat a analysis.­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 176 8.5.2 ‘No­Go’ pilots. ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 181 8.5.2.1 Difference in risk perception of go and no go pilots‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐181 8.5.2.2 Difference in attitude between Go and No‐Go pilots ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐182 8.5.2.3 Difference in participant experience and age Between Go and No‐Go Pilots‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐183 8.5.3 Go pilots. ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 183 8.5.3.1 Risk perception and behaviour of Go Pilots‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐186 8.5.3.2 Attitudes and behaviour of Go Pilots ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐186 8.5.3.3 Participant experience and behaviour of Go Pilots‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐187 8.5.4 Interview results.­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 188 8.6 DISCUSSION ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐194 vii 9 GENERAL DISCUSSION­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 204 9.1 INTRODUCTION ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ ‐‐‐‐‐‐‐‐‐‐‐‐‐204 9.2 THE RELATIONS HIPS BETWEEN ATTITUDES, RISK PERCEPT IONS, AND EXPERE IENC ‐‐‐‐‐‐‐‐‐‐‐‐‐206 9.3 RELATIONSHIPS BETWEEN ATTITUDE, RISK PERCEPTION, EXPERIENCE AND BEHAVIOUR ‐‐‐‐‐‐‐‐‐211 9.4 SYNERGY (OR OTHERWISE) OF THE TWO BEHAVIOURAL METHODOLOGIES USED ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐216 9.5 LIMITATIONS OF THE RESEARCH‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐218 9.5.1 Implications and future research ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 224 9.6 CONCLUSION ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐227 REFERENCES­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 231 APPENDIX 1. PARTICIPANT CONSENT FORM ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 263 APPENDIX 2. DEMOGRAPHIC SURVEY ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 267 APPENDIX 3. HUNTER’S RISK PERCEPTION SCALE ONE­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 269 APPENDIX 4. HUNTER’S RISK PERCEPTION SCALE TWO­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 273 APPENDIX 5. THE AVIATION SAFETY ATTITUDE SCALE ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 275 APPENDIX 6. EXPERIMENT TWO A SCENARIOS ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 277 APPENDIX 7. EXPERIMENT TWO B SCENARIOS ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 299 APPENDIX 8. FLIGHT BRIEFING­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 306 APPENDIX 9. INTERVIEW QUESTIONS ­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­ 308

viii List of Tables

Table 1. Worldview Matrix for Cultural Theory...... 32

Table 2. Experiment One Participant Statistics ...... 111

Table 3. Results for Risk Perception and Attitude Scales in Experiment One...... 115

Table 4. Correlations Between Attitude, Risk Perception and Participant Demographics in Experiment One...... 116

Table 5. Mean Rankings for Scenarios in Experiment Two A...... 141

Table 6. Intra‐class Correlation Coefficients for Expert Ratings...... 142

Table 7. Risk Matrix and Final Scenarios in Experiment Two...... 143

Table 8. Experiment Two (B) Participant Demographic Statistics ...... 145

Table 9. Experiment Two (B) Demographic, Risk Perception and Attitudinal Statistics ...150

Table 10. Breakdown of the Go/NoGo Choice and Risk Rating Breakdown...... 151

Table 11. Correlations Between Behaviour and Risk Perception, Attitude & Experiential Variables ...... 151

Table 12. Experiment Three Participant Statistics...... 171

Table 13. Descriptive Statistics for All Pilots in Experiment Three...... 178

Table 14. Comparison of Descriptive Statistics for “Go” and “No‐Go” Pilots...... 180

Table 15. “Go” Pilot Behavioural Scores...... 183

Table 16. Correlation Analysis Between the Fourteen Predictor Variables and the Two Dependent Variables (Criterions)...... 184

Table 17. Comparison of the Factors that Contributed to the Go/No Go decision in the Simulator ...... 189

ix List of Figures

Figure 1. Risk Perspective Matrix...... 15

Figure 2. The Psychometric Map...... 29

Figure 3. Social Amplification of Risk...... 37

Figure 4. Representation of The Theory of Planned Behaviour ...... 77

Figure 5. Three Risk Perception Hypotheses...... 86

Figure 6. The Role of Risk Perception in Behaviour in Road Safety Literature ...... 88

x • Glossary

o Aerodrome –an area used for aircraft operation o Altitude – height above a datum. Usually sea level, or ground level. o Circuit – a rectangular shape flown by aircraft around airports. It allows for uniform traffic separation and predictable traffic movement. This is used worldwide. Roughly analogous to the use signalised roundabouts. o Flight Hours – the number of hours that a pilot has been in control of an aircraft o General Aviation (GA) – all aviation that is not military, or for revenue. GA excludes sport flying (e.g. hang gliding and ultralights) o General Flight Progress Test (GFPT) – a proficiency test used in Australia to test pilot’s ability to fly in the circuit, and within the local flight training area. o Stall(ed) ‐ the condition where a wing is producing less than its maximum lift as it is being presented to the relative airflow at too high an angle. This usually causes an aircraft to descend relatively rapidly, and is potentially very dangerous. o Unusual Attitude (in reference to aircraft) – when an aircraft is oriented other than within its normal flight domain. For example, when an aircraft is up side down.

xi

1 Introduction

Risk, along with the proverbial ‘death and taxes’, is one of life’s guarantees. All activities undertaken by people will inevitably involve a degree of risk (Hunter, 2002;

Wiener & Rogers, 2002).

The airline industry worldwide can, despite public perception to the contrary, be boastful of the fact that air travel is significantly safer than travel by road (Hawkins,

1993). However, a commercial aircraft accident, whether it is fatal or not, will attract wide media attention (Button & Drexler, 2006). It is within this context that commercial aviation takes place; airlines must outperform other modes of transport in terms of safety by a very wide margin in order that the public confidence in their safety is maintained. Indeed, airline transport operations, otherwise known as airline flying, can boast of an accident rate of less than 1 fatal accident per million departures (Boeing,

2005); gaining the industry the reputation of one of the safest transport modes in existence (Shappell & Wiegmann, 2000).

Other sectors of the aviation industry cannot so boldly claim the level of safety enjoyed by the airlines. For example, in Australia, the General Aviation (GA) sector exhibits a much poorer safety record than the airline industry; it is responsible for 93% of fatal accidents, whilst representing only 55% of the total flight hours. Airline operations, on the other hand, are responsible for 1% of fatal accidents, even though they represent 35% of flight hours in Australia (ATSB, 2006).

1

An improvement is therefore required in order to improve the situation with

regard to GA safety in Australia. Before this takes place though, identification of the

likely problems that face the sector is necessary. In previous research, human error has been implicated as a contributory causal factor in 85% of GA fatal accidents (Lenne,

Ashby, & Fitzharris, 2008). Similarly, aeronautical decision‐making has been identified in the literature as a primary factor in safe operations in GA (O'Hare, 2003; Simpson,

2001), and also one of the main contributors to accident causation (Shappell &

Wiegmann, 2003). Indeed, the improvement of decision‐making in GA, through systematic training may be one of the ways that overall safety is improved (Simpson,

2001).

To address this, a new course has recently been employed to address some

deficiencies that have been of concern to the Australian Civil Aviation Safety Authority

(CASA) and the industry. Threat and Error Management, or ‘TEM’ has been introduced

to address the hitherto lack of training of GA pilots in such areas as threat and error

identification, mitigation of the events, risk assessment and Crew Resource

Management (CRM).

The introduction of TEM to GA has also introduced the concept of risk

management (the management of uncertainty through effective decision making,

Drinkwater & Molesworth, 2010) into the GA sector, which up until recently, had not

explicitly trained pilots in risk management principles (Molesworth, Wiggins, & O'Hare,

2006). Whilst a relative success, it has been suggested that TEM has not clearly

addressed the role of attitudes in effective and safe behaviour (Thomas, 2004). A

2

causative relationship is assumed, but is not explored further, or with scientific rigour.”

This deficiency forms one of the main points of inquiry for the current research project;

which, if any, attitudes are core to safety behaviour in Australian GA?

Risk perception, or the recognition of the risk inherent in a particular activity

(Hunter, 2002), is thought to play an important role in both road (Deery, 1999) and

aviation safety (Hunter, 2005). However, when compared to the relative abundance of

literature concerning the attitude – behaviour relationship, the literature that explores

the relationship between risk perception and behaviour is somewhat scarce (see

Molesworth & Chang, 2009). Again, this deficiency in the literature is core to the

reasoning behind the current research, namely what role does risk perception play in

pilot behaviour in Australian GA operations?

Previous research has posited that individual differences, such as attitudes, risk

perception and experience affect behaviour (Molesworth & Chang, 2009). This

relationship has not been tested at length in the Australian General Aviation industry to

the knowledge of the author. Therefore, the main focus of this thesis is to examine the

relationship between behaviour of pilots in the Australian GA sector and their safety

attitudes, experiential factors like age and flight hours, and individuals’ risk perceptions.

1.1 The Australian Aviation Industry

The aviation industry is comprised of many sectors. In the civil arena the two

broadest are the airline industry (commercial aviation), and General Aviation. The airline industry is the most visible sector of aviation to consumers. In the context of this

3

thesis, any regularly scheduled aircraft operation that is undertaken to gain monetary

remuneration, whether this be through carriage of passengers, freight, mail or otherwise

is part of the airline industry. This is commonly termed a Regular Public Transport

(RPT) operation.

Australia divides the airline industry into high and low capacity operations based

upon seat numbers and/or maximum payload. High capacity RPT operations are those

that utilise aircraft with more than 38 seats and/or a maximum payload exceeding

4,200 kg. Low capacity operations are those operated by aircraft with less than 38 seats

and a maximum payload below 4,200 kg.

General Aviation is the most varied category within aviation. The International

Civil Aviation Organization’s (ICAO) characterization of GA is very broad; with aircraft

types involved ranging from powerless hang‐gliders to large turbine powered aircraft

(Eggers, 2000). In Australia however, the GA designation does not include hang‐gliders,

or other ‘sport’ aircraft like ultralights, auto‐gyros, balloons or gliders (ATSB, 2005).

There is also much variation in the types of activities undertaken in the GA

sector. Typical activities included in GA are flight training, private flights and charter,

however, other activities like agricultural flights, mustering, business flights, aerobatics

and ferry flights are also included within GA (Archibald & Reece, 1977; BTRE, 2005).

Because of the wide range of activities undertaken, and the many individual aircraft types utilised, the GA industry is complex, with the common ground being the operation of relatively small aircraft in non‐scheduled operations.

4

The GA industry, to complicate matters further, has two distinct sectors within itself; the commercial GA sector, and the non‐reward or non‐commercial sector. The commercial sector operates aircraft in order to make revenue; examples of this are the charter (non‐scheduled) and training sectors. Whilst the non‐reward sector operates aircraft for private purposes, examples of this sector are private flights, business flights and sport flying (aerobatics as an example).

The GA industry in 2008‐9 employed approximately 3,000 people with an annual contribution to GDP of nearly $280 million (Department of Infrastructure, Transport,

Regional Development and Local Government, 2009). Around 1.65 million hours were flown in GA operations in 2004, 55% of the total hours flown in the Australian aviation industry (non‐military) (BTRE, 2005, 2006).

There are few industries as incident critical as aviation, or subject to such heavy scrutiny in the event of an accident or system failure. Because of the potential for harm, the commercial aviation industry is heavily regulated and operations are typically performed by personnel that have been trained extensively, often with the aid of simulators, using well‐defined procedures.

Ab‐initio (the aviation industry term for training brand new pilots with no previous experience) flight training however, does not share this operational environment. Student pilots are inevitably inexperienced; their knowledge of aircraft systems, radio procedures and air law are expected to be representative of this lack of experience. However, these pilots are often in the air, solo, within twenty flight‐hours of commencing their training (CASA, 2010b). This is widely different from the operational

5

limitations exhibited by commercial regulations, yet it is often these same pilots that will be flying commercially in the near future. Indeed, a pilot requires only 30 more hours of practical experience (150 hours total) to gain a commercial licence than is required to obtain a driver licence (which requires 120 hours of practical experience) in

New South Wales (it is noteworthy however, that there are significant and numerous competency checks required for aviation, where driving requires only two checks) .

GA exhibits a much poorer safety record than the airline industry (ATSB, 2006).

The fatal accident rate for GA operations in Australia from 1991 ‐ 2000 was 1.2 fatal accidents per 100,000 hours flown (ATSB, 2006). Whereas the fatal accident rate for high capacity RPT from 1993 – 2003 was zero, with no fatal crashes, and the rate for low capacity RPT was 0.18 fatal accidents per 100,000 hours flown in the same period

(ATSB, 2005). This effectively shows the disparity in safety experienced by the sectors.

1.2 Links Between Psychological, Experiential,

Demographic Factors and Behaviour

One of the most common psychological factors that is ascribed a causal relationship with behaviour is attitude. The link between the construct of ‘attitude’ and one’s overt behaviour has been studied for many years (Glasman & Albarracín, 2006), mainly within the psychological realm. The literature appears somewhat divided when it comes to the final verdict as to the strength of the relationship between attitude and behaviour (Glasman & Albarracín, 2006) (also see section 4.7).

6

The link between risk perception and behaviour has a smaller literature

available (see section 4.7 for a full review). Generally, risk perception is thought to be a precursor of behaviour, where a higher level of risk perception will lead to a lower probability of an individual carrying out a particular behaviour (Machin & Sankey,

2008). Another common presumption is that poor risk perception, or the misinterpretation of actual level of risk by individuals, will consequently lead to inappropriate behaviour by the individual (Arezes & Miguel, 2008).

Poor (or inferior) risk perception and poor attitudes have both been identified as

possible ‘holes’ in the defence of accidents (as part of the ‘psychological pre‐cursors’ to

accidents) in the literature (Reason, 1990). Therefore, the causal factors surrounding

these constructs, and the impact of risk perception and attitude upon behaviour in the

Australian GA context is of interest in the attempt to understand, and ultimately to

minimise the influence of these factors upon poor behaviour.

Examples of such poor behaviour are unfortunately relatively forthcoming. One

particular example involved a brother and sister on their way to her wedding (ATSB,

2009). The brother, a holder of a commercial licence and with significant flight

experience, flew his helicopter into powerlines, fatally injuring both himself and his

sister. Investigation into the crash revealed a pilot that was both experienced and

trained in the specific area of low flight, and who should therefore have been a highly

competent and safe pilot, yet he had still undertaken a flight that was inherently more

risky than necessary. The decision (which in hindsight was obviously poor) to fly low to

the ground proved fatal. Another example of poor behaviour is found in the case of a

7

pilot that undertook very low flight near Narrandera in New South Wales (with wheels

touching the surface of a river), and also aerobatic manoeuvres at low level (which the

pilot was untrained in and not approved for). The pilot was further known to conduct

other high risk activities, for which he was not approved (ATSB, 2009).

In both cases, there were no mechanical, or outstanding external factors that

could be attributed as a/the primary causal factor, the inference being that it was an

internal individual or psychological factor, like one’s attitude or perception of risk that

was responsible for the behaviour. The constructs of ‘Risk’, ‘Risk Perception’, ‘Attitude’

and the link between these constructs and behaviour will be reviewed in greater depth

in the following chapters.

Both demographic and experiential values are used in the aviation industry as

discriminators for licensing and employment (CASA, 2008; Virgin Blue, 2010). Age is

used as a lower limit for licence holders by aviation authorities around the world.

Experience, in the terms of flight hours is also used as a lower limit for licensing, and

also as a lower limit for employment. A review of the academic literature failed to reveal an empirical basis for the usage of discrete hour limitations or requirements for the achievement of a particular licence type. It is possible therefore that these limitations may be in place largely through a corporate risk management process, that is a table‐top exercise that attributes likely risk given discrete circumstances, rather than through the findings of formalised research. The lack of research regarding the relationship between age, experiential factors and risk taking behaviour in GA stands in contrast to the

8

relatively large literature regarding the interaction of attitude, risk perception and

behaviour in the automotive industry.

The relationship between these demographic and experiential factors and

behaviour will be explored further in the current research.

1.3 Aims of the Present Research

The main aim of this research is to explore the link between aviation safety

attitudes, risk perceptions of a range of aviation based situations and the behaviour of

pilots in GA operations. From this, it is intended that a more thorough understanding of the drivers of pilot behaviour will be obtained, and that individual attitudinal and perceptual markers signifying risky behaviour will be obtained.

Specifically, the aims of the current research are:

• To determine the relationship between attitude (towards safe flight) and risk

perception of pilots in the Australian general aviation training sector,

• To determine whether pilots’ attitude (towards safe flight) and risk perception are

related to the traditional predictors of pilots’ competencies such as total hours flight

experience, recency (hours flown in past 90 days), and age,

• To determine the relationship between the attitudes, risk perception and self‐

reported behaviour of pilots in the general aviation training sector, and

• To determine the relationship between a pilot’s experience, as measured by age,

9

flight hours, and recent flight time, and self‐reported behaviour in the general

aviation training sector.

In order to achieve this, a series of three experiments will be employed in which

pilots reveal their experience levels, attitudes and risk perceptions (Experiment One), and where these factors are compared to self‐reported behaviour (Experiment Two) and to revealed behaviour (Experiment Three)

1.4 Summary

In summary, the vast and complex research examining predictors of behaviour

can be categorised as follows: behaviour is influenced by attitudes (Glasman &

Albarracín, 2006), by risk perception (Machin & Sankey, 2008) and by experience

(Thomas, 2004). Similarly, inappropriate risk perception and inappropriate attitudes

have both been identified as possible ‘holes’ in the defence of accidents (as part of the

‘psychological pre‐cursors’ to accidents) in the literature (Reason, 1990). This link

between attitude and behaviour is supported by the limited previous research in this

area, which is all derived from the US. Hence, the relative importance of attitude and

risk perception to the effective management of safe flight has not been fully explored in

the context of Australian GA (Thomas, 2004). Therefore, the main aim of the present

research is to explore the relationships between the behaviour of pilots as revealed by

self‐reports and in a flight simulation, and the factors of attitude, risk perception and

experience.

10

2 Risk and Risk Perception

2.1 Introduction

The aim of this chapter is to examine the constructs of risk and risk perception.

This will be achieved by first, defining ‘risk’ and second, exploring how risk is

approached in varied contextual settings. The chapter will also examine risk perception

and the various factors that influence some individuals to perceive risk more or less

keenly than others.

The primary reason for the in‐depth exploration of risk in this chapter, and

indeed, for the thesis as a whole is based on simple logic. Risk management is an

important contributor to safe aviation operation (Drinkwater & Molesworth, 2010).

Before one can manage risk/s, logically, one must perceive them (risk perception), and

therefore the construct of risk perception requires examination. Before risk perception

can be discussed however, ‘risk’, as the central factor of the construct, must be defined

and understood.

2.2 Risk

‘Risk’, although a commonly used term in the population, technical circles and academic literature, is rarely used by each of these groups to describe the same thing. It is clear from research that in general, the public will define risk in a broader sense than

professional risk assessors will (Beckwith, 1996).

11

Risk is characterised by a lack of clarity; from the very basic level of the variety in nomenclature that is used in its description (Aven & Kristensen, 2005), to the fact that there are at least eight different approaches to risk from a theoretical perspective

(Renn, 1998). These theoretical perspectives are further explored below.

Risk is a construct that suffers from multiple conceptions (Slovic & Weber,

2002). It has been variously conceived as a descriptor for a hazard (‘that dog is a risk’),

potential damage (‘what’s the risk of losing?’), threat (‘what’s the risk of terror attack in

India?’) or probability (‘what’s the risk of losing on the property market?’) (Slovic &

Weber, 2002). A gambler that is betting thousands of dollars is likely to describe their

wager as a risk, a sportsman attempting a move to the highest level of their game may

describe their efforts as a risk, a mother may be undertaking a home birth, which she describes as a risk she is willing to take. In other words, ‘risk’ is a fluid construct, and does not simply mean ‘expected fatalities’ to the public (Pidgeon, Hood, Jones, Turner, &

Gibson, 1992). It is a complex construct that has been the subject of research for nearly four decades.

One certainty though, is that risk concerns only those events that are in the

future. That is, the construct is only concerned with events that have not yet occurred.

Otherwise risk would be a measure of frequency or outcome, not one of uncertainty and unknowns. The uncertainty to the value of risk is therefore caused, at least in part, because of the obvious inability of humans to accurately predict the future (Wiener &

Rogers, 2002).

12

It follows then that there are many definitions of risk available in the literature

(see Aven, 2003; Fischoff, Watson, & Hope, 1984; Kaplan & Garrick, 1981; Renn, 1998;

Vriljing, van Hengel, & Houben, 1995; Wiener & Rogers, 2002). The usage of these definitions and their associated classifications are limited to usage in their respective niches because of the differing terminology, assumptions and conceptual understanding of risk (Kristensen, Aven, & Ford, 2006).

Often, risk is associated with the possibility that an action may lead to an adverse effect (Beckwith, 1996). However, this is not always the case. There are many examples of risks that are considered attractive and are actively sought by the public. Gambling, shoplifting, skydiving and hang‐gliding are all examples of activities designed or undertaken to actively seek risk (Machlis & Rosa, 1990).

The many definitions of risk do have a uniting element though, each definition will feature a distinction between possibility and reality (Renn, 1998). Aven and

Kristensen’s review on risk (2005) suggests that risk is the composition of the two basic dimensions of ‘possible consequences’ and ‘associated uncertainties’. In other words, risk is the combination of two things; the probability of an event taking place and the magnitude of the consequences of the event. Similarly, Renn (1998) and Klinke and

Renn (2002) have defined ‘risk’ as the possibility of human involvement in events that lead to an impact upon what they value.

According to Renn’s analysis, a definition of risk contains three elements; (1) outcomes that have an impact on what humans value, (2) possibility of occurrence

(uncertainty) and, (3) a formula to combine both elements. There are up to eight major

13

approaches to the explanation of risk, each of which has been derived from different

disciplines, and has its supporters and application. These are:

• The actuarial approach,

• Toxicology/Epidemiology,

• Probabilistic risk analysis (also called the engineering approach),

• Economics of risk,

• Psychology of risk,

• Social theories of risk,

• Cultural theory of risk, and

• Integrated approaches.

The features of each of these approaches to risk are shown in Figure 1 below. In the figure, the approaches to risk are described as per the six major descriptors used by

Renn. That is, by the base unit of measurement, the style of measurement (e.g. experiments, desktop analysis), the scope of risk covered by the theory/ies, by the basic problem that each theory suffers from (e.g. too much complexity, lack of predictive power), the industry/ies in which the theory of risk is mostly utilised, and the normative function of each risk theory (e.g., policy making, systems safety).

14

Figure 1. Risk Perspective Matrix

(Source ‐ Renn, 1998)

15

Barnes (1996) also provides a detailed overview of the above approaches to risk, in which risk definitions are divided into two basic ‘camps’, the positivist and the non‐ positivist camps. The positivist view of risk is sometimes alternately known as the classical view of risk. In this paradigm, risks are thought to exist objectively, be observable and measureable. Alternately, the non‐positivist, or Bayesian view of risk postulates that risk is simply a way of expressing uncertainty (Aven & Kristensen,

2005).

The positivist or classical approach to risk is often seen as synonymous with the

‘engineering’ approach to risk. In this approach, the prevailing idea is that risk exists objectively and that it is possible for a risk analyst to produce an accurate estimate of the risk. The definition of risk used in most engineering approaches to risk will be expressed through a probability; the number of times an event will happen if the situation is repeated an infinite (hypothetically) number of times, and the expected outcomes of the event. These estimates will generally be made based on ‘hard data’ or the opinions of experts in the field (Aven & Kristensen, 2005). It is the latter trait that draws the most widespread criticism, that the risk estimates made by risk assessors do not reflect an objective probability, but rather are opinions held by the elite in the field, with little or no more validity than the estimates made by non‐experts who work in the field or even the lay‐public (Aven & Kristensen, 2005; Fischoff et al., 1984; Klinke &

Renn, 2002; Renn, 1998).

Renn (1998) has summarised the criticisms of the engineering approach to the following;

16

• The probabilistic analyses used in engineering risk approaches are too simplistic to

capture the complex and unique interactions between human actions/inaction and

the related consequences.

• When faced with immeasurable factors or uncertainty, the approach requires

intervention from the expert/s, which will depend on the individuals’ risk attitudes

and preferences for utility versus resilience maximising strategies.

• The engineering risk is generically prone to systems‐based errors and

organisational failure, that is, the approach does not usually account for the risk of

organisational failures in the measurement of individual risks.

• Expert risk analysis is influenced by values and is not a strictly scientific endeavour.

Individuals’ values, beliefs, perceptions or attitudes may influence the

characterisation, measurement and interpretation of risks.

• The engineering style definition of risk, where there is a combination of magnitude

and probability for an event taking place, assumes equal weighting and importance

for both components. This would mean that there is no difference between an event

that has severe consequence but is unlikely to occur, and an event with minor

consequences, but is likely to occur. In plain terms, it is the total impact of the risk

that is treated as consequential, therefore 1,000 minor incidents that cost $1,000

each would be treated as a similar problem to a risk that as likely to result in a

single accident that cost $1,000,000.

17

• Engineering risk, because of its use of population level statistics, can only provide

an estimate of risk at the population or macro level. This is not representative

therefore, of the risk experienced by an individual. Those that are exposed to a

higher level of risk than a macro analysis has predicted would have a legitimate

objection to policy based upon this analysis.

Despite these criticisms, the engineering approach to risk has proven itself to be

useful in the detection of deficiencies in complex technical systems and also, in the

subsequent improvement of safety of the analysed system (Bohnenblust & Slovic, 1998;

Faber & Stewart, 2003; Renn, 1998).

Aviation, given its origins in engineering often refers to risk using an engineering

style definition. For instance, Hunter (2002) in his study of risk perception and risk

tolerance of aircraft pilots uses ’risk’ to mean the possibility of loss of life or injury, encompassing probability and severity; an engineering style definition of risk.

To this end, research in the GA industry has identified flying as a high‐risk

activity, characterised by the potential to result in fatality or injury, loss or damage to

one’s reputation and/or the loss of significant amounts of money (Green, 2001). This

conclusion is based upon the potential losses involved in an accident, but importantly, it

does not take into account the likelihood or frequency with which the event will take

place. Aviation accidents are extremely rare events (Janic, 2000), meaning that whilst

the potential losses in an aviation accident are high, the relative frequency with which

they occur shows that the probability of an accident occurring on any given sector is

low. Therefore, it is more accurate to describe aviation as a hazardous, or high hazard

18

industry (Molesworth, 2005), rather than as a high‐risk industry, which it is commonly

referred to as.

The notion that risk is a discrete, perceivable and measurable construct is a pre‐

requisite for the positivist view of risk (Aven & Kristensen, 2005), which is a frequently

used method of defining risk in systems safety applications. A common denominator in

many risk definitions is that there is an underlying possibility that human actions may

lead to consequences that have an impact on what people value (Renn, 1998). In light of

this, the perception of risks by individuals that have the power and ability to minimise

risks becomes increasingly relevant. Risk perception is therefore seen as important in

the risk minimisation process, and will be addressed below.

2.3 Risk Perception

As a concept in safety critical domains, risk perception is relatively easy to

define, it is the recognition of the risk inherent in a particular activity (Hunter, 2002).

Risk perception has been a subject of much interest and debate in the literature

(Farrand & McKenna, 2001) and is thought to play an important role in both road

(Deery, 1999) and aviation safety (Hunter, 2005).

Individual’s risk perception is not only a matter of basic sensory perception, but

involves people’s beliefs, feelings, judgements, attitudes, expectations, social and cultural situation (Sjöberg, 2000). Risk perception is therefore the cognitive interface between people and their environment with regards to hazards and their associated consequences (Hunter, 2002).

19

It is important to note that risk perception research is often undertaken at the societal level, in the context of political decision or policy making, and is therefore often concerned with societal risks, or risks that are in place that shall affect all of (or part of) a population (Sjöberg, Moen, & Rundmo, 2004). Examples of this include nuclear power production, alcohol, handguns and smoking (Slovic & Weber, 2002). The findings from this research, whilst not overtly focussed upon individuals, are a starting point from which the likely modifiers for individual’s risk perception can be gleaned. To be clear, most risk perception research appears to be concerned with the political implications of hazards and risks being present within a jurisdiction, whereas the current research is focussed upon the behaviour of the individual, and the reasons/ing behind this behaviour. The implication of this is that operationalised risk perception research, or that which is pertinent to individuals’ risk taking behaviour in a discrete context, appears to be relatively poorly represented in the literature.

Before a thorough understanding of risk perception can be achieved, one must

understand the process through which an individual arrives at a perception of risk.

Remember from above that risk can be defined as the possibility of a hazard impacting

upon things that humans value. Therefore, in order to perceive risk, one must first perceive a hazard in their environment. A hazard in this case is the agent that may or may not cause loss to something humans value. For example, imagine that an office has asbestos roofing. The risk in the office would be that workers may be exposed to the asbestos, which may cause health problems. The hazard in this example is the asbestos itself (the agent causing the problem).

20

The first step in the process of risk perception, the perception of hazards, is

discussed below.

2.3.1 Hazard perception.

Hazard perception may be defined as the identification of physical object/s and

or environmental circumstances that have the potential to be hazardous, and correctly

perceiving them as posing a hazard (Barowsky, Shinar, & Oron‐Gilad, 2010; Brown &

Groeger, 1988). Accurate hazard perception is important; a flawed perception of the

hazards present will lower the number of, or present an inaccurate view of the hazards

that are known by the individual. This in turn will lead to an incomplete or inaccurate

‘picture’ of the situation. Only those hazards that are known to the individual can be

used in any further analysis of the situation (Faber & Stewart, 2003), hence, a flawed

perception of hazards will lead to a flawed perception of the risks in a particular

situation.

In aviation, a hazard may be physical, as in the case of poor meteorological

conditions, other aircraft, terrain, or poor physical condition of pilots. A hazard may also

be somewhat more cognitive, as in the case of flawed knowledge of the aircraft,

regulations or physics of flight. A hazard may also be somewhat more abstract, as in the

case of a regulation that causes a hazard (Molesworth, 2005), an example of which may be regulated minimum rest periods, which may lead to fatigue.

Hazard perception is influenced by fundamental perceptual factors. It has been

found in road safety research that novice drivers will have less efficient situational

perception caused by inefficient visual scanning, inappropriate visual fixation and aid

21

usage and through a lack of contextual knowledge (Barowsky et al., 2010; Chapman &

Underwood, 1998; Deery, 1999). There is no reason that these perceptual deficits are not applicable to novice pilots also.

In comparison to older individuals, it is thought that younger novices will have trouble with perceiving their situation holistically (Barowsky et al., 2010; Wallis &

Horswill, 2007). Importantly, hazard perception appears to be improved by training

(Wallis & Horswill, 2007). This is interesting to note, as the implication of this is that the overall speed of hazard perception (i.e., how quickly one perceives hazards) by relatively inexperienced individuals could possibly be bettered through the use of training aimed specifically at the improvement of hazard detection and perception skills.

Wetton et al., (2010) contend that hazard perception is a three‐step process.

They are:

1. Hazard detection – the detection of a possible hazard by an individual with their

basic perceptual faculties, like their vision, hearing, touch etc.

2. Conflict judgment – the judgement as to whether the potential hazard that was

detected by the individual will indeed be in the correct time and space to cause a

conflict. In other words, is it going to be close enough to affect them?

3. Hazard classification – the judgement as to whether the hazard that has been

detected, and will conflict with them is going to cause harm and deserves a response.

In other words, is the object that will hit me a car, or a feather?

22

The three steps above are dependent and sequential steps. That is, a hazard must be detected before a conflict can be judged, which must be judged before the hazard can be classified.

2.3.2 The theories arising from societal risk perception.

Early risk perception work was designed to explore the risk judgements of laypeople in comparison to experts in the field and/or statistical data. This work was designed to measure the perceptions of risk in a societal sense, such that the studies were measuring the risk perceptions of factors like nuclear power. The most famous and influential of these studies was that conducted by Lictenstein, Slovic, Fischoff,

Layman and Combs (1978) in which laypeople were asked to judge the annual frequency of death in the United States from forty distinct hazards.

Two major findings came from this study, both showing systematic differences between the lay‐people’s judgements and the ‘real risks’ or statistical estimates by experts. The first was that, when compared to the real risk, the laypeople overestimated the fatality rate from infrequent hazards but underestimated the fatality rate from relatively common hazards. The second finding was that the variance shown in the lay‐ people’s risk judgements followed a pattern, the more spectacular or ’vivid’ the type of death, the greater the estimation of fatalities by the laypeople (Lichtenstein et al., 1978).

This pattern occurred for both the frequent and infrequent causes of death, meaning that laypeople overestimated the infrequent and vivid causes of death and only slightly underestimated the more frequent and vivid causes of death (Pidgeon et al., 1992).

23

The pattern shown by the lay‐people’s responses fit in with the pattern expected

with the ‘availability heuristic’ in mind (Lichtenstein et al., 1978). The availability

heuristic essentially suggests that the more readily identifiable or imaginable an event

is, the more probable or frequent that event is likely to be judged (Tversky &

Kahneman, 1973).

This makes sense under many circumstances, however there have been valid

arguments raised against the availability heuristic. The available information to an

individual may be flawed, as in the case of incomplete disclosure of information or

inaccurate perception of one’s environment. In this case, certain events or hazards may

be over or under‐represented, resulting in a flawed perception of frequency. One

avenue through which individuals may receive information regarding their

environmental risks is through their local media outlet. The role of the mass media in

the dissemination of information, and therefore their role in the influence on public risk

perception is debated (Freudenburg, Coleman, Gonzales, & Helgeland, 1996; Sjöberg,

2000; Wahlberg & Sjöberg, 2000).

The heuristics based research is no longer considered to be of significant

importance in the risk perception literature, as other studies, for example Fischoff,

Slovic, & Lichtenstein (1982) have shown that people’s risk perception was influenced

by dimensions other than the judged probability of an event, such as question framing, semantics, the style of risk information given, and the individuals’ judgement as to what the questioner is actually asking (Fischoff et al., 1982). Individuals may also give an

‘opinion’ that is not actually theirs. ‘No opinion’ responses to surveys are relatively

24

infrequent when the range and scope of surveys that are conducted is taken into

account. This suggests that people are able to form an answer to any question asked of them. The implication is that these answers are not one’s considered opinion, but rather are stereotypical or associative answers given to satisfy the desire to ‘be counted’

(Fischoff et al., 1982).

An interesting phenomenon that is evident in risk perception research is termed

‘Risk Target’ (Sjöberg, 2000). One’s estimates of risk posed to them from a particular

hazard will be different than that individual’s estimate of the risk posed to their

immediate relatives, and to society in general. The risk rating given by individuals for

themselves is lower than that given for immediate family, which in turn is lower than

the general societal risk rating (Sjöberg, 2000). Sjöberg’s, (2000) finding has been

replicated by Bronfman and Cifuentes (2003) in their study of Chilean risk perceptions.

It is obvious though, that it is not possible for every individual to be at less risk for any given event than the average.

Following on from this, there is a similar phenomenon that is widely observed in

the literature. ‘Optimistic bias’ refers to the perception that oneself is less likely to

experience a negative outcome and more likely to experience a positive event than the

average person (Kos & Clarke, 2001; Weinstein, 1980). Weinstein (1980) in a study of

college students found that individuals believed that they were more likely to own their

own home, like their future job and start their career on a good wage than were their

peers. These students also believed that they were less likely than their peers to

experience negative events such as becoming an alcoholic, being fired or attempting

25

suicide. Weinstein contended that the bias was strongest when the event was perceived

to be controllable and commitment to the event or level of emotional investment was

higher (1980).

McKenna (1993) found that there was a link between optimistic bias and control

in the automotive context. People judged themselves to be at a lower risk of having an

accident if they were driving than if they were a passenger (McKenna, 1993). This is

supported by the automotive literature that has found that drivers perceived

themselves as being exposed to less risk than average (Groeger & Brown, 1989; Job,

1990; McKenna, 1993; Svenson, Fischoff, & MacGregor, 1985).

An explanation for this perception is that people overestimate their skill level

(Deery, 1999); an explanation that is backed up by a relatively large body of literature

(see for example Deery, 1999; DeJoy, 1989, 1992; Delhomme, 1991; Guppy, 1993;

Matthews & Moran, 1986). The major theme being that a majority of people rate their

driving skill as greater than that of the average motorist.

The findings in the optimistic bias and ability bias literature have been replicated in the aviation field by Wilson and Fallshore (2001). In their study, fifty‐seven pilots from Central Washington University and 103 General Aviation pilots attending safety seminars were administered a questionnaire designed to measure optimistic and ability bias. Their results showed that, compared to other pilots, individuals underestimated the likelihood of their being involved in an accident and overestimated their ability to both avoid entry into Instrument Meteorological Conditions (IMC) and avoid an accident following an inadvertent entry into IMC (Wilson & Fallshore, 2001).

26

The major theories of societal, or non‐operationalised risk perception are

discussed immediately following.

2.3.2.1 The psychometric model.

One of the earliest, most influential and widely cited attempts to explain the complex nature of public risk perception is the ‘psychometric’ model. The psychometric model was designed as an attempt to identify the perceived qualities of a particular hazard and from this to understand the linkage between these characteristics and the

endpoint of the perception of a particular risk (Pidgeon et al., 1992). The model finds its

roots in the work of Fischoff, Slovic, Lichtenstein, Read, and Combs (1978) in which

thirty activities were evaluated with respect to 1) the activities’ benefit to society 2) its

perceived risk 3) the acceptability of the current level of risk and 4) the risk’s position

on the nine ‘dimensions’ of risk which would later become 18 (with the reciprocal of

each of the nine dimensions added) (Sjöberg, 2000).

The model shows that there are certain properties of a hazard that will influence

people’s risk perceptions. According to Oltedal, Moen, Klempe, and Rundmo (2004), the

specific properties of a hazard that are thought to influence people’s risk perception

are;

• Voluntariness of risk ‐ are the risks undertaken voluntarily or not?

• Immediacy of effect ‐ the extent to which the risk of death is immediate

• Knowledge of risk ‐ are the risks known to science?

• Control over risk ‐ the extent to which one can, by personal means, avoid death

27

whilst engaging in an activity or being exposed to a risk factor

• Newness ‐ the extent to which a risk is novel

• Chronic‐catastrophic ‐ on a plane from chronic (killing one person at a time) to

catastrophic (killing many people in one event)

• Common ‐ dread ‐ the extent to which people have become used to the hazard

(common) or the opposite, the extent to which people are unused to a hazard and

experience a ‘gut’ reaction (dread)

• Severity of consequences ‐ if a risk is realised (an event occurs), what is the likely

hood that the consequences will be fatal?

These properties in turn can be reduced to two factors ‐ ‘Dread’ and ‘New/Old’

(also called ‘unknown risk’). These two dimensions form the basis of the ‘map’ of

hazards which shows where hazards lie on a four‐quadrant table, showing their relative positions and their overall rating of perceived risk. The map appears below in Figure 2

below. As an example, the risk of a car crash is relatively known and accepted, and

therefore presents in the bottom left quadrant, whereas the risk of meltdown at a

nuclear power station is relatively unknown and dreaded, hence presented in the upper

right quadrant.

28

Figure 2. The Psychometric Map

Authors in the psychometric literature have reported that 60‐70% of variance is accounted for in the above factors (Fischoff et al., 1978; Siegrist, Keller, & Kiers, 2005;

Sjöberg, 2000) giving an indication that the model is relatively strong and that policy decisions might use it to predict acceptance of a risk by the public (Fischoff et al., 1978).

This work is widely represented in the literature, and it has been used to measure the perceived risk in many contexts, including, but not limited to; rail systems

(Nighswonger, Kraus & Slovic, 1988), electromagnetic field exposure (MacGregor,

Slovic, & Granger Morgan, 1994) and food hazards (Sparks & Shepherd, 1994).

However, this model is criticised in the literature (Marris, Langford, & O'Riordan,

1998; Siegrist et al., 2005; Sjöberg, 1996, 2000, 2002) for several reasons. The first is simply that the model is not as strong as its proponents state. The reason for this is that

29

mean ratings are used in psychometric analysis, not raw data. Raw data are subject to more error than mean data and models based on raw data are less likely to fit neatly

(Sjöberg, 2000). Moreover, the usage of mean data has lead to much higher correlations being reported than would be the case with raw data usage. Correlations among means are generally much higher than those found using raw data (Sjöberg, 1996). In fact, raw data analysis of that used in psychometric studies yields an explanation of only about

20‐30% of variance (Sjöberg, 2000).

The mean data usage criticism is made for two reasons. First, the psychometric paradigm inherently studies the average response, rather than the actual responses given by subjects, which is less than ideal if a ‘real‐world’ picture of risk perception is the goal. Secondly, the apparent strength of the model is misleading to the end‐user, giving them a false idea of the extent to which the model explains risk perception in their context (Sjöberg, 2000).

The second criticism of the psychometric model is that the model may have missed important factors that influence risk perception. Sjöberg (2000) contends that when a new factor, ‘unnatural and immoral risk’ (tampering with nature) is added to the traditional psychometric model, the performance of the model is strongly improved.

If this is correct, then the model (without unnatural and immoral risk added) is incomplete and any findings from it are also incomplete.

2.3.2.2 The cultural theory of risk perception.

The cultural theory of risk is a slightly newer approach to risk, dating to Douglas and Wildavsky’s research in the early 1980s (Douglas & Wildavsky, 1982). As per the

30

other risk perception theories, cultural theory attempts to explain how people perceive

their world and their subsequent actions. Cultural theory differs from other risk

perception paradigms in that it claims that this process is primarily determined by

social factors and adherence to cultural norms (Oltedal et al., 2004). In essence, the

theory holds that people will decide on what constitutes a risk, and how to treat those

risks based on an effort to sustain their preferred way of life. Risk is therefore socially

constructed (Steg & Sievers, 2000).

The theory is in itself derived from Douglas’ grid‐group typology (Douglas, 1978;

Oltedal et al., 2004) in which ‘group’ refers to the membership and depth of absorption or commitment to a social unit. ‘Grid’ refers to the level of regulation and restriction that a particular social setting places on the individual (Oltedal et al., 2004).

‘High‐grid’ situations will be very restrictive on behaviour. An example given in

the literature is a court trial, where behavioural options are severely limited. ‘Low‐grid’

is therefore a situation in which behaviour is not restricted and people are free to

interact with each other as they wish. In other words, the grid‐group analysis describes

different examples of social control (Oltedal et al., 2004).

The grid‐group analysis is often represented in a two axis ‘map’. Each of the four

areas of the map correspond to the four ‘worldviews’ or ‘ways of life’ that are a

cornerstone for cultural theory. These are individualists, hierarchists, egalitarians, and

fatalists. Each of these groups will feature different social commitments from those

associated with the group, this commitment is thought to predispose members to

adopting a particular view of society, nature and the world in general (Oltedal et al.,

31

2004). The characteristics of the worldviews as described by Oltedal et al. (2004) are

shown in Table 1 below. These are the fears of those with a worldview, the societal

preference (e.g. capitalism) for each worldview, the political alignment of those in the

worldview group (left to right), the view of nature (in terms of nature’s natural position

and regenerative ability) of those with the worldview, and the view of what risk is to

those that have the particular worldview.

Table 1. Worldview Matrix for Cultural Theory

Fears Support Political View of View of Risk Alignment Nature

Individualists Obstruction of Free Market Right wing Self‐preservingOpportunity individual (so long as no freedom threat to freedom)

Egalitarians Increased Socialist Left‐Wing Fragile, Generally risk inequality Government vulnerable averse among people

Hierarchical Crime, civil Experts/ In power Mostly Self Accepting of disobedience Government preserving, risk, with within limits appropriate justification Fatalists Nothing Nothing No alignment Random, Indifferent uncontrollable

There is a fifth group in the cultural theory that fits with none of the above groupings, this group is often termed the ‘hermit’ after their style of living outside of society (Oltedal et al., 2004). This group is rare and is fundamentally outside of society, and is therefore not included in the tabulation.

32

The amount of empirical research utilising cultural theory is relatively low, with

few studies available in the literature. Wildavsky and Dake (1990) are acknowledged as

the first to attempt to quantitatively study cultural theory (Sjöberg, 2000). Their study

was a psychometric style study, used to measure how respondents assessed the

magnitude of 36 risk factors associated with ‘technology and the environment’, ‘war’,

‘social deviance’, and ‘economic trouble’. Of these, 25 of the factors were used. The study

set out to explain the risk perceptions given in the psychometric test using several

factors, these were: knowledge (self‐reported), personality, political orientation, and

cultural biases (the worldviews above). Of these, cultural biases were found to be the

best predictor of risk perceptions found in the study. The correlation between cultural

bias and risk perception was reported to be statistically significant (largest r=0.51).

Marris et al., (1998) attempted to replicate the work of Wildavsky and Dake in

England. This study utilised face‐to‐face surveys of residents in their homes, a total of

129 respondents were studied. The study differed from Wildavsky and Dake in that it overtly sought to compare the psychometric and cultural models. There are some interesting results from this study; first, only 32% of the studied population displayed a strict adherence to the worldviews of cultural theory. Second, the sample group was biased towards egalitarians, that is, there were more egalitarians than any other worldview, a finding backed up by a French study completed at a similar time (Brenot,

Bonnefous, & Mays, 1996). The study by Marris found that a reasonably significant

(maximum of 41%) proportion of the variance in risk perception was explained by the psychometric model, whilst a relatively low (12% maximum) amount was explained by the cultural bias scales. This meant that the cultural bias scales explained no more of the

33

variance than socio‐demographic variables (Marris et al., 1998). Their theory though

was that it was not the strength of correlation that was important, but rather the

number of significant correlations that are to be found that may form a significant

pattern.

The cultural theory has been described as important and indispensable to the

study of risk perception (Oltedal et al., 2004; Wilkinson, 2001). However, the over‐

riding theme in these quantitative studies though is of a lack of explanatory power of

the cultural theory of risk (Sjöberg, 2000). There appears to be multiple problems with

the theory itself (Marris et al., 1998). The first is that the theory has two versions,

labelled the ‘stability’ and ‘mobility’ versions of the theory by Marris et al. (1998).

In the stability version of the cultural theory, people will align themselves with

social groups and institutions that share their cultural bias, it is inferred that their

cultural bias is lasting and consistent in their lives. The mobility version of the theory sees people as flexible and maintains that they can and will change their cultural bias in different settings or over time, as in a person changing their bias at work compared to home, or from the time they are a teenager compared to their middle aged bias. The rigid nature of the stability version of the theory, that people must adhere to a single

worldview and that their bias will not change is often cited as a problem for cultural

theory. Indeed as was found by Marris et al. (1998) and Brenot et al. (1996) the

majority of a sample will not conform to a single worldview

Another problem for cultural theory is the ambiguity associated with ‘culture’.

Oltedal et al., (2004) suggest that this factor may be a reason for the lack of explanatory

34

power of the theory. Sjöberg (2000) however, is more severe in his appraisal of the theory. He suggests that the biggest problem with cultural theory is that it is only speculative in nature, with little empirical evidence, rendering it a failure.

The review of risk perception literature thus far has shown that the most popular theories of risk perception have displayed relatively little predictive power.

2.3.2.3 Other theories of risk perception.

The need for a theory of risk that displays more predictive power has driven some authors to develop some alternative theories to risk perception. Kasperson et al.,

(1988) developed their ‘Social Amplification of Risk’ theory as an attempt to unify the psychological and cultural theories of risk (Kasperson et al., 1988; Pidgeon et al., 1992).

The theory is based on the metaphor of amplification, adapted from signal amplification in communications theory (Machlis & Rosa, 1990; Pidgeon et al., 1992). In this theory, risk is seen as a ‘signal’ that, like a radio signal, will be amplified before reaching the desired receiver (Machlis & Rosa, 1990). The ‘amplifiers’ in the theory are both the individual and social groups that receive and disseminate information about a risk.

Social Amplification of Risk theory is reliant upon the fact that people learn a majority of their knowledge from outside sources rather than through intimate experience (Pidgeon et al., 1992). Because of this, knowledge and perception of the world can be shaped by the opinions or judgments of others who are responsible for the communication of critical information. In the theory, risks can be ‘amplified’ in two ways (Kasperson et al., 1988) just as a radio signal might be:

35

• Through the intensification or weakening of signals that are passed on to social groups or individuals

• Through filtration, in which multiple signals are cut to one.

The amplification of these signals will be carried out by many different

amplification ‘stations’ (Kasperson et al., 1988), such as scientists that specialise in the

area, risk‐management institutions, media, social activist organisations, social leaders,

personal social groups, and public agencies.

Transmission of information from each of these stations will be through the

normal communication channels, like the media, telephone conversations, direct

conversations and the Internet. Recipients will also form part of the amplification

process, acting as another amplification station (Kasperson et al., 1988).

The key processes of amplification are thought to be (Kasperson et al., 1988):

• Filtration of information

• Decoding of signals

• Processing the signal

• Attachment of social values to the information

• Interaction with peers to aid interpretation and validation of signals

• Engaging in actions (as groups or individuals) in response to the signal

A graphic representation of the model can be found in Figure 3 below.

36

Figure 3. Social Amplification of Risk

(Source ‐ Kasperson et al., 1988)

It is also theorised that there will be secondary effects from social amplification of risk. These are thought to be: enduring perceptions and attitudes, local economic impacts, political and social pressure or changes, changes in the original risk source, changes to education and skills of emergency response personnel, civil disobedience, changes to regulation and monitoring, increased insurance cost, repercussions for other

37

technologies/risk sources (lower acceptance) and for institutions (lower public trust)

(Kasperson et al., 1988).

These secondary effects are themselves perceived by people and amplified,

which in turn may be perceived and amplified. The effect of this is described as

‘rippling’ just as one sees in a pond when a rock is thrown in. The ripple effect illustrates

how the impact of a risk can quickly spread from a local problem to distant parties like

other industries or institutions (Kasperson et al., 1988).

The theory is not without its problems though. Social amplification, like the

cultural theory, is very general and is an observation of reality, rather than an empirical

measure of the situation. This presents problems for researchers wishing to carry out

empirical research. Also, the behavioural link made in the theory may not be as simple

as is implied in Figure 3 above. As an example, people may be faced with a signal or risk,

and instead of allowing the signal to influence them negatively (possibly leading to

avoidance), they might use the signal as a warning of further risks and attempt to

improve the relevant technology or to mitigate the hazard (Pidgeon et al., 1992). For example, a mother hearing that warm liquid in a plastic bottle may cause the plastic to leach harmful chemicals, may chose not to provide such liquid to her children in these bottles, rather than not to use them at all.

The social amplification theory has contributed to the study of risk perception in

an important way in that it has highlighted the need for, and advantages of, cross‐

disciplinary research in a field as broad as risk perception (Pidgeon et al., 1992).

38

One other alternative perspective on risk comes from the work of Sjöberg. His

theory on risk is based upon the influence of three factors; risk sensitivity, attitude and specific fear (Sjöberg, 2000). Risk sensitivity is a rating of an individual’s general sensitivity to risk. This is measured by looking at the individuals rating of risks from many different types of hazards and determining whether or not they are more or less sensitive to risk than the average. Attitude is a complex construct, and will be explored in detail in the following chapter, however in Sjöberg’s theory, attitude is assumed to drive risk perception (attitude influences beliefs). Specific fear is the fear aroused in an individual concerning the particular hazard or risk that is being studied (Sjöberg, 2000).

For example, perceived risk of nuclear power is affected by the specific fear of radiation accidents like Chernobyl or Three Mile Island. According to Sjöberg, his theory accounts for more of the variance in risk perception than the psychometric and cultural theories

combined, at approximately 46% (Sjöberg, 2000).

Sjöberg (2000) found that attitude plays a significant role in risk perception, and

is in fact the most significant of the three factors in his theory explaining between 30

and 40% of the variance in risk perception. The construct of attitude will be explored in

detail in the following chapter.

2.4 Conclusion

‘Risk’ is a very difficult construct to understand. There are many different

explanations and definitions of risk available in the literature, all of which have their

appropriate usages and contexts. The unifying elements of a definition of risk are;

39

outcomes that have an impact on what humans value, possibility of occurrence

(uncertainty) and a formula to combine both elements (Renn, 1998).

Risk perception, just like risk itself, is a difficult construct to understand. Before risk can be managed, it must first be perceived, and before it can be perceived, the relevant hazard must be identified or perceived. Wogalter, Brelsford, Desaulniers and

Laughery (1991) found in their studies that familiarity, severity of injury, technological complexity, confidence in knowledge of the hazard and the individual’s stereotypes associated with the hazard all had an influence on an individual’s hazard perception.

Sjöberg’s theory on risk perception appears to be among the most explanatory of those theories in the literature. It holds that attitude, risk sensitivity, and specific fear are important factors in risk perception. Attitude appears to be the most important factor, accounting for between 30 and 40% of variance (Sjöberg, 2000). Accordingly, attitude will be explored in depth in the following chapter.

40

3 Attitudes

3.1 Introduction

Attitude is one of the most studied constructs in social psychological literature

(Ajzen, 2001), and is thought to affect risk perception (Sjöberg, 2000) and behaviour

(Glasman & Albarracín, 2006). Despite the intense focus on attitude, the concept of

attitude has not significantly changed over the period it has been researched (see Eagly

& Chaiken, 1993). It is the aim of this chapter to explore the concept of attitudes, so as to

allow for a more accurate understanding of the construct, and to allow for later

discussion on the possible link/s between attitude and behaviour.

3.2 Attitude: Definitions and Components

Attitude, just like risk, is a word used widely in society. It is also a hypothetical

construct that is not directly measurable (Ajzen, 1988; Deaux, Dane, Wrightsman, &

Sigelman, 1993). The concept of attitudes has been the subject of study for the best part of a century (Crano & Prislin, 2006), and accordingly there are many various and overlapping definitions available in the literature (Åberg, 1999). A definition of attitude in the recent literature proposes that:

“Attitudes are the evaluative judgements that integrate and summarize these

cognitive/affective reactions. These evaluative abstractions vary in strength, which in

41

turn has implications for persistence, resistances, and attitude‐behaviour consistency”

(Crano & Prislin, 2006 p. 347).

In other words, an attitude is the product of what one thinks of, and how one

feels towards a certain object or thing. This product can vary in strength both in the

short and long term, and may affect how one behaves.

3.2.1 Classic definitions and components of attitudes.

There are at least two distinct approaches in the classic attitudinal literature in the attempt to define what specifically comprises an attitude; the ‘three component model of attitudes’ and the ‘uni‐dimensional’ approach to attitudes (Stahlberg & Frey,

1996). The three component model of attitudes is founded on the theory that attitudes are formed as a combination of three discrete components, the affective, cognitive and behavioural (or conative) component (Eagly & Chaiken, 2007). Each of these components can be manifested in a verbal and non‐verbal fashion (Ajzen, 1988). A classic definition of attitude according to the three component model is proposed by

Eagly and Chaiken (1993, p. 1)

“Attitude is a psychological tendency that is expressed by evaluating a particular

entity with some degree of favour or disfavour . . . Evaluating refers to all classes of

evaluative responding, whether overt or covert, cognitive, affective or behavioural”.

Indeed, research in the aviation arena has utilised a similar definition of attitude

in its formation, e.g where attitude can be overt or covert, and towards all classes of

object or otherwise (see Hunter, 2005).

42

The affective component is the emotional response to the object or entity in

question, typically the affective component is described with terms like ‘love’ or ‘hate’,

‘like’ or ‘dislike’ (Stahlberg & Frey, 1996). The cognitive component is comprised of the

opinions, beliefs and ideas about the object/entity. The cognitive component, as

suggested by its name, is the thought process or thought‐out response, which reflect the

perceptions of the individual (Ajzen, 1988). The behavioural component refers to the

behaviour (intended or actual) or the tendencies for specific action displayed by an

individual (Stahlberg & Frey, 1996). However, one must question exactly what

constitutes a ‘tendency’ to evaluate in a particular fashion, except for the interpretation

by an experimenter of a relatively consistent response to a specific set of stimuli by an

individual. According to Schwarz, (2007) an inherent limitation of this model is that it

is essentially a case of fundamental attribution error, however, instead of a layperson

committing this error, it is the experts within the attitude field.

The uni‐dimensional approach to attitude was developed in response to the

inconsistency in cognitive, affective and behavioural actions in the behavioural

construct. In other words, the uni‐dimensional model was developed in response to the

fact that people do not always act, react or think in accordance with how they feel.

Factor analysis has also not provided support for the model, as the three components were too highly correlated as to be treated as conceptually different (Stahlberg & Frey,

1996).

In the uni‐dimensional approach to attitude, the affective (emotive) component

of an attitude is held to be the only viable indicator of an attitude (Stahlberg & Frey,

43

1996). In other words, if one reveals their feelings, and by extension, their evaluation of

an object (e.g. ’I like my bike’) then they have revealed their attitude.

3.2.2 Recent theories of the nature of attitude/s.

Within the attitudinal literature, there is still debate as to a singular,

comprehensive, and universally accepted definition of attitude (for recent discourse see

Conrey & Smith, 2007; Cunningham, Zelazo, Packer, & Van Bavel, 2007; Eagly &

Chaiken, 2007; Fazio, 2007; Gawronski & Bodenhausen, 2007; Schwarz, 2007).

Most professionals in this area would, at least in part, agree with the

multidimensional view of attitude, which according to Crano and Prislin, (2006) is that

an attitude is a representation of the evaluative integration of a cognitive and affectual

component of the individual’s experience towards an attitude object. In other words, an

attitude is the amalgamation of what one thinks and feels about an object or situation

(Ajzen, 2001; van Overwalle & Siebler, 2005). The multi‐dimensional view is therefore

effectively half way between the three‐component paradigm, and the uni‐dimensional

paradigm. Instead of attitude being influenced by affect, cognition and behaviour, it is

influenced, theoretically, by affect and cognition only (behaviour is removed).

This is not to say however, that the concepts and definitions of ‘attitude’ are not

being actively questioned in the academic literature. Indeed there has been at least one new set of theoretical conceptions of attitude proposed that differ markedly from the

‘classic’ or ‘legacy’ theories of the past. These are the constructionist models (see for example Conrey & Smith, 2007; Gawronski & Bodenhausen, 2007; Schwarz, 2007). The constructionist models differentiate themselves from the attitude theories of old,

44

through the willingness to engage in open questioning of whether or not ‘attitudes’ exist

at all (Schwarz, 2007). There have also been multiple new theories emerge concerned with the nature of attitudes. These theories are generally extensions of existing theories, or composite theories that utilise existing theoretical frameworks and will be discussed

below.

There are therefore now three major theoretical streams in the modern attitude

literature, the Unitary or Single Attitude view, which hold that attitudes are an

individual’s association between an attitude object, the Dual Attitude Theory, in which it

is thought that an individual will hold (at least) two attitudes, one that is easily, readily

and automatically accessible, and one that is more difficult to access, and the

Constructionist Theories of attitude, in which attitudes are constructed in an ad‐hoc

basis, as required by the individual (see Chaiken & Stangor, 1987; Conrey & Smith,

2007; Eagly & Chaiken, 1993; Petty, 2006; Petty, Briñol, & DeMarree, 2007).

3.2.2.1 Single attitude theories.

The Single Attitude theories hold that attitudes are an individual’s association

between an attitude object (which can range from purely physical to purely abstract)

and their evaluation of that object (Fazio, 2007; Petty et al., 2007). The evaluation of the

object may stem from multiple sources, including, but not limited to, an individual’s

fundamental beliefs or values, their affect, either past or present feelings, or behaviour

(past or present), or past experience with that object (Fazio, 2007). In other words, an

attitude is what one thinks of a certain object, real or abstract. The associations between

45

attitude object and evaluation are variable in terms of strength, and this affects the

accessibility of the attitude (Fazio, 2007).

According to Fazio (2007), attitudes can be thought of as a form of knowledge.

That is, one’s association between an attitude object (such as beer) and their evaluation

(that beer is delicious) is represented in the same fashion as other knowledge in one’s

memory. This theory of attitude has been in existence for twenty‐five years, and has

been influential in Social Psychology in the pursuit of a working definition of attitude

(Petty et al., 2007).

The Fazio model has come under some pressure in the literature because the

theory requires one to make an evaluation of an attitude object in order to have an

attitude. This evaluation is thought not to be the attitude, but rather an expression of

the underlying attitude (Eagly & Chaiken, 1993). In other words, Fazio’s model is measuring the expression of the attitude, rather than the attitude itself. An analogy would be an attempt to measure someone’s body using their preferred clothing size.

One might be quite thin, but prefer very baggy clothing, or they may have recently lost a great deal of weight. In either of these cases, the inferences made to an individual’s body size would be incorrect.

3.2.2.2 Dual attitude model.

The Dual Attitude model of attitude is a more recent concept to the literature

than the Single Attitude theories (Petty et al., 2007; Wilson, Lindsey, & Schooler, 2000).

There are multiple theories that broadly conform to the dual attitude model (Petty et al.,

46

2007), however they will not be dealt with at length here. Rather, the model itself shall be analysed.

According to the Dual Attitude theory, an individual will hold (at least) two attitudes; one that is easily, readily and automatically accessible, and one that is more difficult to access, hence requiring more effort and motivation to retrieve from one’s memory. These are termed the implicit and explicit attitudes respectively (Wilson et al.,

2000).

Implicit attitudes are characterised as being like habits, in that they are enduring and automatic to the individual given a specific contextual situation (Wilson et al.,

2000). In contrast, explicit attitudes are characterised as being like a skill that has been learnt. Under time or performance pressure this skill (and therefore in the analogy the explicit attitude) may be negated by the old habit (or the implicit attitude). An appropriate example would be a pilot who has learnt to operate Airbus aircraft reverting to Boeing procedures (the original aircraft that they trained on) if given a last minute change to their approach. In this way, it is seen that under time pressure or other stressor, it is possible that an individual will revert from their explicit attitude, to their implicit one.

Another interesting point to note regarding the difference between explicit and implicit attitudes relates to how the two differ with regards to attitudinal change. It appears that explicit attitudes are much more readily changed than implicit attitudes

(Rydell & McConnell, 2006).

47

Petty et al. (2007) have suggested three common elements amongst the multiple theories that have arisen in the dual attitude paradigm. They are:

1. The two attitudes held will be separate representations in the mind.

2. The two attitudes will stem from independent cognitive processes. Implicit

attitudes from processes like evaluative conditioning, explicit attitudes from

processes like overt cognitive involvement.

3. The two attitudes are relatively independent of one another, and operate in

different situations from one another. They are not expected to show any conflict.

Implicit attitudes are expected to be utilised in spontaneous situations: explicit in more deliberative situations (Wilson et al., 2000). It is thought that the two attitudes co‐exist, and one will not cancel the other out. Each is theorised to be stored in a separate area of the memory, to be used when appropriate (Wilson et al., 2000).

3.2.2.3 Constructionist models.

In the constructionist model, attitudes are constructed in an ad‐hoc basis, as required by the individual (Petty, 2006). They are based upon the immediate and broader context in which the individual finds themselves, salient beliefs, affects and behaviours (Cohen & Reed II, 2006; Petty et al., 2007). In the theory, individuals therefore do not possess stable attitude/s towards anything. Rather, they evaluate the object within its context with specific goals in mind (Schwarz, 2007). That is, unlike the

48

dual‐attitude models above, no attitudinal evaluation is stored in memory by an

individual.

This theoretical paradigm has been criticised on two major fronts. First, it is criticised on the apparent lack of ability to explain attitudinal stability (Eagly & Chaiken,

2007). Second, it has been criticised because research appears to indicate that in some situations, individuals do not make new evaluations of the attitude object (Petty et al.,

2007) but rather rely on pre‐existing attitudes stored in memory (Fazio, 2007).

With regards to attitude stability, if one is presented with the same information,

in the same context, constructionist theory would conclude that the individual would

present the same attitude (Schwarz, 2007). This is true of the other attitude models,

except they attribute that consistency to attitude stability, rather than to contextual

stability. It appears problematic to criticise a theory of attitude on its apparent lack of

ability to predict stability of attitude, when in fact it does, in the appropriate

circumstance. Indeed, if one’s ‘attitude’ remains the same, regardless of context, then

one might question if the individual is expressing an attitude, a value/ethical statement,

or something different altogether. For example, consider theft. It is common in the

current climate to share movies, music, and photographs. This is strictly (in certain

circumstances) theft, however, many people would not treat it as a serious problem. On

the other hand, theft of motor vehicles, or money from a bank is considered a serious

crime by most. Although the subject of the attitude in the example above is constant, the

context has changed.

49

3.2.2.4 Other models.

Of late, there has been an influx in the literature of new models of attitude. An in‐

depth review of each of these is beyond the scope of this chapter, as they are generally

amalgamations, or modifications of the existing theories. However, each will be

summarised below, as the theories each add to the current understanding of attitude as

a construct.

The Attitude Representation Theory (ART) holds that when people encounter an

attitude object, they will activate a stored mental representation of that object (Lord,

Paulson, Sia, Thomas, & Lepper, 2004). The individual may also take into account contextual factors and their attitude will change accordingly. The theory therefore appears to be an amalgamation of the single attitude theory and the constructionist theory. It appears to posit that attitude is a mental representation (as per the Fazio model) that is explicitly affected by contextual factors (as in the constructionist theories). This in effect gives the best of both worlds, familiarity for the attitude theorist, and more predictive power than in past (Schwarz, 2007).

The Associative‐Propositional Evaluation Model (APE) is grounded in the dual

attitude theory (Gawronski & Bodenhausen, 2007). According to the model, an attitude

is composed from one of two different evaluative responses. The ‘associative’ process

gives responses for immediate and affectually‐based evaluations; these are not checked

by the individual for truth or falsehood (Gawronski & Bodenhausen, 2007). The

‘propositional’ process represents the more thought about and factually based

evaluations undertaken by the individual. In the propositional process, the individual

50

will exhibit more careful thought and the use of multiple salient pieces of information

(Gawronski & Bodenhausen, 2007; Petty et al., 2007). This echoes the dual attitude theory’s implicit and explicit attitudes.

The Iterative Reprocessing Model (IR) (Cunningham et al., 2007) is an extension of the dual attitude model, in so far as the model allows for attitudes to be determined by more than one process. Interestingly, it has been informed by neuroscience, rather than a social science or psychology, which is more the norm. This adds a new element and discipline to the attitudinal literature. In the IR model, attitudes are formed as a result of constant reprocessing of the attitude by the individual. This reprocessing is termed iterative reprocessing (Cunningham & Zelazo, 2007). An initial iteration of the attitude would be similar to the implicit attitude in the dual attitude model, however, this is reprocessed using a more thorough evaluation to produce a more meaningful or accurate attitude (second iteration). Reprocessing can continue for many iterations if the reprocessing is necessary or relevant for the situation (Cunningham & Zelazo,

2007).

The Multiple Pathway Anchoring and Adjustment (MPAA) Model combines the more traditional attitudinal theory and constructionist theory (Cohen & Reed II, 2006)

In this model, attitudes are formed through two separate mechanisms, the ‘Outside‐In’ mechanism for external objects, and the ‘Inside‐out’ for internal objects. These attitudes are then stored, and are available for the individual to recall, should a similar context arise in which the attitude is required. If an attitude is not available that is relevant, a new attitude is constructed as needed. This model appears valid, however the level of

51

complexity in the model is problematic, in that a casual reader, or possibly even the intended users of the model are unlikely to fully appreciate or understand the full implications and findings of the model easily or quickly.

The Connectionist theory is an extension of the constructionist theory. In the

Connectionist models, attitudes are represented using a loose analogy of the human brain. Attitudes are not whole ideas that are well formed and in memory, rather they are actually patterns of activation of very small information processing units (Conrey &

Smith, 2007). In this way, attitudes are the product of many small pieces of information coming together to form a whole picture. An appropriate analogy is the computer monitor; each pixel will be a discrete small piece of information, which when combined with many others, forms a picture that is interpretable and meaningful. Strong or stable attitudes are formed when the pattern of activation is frequently experienced, learnt, and becomes what is termed an attractor, or favoured response (Conrey & Smith, 2007).

In this model, processing of information is undertaken in a distributed manner.

Put simply, there is not a single central processor undertaking the work. Rather, each of the individual nodes will receive external and internal stimuli, which they will process and then in turn distribute internally for further processing. The overall attitude will be the result of the brain processing both external and internal information. The model deals with novel or unexpected information by introducing an error reduction capability; this is the known as the delta algorithm (van Overwalle & Siebler, 2005).

Because of the distributed style of information processing, this model predicts that

52

attitude modification will be undertaken in small steps, by small changes in inter‐nodal connections (van Overwalle & Siebler, 2005).

The Meta‐Cognitive Model (MCM) (Petty et al., 2007) draws from two classic attitudinal theories to create a new theory of attitude. In the theory, attitudes can be stored, and are linked to object associations. There can also be two attitudes held, that can be opposite in valence (e.g. positive or negative). The theory differs from the Dual

Attitude theory in that opposing attitudes are not necessarily thought to come from a different cognitive process, rather, an individual can hold both a negative and a positive attitude about something which is based in associative evaluation (or based on propositional processes depending on the context) (Petty et al., 2007). The model draws its name from the notion that individuals can tag their original attitudes with additional information in much the same way as one may tag a photo on a computer with a date, location and caption to aid in subsequent use of the photo. In this way, one may have an

‘attitude’ of “I like cola” which may be further tagged with a measure of confidence, “I am confident I like cola” (Petty et al., 2007).

Of the models in the literature, the most useful to the current research are those that follow the classic three‐dimensional view of attitude. This is because the theory holds that attitudes are enduring, and are therefore measurable at different points in time (as opposed to being contextual and therefore only measureable at the time of attitude formation). The three‐dimensional view has also previously been used in aviation safety research (Hunter, 2005), and measurement tools are readily available because of this.

53

3.2.3 Function of attitudes.

Understanding how and why people hold the attitudes that they do has been the

subject of research for the best part of the last century. The knowledge of the function of

holding a particular attitude will allow one to understand more accurately the

motivations of the individual and the likelihood of the individual changing their attitude

(Katz, 1960). The early work in attitude in the psychological literature gives the reader

a solid, and even now, nearly 50 years later, relevant account of the function of

attitudes. Katz (1960) proposed four functions of attitudes;

• The Utilitarian Function,

• The Ego‐Defensive function,

• The Value Expressive function, and

• The Knowledge‐Based Function.

The utilitarian function is shown by attitudes that attempt to maximise rewards

and minimise potential punishment or penalty from the external environment of the

individual (Katz, 1960; Maio & Olson, 1995). This is seen in everyday life, and is the acknowledgement that people like to feel good, not uncomfortable. The utilitarian function of an attitude can be illustrated by the example of an individual who holds a favourable opinion of their government as they are helping the individual’s economic situation through their policies (Katz, 1960). The individual in this illustration is showing the classic utilitarian attitude, in that their attitude serves a purpose. Due to its being utilitarian, the attitude will change according to the dynamic nature of the relative

54

utility that the attitude holds in relation to the attitude object. In other words, if the

government reverses their policies that currently help the individual, he or she will

likely change their attitude towards the government to a more negative stance, as there

is not longer a utilitarian reason to hold a positive attitude.

Utilitarian attitudes can also serve individuals in social settings. Adopting an

attitude that is similar to a potential friend or mate may help the individual to attract

that person (Stahlberg & Frey, 1996).

Utilitarian attitudes are dependent on many factors. As described above, the past

and present perceptions of the attitude object will mould the utility of the attitude

object to the individual, which will affect the attitude held by the individual (Katz,

1960). Other factors that will affect utilitarian attitudes are the nearness of reward or punishment to the attitude object, for example, whether or not the linkage between attitude and outcome is close in proximity or time, the clarity of the linkage between the attitude and the outcomes and the consistency between the attitude object and any outcomes (Katz, 1960).

The ego defensive function of attitudes can aid one in protecting themselves

from negative emotions towards themselves, or their peer group, by projecting their

emotions onto others or other groups (Stahlberg & Frey, 1996). People will invariably

spend a sizable amount of time and energy reconciling their actual persona with their

intended one (Katz, 1960). By holding attitudes that are ego defensive an individual can

reduce the level of anxiety felt by avoiding the reality of their persona or the reality of

their environment (Katz, 1960).

55

The Value Expressive function of attitudes can aid an individual to accomplish the need to confirm their own self‐concept (Stahlberg & Frey, 1996). That is, the expression of an attitude may make an individual feel more positively about their personality. This is the other side of the coin as it were to the ego defensive attitude.

Instead of mitigating negatives, the value expressive is an active and positive expression of one’s self.

As an illustration, if one’s ego is seen as a bank, the ego‐defensive attitude is like

the security system. It guards the bank from loss. However, a bank with no money is not

particularly useful. Therefore, the value‐expressive attitudes are required as ‘deposits’

in the bank. These two functions together allow the ego to be both defended and fed.

The Knowledge function of attitudes aid the individual in the organisation and

cognitive structuring of their world, which would otherwise prove near impossible

(Stahlberg & Frey, 1996). People seem to require frames of reference and mental

shortcuts to cope with their environment, and holding specific attitudes appears to aid

in this (Katz, 1960).

The functions of attitudes, in a generic sense, are therefore reasonably well

known, and have been for a long while in the literature. Of interest are the variations in

attitudinal function between individual persons and situations. People with different

personality types are likely to vary in respect to their attitudinal function in a given

situation. According to Tesser and Shaffer (1990), people who are low self‐monitors,

that is, people who exhibit less awareness of self, are more likely to exhibit the following

56

traits when compared to people who are high self monitors. Low self‐monitors are likely to:

• Appeal to values when justifying attitude

• Rely on personal evaluations at the expense of social or situational norms in decision making

• Evaluate people on attitudinal and value compatibility, rather than social or situational activities when meeting for the first time.

Tesser and Shaffer (1990) also point out that attitudes may be modified by individuals due to situational or contextual differences. Some possible situations are as follows:

• Anonymity will encourage value expressive attitudes to be expressed. Public expression of an attitude on the other hand will encourage utilitarian attitudes to be expressed

• Threats to one’s personality or peer group will encourage ego defensive attitudes to be expressed

• The need for rapid judgments or decision‐making will encourage knowledge based attitudes to be expressed.

3.3 Attitude Change

The factors that are responsible for attitude change have been a staple of attitudinal research since the very earliest days of the discipline (Crano & Prislin, 2006).

They have summarised the initial models of attitude change eloquently.

57

“messages are presented, processed, and if successful, move recipients’ attitudes

toward the advocated position.” (p. 348).

These classic models of attitude have been expanded upon and altered

somewhat in the years of research. The most popular and influential works on attitude

change are the dual process theories of attitude change, the Elaboration Likelihood

Model (ELM) from Cacioppo (1984) and the Heuristic‐Systematic Model (HSM) from

Chaiken (1980).

3.3.1 Dual process models of attitude change.

The ELM and the HSM have both been influential throughout the last three

decades due, in part at least, to the fact that they are flexible enough and sufficiently wide in scope to explain the effects of a large number of persuasion variables, processes of persuasion and outcomes of persuasion (Petty, Wegener, & Fabrigar, 1997).

The two models do exhibit differences, the importance of which will depend upon contextual factors, however the two theories are more similar than not. They will typically explain the same results, though there may be some nomenclature differences and the processes assumed to have driven persuasion may also differ (Petty et al.,

1997). Because of these similarities, these two models are often discussed together in the attitude change and persuasion literature. The central idea of both models is that the process of attitude change is dependant upon individuals’ ability and motivation to process information presented to them (Wood, 2000).

58

Individuals who have both the ability and the motivation to consider

attitudinally relevant information provided, may also have attitudes based upon a more

thought out, or systematically processed assessment of the information available. This

is called the systematic or central route to persuasion (Tesser & Shaffer, 1990; Wood,

2000). Alternately, people who lack the motivation (e.g. the message is not relevant to

their situation, or they do not wish to do anything at that point in time) or ability (e.g.

they do not believe that they have the time to think carefully) to process the information

given to them will likely adopt attitudes based upon readily available cues about the

source of the information, the message itself, or the context in which information is

provided (Tesser & Shaffer, 1990; Wood, 2000). These attitudes are relatively

insensitive to message quality (Tesser & Shaffer, 1990).

3.3.1.1 Systematic or central route to persuasion.

The systematic or central route to persuasion, as has been mentioned above, is the route that is taken by people who are able and motivated to undertake more in‐ depth processing of attitudinally relevant information. Some of the more generic factors that have been found to increase the incidence of systematic thinking are framing of persuasive information in an unexpected manner, the relevance of the message to the individual and the use of catch phrases that indicate more wide reaching implications.

(Wood, 2000). However, there are more specific factors that will influence systematic processing. Both the ‘ability’ and ‘motivation’ of individuals that will guide processing in the theories are not simple constructs in themselves; they both are affected by multiple factors, both internally and externally based.

59

The factors that have been identified that will influence the ability of an

individual to undertake systematic processing according to Tesser and Shaffer (1990)

are: repeated exposure to information, absence of distractions, neutral mood (i.e. not

happy or sad), knowledge of the topic area, direct experience with the topic area, and a

high need for cognition by the individual

To build on this, recent authors have identified other factors that are thought to

enhance systematic processing (Wood, 2000). These are: unexpected question framing

(framing of questions away from the norm) (Smith & Petty, 1996), relevance of message

to the individual (Turco, 1996), attitudinal strength, both strong (Fabrigar, Priester,

Petty, & Wegener, 1998) and ambivalent (Fabrigar et al., 1998; Maio, Bell, & Esses,

1996), when individuals believe in their ability and efficacy with regard to their

evaluation (Bohner, Rank, Reinhard, Einwiller, & Erb, 1998), and when the message is

presented in normal and accessible language (Hafer, Reynolds, & Obertynski, 1996).

Additionally, situational variables can affect the motivation of individuals to

undertake more deep processing of persuasive information. According to Tesser and

Shaffer, (1990) these are: the use of interrogative formats to asses individual’s opinions,

the delivery of arguments by more than one presenter, in written form and in a

rhetorical fashion, and having individuals sit during delivery of information.

Attitudes that are formed by individuals using systematic information processing are generally thought to be stronger, more durable, more persistent and also more likely to be predictive of future behaviour (Chaiken & Stangor, 1987; Tesser & Shaffer,

1990) than those based on the use of heuristic style processing.

60

3.3.1.2 Heuristic or peripheral route to persuasion.

According to the dual process models, when an individual is not motivated or

able to undertake systematic processing of attitudinally relevant information, the

individual will engage in heuristic style processing of information by using peripheral or

auxiliary cues (Chaiken, 1980; Chaiken & Stangor, 1987; Petty et al., 1997; Tesser &

Shaffer, 1990; Wood, 2000).

The kind of cues used in the heuristic or peripheral route to persuasion are

heuristics (simple decision rules e.g. ”I am always wrong” or ”The Government know

what they are doing”), classical conditioning effects (multiple experiences with an object leading to an attitude), mere exposure effects (exposure to an object leading to an attitude), and expertise of the presenter of the information (more expertise is better)

(Chaiken & Stangor, 1987). Other factors that have been identified are source credibility, likability, physical attractiveness, message length and audience reaction

(Chaiken & Stangor, 1987; Tesser & Shaffer, 1990).

It should be noted that the two routes (systematic and heuristic) described

above are not mutually exclusive, that is, systematic and heuristic processing can co‐

exist for a particular attitudinal entity (Chaiken & Stangor, 1987; Tesser & Shaffer,

1990). In fact, Petty et al. (1997) have described the relationship between systematic

and heuristic as being at opposite extremes of a continuum. This description may help

to provide a more clear idea of the interplay between the two.

61

3.3.2 The unimodel of persuasion.

The unimodel of persuasion is based upon work undertaken by Kruglanski in the

late 1980s (Kruglanski & Thompson, 1999). The model shares many features with the

dual‐route models of persuasion, but differs on the crucial point of the process (or

route) of persuasion.

As the name suggests, the unimodel posits that there is a single ‘route’ through

which persuasion takes place. The unimodel holds that persuasion is “a motivated

process of hypothesis testing and inference dependent on individuals’ cognitive

availability and accessibility of pertinent information. More generally speaking, it is a

process during which beliefs are formed on the basis of appropriate evidence”

(Kruglanski & Thompson, 1999, p. 89). This indeed sounds very similar to the dual‐

process model definitions in that an individual’s cognitive ability plays a major role.

Inherent in the model though is the assumption that the two persuasive

processes in the dual‐process models are based upon two independent content factors.

That is, that the source of the information and the message found in that information

are independent and that they will affect the persuasive process independently. This

has been criticised as inaccurate and has been a criticism of the model (Petty, Wheeler,

& Bizer, 1999).

The criticisms and apparent short‐fallings of the model (for example, see Petty et

al., 1999) are such that the model has apparently not gained widespread following from authors other than Kruglanski.

62

3.4 Conclusion

Attitude remains a concept under wide academic study (Gawronski &

Bodenhausen, 2007). As can be seen by the above review, the concept of attitude and its

implications are not a simple or unified subject. Given that the best part of a century of

academic interest and inquiry has been spent on attitudes, the understanding of the

construct may very well be beyond the abilities of the scientific community for some

time to come.

Attitude remains a concept without a universally accepted and wholly workable

definition. That said, the definition that will be adopted for this thesis is that an attitude

is the product of what one thinks of, and how one feels towards a certain object or thing.

This product can vary in strength both in the short and long term, and may affect how one behaves. This is adopted as it represents the definition of attitude utilised in most academic material within the attitude research area.

Of less contention than the definition of attitude itself, is the notion that attitude

influences or is related to behaviour (for a review, see Glasman & Albarracín, 2006).

This link will be explored in the next chapter, as the implication for aviation is

significant. If one wishes to positively affect behaviour of pilots or other technical staff,

then attitude change, or hiring staff based on attitudinal measures may be an

appropriate method of attempting to increase the safety levels of the industry.

63

4 The Relationship Between Attitude, Risk

Perception and Behaviour

4.1 Introduction

“It is an Article of faith in psychology that human behaviour is complex and

therefore, very difficult to explain and predict” (Ajzen & Fishbein, 2005, p. 178).

However, there still exists much in the way of interest in the relationship between attitude and behaviour. In the simplest terms, attitudes are generally assumed to influence behaviour (Iverson, 2004) (although it is noted that this assumption may be flawed). Therefore, it is logical to study the link between the two in an attempt to reach

a more thorough understanding of the situation and possibly to gain insight into the

most efficient ways in which to elicit discrete responses from individuals regarding

attitudes (Glasman & Albarracín, 2006).

In contrast to attitude, the risk perception to behaviour relationship has not

attracted the amount of attention, and hence literature, as is evident in the attitude‐

behaviour relationship. Generally, risk perception is thought to be a precursor of

behaviour, where a higher level of risk perception will lead to a lower probability of an

individual carrying out a particular behaviour (or potentially carrying out alternative

actions that are lower in risk or precautionary) (Machin & Sankey, 2008). Another

common presumption is that poor risk perception, or the misinterpretation of actual

64

level of risk by individuals, will consequently lead to inappropriate behaviour by the

individual (Arezes & Miguel, 2008).

In light of this, the following chapter shall explore the predictors of behaviour,

with a particular emphasis on attitude, and risk perception.

4.2 A History of the research on the Attitude – Behaviour

Relationship

In the initial stages of social psychological research, it was believed that social

psychology was in fact the study of attitudes, as they were assumed to be the

determinant of human behaviour (Ajzen & Fishbein, 2005). This tenet has been

subsequently shown to be inaccurate, with very few studies in the literature achieving

results that give evidence for a strong and stable relationship between ‘attitude’ as a

pre‐existing and enduring entity, and the behaviour of individuals (Schwarz, 2007;

Wicker, 1969) such that attitudes could be described as the sole determinant of human

behaviour.

One of the first pieces of research that tested the theory that attitude was linked

to overt behaviour was the work of LaPiere in the 1930s. In this research he tested the link between expressed attitudes, and overt behaviour by covert means. LaPiere travelled around the United States (U.S.) with a Chinese couple, eating at restaurants and gaining accommodation overnight. The Chinese couple were refused only once out of over 200 visits to public establishments. Subsequent to these visits, LaPiere mailed a

65

questionnaire to all establishments asking if they would serve a Chinese person. In excess of 90% indicated that they would not (Dockery & Bedeian, 1989). This work was the first to produce evidence that self‐reported attitude may not be congruent with actual behaviour. Subsequent research by Wicker (1969) went as far as to conclude that it was more likely that attitude was not related to, or at best, weakly related to behaviour, than it was that attitude/s was strongly related to behaviour.

The link between attitude/s and behaviour/s remains complex and is not comprehensively understood, despite the amount of work in the arena (see Glasman &

Albarracín, 2006 for a comprehensive review). Recent research has shed light on some facets of the interaction, and has pointed toward the existence of a measurable link between attitude and behaviour (Glasman & Albarracín, 2006), although certainly not one as strong or stable as was previously thought when attitudes were assumed to govern behaviour. This research has also raised questions that necessitate further research if they are to be answered sufficiently.

The attitude – behaviour relationship has been found to be stronger and more predictable if the ‘principle of compatibility’ is satisfied (Ajzen & Fishbein, 2005). In the principle of compatibility, a behaviour is viewed as a specific action undertaken by an individual, towards a target in a given context and time. The principle holds that for the attitude‐behaviour relationship to be at its most consistent, the attitude and behaviour that are measured must share the same action (the same act or omission), target (the same entity), context (the same environmental situation) and time (the same time in terms of the attitude being measured at the time of the behaviour, as opposed to before

66

or after the behaviour) (Ajzen & Fishbein, 2005). This principle holds within it some important definitions, namely that a behaviour is a discrete action that is performed by an individual in relation to their given environment.

Research that satisfies the principle of compatibility has delivered more robust

results than those found previously (Ajzen & Fishbein, 2005). This result has been

encouraging to the field, and many other research projects have utilised a similar

methodology in order that meaningful and significant results are to be found.

An interesting field of study that is not as extensively cited in the literature as the

attitude‐behaviour relationship in a generic sense is the direction of influence between

attitudes and behaviours. It has been suggested that there are at least four possible

ways in which attitude and behaviour may be related (Bentler & Speckart, 1981).

Attitude may have a causal relationship with behaviour, behaviour may have a causal

relationship with attitudes, attitude and behaviour may have a reciprocal affect on each

other, and that there is no relationship between attitude and behaviour (Kahle &

Berman, 1979). The notion that there is no relationship between attitude and behaviour

has been given little credence in the literature, with Hini, Gendal and Kearns (1995)

stating that it is clear that such a notion is to be rejected. Another possibility that is

logically possible is that an external factor may be acting upon both attitude and

behaviour.

67

4.3 Factors That Influence the Attitude­Behaviour

Relationship

There has been a large variability evident in the literature with regards to the

degree with which attitudes predict behaviour. Correlations performed in research

projects have ranged widely, with results of a strong positive correlation (.73) (see

Fazio & Williams, 1986) to weak (negative) correlation (‐.20) (see Leippe & Elkin,

1987).

Whilst an average correlation of attitude and behaviour is difficult to uncover,

general attitudes are typically described as being poor predictors of behaviour (Ajzen &

Fishbein, 2005), and the correlation between the two is usually low. This phenomenon, and the subsequent research undertaken in order to further understand the relationship between more specific attitudes and behaviours will be discussed below.

Glassman and Albarracin (2006) and Ajzen and Fishbein (2005) suggest that

there are three broad areas of the attitudinal arena that affect the attitude‐behaviour

relation. These are attitude aggregation and specificity, attitude accessibility and

attitude stability. These will be discussed below.

4.3.1 Attitude aggregation and specificity

It has become apparent that broad attitudes, or those that are related to general

ideals or objects, are not strong predictors of behaviour (Ajzen & Fishbein, 2005).

Therefore, in order to further test the theory that attitudes and behaviours are linked,

68

researchers have attempted to refine their definition of attitudes, behaviours and also

minimised the scope to which experiments were designed (Schwarz, 2007).

Stronger correlations between attitude and behaviour have been reported in experiments that utilise aggregate behavioural intentions and compare them to broad attitudes, than those that measure broad attitude and compare them with specific single

behavioural measurements (Ajzen & Fishbein, 2005). That is, a large number of

behaviours that are relevant to the attitude are added together in order to have a broad

instrument that is designed to give a wider picture of the individuals’ intended

behaviour. These aggregated behaviours are then compared to broad attitudes in order

that the research compares the two on a similar level of generality and specificity.

As an example, in a research project concerned with attitude and behaviour in a

religious context, Fishbein and Ajzen (1974) measured general attitudes, and then

asked participants about their behaviour in 100 discrete circumstances. General

attitudes did not typically correlate well with individual behaviours, but when

behaviours were aggregated a stronger correlation was found. Results similar to this

have also been found by Werner (1978) in a project on abortion activism, and by Weigel

and Newman (1976) in research concerned with environmental protection.

To predict behaviour in a more specific context, or to predict a specific single

behaviour, aggregate measures of behaviour and broad attitudes will likely be of little

use. As stated above, researchers have discovered that attitudes and behaviours are

most related when both the attitude measured and behaviour revealed are concerned

69

with the same action, target, context and time based elements; the principle of compatibility (Ajzen & Fishbein, 2005).

There are many domains in which more specific (in terms of the principle of compatibility) methodologies have been utilised in order to explore the relationship between attitude and behaviour. These include (but are not limited to) the use of contraception (Kothandapani, 1971), breast feeding (Manstead, Proffitt, & Smart, 1983), illicit drug use (McMillan & Conner, 2003), and the propensity to exercise (Terry &

O'Leary, 1995). These experiments have found correlations ranging from .35 to .67, a much higher level than those found in experiments that compare general measures of attitude and specific behaviours, which are often extremely low, and/or non‐significant

(Ajzen & Fishbein, 2005).

The necessity and importance of compatibility is backed up by evidence from a meta‐analysis undertaken by Kraus (1995). In this study it was found that compatible attitudes and behaviours exhibited a correlation of .54, whereas incompatible (broad attitudes and specific behaviours) attitudes showed a correlation of .13.

4.3.2 Attitude accessibility.

Glasman & Albarracin (2006) state that people will often utilise attitudes that are held from previous experience as a basis for their behaviour in the present. It follows then that attitudes must be accessible for them to have a discernible affect on an individual.

70

Attitudes that are easier to retrieve from an individual’s memory are thought to be more likely to affect behaviour than those that are difficult to retrieve (Fazio, Powell,

& Williams, 1989; Glasman & Albarracín, 2006). Two reasons are theorised for this. The first is that attitudes that are readily accessible are easier for the individual to utilise and are more likely to be available and used as an input into behavioural decisions by the individual (Glasman & Albarracín, 2006). The second is that more accessible reasons may influence the individual’s perception of relevant information regarding the attitude object, thereby modifying behaviour. An example would be an individual that has strongly negative attitudes towards alcohol seeing a person drinking ginger beer, but assuming that it is in fact real beer and therefore judging the person in a negative fashion.

Increasing the level of attitudinal accessibility through external stimulation should therefore increase the strength of the attitude‐behaviour relationship, just as if the individual held an accessible attitude before the intervention. In this vein, researchers have found that the act of priming, or thinking carefully about a particular subject can produce particularly strong attitude‐behaviour relationships (Cacioppo,

Kao, Petty, & Rodriguez, 1986).

Similarly, repeated direct exposure to a particular attitude object, or repeated expression of an attitude that one holds has been found to increase the attitude‐ behaviour correlation (Fazio, Chen, McDonel, & Sherman, 1982). This is also thought to be because of increased attitude accessibility (Glasman & Albarracín, 2006). Attitude

71

accessibility can also be affected through direct exposure to the object or repeated

expression of an individuals’ attitude (Glasman & Albarracín, 2006).

4.3.3 Attitude stability and contextual factors

Attitude stability, or the degree to which an individual is willing to modify their attitude, is thought to affect the attitude‐behaviour relationship (Glasman & Albarracín,

2006). More stable attitudes are thought to be more related to behaviour than those that are open to change (Glasman & Albarracín, 2006). The relevance of an attitude towards a behaviour is thought to affect the relationship between the two. That is, if the

information upon which attitudes are formed is proximal to, or accurate regarding the

behaviour that is to be undertaken, than the relationship between one’s attitude and

their revealed behaviour is likely to be stronger than if the attitude is formed on non‐

proximate or relevant information (Glasman & Albarracín, 2006). This may be related to

the effect of direct experience with an attitude object, as an individual may gather more

relevant and accurate information through their experience with the attitude object,

further increasing the strength, relevance and stability of their attitudes held (Glasman

& Albarracín, 2006).

Contextual factors are also thought to affect the attitude‐behaviour relationship.

Many actions undertaken by individuals are done so in private circumstances. When

attitudes are reported in public, but undertaken in private, the attitude‐behaviour

relationship is likely to be weaker than when both are given in the same context (Kraus,

1995).

72

Another contextual factor that may affect the attitude‐behaviour relationship is

the correspondence of the type of behaviour to the type of attitude that is being utilised.

This is termed the ‘hedonic‐instrumental correspondence’ by Glassman & Albarracin

(2006), in which hedonic behaviours, or those that are concerned with enjoyment and

instrumental behaviours, or those that are driven by cognitive or practical reasoning

are compared to the type of attitude that is activated at the time. If affectively based

attitudes are activated, correspondence with hedonic behaviours is likely to be high

(Millar & Tesser, 1986). Similarly, if the attitudes activated are cognitively based or practical in nature, correspondence with instrumental behaviours is likely to be high

(Millar & Millar, 1998). The reverse of these is also true; if affectively driven attitudes are activated, correspondence with instrumental behaviours is likely to be low and vice versa (Glasman & Albarracín, 2006).

In simple terms, if one holds an attitude toward a behaviour that is concerned

with enjoyment, they are unlikely to behave in a way that will diminish that enjoyment.

In the same way, if one holds an attitude that is concerned with maximising practical

utility, it is unlikely that the individual will engage in behaviour that focuses upon the

maximisation of enjoyment (assuming enjoyment and practicability are different).

One interesting aspect of the attitude behaviour relationship lies in the ability of

external information to guide the process. If an individual is presented with one‐sided

information about an attitude object, and does not engage in significant thought about

the information, then the attitude formed from the information is likely to have a higher

73

level of correlation with behaviour than if two‐sided information is presented (Conner,

Povey, Sparks, James, & Shepherd, 2003; Conner et al., 2002).

Similarly, if an individual is confident in their attitude, then the relationship

between their attitude and their behaviour is likely to be higher than if they are not

confident (Albarracín, Wallace, & Glasman, 2004). This is thought to stem from the

instability of attitudes that are not held strongly, as individuals are more prone to

attempt to change these attitudes (Glasman & Albarracín, 2006). Attitudinal confidence

can be in turn influenced by having direct experience with the attitude object (Fazio &

Zanna, 1978), the amount of thought given to the attitude (Krishnan & Smith, 1998) and the nature of the information that one holds with regard to the attitude object (one versus two‐sided information) (Glasman & Albarracín, 2006; Prislin, Wood, & Pool,

1998).

4.4 Attitude­Behaviour Models

The research into the attitude‐behaviour relationship is vast and of high quality

(Glasman & Albarracín, 2006). Two of the more successful models that attempt to

describe the attitude‐behaviour relationship are the Theory of Reasoned Action (Ajzen

& Fishbein, 1980) and the Theory of Planned Behaviour (Ajzen, 1991). These models

have demonstrated an ability to describe the relationship between attitudes and

behaviours in the past (Cook, Moore, & Steel, 2005).

According to the TRA, behaviour is influenced by the individual’s attitude (as

discussed above), and by their subjective norms (the perceived social norms that one

74

would conform to in order to behave appropriately; Albarracín, Fishbein, Johnson, &

Muellerleile, 2001).

The TPB is an extension of the TRA (Ajzen, 1991). According to the TPB,

behaviour is directly affected by an individual’s intention to undertake the said

behaviour. This intention is in turn affected by the individual’s attitude (as discussed

above), by their subjective norms (the perceived social norms that one would conform

to in order to behave appropriately), and by the perceived behavioural control over the

situation (Ajzen, 1991; Cook et al., 2005), when perceived control can be expected to

substitute for, or correspond relatively well with actual control (Bagozzi, 1992).

Underpinning all of these relationships are the beliefs of the individual. That is,

one’s attitude, behaviour and control are thought to be influenced by ones’ beliefs about

these factors (Ajzen, 1991). As an indicative example of this, consider the following. If

one believes that safety is unimportant because of culture (norms), then they are less

likely to hold positive normative beliefs with regard to safety initiatives, and therefore

less likely to comply with the required acts within the initiative.

There is a general support for the theory in the literature as evidenced in part by

the breadth of topics and industries in which the model has been used to predict and

model the behaviour of individuals. These include, but are not limited to; speeding

(Paris & Van Den Broucke, 2008), driver behaviour (Forward, 2009), pedestrian rule violation (Moyano Dìaz, 2002), truck driver behaviour (Poulter, Chapman, Bibby,

Clarke, & Crundall, 2008), hearing protection usage (Quick et al., 2008),

environmentally friendly hotel choice by tourists (Han, Hsu, & Sheu, 2010), the use of

75

internet shopping (Crespo & del Bosque, 2008), manual handling (Johnson & Hall,

2005), HIV testing (Kakoko, Astrom, Lugoe, & Lie, 2006), why women choose to use dietary supplementation (Conner, Kirk, Cade, & Barrett, 2001), adult reading behaviours (Miesen, 2003), and recycling behaviour (Tonglet, Phillips, & Read, 2004).

A simplified illustration of the TPB is included below.

76

Figure 4. Representation of The Theory of Planned Behaviour

(adapted from Ajzen & Fishbein, 2005)

The solid arrows represent relationships in the TPB, the broken arrows represent factors that can be taken as proxies (that is, they can be treated as analogous) when they are sufficiently consistent with the factor that they are replacing. This is similar to Bandura’s (1977) self‐efficacy, which refers to an individual’s confidence in their ability to perform the intended behaviour (Bagozzi, 1992). Armitage and Conner

(2001) argue that perceived behavioural control was added to the model to account for situations in which actual volitional control was not present, such that in complex situations in which the intention to perform a behaviour was not the limiting factor, the model could still be used to predict an individual’s behaviour. In other words, when the

77

performance of a behaviour is not empirically under the direct volitional control of an individual, the perception of control that the individual held would influence behaviour.

Interestingly, in a meta‐analysis of the theory, Armitage and Conner (2001) found that intentions and self‐predictions were more closely related to behaviour than the other factors (i.e. behavioural beliefs and normative beliefs) in the TPB.

Ajzen and Fishbein (2005) point out that the TPB as depicted in Figure 4 above is somewhat simplistic, and that there are many relationships that are left out of the diagram to aid in readability. These relationships are discussed below.

The first relationship identified by them is the feedback of performance of a behaviour back into the inputs for future behaviour. That is, undertaking a certain behaviour in the past may influence the likelihood of performance of a similar behaviour in the future. This could be accounted for in the TPB diagram above by including ‘past behaviour’ in the background factors box (Ajzen & Fishbein, 2005).

The second relationship identified is concerned with the direction of the relationships in the diagram. Both attitude and subjective norms may influence one‐ another in both directions. That is, attitudes may influence both intention (forward) and behavioural beliefs (backwards). Similarly, subjective norms may influence intention

(forward) and normative beliefs (backwards) (Ajzen & Fishbein, 2005).

The third relationship is the possible inter‐correlation between attitudes, subjective norms, and perceptions of control (Ajzen & Fishbein, 2005). An example of

78

this would be a person who expects to succeed in performing a given behaviour. They

anticipate success, and therefore develop a positive attitude towards the behaviour.

The fourth factor identified is concerned with the relative weights (or

importance) of attitudes, subjective norms and perceived control in the relationship

between these and the intention to perform a specific behaviour. The relative

importance of these factors in the relationship between attitude and behaviour are thought to change dependant on external and internal factors, such as the function of the behaviour to be undertaken and the population studied (Ajzen & Fishbein, 2005).

Ajzen and Fishbein (2005) state that ones’ behaviour shall ultimately rest on the

information present and available to the individual that is relevant to the behaviour.

They also point out that although behaviour is dependent upon the presence of

information, it is not correct to assume that individuals will utilise or consciously

review each piece of information available to them in the chain of behaviour causation.

They argue that once attitudes are formed and become readily available, individuals will

not necessarily consult the building blocks of these attitude (their behavioural,

normative and control beliefs) in order to undertake a behaviour (Ajzen & Fishbein,

2005). In other words, once one has a positive attitude towards playing golf, this

attitude can be readily accessed and utilised when considering the decision of whether

or not to play a round of golf on the weekend.

79

4.5 Critiques of and Additions to the TPB

The TPB is one of the more popular and enduring theories to attempt to explain

the attitude‐behaviour relationship, however, the theory has not been without criticism

since its inception. Some criticism has been directed towards the theory due to a

perceived lack of explanative power and causal explanation (Cook et al., 2005), others

have suggested modifications that they perceive necessary to increase this explanative

power (Bagozzi, 1992). At the more extreme end of the spectrum some researchers have even questioned the utility and validity of the TPB, and whether the theory increases the understanding of the attitude‐behaviour relationship beyond common‐ sense presuppositions (Cook et al., 2005).

Another criticism of the TPB is the lack of a causal link between evaluation of an

attitude object, and the intentions concerned with said object (Bagozzi, 1992). Bagozzi

hypothesises an emotive self‐regulatory process is involved with this link. That is, the

link between evaluation and intention is governed by the emotional response to the

likely or actual outcomes of performing a behaviour.

Another criticism of the TPB is the use of self‐report, which is prevalent in

attitude‐behaviour research (Armitage & Conner, 2001; Iverson, 2004; Rothengatter,

2002) as it is perceived by some to be unreliable. This is explored in further detail

below.

A common criticism of the TPB (and the TRA by way of the fact that the TPB is

based upon it) is that the link between intention and behaviour is assumed, but not

80

explored and directly attributed (Ajzen & Fishbein, 2005; Armitage & Conner, 2001;

Bagozzi, 1992; Cook et al., 2005). That is, the major theories of the attitude‐behaviour

relation do not explore how attitudes and intentions are related, or the conditions

under which attitudes may trigger intention (or vice‐versa) (Bagozzi, 1992).

Some research in the TPB paradigm has been criticised for not utilising measures that clearly and explicitly measure intention, as it is defined in the TPB (Armitage &

Conner, 2001). This forms the basis of another criticism of TPB, that the factor of

‘intention’ is not a fixed factor, and its meaning is changeable between the theoretical

definition, and that measured in experiments (Cook et al., 2005). In other words,

intention is thought of as the motivation behind behaviour in the TPB, whereas

intention, as measured in TPB based experiments, is generally an account by the

individual as to what they intend to do in a given situation (Cook et al., 2005). This

difference appears subtle, but is not. Intention, as measured in experiments, is often an

account by the individual of their planned behaviours, based upon their beliefs and

attitudes. This is not an account of a motivational factor, but a plan of action. Therefore,

it is likely that a relationship should be found between attitudes, beliefs and intentions to act, as the individual is effectively providing a linked plan.

Some researchers have questioned the utility of the TPB, due to the lack of

evidence that it increases the understanding of the attitude‐behaviour relationship

beyond common‐sense presuppositions (Ajzen & Fishbein, 1980; Cook et al., 2005;

Greve, 2001). Moreover, given that individuals are generally reporting their intended

behaviours (not intentions as defined in the TPB – they are motivators for action, not

81

actions themselves) it makes common sense that these would be related to the salient

information to individuals, their attitudes, beliefs and social norms (Cook et al., 2005).

Indeed, the authors of the TRA give a similar appraisal of the theory (Ajzen & Fishbein,

1980, p. 6);

“So far we have said little that does not conform to common sense, but even at

this simple level our analysis raises some interesting questions”

Researchers have also questioned the TPB’s ability to predict and describe causal relationships (Cook et al., 2005). Cook et al., (2005) have proposed three rules for proof of causality, which were adapted from earlier work by Hume (see Goldvarg & Johnson‐

Laird, 2001). The rules are as follows

1. Given that A and B are present, A should precede B in time

2. Both A and B should be observed and documented

3. A and B have a necessary connection

Cook et al (2005) state that whilst the TPB meets the criteria of establishing the

necessary connections between factors, the temporal order of the attitude‐behaviour

relation is not proven. That is beliefs do not necessarily precede attitudes in time. They

go on to say that the TPB does not meet the requirements of rule 2 above, because of a

lack of evidence that modification of beliefs in experimental situations will alter

attitudes in the experiments (Cook et al., 2005).

82

Of the three rules above, the TPB generally satisfies the third, and Cook et al.

(2005) ascribe this as one of the reasons behind the popularity and endurance of the theory. However, popularity it is argued, does not lead to causality. Cook et al. (2005) ultimately condemn the theory with this statement

“In short, despite good statistical results the TPB is not aligned to the phenomenon it purports to model” (Cook et al., 2005, p. 149)

4.6 Summary of the Attitude Behaviour Models

The attitude behaviour relationship has had significant attention in the academic literature. Many models have been proposed, although the most enduring and popular of these have been the Theory of Reasoned Action, and the Theory of Planned

Behaviour. These models, as discussed above, illustrate that behaviour is governed by beliefs, norms, attitudes and perceived control. The theories, despite their popularity, have not gained absolute acceptance in the academic community, mostly due to concerns regarding their efficacy, and accuracy with regard to the ability to model actual behaviour (Cook et al., 2005).

4.7 Risk Perception and Behaviour

Risk perception has not attracted the amount of attention and hence literature as is evident in the attitude‐behaviour relationship. Remember risk perception is defined as the recognition of the risk inherent in a particular activity (Hunter, 2002). Generally, risk perception is thought to be a precursor of behaviour, and is thought to have a

83

negative relationship with behaviour (Machin & Sankey, 2008). That is, a higher level of

risk perception will lead to a lower probability of an individual carrying out a particular

behaviour. Another common presumption is that poor risk perception, or the

misinterpretation of actual level of risk by individuals, will consequently lead to

inappropriate behaviour by the individual (Arezes & Miguel, 2008)

Rundmo (1996) suggests that there are three possible ways in which risk

perception may affect performance and accident causation (and vice versa) in the

workplace. First, and perhaps the most obvious, is that accidents are thought to affect

an individual’s risk perceptions. This may be because those that have experienced accidents have greater knowledge concerning the hazards and subsequent risks associated with the activity under examination than those that have not. Second, risk

perceptions are thought to be a possible causal factor in accidents. This is thought to be

related to the increased workload and strain experienced by those that perceive their

risk exposure in a given activity as higher, or those that generally perceive risks in a

more sensitive fashion. The third option is that risk perception and accidents are

independent of each other (Rundmo, 1996).

There are, to the knowledge of the author, few models of the relationship

between risk perception and behaviour in the literature. Brewer Weinstein, Cuite and

Herrington (2004) argue that there are three possible risk perception/risk behaviour

hypotheses. The first is known as the ‘accuracy hypothesis’. This hypothesis holds that

an individual that engages in a higher level of actual risk (actual risk is not defined)

should logically also have a higher level of perceived risk. Therefore, an individual that

84

was undertaking a skydive should perceive a higher level of personal risk than a person

sitting on their couch at home, because of the logical and measurable difference in risk

level that is evident. In other words, risk perceptions are based on the accuracy of the

appraisal of risk that is undertaken by the individual. Brewer et al. (2004) do point out

that this hypothesis is descriptive in nature, and therefore says little or nothing about

the causality of any relationship.

The second hypothesis is termed the ‘behaviour motivation hypothesis’. This

hypothesis holds that risk perception at time ‘A’, will influence risk behaviour at time

‘A+n’. This relationship is thought to be causal and in a negative direction, that is, a high

level of perceived risk shall produce a lower level of risk undertaken by the individual in

their behaviour (Brewer et al., 2004). This hypothesis is effectively saying that

individuals are motivated to behave so as to satisfy their optimal or preferred level of perceived risk.

The third hypothesis, the ‘risk appraisal hypothesis’ holds that changes in

behaviour in response to risk perceptions will change the subsequent risk perceptions

held by the individual. If an individual takes action to reduce their perceived risk, then

their risk perception after this action shall be reduced. It is important to note that

anticipation of behaviour is also thought to modify the risk perceptions of the individual

(Brewer et al., 2004). The three hypothesis presented by Brewer et al., are shown in

Figure 5 below.

85

Figure 5. Three Risk Perception Hypotheses

(adapted from Brewer et al., 2004)

As can be seen in the model above, there are three main paths through which risk perception is thought to modify behaviour. The accuracy hypothesis will contribute to an affect on behaviour at the time in which the risk perception occurs. That is, once an individual perceives a risk, they are likely to consider their behaviour immediately.

The behaviour motivation hypothesis posits that risk perception at time one, will influence behaviour at time two, and the risk re‐appraisal hypothesis posits that behaviour will modify risk perceptions at the time of the behaviour.

86

At least one model is available in the road safety literature that describes the

influence of risk perception on behaviour. In the model outlined by Deery (1999), given

below, risk perception is a component of the relationship between hazard identification

and behaviour, rather than an exclusive antecedent to behaviour.

In the model, risk perception has two inputs, and a single output. The first, and

presumably major input, is hazard perception. As previously discussed in section 2.3.1

above, hazard perception is the identification of physical object/s and or environmental circumstances that have the potential to be hazardous, and correctly perceiving them as

posing a hazard (Barowsky et al., 2010; Brown & Groeger, 1988). The second input to

risk perception is the self‐assessment of skills by the individual.

The output from the individuals’ risk perception is risk acceptance (or non‐

acceptance), which leads directly to the behaviour that is undertaken by the individual.

87

Figure 6. The Role of Risk Perception in Behaviour in Road Safety Literature

(Adapted from Deery, 1999).

Despite the lack of a popular model describing the relationship between risk perception and behaviour, there is some research available that examines the link between risk perception and research in various contexts.

In the context of high risk workplaces, specifically on board oil rigs in the North

Sea, Rundmo (1996) found that there was no statistically significant relationship evident between risk perceptions and behaviours. It is prudent to mention though, that the measurement instrument used by Rundmo was concerned with generic risk perceptions of individuals. That is, participants were asked to rate their perceptions of the risk of injury from a wide variety of sources, for instance, the perceived risk from

88

explosion. This is in comparison to asking for their perception of the risk from a very specific and finite situation, such as the risk of an explosion during gas ventilation procedures. Therefore, it would appear that Rundmo found that generic risk perceptions have little affect on specific risk behaviours in high‐risk workplaces.

In contrast, Arezes and Miguel (2008) found that risk perception associated with the risks surrounding exposure to high ambient noise levels affected the behaviour of individuals in that those with a higher perception of these risks were more likely to wear hearing protection.

In the aviation context, Hunter (2002) found some support for the hypothesis that the relationship between risk perception and behaviour is negatively correlated

(higher level of perceived risk is related to a safer behaviour) in General Aviation in the

United States. In his experiment, Hunter (2002) asked pilots about their risk perceptions of specific aviation‐based situations that are likely to be faced by pilots, or experienced by colleagues at some point during their flying careers. He then compared their risk perceptions to their tolerance of similar risk, as measured through the use of his Hazardous Events Scale, which measured the number of times in the preceding two years that a pilot had been exposed to (either from personal action, or external factor) a hazardous situation. Therefore, Hunter effectively compared individuals’ risk perceptions to their past risk‐related behaviour.

Hunter found that for weather related scenarios, a higher level of perceived risk lead to a lower level of risk tolerance by pilots. In other words, those that perceived higher risks would behave in a less risky fashion than those that perceived a lower level

89

of risk (Hunter, 2002). Interestingly though, research in Australian GA has found some divergent evidence. Molesworth and Chang (2009) found that as the perception of risk of general life activity (like driving a car or riding in an elevator) was positively related to risky flight behaviour (r=0.35, p=0.47). Therefore, those that perceived more risk in everyday life, were also more prepared to undertake risk in their aviation behaviour.

The differences between the two studies may be explained in part by the two different measures of pilot behaviour. In contrast to Hunter, Molesworth and Chang examined pilots’ behaviour through the use of a flight simulator opposed to self‐reported experience with hazardous events.

One area of research that is relatively common in the road safety arena is the efficacy of enforcement on behaviour. Enforcement at its basic level is designed to discourage illegal behaviour by creating a risk that individuals will be punished.

Therefore, enforcement is essentially a case of risk perception influencing behaviour.

Research in the area has found that speed enforcement is related to a reduction in mean speeds and a reduction in the number of serious accidents (Ryeng, 2012). This area of study is less relevant to GA, as systematic enforcement of a single behaviour (like speeding) is extremely uncommon. Pilot behaviour is scrutinised through other means in the GA environment, from peer reviews through to intervention from air traffic control, should they see any major deviation from safe operating practice. Therefore, enforcement related perceptions were not further included in the current research.

Given the divergence in findings in different contexts, the relationship between risk perception and behaviour remains a field in which more research is required.

90

4.7.1 Summary of risk perception

The relationship between risk perception and behaviour has not gathered the level of attention in the literature, as has the attitude‐behaviour relationship. In aviation, the general hypothesised relationship between risk perception and behaviour is a negative causal one. That is, as risk perception increases in magnitude, risk taking behaviour decreases.

4.8 Summary

The attitude‐behaviour and risk perception‐behaviour relationships have both gathered some attention in the academic literature. Of the two however, the attitude‐ behaviour relationship has by far the greater amount and variety of literature available.

Attitudes are thought to have a positive causal relationship with behaviour. That

is, the direction (positive or negative) of the attitude held is thought to influence behaviour in that direction. Therefore, a negative attitude towards speeding, may lead to a decreased likelihood of the person speeding in their car.

Many models to describe the attitude‐behaviour relationship have been

proposed, the most enduring and popular of these has been the Theory of Planned

Behaviour. The model posits that behaviour is governed by beliefs, norms, attitudes and

perceived control.

Similar to attitude, risk perception is thought to have a causal relationship with

behaviour. In the case of risk perception though, it is thought to be a negative

91

relationship. That is, as the perception of a risk increases in its magnitude, it is less

likely that the individual will behave such that they will be exposed to that risk. In other

words, if an individual perceives the risk of being caught speeding in their car as

extremely high, then they are less likely to speed than if they perceive the risk as

minimal.

One point of note is that the popular models of attitude‐behaviour and risk

perception‐behaviour (the models discussed above are included in this group) are

relatively exclusive of one another. That is, attitudes have rarely been recognised as a significant causal component of the linkage/s between risk perception and behaviour and vice‐versa (Finucane, Alhakami, Slovic, & Johnson, 2000). This has therefore

resulted in the literature ascribing the constructs of attitude and risk perception separate roles in the link to behaviour. It would be interesting to test if these two

constructs were indeed separate, and impacted behaviour in a separate fashion, as is

the de‐facto position of the popular literature, or if they are related and interdependent.

This will be tested in the experiments below.

92

5 Methodological Overview

5.1 Introduction

The current study employed three experiments in order to answer the proposed research questions, which were. The aim of the present research is to investigate if:

1. There is a relationship between attitudes, risk perceptions, and experiential

values (flight hours, recency, and age)

2. There is a relationship between the attitudes, risk perception and the behaviour

of pilots in the General Aviation training sector

3. There is a relationship between a pilot’s experience, as measured by age, flight

hours, and recent flight time, and behaviour in the general aviation training

sector

This multi‐experimental process was designed to provide data from participants

in three different manners. First, experiential data, attitudinal data and risk perception

data was collected utilising pen and paper surveys. Second, self‐reported behaviour in a

variety of risky situations ranging from low to high‐risk was gathered utilising

specifically constructed surveys. Thirdly, actual flight behaviour was observed utilising

a flight simulator. This experiment was augmented with a post simulation interview in

order to ascertain the participant’s intentions, decision‐making style and hazard

awareness during the experiment.

93

In lay‐person's terms, the methodology was designed as a three‐part process in which participants would state their thoughts about safety, and then reveal what they thought that they would do in a risky situation, or alternately illustrate what they would do in a risky situation.

5.2 Background to the Methodology

The current study sought to measure three constructs; (1) Individuals’ attitudes towards safety in aviation settings; (2), individuals’ perceptions of risk in aviation settings; and finally (3) the relationship between attitudes, risk perceptions, experiential factors and an individuals’ behaviour (both observed and self‐reported) in aviation settings. Each of these constructs were measured using separate methods, as discussed below.

When compared to the literature on the attitude‐behaviour link, there is relatively little information regarding the link between experiential variables and behaviour in the GA sector. Of ten factors identified that potentially modify behaviour in a review of the literature, only one was experiential – age; further, the literature in the area appears to be contradictory in findings (Drinkwater & Molesworth, 2010). There is a relative dearth of literature in this area, and therefore the current research will assess the impact of experiential factors upon behaviour in low medium and high‐risk aviation scenarios.

Attitudes are thought to have a positive causal relationship with behaviour. That is, the direction (positive or negative) of the attitude held is thought to influence

94

behaviour in that direction. The current research will explore this assumption in the GA sector of Australian aviation.

Work within the field of the attitude ‐ behaviour relationship has traditionally utilised self‐report style methodologies to arrive at conclusions regarding the relationship between constructs (Armitage & Conner, 2001). ‘Attitudes’ have typically been determined using attitudinal inventories or questionnaires that are administered in pen and paper format. This tradition has been followed in the aviation based academic field, with many attitudinally‐based experiments utilising pen and paper (or computer based, pen and paper style) inventories in order to measure individuals’ attitude/s in an empirical fashion (i.e. Hunter, 2002, 2005, 2006; Molesworth & Chang,

2009).

Similar to attitude, risk perception is thought to have a causal relationship with behaviour. Risk perception, in much the same way as attitudes, has traditionally been measured using self‐report methodologies. In the aviation industry this tradition has also been followed, with research into the risk perceptions of those involved in aviation relying upon the use of self‐report surveys in order to gather data on risk perceptions

(Hunter, 2002). As above, these surveys are typically pen‐and‐paper exercises. They will usually utilise fictional aviation scenarios in order to gather individuals’ risk perceptions as measured by a rating given by participants.

In the attitude‐behaviour field, the use of behaviour as an experimental variable

(self‐report and actual) in peer‐reviewed papers is surprisingly rare (Armitage &

Conner, 2001; Paris & Van Den Broucke, 2008). This is less than ideal; conclusions

95

cannot be drawn if measurements are not taken. One meta‐analysis undertaken by

Glassman and Albarracín (2006) in which only studies that measured attitude and overt behaviour (as opposed to intended behaviours) were included, revealed only 29 suitable research publications from 41 studies. This was distilled from in excess of

25,000 citations initially found.

Those studies that do use behaviour as a variable, often use self‐reports of

behaviour (i.e. Iverson, 2004; Ulleberg & Rundmo, 2003), or alternately, some use self‐

reports of the intention to behave as a variable (Armitage & Conner, 2001; Paris & Van

Den Broucke, 2008). This self‐reported behaviour information is then compared to

attitudinal or risk perception data and relationships are typically examined using

correlations, regressions or ANOVAs.

This style of methodology has been used extensively in the attitude‐behaviour

literature, and will likely continue to be used as it is one of the few ways that

experimenters can ascertain from individuals their thoughts, attitudes and desires. In

plain terms, only the individual knows what they are thinking, one must therefore ask

the individual for that information.

It is acknowledged that this style of methodology, and other studies that are reliant solely upon self‐reporting for data acquisition may suffer from a range of problems, which broadly fall under the area of recall inaccuracies (in the case of factual responses or those related to behaviour) or reporting inaccuracy (in the case of factual response and for attitudes and perceptions) (Hatfield, Fernandes, Fuance, & Job, 2008).

More precisely, self‐reports may suffer from distortion through response biases which

96

include phenomena such as acquiescence, extreme or moderate responding, negative

affectivity bias, socially desirable reporting, the perception of the researcher’s expectations and participants’ self‐image protection (Hatfield et al., 2008; Razavi, 2001;

Rothengatter, 2002).

In recent research, the problems surrounding the use of self‐report have been

addressed with the use of an Implicit Association Test (IAT) (Greenwald, McGhee, &

Schwartz, 1998) in which individuals’ attitudes are inferred by the difference between

response times given when compatible (in the sense of the evaluation direction

[positive or negative]) stimuli, compared to those that are not compatible. For example,

it is theorised that the response given by an individual for the stimuli of a picture of an

apple, with the word ‘nice’, would be faster than that shown if the picture of the apple

was paired with the word ‘death’ (Greenwald et al., 1998). The IAT has not however

gained full acceptance in the academic community with regards to its ability to measure

attitude (Arkes & Tetlock, 2004; Blanton, Jaccard, Gonzales, & Christe, 2006; Chang &

Mitchell, 2009; Gawronski, Lebel, & Peters, 2007). It is for this reason that the IAT is not

used in the current research.

In order to address the potential limitation with self‐report data when concerned

with behavioural survey, the current research utilised two forms of data capture to

gather data about participants’ behaviour – a survey in which participants indicated

their preferred options for behaviour in a given scenario, and actual behaviour captured

through a flight simulator. This is similar in design to Paris and Van den Broucke’s

97

(2008) experiment in the road environment, in which attitudes were related to actual behaviours as measured through GPS tracking devices fitted to cars.

The use of two forms of data capture for behaviour served two main purposes.

First, the use of two types of data capture was thought to allow for a more accurate understanding of the situation with regards to any relationships present. This is

because this type of methodology allows for comparison between what people report

that they will do (or what they think, i.e. attitudes), with their behaviour.

Second, self‐report style surveys allow for significantly more data to be gathered

in a shorter period of time than the use of a flight simulation. The methodology

therefore, was designed to provide as much data for the research as possible, given the

inherent restraints. This allowed for the most accurate picture of the situation to be

found with regards to the relationship between attitude, risk perception, experience

and behaviour.

Further, the proposed methodology was also designed to address a common

shortcoming in the area of risk‐taking within the field of aviation. Specifically, prior

research has focused on examining risk management behaviour through the use of a single scenario (see, for example Goh & Wiegmann, 2001; Molesworth et al., 2006;

O'Hare & Smitheram, 1995). By relying on one scenario, researchers appear to assume that risk management is constant across all situations, irrespective of the type or number of hazards presented and the subsequent risks involved in the situation. As discussed in Chapter 1 above, effective risk management involves hazard identification

and risk assessment (Drinkwater & Molesworth, 2010). Therefore, in order to examine

98

if predictors of risk management behaviour varies based on the hazards and risks involved in a scenario, the present research employed three uniquely different risky scenarios.

As a secondary outcome, the simulation data was to be compared to the self‐ report surveys, such that any differences in the two forms of data gathering were revealed. This has the benefit of providing evidence to the academic community of the appropriateness or otherwise of the use of self‐report methodologies in an aviation context.

All participants undertook Experiment One. This group was then split into two non‐equal groups to undertake Experiments Two and Three. The reason for this split was due to participants’ ability to engage in a simulated flight at a later date. If participants could not undertake the secondary flight simulation (Experiment Three), then they were put into Experiment Two. If they indicated that they were willing to undertake a flight simulation, then they were put into Experiment Three. In this way, group numbers were not directly controlled by the experimenter. Nor was there any other control exerted by the experimenter in terms of selection for different experiments

While the benefits of the present approach have been highlighted, it is not without its limitations. Specifically, the present study utilises a flight simulator to examine pilot behaviour. While this approach is not uncommon (Drinkwater &

Molesworth, 2010; Goh & Wiegmann, 2001; Molesworth & Chang, 2009; Molesworth et

99

al., 2006), and from a safety and ethical perspective has merit, any conclusions drawn

from the research need to be interpreted within context.

5.3 Summary

In summary, the current study utilised multiple methodologies that have been

used previously in academic research so as to provide a detailed, methodologically

sound and accurate examination of the topic under investigation. While there are a number of benefits to the chosen method (i.e. the conclusions drawn come from both self‐reported and revealed behaviour in a simulation), any conclusions drawn from the research need to be interpreted within the appropriate context – that is, the study utilised a specific subset of the aviation population, and was reliant upon self‐reports of behaviour and the use of a simulated flight.

The present study consisted of three experiments. The first was designed to elicit

the attitudes towards aviation safety, the risk perceptions of aviation centric scenarios

and the experience of the participants. The second was designed to compare the

attitudes, risk perceptions and experience factors to self‐reported behaviour in a variety

of risky situations. The third and final experiment was designed to compare the attitude,

risk perceptions and experience to actual behaviour as measured in a flight simulation.

In simple terms, the methodology was designed as a three‐part process in which

participants would state their thoughts about and perceptions of safety, reveal what

they thought that they would do in a risky situation, or illustrate what they would do in

a risky situation.

100

6 Experiment One ­ Risk Perceptions and Attitudes

of Pilots

6.1 Introduction

Experiment One was designed to provide two distinct sets of information. First, the experiment was designed to identify links between a pilot’s risk perception, attitude

(towards safe flight operation), age, and their aviation related experience in order to determine the existence or extent of the relationship with one another. Second, the experiment was designed to reveal the risk perceptions and attitudes of the pilots in the study in order to allow for a comparison with behavioural measures (see Experiment

Two and Three).

Pilots’ attitude (towards aviation safety) was measured using Hunter’s Aviation

Safety Attitude Scale (ASAS). This scale is one of very few that is specifically designed to measure attitude in an aviation‐centric setting. In the current research, attitude is defined as the product of what one thinks of, and how one feels towards a certain object or thing.

Therefore the attitude object in the ASAS relates to their own skill or ability.

Pilots’ risk perception was measured using Hunter’s Risk Perception Scale 1 and

2. All scales have been used previously within the general aviation industry (Drinkwater

& Molesworth, 2010; Hunter, 1995, 2002, 2006; Molesworth & Chang, 2009), albeit mostly in the United States. Based on past studies examining cultural differences in aviation (see Hutchins, Holder, & Pérez, 2002 for a review), it is conceivable that the

101

populations will differ to some degree as a result of the different operating

environments and different curriculum that is studied as part of the licensing

requirement for an Australian pilot’s licence.

Previous research in Australian GA that utilised the ASAS (as used in the current

research) has uncovered a positive relationship between flight hours and self‐

confidence, such that a pilot with more hours is likely to exhibit more self‐confidence

(Molesworth & Chang, 2009). Hunter’s (2006) findings give mixed conclusions as to the

likely relationship between flight hours and the risk perceptions of US based pilots.

Hunter’s research found a negative relationship between flight hours and the majority

of the risk perception factors utilised (the relationship between flight hours and

Altitude Risk, Driving Risk and Everyday Risk were negative, but not statistically

significant) (Hunter, 2006). Therefore, the more hours a pilot has, the lower that they

were likely to perceive risks in these particular areas. This finding was not replicated

however, in a recent study that utilised Australian general aviation pilots (Molesworth

& Chang, 2009).

Of note is that the two risk perception scales measure slightly different perceptions. Risk Perception Scale One measures the perception of risk when another pilot is involved in the situation. Risk Perception Scale Two measures the perception of of risk when the individual themselves is involved in the situation.

Previous research into the relationship between age and attitudes has found

generally consistent results; that age appears to be related to holding safer attitudes

(Deery, 1999; Siu, Phillips, & Leung, 2003). The expected relationship was therefore

102

that older pilots would exhibit ‘better’ (more conservative) attitudes towards safe

operations. The relationship between age and risk perception, on the other hand, has

not followed a clear pattern in the literature. Hunter’s (2002) research suggests that age

is related negatively to risk perception, such that older pilots perceive the risks in

aviation as lower when compared to their younger counterparts. On the other hand,

research in road safety has found that younger drivers are likely to perceive the risks in

a driving situation as lower (Deery, 1999).

In the aviation industry, a pilot must satisfy minimum standards in terms of

flight hours and tested proficiency in a discrete period of time for them to be allowed to

be in command of an aircraft. This is called recency, and is typically measured in terms

of take‐offs and landings completed in the past 90 days in the GA sector (CASA, 2008).

Recency is therefore seen as being important in aviation from a safety perspective. But

it may also prove important from an attitudinal perspective. Having flown recently (in

terms of hours) will likely provide participants with a greater amount of recent direct

experience of the experiments attitudinal object, which was expected to make the

attitudes associated with flying more accessible, and might therefore affect their

attitudes (Glasman & Albarracín, 2006). However, in aviation specific research, recency

was not found to be related to attitudes (Hunter, 2006; Molesworth & Chang, 2009).

Since there is much debate about the importance of recency as a predictor of pilots’ risk taking behaviour, the present research will investigate whether there is a link between recency and attitude or recency and risk perception

103

In sum, the relationships found in the existing literature between experiential factors and attitudes and risk perceptions were as follows:

• As age increases, attitudes are expected to become more conservative (lower level

of agreement with the attitudinal scale), and the relative risk perceptions

exhibited are expected to be higher (in line with aviation research).

• As flight hours increase, participants are expected to display less conservative

attitudes, and a lower perception of risks (in the US)

• Recent flight hours were not related to attitudes and risk perceptions

The present research will examine the effect/s of experience (as measured by total flight hours, recent flight hours, and age) on attitude and risk perception in this cohort, to test if it is similar to the previously uncovered relationships from both aviation and wider attitudinal research.

As above, there were two intended outcomes from Experiment One. The first was the revelation as to whether there exists a relationship between the common experiential factors used as discriminators in aviation (total flight hours, recent flight hours, and age), and pilots’ attitudes and risk perceptions. The second was to the production of a set of baseline data that could be compared against actual behaviour to test for the relationship between attitudes, risk perceptions and behaviour (in

Experiment Three).

104

In order to satisfy the intended outcomes of the experiment, Experiment One

utilised two research questions.

• What is the relationship between attitude and risk perception of pilots in the

general aviation training sector?

• What is the relationship between experience, as measured by total hours flight

experience, a pilot’s level of recency, and pilot age, and the attitudes and risk

perceptions of pilots in the general aviation training sector?

6.2 Method

6.2.1 Measures

Experiment One was conducted utilising three separate pen‐and‐paper

questionnaires. These were comprised of one demographic survey, one attitudinal

survey, and two risk perception surveys.

The first questionnaire (pilot demographics and experience) was concerned with

background information regarding participants, age, gender, and flight experience

including total flight time, and number of hours flown in past 90 days (the measure for flight recency in the current research).

105

The second questionnaire was concerned participants’ attitude towards safe

flight. Specifically, Hunter’s (1995) Aviation Safety Attitude Scale (ASAS) was employed.

This scale has been used previously within general aviation.

The ASAS consists of 26 short statements and is designed to measure the attitude of pilots towards aviation safety. Participants were asked to rate, according to the level

in which they agree, or lack thereof, as the case may be, on a five point Likert scale

ranging from 1 (strongly disagree,) to 5 (strongly agree). Examples of statements in the

ASAS include:

“I am very skilful on controls”.

“I am so careful that I will never have an accident”,

“I would duck below minimas to get home”, and

“The rules controlling flying are much too strict”

Three factors have been identified within this scale: ‘Self Confidence’. ‘Risk

Orientation’, and ‘Safety Orientation’. A measure of internal consistency reliability was

computed by Hunter (2005) for each of the factors. These were .76, .59, and .40,

respectively. It is acknowledged that the values for the reliability tests are very low, and

this may mean that the scale is measuring something other than the factors attributed

them. The scale has been used previously in aviation research however (Drinkwater &

Molesworth, 2010; Hunter, 2005; Molesworth & Chang, 2009), and was used here

because of the previous use and acceptance of the scale.

106

‘Self Confidence’ consists of 14 statements, which deal with a participants’

opinion of their ability and their confidence in their ability to operate an aircraft safely

(i.e., “I am a very capable pilot”). ‘Risk Orientation’ consists of 8 statements. These are

concerned with a pilots’ attitude towards and appraisal of the inherent risk in aviation

(i.e., “I would duck below minimums to get home”). ‘Safety Orientation’ consists of 4

statements. These are concerned with a pilots’ attitude towards their behaviour and

abilities in aviation centric situations (i.e., “I am a very careful pilot”).

The third and forth questionnaires were the two risk‐perception survey

measures, Hunter’s (2002) Risk Perception Scale 1 and 2.

Risk Perception Scale 1 consists of 17 short scenarios in which pilots are asked

to rate the risk to a third party pilot. Ratings were given by pilots on a scale of 1 to 100.

1 being indicative of almost zero risk undertaken, like sitting in front of the television at home, 100 being indicative of extremely high risk‐taking, akin to an almost certain

chance of death.

Three factors have been identified in this scale – ‘Delayed Risk’, ‘Nominal Risk’

and ‘Immediate High Risk’. A measure of internal consistency reliability was computed

by Hunter (2006) for each of the factors. These reliability values were .81, .75, and .32

respectively. The relatively low value for ‘Immediate High Risk’ indicates that the

scenarios presented that are assumed to be measuring the perception of Immediate

High Risk, may in fact be measuring constructs other than this (Hunter, 2006).

107

The ‘Delayed Risk’ factor is comprised of 8 scenarios. These scenarios are characterised by being high‐risk situations that do not involve significant time pressures (i.e., “It is late afternoon and the VFR pilot is flying west into the setting sun.

For the last hour, the visibility has been steadily decreasing, however this arrival airport remains VFR, with 4 miles visibility and haze. This is a busy uncontrolled airfield with a single East‐West runway. He decides to do a straight‐in approach”).

The ‘Nominal Risk’ factor consists of 5 scenarios. The scenarios are indicative of normal operations, and therefore analogous to nominal risks (i.e., A pilot is cruising in good weather to a destination airport about an hour away. It is midday, and there are three hours of fuel on board”).

The ‘Immediate High Risk’ factor consists of the 4 remaining scenarios. This risk perception factor is characterised by being both a high‐risk scenario, and also having significant time pressures involved (i.e., “Low ceilings obscure the tops of the mountains, but the pilot thinks that he can see through the pass to clear sky on the other side of the mountain ridges. He starts up the wide valley that gradually gets narrower.

As he approaches the pass he notices that he occasionally loses sight of the blue sky on the other side. He drops down closer to the road leading through the pass and presses on. As he goes through the pass, the ceiling continues to drop and he finds himself suddenly in the clouds. He holds his heading and altitude and hopes for the best”).

Risk Perception Scale 2 consists of 26 scenarios. In this measure, pilots are asked to rate the risk of the situation if they themselves were involved. Ratings were given by pilots on a scale of 1 to 100. 1 being indicative of almost zero risk undertaken, like

108

sitting in front of the television at home, 100 being indicative of extremely high risk

taking, akin to an almost certain chance of death

Five underlying factors have been identified in this scale, ‘General Flight Risk’,

‘High Flight Risk’, ‘Altitude Risk’, ‘Driving Risk’, and ‘Everyday Risk’. A measure of

internal consistency reliability was computed by Hunter (2006) for each of the factors.

These were .93, .87, .87, .79, and .63, respectively.

‘General Flight Risk’, consists of 10 items that are descriptive of normal and

higher‐risk flight operations (i.e.,” During the daytime, fly from your local airport to

another airport about 150 miles away, in clear weather, in a well‐maintained aircraft“)

The second factor, ‘High Flight Risk’, consists of 10 items. These are concerned

with high‐risk flight conditions (i.e., “Fly in clear air at 6,500 feet between two

thunderstorms about 25 miles apart”)

The third factor was ‘Altitude Risk’. This factor consists of 7 items in which

altitude was the dominant risk factor (i.e., “Fly across a large lake or inlet at 500 feet

above ground level.”)

The fourth factor was ‘Driving Risk’. This factor consists of 3 items that are

concerned with driving a motor vehicle in different situations (i.e., “Drive your car on a freeway near your home during the day, at 65 MPH in moderate traffic” [changed to 110

KPH for Australian students])

109

The final factor is ‘Everyday Risk’. This factor consists of 4 items that are

concerned with normal and everyday life situations (i.e., “Ride an elevator from the

ground floor to the 25th floor of an office building”)

Of the 26 scenarios in the Risk Perception Scale 2, seven are non‐aviation scenarios

As has been said above, Experiment One was designed to gather data and then

compare the possible permutations between the variables (attitudes, risk perceptions

and experiential factors), in order to reveal any possible relationship.

6.2.2 Participants.

One hundred and seven participants were recruited from various flight schools

located at the Bankstown and Camden aerodromes in Sydney, Australia. These included

current students in the Bachelor of Aviation (Flying) degree at the University of New

South Wales, graduates of the program, and from other smaller flight training schools.

Students of the University were recruited through a briefing given in class sessions.

Those participants that were graduates, or those pilots from private flight schools were

recruited by individual conversations and briefing sessions with the researcher at their

current flight school.

All participants were aware that the research was focused on the perceptions,

attitudes and behaviours of general aviation pilots, but were unaware of any further

details of the research project’s purpose. Table 2 below outlines the demographic and

experiential statistics of the participant population.

110

The participants were largely recruited from schools dealing with ab‐initio training, or specialising in school‐leaver course. Therefore, the cohort in this study is not representative of the wider aviation population, but of pilots attempting to gain employment as a pilot. No inducements were offered for undertaking the research. The total time to complete the research was approximately 30 minutes

Table 2. Experiment One Participant Statistics

Participants Participant Age in Participant Flight Flight Hours in the Last 90

years (SD) Hours (SD) Days (SD)

107 22.46 (9.58) 308.22 (1276.45) 20.95 (27.27)

There were 18 female pilots in Experiment One. This included 20 private pilots,

31 Commercial pilots, and the remainder not holding a specific licence. There were seven pilots that held instructor ratings, and nine with instrument ratings.

111

6.2.3 Procedure.

Participants were informed of the nature of the experimental process verbally, and in writing from the information sheet provided them (See Appendix 1). Participants

were briefed that the experiment was part one of a two‐part experimental process, and

that participants would be required to commit to a second session (Experiment Two or

Three).

Participants were asked to complete a consent form prior to being issued with their surveys (See Appendix 1). Upon completion of the consent form, participants were

given a unique participant number in order to remove the issue of identification of

participants from survey data. Participant details were also collected to allow for the

arrangement of the participant’s participation in the second session of the experimental

process.

Participants were then issued with a booklet containing the demographics

questions and the surveys. This was issued as a single booklet so as to ensure that

participants completed the survey in one session, and also to reduce the chances of lost

data through misplaced survey sheets. Once complete, the surveys were returned to the

researcher directly. The survey data was then entered into SPSS version 13 for

Macintosh manually. The research was approved in advance by the UNSW Ethics

Committee.

112

6.2.4 Data analysis.

The main objective of the experiment was to determine the relationship between

pilot demographic measures and the attitudes and risk perceptions of pilots. The mean

scores for each of the attitude and risk perception factors are shown in 6.3 below. For

the risk perception scales (maximum of 100 ‐ range 0 ‐ 100) and the attitudinal scale

(maximum of 5 – range 1‐ 5), higher scores indicate a higher level of perceived risk for

the risk perception scales, or greater agreement with the attitude factor on the attitudinal scale. Note that the scale for the ASAS was reversed in direction from the direction utilised on the questionnaire provided to participants (as shown in Appendix

5), so as to allow for correlations to be explained in the intuitive direction; positive

correlation meaning that both measures get ‘bigger’ (e.g. if the correlation was between

flight hours and self‐confidence, a positive correlation would mean that more flight hours meant a higher level of agreement with the statements in the self‐confidence

scale).

In order to investigate the relationship between risk perception, attitude and demographic variables in pilots, a correlational analysis was performed. Alpha was set at .05. Table 3 below shows the correlation matrix that was formulated from the results

(with alpha set at .05, and test assumptions of normality and homogeneity of variance

satisfactory, a correlational analysis was performed}.

113

6.3 Results

The results from Experiment One are reported below in sub‐headings in order to aid in interpretability. Table 3 below shows the results of the surveys in Experiment

One. Remember that higher scores on the ASAS show a higher level of the attitude in question (e.g. a higher level of self‐confidence), and that higher scores on the Risk

Perception Scales show a higher level of perceived risk. Table 3 below shows the correlation matrix that was evident from the analysis undertaken.

114

Table 3. Results for Risk Perception and Attitude Scales in Experiment One

Source Measure Number of Mean SD

items Risk Perception Delayed Risk 8 64.85 10.69

Scale One Nominal Risk 5 35.27 14.39

(0 – 100) Immediate High Risk 4 79.31 10.52

Risk Perception General Flight Risk 10 40.16 17.25

Scale Two High Risk 10 56.90 16.84

(0 – 100) Altitude Risk 7 50.53 16.70

Driving Risk 3 54.49 17.42

Everyday Risk 4 33.04 14.76

ASAS Self Confidence 14 2.41 0.52

(1 – 5) Risk Orientation 8 2.29 1.25

Safety Orientation 4 2.54 0.53

Remember that in Table 3 above, a higher figure for the Risk Perception Scales

indicates a higher level of perceived risk. A higher figure for the ASAS indicates a higher level of agreement with the statements in the scale, which would be interpreted as a higher level of the attitude in question (e.g., a higher level of self‐confidence)

115

Table 4. Correlations Between Attitude, Risk Perception and Participant Demographic s in Experiment One Participant Total Flight Hours in Delayed Nominal Immediate General High Risk Altitude Driving Everyday Self Risk Safety Age Hours the Last Risk Risk High Risk Flight Risk Risk Risk Confidence Orientation Orientation 90 Days Risk Participant 1 Age Total Flight 0.68* 1 Hours Hours in the 0.16 0.26** 1 Last 90 Days

Delayed Risk ‐.01 .02 ‐.09 1

Nominal Risk .03 .02 ‐.06 .37** 1

Immediate .17 .12 .15 .40** .08 1 High Risk General Flight .00 ‐.03 ‐.10 .42** .62** .12 1 Risk High Risk .14 .08 ‐.08 .38** .41** .33** .74** 1 Altitude Risk .12 .05 ‐.09 .45** .54** .29** .82** .73** 1 Driving Risk .04 .04 .09 .33** .29** .39** .37** .45** .48** 1 Everyday Risk .11 .12 .02 .40** .59** .26** .66** .53** .60** .50** 1 Self ‐.27** ‐.22* ‐.39** .12 .05 ‐.27** .07 ‐.29** ‐.01 ‐0.17 ‐0.01 1 Confidence Risk ‐.08 ‐.12 ‐.09 .10 .03 ‐.27** .02 ‐.25** ‐.05 ‐.10 ‐.08 ‐.52** 1 Orientation Safety ‐.26** ‐.12 ‐.10 .12 .12 ‐.22* ‐.03 ‐.30** ‐.01 ‐.11 ‐‐.08 ‐.61** ‐.42** 1 Orientation

*p<.05 **p<.01

116

As illustrated in Table 4, there are a number of inter‐correlations within the risk perception and attitudinal scales employed. This is consistent with recent research in

Australia that has utilised the scale (Drinkwater & Molesworth, 2010; Molesworth &

Chang, 2009), and also with Hunter’s original results, where all risk perception factors were inter‐correlated (Hunter, 2006), and where Safety Orientation was correlated with

Risk Orientation, albeit weakly (Hunter, 2005). Despite the risk perception and attitudinal scales being correlated, Hunter found that internal reliability of the scales was generally acceptable (Hunter, 2005, 2006). As a secondary consideration, the ASAS and Risk Perception Scales One and Two are relatively unique, in that they are scales designed specifically for General Aviation. The ability to use a scale that was previously proven, as opposed to one that required validation, was judged to be a significant advantage to the current research.

The correlation analysis revealed six statistically significant relationships between attitudinal factors, and risk perception factors.

A weak, (but statistically significant) negative relationship was evident between the risk perception factors of Immediate High Risk and High Risk, and all of the attitudinal factors. Specifically, the relationships found were between Immediate High

Risk and Self Confidence r(107) = ‐0.27, p = .004., Immediate High Risk and Risk

Orientation r(107) = ‐0.27, p = .005., and Immediate High Risk and Safety Orientation r(107) = ‐0.22, p = .02., and further, relationships were found between the risk perception factor of High Risk and Self Confidence r(107) = ‐0.29, p = .003, High Risk

117

and Risk Orientation r(107) = ‐0.25, p = .008, and High Risk and Safety Orientation

r(107) = ‐0.30, p = .002

The results of the correlation analysis failed to reveal any statistically significant

relationships between the experiential factors used, including Participant Age, Flight

Hours and Recent Flight Time, and Risk Perception, largest r, r(107)= 0.17.

The results did however reveal two statistically significant relationships

between Participant Age and Attitude. Specifically, a weak negative relationship was

evident between age and Self Confidence r(107) = ‐0.27, p < .01. Also, a weak negative

relationship was evident between age and Safety Orientation r(107) = ‐0.27, p = .006.

Together these results suggest as a pilot ages, the less self‐confidence they exhibit, and

their attitudes towards safety become more conservative (e.g., pilots are less likely to

agree with statements like ‘I am a very capable pilot’ and ‘I am a very careful pilot’).

The analysis revealed one statistically significant relationship between flight

hours and attitude, a weak negative relationship between Flight Hours and Self

Confidence r(107) = ‐0.22, p = .02. This result suggests that the more experienced pilots

in terms of flight hours will exhibit a lower level of self‐confidence (e.g., pilots with

more flight hours are less likely to agree with statements like ‘I am a very capable

pilot’).

The analysis of the data also revealed one statistically significant relationship

between Recency (in terms of number of hours flown in the last ninety days) and

Attitude. Specifically, a weak negative relationship was evident between Hours in the

118

Last 90 Days and Self Confidence r(107) = ‐0.39, p< .01. This result suggests that the

more a pilot has flown recently (in terms of flight hours), the more likely it is that the

individual will exhibit less self‐confidence than those that have not flown in recent days.

6.4 Discussion

The first research question in Experiment One was concerned with the

relationship between attitude and risk perception of pilots in the General Aviation training sector.

Few statistically significant relationships were found between attitude and risk

perception. The relationships that were found were weakly negative relationships

between the risk perception factors of High Risk and Immediate High Risk

(representing ¼ of risk perception factors) and all of the attitudinal measures used (Self

Confidence, Safety Orientation and Risk Orientation).

The direction of the Likert scale used in the ASAS is such that a higher score will

indicate higher level of agreement with the statements made (i.e., a higher score on Self

Confidence will mean that pilots agree more with statements like ‘I am a very capable

pilot’). Therefore, it appears from the evidence derived from this research, that the less

one displays self‐confidence, risk or safety orientation (which is interpreted as the more

conservative one’s attitudes), the better one’s ability to perceive the risks posed in highly hazardous situations.

119

Immediate High Risk, and High Risk, as well as being related to attitudes in the

experiment, were the two most extreme in terms of the consequences of the risky

situations portrayed in the survey. To qualify this, compare an Immediate High Risk

situation to an Everyday Risk situation. In one Immediate High Risk situation (e.g.

question 4 of Risk Perception Scale One) a pilot who is not capable (it seems) of

instrument flight is undertaking low flight under cloud cover, in a valley that is likely to

have ground above the cloud base, when s/he enters cloud and decides to hold his/her

altitude and ‘hope for the best”. This contrasts with an Everyday Risk (e.g. question 25

of Risk Perception Scale Two) situation where a person rides in a normal elevator. The

two situations are obviously different in terms of the risk type and likelihood of an

adverse event taking place, with the Immediate High Risk situation significantly more

likely to result in a serious adverse outcome for the pilot.

It is only in situations that feature highly hazardous circumstances (High Risk)

and those that are highly hazardous and have time pressures associated (Immediate

High Risk) that risk perception and attitudes appear to be related. Therefore, it can be

inferred that the attitudes and risk perceptions, as measured by the battery of surveys

utilised in this experiment, are not generically related to one another. Because of this

lack of relationship, it cannot be stated categorically that a ‘stronger’ or ‘better’ attitude

(or indeed a weaker or worse attitude) would generically be accompanied by a superior

level of risk perception by a pilot. What a more conservative attitude towards safety in aviation appears to be, in Australian GA, is related to an increase in pilots’ level of perception of risk in situations involving high risk/s.

120

Recent research in the area of the relationship between attitude and behaviour

has pointed toward the existence of a measurable link between attitude and behaviour

(Glasman & Albarracín, 2006). In aviation, this has also been the case, with attitudes

found to predict risk management behaviour of pilots (Molesworth & Chang, 2009)

Therefore, the current data (the attitude, risk perception and demographic data) were

to be used in Experiment Two to determine if this relationship was replicated in self‐

reports of pilot’s behaviour in Australian General Aviation.

The second research question was concerned with the relationship between

experience, (as measured by total hours flight experience, recent flight hours, and pilot

age) and the attitudes and risk perceptions of pilots in the GA training sector.

Remember that as per Molesworth and Chang’s (2009) findings, flight hours were expected to exhibit a positive relationship with attitudes, as measured by the ASAS

(that is, having more flight hours experience was expected to be related to a greater level of agreeance with the statements in the attitudinal scale) and as per Hunter’s

(2006) findings, a negative relationship with risk perception (such that more flight

hours was related to lower perceptions of risk).

It was found in the current experiment none of the experiential factors utilised

(flight hours, recent flight hours, and age) were related to risk perception. This was true

for both situations in which participants were the subject of the question (the individual

taking the risk), and also when the participants were giving their perceptions of the risk

of a situation based on another individual being involved in the situation. Studies

outside the aviation industry (i.e., road safety) have found a significant difference in the

121

risk perceptions of individuals when answering for themselves compared to others

(Sjöberg, 2000), however, this was not replicated in this experiment. It is possible that

this could be explained by pilots expecting their colleagues to behave in much the same

way as they would. That is, pilots may expect (rightly or wrongly) that other pilots are

skilled and capable, and will not behave in a manner that is dissimilar to their own actions. No testing for this was undertaken however; it is therefore a proposition as to the reasons behind the phenomenon.

The finding that experience appears unrelated to risk perception supports

previous findings in this area (Molesworth & Chang, 2009) where risk perception, as

measured utilising Risk Perception Scale Two, showed no relationship with experience.

These findings however, are in contrast to those found by Hunter (2006), where weakly

negative (largest r = ‐0.19) relationships were evident between flight hours and 5 of the

8 risk perception factors. Possible explanations of this difference are that the studies

were undertaken in different countries (Hunter’s in the U.S.), and that Hunter’s study

featured only private pilot’s licence holders. From the findings of the current research

therefore, it appears that the perception of risk in a variety of situations from everyday

life situations, to extremely high‐risk aviation situations, is not influenced by

experience.

In terms of attitude however, there were some findings of interest in the current experiment. Pilots with more flight hours displayed a lower level of self‐confidence than their less experienced counterparts in the current research. This finding was in the opposite direction from that expected. It also provides a counterpoint to the findings of

122

Molesworth and Chang (2009), upon which the expectation was based. This finding is, however, in agreement with findings in the road safety domain; in which more experienced drivers tend to exhibit lower level of confidence in their skill levels and abilities compared to those with less experience (Deery, 1999). The differences evident between the current results, and those found by Molesworth and Chang (2009) may indeed be explained by different cohort demographic; the current experimental cohort having a broader range of pilots in terms of age, experience and career goals.

It was found that a weak negative relationship existed between the experiential factor of recency and the attitudinal factor of Self Confidence. This relationship is such that those that flew more in the preceding 90 days were likely to exhibit a lower level of self‐confidence than those that flew relatively less. This result provides some corroborating evidence for the finding that flight hours are related to attitude, as recency was related (albeit weakly, r = ‐0.26) to total flight hours in this cohort.

As stated above, no relationships were evident between recent flight and risk perceptions, which again is counter to the hypothesis presented earlier that risk perceptions would be reduced by increased experience. This is interesting and important, as an implication of this is that recency (as measured in this experiment ‐ with hours rather than by take‐offs and landings) cannot be seen as a guarantee of more proficient risk perception by pilots; that is, recent flight hours cannot be used to ensure that a pilot will perceive the stimuli required to make safe decisions regarding the strategic choices required in aviation. This finding introduced an interesting area of

123

research for Experiment Two and Three, to find if an increased level of flight recency would affect the behaviour of pilots.

In terms of the relationship between a pilot’s age and their attitudes, two of the three attitudinal factors of the ASAS were correlated with Participant Age. Specifically, there was a weak negative relationship between age and both Self Confidence and

Safety Orientation. This finding suggests that older pilots not only exhibit less self‐ confidence than their younger counterparts, but are also less likely to describe themselves as safety oriented pilots. This is tempered by the knowledge that the relationship between age and attitude was weak in the current experiment, as well as that the knowledge that older participants in this research were also likely to have more flight experience, which as stated above, was characterised by a lower level of self‐ confidence in this cohort.

The finding that age appears to be related to attitude is backed by previous findings in the literature. In the aviation field, Popa, Rotaru and Oprescu (2005) found that age was related to more conservative attitude. Similarly, Deery’s (1999) research in the road safety field found that older drivers had lower levels of self‐confidence

(considered a more conservative attitude). In research that addressed the safety‐related attitudes of construction workers, it was found that older participants held a better overall attitude towards safety related factors and behaviours (Siu et al., 2003).

As stated above, no relationship was identified between age and risk perception.

This finding fails to support Hunter’s (2002) findings, where risk perception was negatively related to age. This may be explained in part by the lack of strength of

124

Hunter’s findings (largest r = ‐0.106). Another possible reason for the apparent difference is that Hunter’s research utilised a cohort that differed from that of the current research on two fronts; nationality (U.S. compared to Australian), and also on experience (Hunter’s cohort contained airline pilots as well as the types of pilots in the current research).

An implication of the finding of a lack of relationship between risk perception and age is that it suggests that during normal flying operations, which will (hopefully) represent the majority of operational situations that a pilot will encounter, older pilots will not exhibit superior perception of the risks in a given operation. In basic terms, the findings in the current research suggest that experience does not appear to alter the ability of pilots to perceive ‘normal’ or everyday operational risks, or affect the pilots’ attitudes toward them. Older pilots will be just as likely as younger pilots to perceive or misperceive risks in normal operations.

It is important to note that Experiment One utilised correlational analysis to examine the relationship between attitude, risk perception and demographic variables such as age and experience. As with all research in this area, correlational relationships do not infer causation. Similarly, no conclusion can be drawn about which factor, if any, preceded the other.

6.5 Conclusion

Experiment One was designed to achieve two outcomes. First, to identify links between a pilot’s risk perception, attitude (towards safe flight operation), age, and their aviation

125

related experience in order to determine the existence or extent of the relationship with one another. Second, the experiment was designed to reveal the risk perceptions and attitudes of the pilots in the study in order to allow for a comparison with behavioural measures (see Experiment Two and Three).

The analysis revealed that there were very few links between experience and attitude, experience and risk perception and attitude and risk perception.

126

7 Experiment Two – Self­Reported Behaviour of

Pilots in High­Risk Situations

7.1 Introduction

Experiment Two was undertaken as an extension of Experiment One, in that it utilised the data captured in Experiment One, and compared the data to self‐reports of behaviour in aviation specific situations. The experiment was thus primarily designed to examine the relationship between the self‐reported behaviour of pilots and their experience levels, attitudes and risk perceptions.

Experiment Two was designed as a pen‐and‐paper survey to capture self‐ reported behaviour of pilots that had previously undertaken Experiment One. The experiment was undertaken in two parts: a pilot study that utilised expert opinion in order to aid in the design of the stimulus material (Experiment Two A), and the main study (Experiment Two B).

Experiment Two compared the findings from Experiment One with regards to the demographic, experiential, attitudinal and risk perception factors that were measured, to the self‐reported behaviours revealed in this experiment. This study is

127

uniquely different from those of past (see, for example Goh & Wiegmann, 2001;

Molesworth et al., 2006; O'Hare & Smitheram, 1995) as three different risky situations

were presented, reflecting different levels of risk, as opposed to a single risky situation

being used. Also, the experiment was designed such that pilots could choose whether or

not they would undertake the scenarios; they have what amounts to total control over

the situation, as opposed to being asked to assume that they have behaved in a specific

way preceding the experimental situation. In this way, the self‐reported behaviour of

pilots would be revealed for three risk levels, in what was hoped to be a relatively

naturalistic fashion, as opposed to a situation where pilots were forced to fly in a

situation they would not normally fly in. That is, pilots were not asked to assume that

they were already in a situation that they would not wish to be, like flying near

thunderstorms, or attempting to land an unfamiliar aircraft after a fatiguing flight.

Rather, they were asked to decide upon their actions at an earlier stage, before major risks were encountered.

The participants in this experiment include private pilots, qualified instructors, and pilots undertaking commercial licence training. In this way, the current experiment differs somewhat from other research in the Australian GA field, where the majority of participants are student pilots (that is, pilots that are undergoing instruction to gain a higher level of licence) (see for example Drinkwater & Molesworth, 2010; Molesworth &

Chang, 2009). Therefore, this experiment, when compared to other research in the area, and Experiment Three of the current research, has a wider cohort in terms of age, employment/position held (with respect to the usage of their aviation‐based skills) and

128

experience levels. The experiment consisted of a single session, which involved completing a battery of pen‐and‐paper surveys.

All participants that were recruited to undertake Experiment Two had previously undertaken Experiment One. None of the participants that undertook

Experiment Two however, were asked to complete Experiment Three. The reason for this split was due to participants’ ability to engage in a simulated flight at a later date. If participants could not undertake the secondary flight simulation (Experiment Three), then they were put into Experiment Two. If they indicated that they were willing to undertake a flight simulation, then they were put into Experiment Three.

7.2 Background

Experiment One revealed few significant correlations between the experiential factors, risk perceptions and attitudes that were examined. There were however, some findings from Experiment One that warranted further examination. One (of the three used) attitudinal measure, Self Confidence was related to the experiential factors measured. Specifically, there was a weak negative relationship found between age and

Self Confidence, flight experience and Self Confidence, and recency and Self Confidence.

In other words, it appears that more experience one gathers, the less self‐confidence

(which is seen as a more conservative attitude) that one will exhibit.

Another interesting finding from Experiment One was that there was no relationship evident between risk perception and the experiential factors utilised. In other words, it appeared from the experiment that older or more experienced pilots will

129

be just as likely as their younger or less experienced counterparts to perceive or misperceive risks in normal operations.

As has been stated above, it is thought that attitudes, risk perceptions and experience influence risk‐taking behaviour (Molesworth & Chang, 2009), therefore, the relationship between attitude, risk perception, experiential variables (i.e., age, flight experience, and recent flight time) and behaviour were tested in this experiment

(Experiment Two)

Two questions were designed for Experiment Two, such that the second and third research question (from Chapter 5) was explored. Specifically, the following relationships were to be explored: (1) the relationship between attitudinal and risk perception factors and self‐reported behaviour, and (2) the relationship between experiential/demographic factors and self‐reported behaviour.

The questions posed in Experiment Two are given below:

• What is the relationship between the attitudes, risk perception and self‐reported

behaviour of pilots in the general aviation training sector?

• What is the relationship between a pilot’s experience, as measured by age, flight

hours, recent flight time, and self‐reported behaviour in the general aviation training

sector?

A secondary outcome of the experiment was the analysis as to the relationship between an individual’s risk perception, as measured by existing risk perception scales,

130

and their rating of risk in situations that present multiple risk factors. That is, Hunter’s

(Hunter, 2006) risk perception scales utilise situations in which there is a single risk factor present, pilots are then asked to rate the risk of these situations. The results from the scales will be compared with the risk scenarios in the current experiment (which feature more than one risk factor in each scenario) to determine if risk perception of single risk factor situations is related to risk perception of multiple risk factor situations. This was augmented with an analysis of how the other measures used in the current research (attitudes and experiential factors) were related to the ratings of risk given.

7.3 Experimental Design

As stated above Experiment Two was designed as a pen‐and‐paper survey to capture self‐reported behaviour of Australian General Aviation pilots. The experiment was undertaken in two parts, the pilot study (Experiment Two A), and the main study

(Experiment Two B).

Experiment Two was designed such that pilots could choose the level of risk that they were willing to undertake in the full range available. That is, pilots could chose to minimise their risk taken to the lowest level possible for all scenarios, and report that they would not undertake the flights. Conversely, the pilots could report that they were willing to undertake all flights, and were willing to fly recklessly (to give an extreme example) in order to achieve a goal. Although the likelihood of pilots reporting intentionally reckless behaviour was relatively low, it was possible for pilots to report

131

it. This is one ability that most methodologies utilised in aviation risk management

research, and those within the attitude‐behaviour research do not allow. This freedom to choose a target risk level served a further purpose, in that it was designed to echo the same ability of pilot to choose their risk level taken on a full spectrum (very low to very high risk) in both Experiment Three (described below) and in real‐world operations.

In addition to this, there were three distinct risk levels (i.e., there was a low‐risk,

medium‐risk, and high‐risk situation) and also two risk factors (as opposed to a single

risk factor being used) in the three scenarios presented. In this way, a range of potential risk levels were presented to pilots, and they were free to choose their behaviour in the scenarios presented. The use of more than one risk factor in each scenario was hoped to aid in the reality of the scenarios used. This was because the scenarios utilised were based upon real‐life situations, all of which featured more than one risk factor. This is dissimilar to previous research in the field (see, for example Goh & Wiegmann, 2001;

Molesworth et al., 2006; O'Hare & Smitheram, 1995) where a single scenario is utilised in order to gauge behaviour.

The final experimental scenarios were designed semi‐collaboratively, such that

experts in the aviation industry were utilised to give a baseline rating of risk to each

scenario. This is a similar design tool as previously used in methodologies in the

aviation industry (Hunter, 2003; Hunter, Martinussen, & Wiggins, 2003). Nine scenarios

were initially designed for the experiment. The scenarios were presented to a group of

seven expert pilots in a pilot test (called Experiment Two A). Their status as an expert

was determined by the fact that they held either an instructor rating, or had over 500

132

flight hours within General Aviation. In this way, the current methodology was similar

to previous research undertaken in the U.S. (see Driskill et al., 1997) in that experts

were used to rate multiple scenarios that varied based on risk levels presented.

7.3.1 Experiment Two A – Pilot Study to Determine Flight Scenarios.

Experiment Two A was designed as a pilot study in which the 9 scenarios

initially designed to be used for Experiment Two were reduced to 3 scenarios that

would be used in the final experiment. The 3 scenarios chosen were representative of the 3 levels of risk that were targeted in the experimental design. The aim of the experiment was therefore to produce a set of 3 scenarios that were representative of normal, medium and high‐risk operations.

In the pilot experiment (Experiment Two A), experts were asked to rate on a

scale of 1 ‐100 the level of risk in each scenario (consistent with Hunter’s Risk

Perception Scales). The scenarios where purposely constructed so that they systematically varied based on altitude, duration of flight, aircraft type, aircraft

serviceability, weather and fuel, reflecting various levels of risk. These variables were

then injected into real‐life scenarios. All scenarios were derived from either personal experience, the experience of colleagues (relayed to the author verbally or through

email), or from incident/accident reports. A short synopsis of each is given immediately

below.

The “Cross‐country” scenario: a situation where a pilot was to fly a routine cross‐

country flight. The scenario was designed to be low risk, with only slight time pressure

133

(landing would be less than an hour before last light) in a relatively new and familiar

aircraft.

The “Ferry to LAME” scenario: a situation where a pilot was to fly an aircraft

nearing the end of its maintenance cycle to an airport that was in an area of

deteriorating weather conditions. The scenario was designed to be high risk, with time

pressure due to legal requirements for maintenance and the prevailing weather, and

contain a low flight risk due to the weather requiring flight at or less than 1,000 feet

above ground level.

The “Sightseeing” scenario: was a situation where a pilot was to fly their friends

on a local low‐level sightseeing flight. The scenario was designed to be medium risk, as

the flight involved low flight, and the passengers had been drinking.

The “Sydney‐Melbourne” scenario: a situation where a pilot was to fly an aircraft

between Sydney and Melbourne, with no time or economic pressures attached. The

scenario was designed to be low risk, as the flight was being conducted with more than sufficient time, the aircraft type was familiar and alternates were available (i.e., there

was no enforced flights over water).

The “Hunter Valley” scenario: a situation where a pilot was flying his friend to

the Hunter Valley from Sydney. The scenario was designed to be low risk for similar

reasons to those in the Sydney‐Melbourne scenario above.

134

The “Victor One” scenario: a situation where a pilot was to fly his friends on a

dusk Victor‐One (a popular scenic flight along the Sydney coast at 500 ft AGL). The flight

was designed to be medium risk, as the pilot was likely to be fatigued, there was poor

weather in the area, there was pressure to complete the flight because the passengers

were soon to return home and finally the pilot had some personal issues occurring at

home.

The “Moruya” scenario: a situation where a pilot was asked to attempt a search

and rescue flight for a wayward skydiver in an aircraft with 23 minutes of fuel

remaining. This scenario was designed as a high‐risk flight, as there was very little fuel

remaining, there were no alternate aerodromes in the area, and the pilot was unfamiliar

with the area.

The “USA to Australia” scenario: a situation where a pilot was asked to ferry a small aircraft from the USA to Australia in which one particular leg would require

landing with approximately 50 minutes of fuel remaining. This flight was designed as a

medium risk scenario, as the fuel state was a hazard, as was the lack of experience of the

pilot in the scenario.

The “Mount Hotham” scenario: A situation where a pilot was asked to fly an

unknown package of unknown weight and contents from an airport that is at altitude

and in hot weather (which both reduce aircraft performance). This flight was designed

as a high risk scenario because of the potential for the package to cause the aircraft to

be over its maximum takeoff weight, the personal and financial pressure to complete

the flight, and the lack of alternates available to the pilot.

135

The scenarios are available in full in Appendix 6.

Inter‐rater reliability testing, through the use of inter‐class correlation was employed to ensure that the scenarios used were each measuring the risk level intended. A single Cronbach’s Alpha test was not appropriate as a measure of reliability, as there were three distinct levels of measurement in the items under test.

7.3.2 Experiment Two B – The Full Self­Report Study.

In this part of the experiment (Experiment Two B), participants were asked to rate on a scale of 1 ‐100 the level of risk in each scenario (consistent with Hunter’s Risk

Perception Scales). Participants were presented with a booklet containing the three written aviation scenarios. All scenarios were derived from Experiment Two A. These scenarios were the Hunter Valley Scenario, the Ferry to LAME scenario, and the Moruya

Scenario. They are described above, and are available in full in Appendix 7.

The three distinct levels of risk were used in the experiment for two reasons. The first was so that any differences in reported behaviour between pilots based on the experiential, attitudinal and risk perception variables were apparent. That is, it would be possible to see if different pilots rated risk in the low, medium and high levels differently depending upon their experience, perception or attitude.

The second reason for the inclusion of multiple risk levels in the scenarios was so that pilots in this experiment would be presented with situations that simulate the levels and types of risk that they are likely to encounter in their career. This allowed for pilots to report behaviours that would vary in a similar way to their real life behavioural

136

choice. The inclusion of multiple risky scenarios allowed for the experiment to simulate real‐world situations more closely,

The design also allowed the risks presented to pilots to be multi‐dimensional, rather than utilising uni‐dimensional risks, as is the norm in the literature (see, for example Goh & Wiegmann, 2001; Molesworth et al., 2006; O'Hare & Smitheram, 1995).

The advantage of which is that this more closely resembles the operational environment, and the findings of a study that is grounded in this may be more representative of the situation with regards to the predictors of risk management behaviour in the industry.

These flight scenarios were designed such that pilots had the ability to choose and report their actions, rather than simply rating the relative risk of the scenario on a scale of 1 to 100, as in the risk perception scales in Experiment One. In this way, the experiment was designed to be more like a flight simulation than a risk‐rating survey, as pilots were asked to make the same decisions as they would be forced to make in a simulation (or real aircraft), but they are not asked to undertake the physical act of operating the aircraft. This rating was also used to compare those that perceived risk like the expert group in Experiment Two A, to those that do not rate the risks similarly.

The procedure of administration involved asking pilots to read each scenario

(presented in a counterbalanced order; see Appendix 7 for the full scenarios). Upon reading the scenario the first question to the pilot was “would you undertake the flight as described?” Allowing for pilots to choose their preferred course of action, from lowest to highest level of risk, just as the participants in Experiment Three would be

137

able. Regardless of their answer, pilots were next asked why they chose their path of

action. Pilots were then asked to give the scenario a rating of between 1 and 100, low to

high risk, which was an identical task to the question posed in the risk perception scales in Experiment One above.

The exception to this was the Moruya Scenario, in which pilots were given more

information regarding the flight, only if they chose first to undertake the flight. That is, if

participants elected not to fly, they were directed to the risk‐rating question, however,

those that elected to go were directed to a question in which details of the flight’s

progress were revealed. This was such that it echoed as closely as possible, the way in which a real flight would take place. In the real world, pilots will not know what they

are going to find on a search and rescue flight until they undertake it. Therefore, details

regarding what the pilots were confronted with upon their arrival at the search site, and

the time taken to reach the site were given to ensure that no external factor like

differing assumptions of the situation specifics affected reported behaviour.

Pilots in this scenario were then asked to describe their decisions and actions

during the search, the turn‐back to Moruya, and the return leg/approach to Moruya.

In order to ascertain the relationship/s between attitude and behaviour and risk

perception and self‐reported behaviour, two research questions were formulated.

These were:

1. What is the relationship between the attitudes, risk perception and the

behaviour of pilots in the General Aviation training sector?

138

2. What is the relationship between a pilot’s experience, as measured by age, flight

hours, recent flight time, and behaviour in the general aviation training sector?

It was thought that attitudes would affect behaviour in line with their valence, that is, a more conservative attitude would lead to a more conservative behaviour, and a less conservative attitude, to a riskier behaviour. This is in line with the Theory of

Planned Behaviour (see for example, Ajzen, 1991; Ajzen & Fishbein, 2005; Armitage &

Conner, 2001)

In a similar fashion, risk perception was expected to effect behaviour in line with its strength and direction. That is, if one perceives a risk to their safety (or other factor/item that they value), then it was expected that the individual would modify their behaviour to minimise the said risk. This is in line with the hypotheses put forward in the risk perception literature (see Brewer et al., 2004).

7.4 Method

7.4.1 Experiment Two (A)

7.4.1.1 Participants.

Seven pilots who were considered experts in their field were recruited from various flight training organisations located at Bankstown aerodrome and Camden aerodrome.

The mean flight experience of the group was 3,677 hours (range = 12,502 hours). These pilots were chosen based upon their employed position as a flight instructor or testing

139

officer, or their experience level of 500 or more hours in command of an aircraft. No inducements were offered for undertaking the research, which was expected to take approximately 50 minutes to complete.

7.4.1.2 Procedure.

Nine flight scenarios were presented to the experts (see Appendix 6). Each flight scenario was based on situations presented in Australian GA that have been reported in aviation publications, the general media in Australia, or have been reported to the researcher as personal experiences. Each scenario was specifically designed so that the risk involved varied. For example the “Mount Hotham” scenario, in which pilots were faced with a possible overweight take‐off from an elevated airport was designed to be significantly more risky than the “Hunter Valley” scenario, in which pilots were faced with a routine and simple flight. This variance was designed such that the scenarios would (hopefully) return distinctly different risk ratings, allowing for scenarios to be chosen that were different from one another based upon risk rating, but had similar risk factors at play (e.g. low flight or low fuel) to allow for comparison between them.

The experts were tasked to rank each scenario on a scale from 0 – 100 with regard to their level of perceived risk involved in the hypothetical flight. Each scenario was to be given a discrete score out of 100, rather than a comparative score between it, and the other scenarios presented. The nine scenarios were presented in a counter‐balanced order.

140

7.4.1.3 Results.

Table 5 below displays the mean ranking for the nine scenarios. The results of inter‐ rater agreement analysis utilising an intra‐class correlation coefficient illustrate good agreement between the raters with respect to the ratings given to the scenarios, R(6) =

.875.

Table 5. Mean Rankings for Scenarios in Experiment Two A

Mean SD Rating Assi gned

Hunter Valley Scenario 31.57 20.91 Low

Cross‐ Country Scenario 38.00 26.83 Low

Sightseeing Scenario 43.00 28.89 Low

Sydney‐Melbourne 46.43 20.15 Low

Ferry to LAME scenario 74.29 15.39 Medium

Mount Hotham Scenario 75.00 15.00 Medium

USA‐Australia Scenario 75.71 13.05 Medium

Victor One Scenario 81.43 14.92 High

Moruya Scenario 86.43 11.80 High

The intra‐class correlation coefficient test results are provided below in Table 6 below.

141

Table 6. Intra­class Correlation Coefficients for Expert Ratings

1. 2. 3. 4. 5. 6. 7. 8. 9. 1. Cross Country 1 2. Ferry To LAM E .63 1 3. Sightseeing .98 .68 1 4. Sydney to .68 .30 .70 1 Melbourne 5. Hunter Valley .99 .66 .98 .65 1 6. Victor One .13 ‐.19 .14 .67 .06 1 7. Moruya .08 ‐.29 .05 .46 .00 .93 1 8. USA – Australia .36 .63 .46 ‐.16 .38 ‐.35 ‐.31 1 9. Mount Hotham .75 .63 .71 .18 .73 .00 .14 .66 1

(scenarios found to be similar are shaded in similar colour)

Values below 0.9 were taken to indicate that the scenarios were dissimilar from

one‐another – that is, they were rated as being different from one another by the expert

group. The opposite was also true. From this the scenarios were grouped as shown in 0 above. As can be seen, there were four scenarios that were rated between 30 and 50,

three scenarios rated at approximately 75, and two scenarios rated above 80. These

were labelled the low, medium and high‐risk levels respectively. The labels are

arbitrary insofar as there is not a linear and empirically based relationship between

them; it is a ranking scale to signify that one scenario is more (or less) risky than

another for convenience.

As the object of Experiment Two A was to establish one scenario from each risk

level as the representative scenario for the final survey, the second part of the process

142

was to decide which of the scenarios that were rated similarly would be used. That is, of

the scenarios that were rated at a similar level of risk, one was chosen to be in the final

survey. For the low‐risk level this was the Hunter Valley scenario, for the medium‐risk rating this was the Ferry to LAME scenario, and for the high‐risk scenario, this was the

Moruya scenario.

Table 7. Risk Matrix and Final Scenarios in Experiment Two

Low Risk Medium Risk High Risk Scenario Scenario Risk Scenario Risk Factors Factors Risk Factors

Hunter Normal Valley Flight Risks Scenario

Low Flight, Ferry to Time LAME Pressure, Scenario Weather

Moruya Low Flight,

Scenario Lack of Fuel

The three final scenarios are given in Appendix 7. The low‐risk scenario was the

‘Hunter Valley’ scenario, the medium‐risk scenario was the ‘Ferry to LAME’ scenario, and the high‐risk flight was the ‘Moruya’ scenario. The scenario risk rating, and relevant risk factors for each are shown in Table 7 above.

143

The reasoning behind the use of these specific scenarios is relatively straightforward. The Hunter Valley scenario was chosen as the low risk scenario, as it is a fairly typical weekend flight undertaken by many pilots in any given year. Most pilots in Sydney would have undertaken the flight, and be comfortable with the requirements.

The Ferry to LAME scenario was chosen as it featured low flight and time pressures – these factors closely resembled those in the Moruya scenario, which was the high risk scenario. This meant that two similar (in hazard type) scenarios could be utilised that were rated differently in terms of risk by experts. The Moruya scenario was also used in

Experiment Three, meaning its inclusion in Experiment Two was a logical step as this would provide the potential for comparison.

7.4.2 Experiment Two (B).

7.4.2.1 Participants.

Thirty‐Eight participants were recruited from flying schools and flying organisations of the Sydney Basin, located at either Camden or Bankstown aerodromes.

These included current students in the Bachelor of Aviation (Flying) degree at the

University of New South Wales, graduates of the program, and from other smaller flight training schools. Students of the University, those participants that were graduates, or those pilots from private flight schools were recruited by individual conversations and briefing sessions with the researcher at their current flight school. No inducements were offered for undertaking the research. The total time to complete the research was approximately 30 minutes.

144

The participants had previously completed part one of the experimental process

(Experiment One). Upon completion of this experiment, their involvement in the experimental process was complete. The participants in Experiment Two had previously undertaken Experiment One, and had indicated that they could not attend a simulation session at a later date. They were therefore asked to complete Experiment

Two instead of Experiment Three, as it required less investment of time by the participant.

All participants were aware that the research was focused on the perceptions, attitudes and behaviours of general aviation pilots, but were unaware of any further details of the research project’s purpose. Table 8 outlines the demographic and experiential features of the participants. The primary target group for recruitment were trainee GA pilots. The data, therefore, was expected to represent the intended pilot demographics, rather than a population wide distribution.

Table 8. Experiment Two (B) Participant Demographic Statistics

Total Hours in the Participants Age in years Flight Hours (SD) (SD) Last 90 Days [Recency] (SD)

38 27.03 (14.90) 599.16 (2102.67) 28.56 (32.95)

There were seven female pilots in Experiment Two. A total of 11 pilots held a

Private Pilot Licence, nine pilots held a Commercial Pilot Licence, the remaining pilots were not licenced at the time of the research. Three of the pilots held an instructor rating, and two Held an instrument rating.

145

7.4.2.2 Design.

The study was designed to examine the relationship between attitudes towards

aviation safety (using Hunter’s ASAS), risk perceptions of aviation‐centric situations

(using Hunter’s Risk Perception Scales 1 and 2), experiential variables (e.g., age, and flying experience in terms of flight hours), reported risk ratings, and self‐reported risky flight behaviour. Reliabilities of Hunter’s scales can be found in section 6.2.1 above. Self‐ reported risky flight behaviour and the reported risk ratings were derived from the three hypothetical flight scenarios that were generated from Experiment Two A. Since it was a pen and paper exercise, the behavioural variable featured was dichotomous – go

(choosing to fly the flight in the scenario) or no‐go (choosing not to fly the flight in the scenario).

Due to the unequal and small sample size, a non‐parametric equivalent to

Pearson product‐moment correlation, namely Spearman’s rho was employed to investigate the relationship between risk perception, attitude, demographic variables, and self‐reported behaviour (in terms of the go/no go decision). These correlations used the dependant variable of Go or No Go, that is, the dependant variable was whether or not pilots indicated that they would undertake the scenario in the surveys provided.

The independent variables consisted of Participant Age, Total Flight Hours, Total

Hours in the Last 90 Days, the three factors in the ASAS (Self‐Confidence, Risk

Orientation and Safety Orientation), the three factors in Risk Perception Scale One

(Delayed Risk, Nominal Risk, Immediate High Risk), and the five factors of Risk

146

Perception Scale Two (General Flight Risk, High Risk, Altitude Risk, Driving Risk, and

Everyday Risk). Alpha was set at .05.

7.4.2.3 Materials.

The three flight scenarios derived from part A of the present experiment were utilised to gather the self‐reports of behaviour that were required. The flight scenarios varied based on the level of risk involved within each.

The three scenarios, in order of risk rating given by the expert group (1 = least

risky) were: (1) the “Hunter Valley” scenario, (2) the “Ferry to LAME” scenario, and (3)

the “Moruya” Scenario.

The first scenario was the “Hunter Valley” scenario. In this scenario, pilots were

told that they had planned a flight from Camden to Cessnock in fine weather, there were

back‐up plans in case the weather turned bad, or the aircraft was not performing

suitably. This was rated as low risk according to the values given by the expert group.

The second scenario presented was the “Ferry to LAME” scenario, in which pilots

were being asked to ferry their friend’s aircraft to an airport in order that maintenance

be performed. In the scenario there existed time pressure and weather related hazards.

It was rated as medium risk according to the ratings given by the expert group

The final scenario was the “Moruya Scenario”. In this scenario participants were

asked to fly an aircraft with critically low fuel to search for a skydiver that had landed

147

away from the normal landing zone for the local skydivers. This scenario was rated as

high risk due to the ratings given by the expert group.

As stated above, these scenarios can be found in Appendix 7 below.

7.4.2.4 Procedure.

Participants were informed of the nature of the experimental process verbally, and in writing from the information sheet provided (See Appendix 1). Participants were

briefed verbally that the experiment was part two of the experimental process, and that

once the experiment was completed, they had completed their commitment to the

research.

The surveys were provided to participants were provided in the following order;

the demographics questionnaire, the ASAS questionnaire (Hunter, 1995), the Risk

Perception 1 & 2 questionnaires (Hunter, 2002), followed by the three flight scenarios

presented in a counterbalanced order. The completed surveys were collected from the

participants directly. All data gathered was entered into SPSS v. 13 for Macintosh.

7.5 Results

7.5.1 Data analysis.

The main objective of the experiment was to determine the relationship between

pilot demographic measures, attitudes, risk perceptions, and the self‐reported

behaviour of pilots. The other objective was to determine the relationship between the

ratings of risk given for the multi‐dimensional risks in this experiment with the

148

attitudinal, risk perception (as measured by Hunter’s scales), and experiential data gathered.

Descriptive statistics of the participants are provided in Table 9.

149

Table 9. Experiment Two (B) Demographic, Risk Perception and Attitudinal

Statistics

Scale Factor Name M SD

Demographics Participant Age 27.03 14.90

Total Flight Hours 599.16 2102.67

Total Hours in the Last 90 Days 28.56 32.95

Risk Delayed Risk 62.01 8.44

Perception Nominal Risk 35.30 15.75

One (0 – 100) Immediate High Risk 83.32 9.28 Risk General Flight Risk 41.00 17.14

Perception High Risk 65.89 12.45

Two Altitude Risk 52.09 15.17 Driving Risk 55.81 15.44 (0 – 100) Everyday Risk 32.74 14.00

ASAS Self Confidence 3.26 0.48

Risk Orientation 2.17 0.51 (1 – 5) Safety Orientation 3.79 0.40

Note: In the table above, a higher score for RP1 and RP2 indicated a higher rating of the risks for this risk type. A higher score for the ASAS indicated a higher level of agreement with the statements that constitute that attitude.

In total, 2 pilots chose not to complete the Hunter Valley scenario (low risk

flight), 35 pilots chose not to complete the Ferry flight (medium risk), and 35 also chose

not to complete the Moruya flight (high risk) (see Table 10).

150

Table 10. Breakdown of the Go/NoGo Choice and Risk Rating Breakdown

Scenario (Expert Risk Level) Go (%) No Go (%) Mean Risk Expert Risk

Rating Rating

Hunter Scenario (low) 36 (95) 2 (5) 27.76 31.57

Ferry to LAME Scenario 3 (8) 35 (92) 79.27 74.29 (Medium)

Moruya Scenario (High) 3 (8) 35 (92) 79.76 86.43

Table 11. Correlations Between Behaviour and Risk Perception, Attitude &

Experiential Variables

Scenario Age Total Recent Delayed Nominal Immediate General High Altitude Driving Everyday Self Risk Safety (Expert Hours Hours Risk Risk High Risk Flight Risk Risk Risk Risk Confidence OrientationOrientation Risk Risk Level)

Hunter .08 ‐.17 ‐.24 .11 .22 ‐.15 ‐.08 ‐.02 .06 ‐.09 ‐.05 ‐.12 .23 ‐.10 (Low) Ferry .04 ‐.11 ‐.23 .36* .24 .28 .33*. 41* .44** .07 .14 ‐.04 ‐.07 ‐.10 (Medium) Moruya .09 ‐.18 ‐.11 .17 ‐.07 ‐.05 ‐.03 .00 .01 .13 .18 ‐.29 .09 .04 (High)

* p < 0.05 ** p < 0.01

As can been seen in Table 11 above, only the medium risk scenario (Ferry to

LAME scenario) generated statistically significant relationships between the

independent variables and the go/no go decision – the self‐reported behaviour variable.

The results failed to reveal any statistically significant relationship between the

attitudinal measures in the study (Hunter’s ASAS) and the self‐reported behaviour of

pilots in the study (largest r=‐.29, p=.08).

151

In contrast, a positive relationship was found between four risk perception

factors and the self‐reported behaviour for one scenario, the ‘Ferry to LAME’ scenario

(medium risk). The relationships were specifically between the decision to go and the

risk perception factors of Delayed Risk r(38) = 0.36, p=.03, General Flight Risk r(38) =

0.33, p=.04, High Risk r(38) = 0.41, p=.01, and Altitude Risk r(38) = 0.41, p=.01.

The positive direction of the relationship means that those that chose to go

(which was coded as one in the analysis) displayed a lower level of risk perception on

these factors than did those pilots that elected not to go (which was coded as two in the analysis) in the scenario. Therefore, it appears that risk perception, as opposed to attitude, affects self‐reports of behaviour in this experiment. Further, it appears that discrete sub‐types of risk perception are responsible for the modification of behaviour,

rather than risk perception as a broad concept being responsible for the behaviour

modification.

The results failed to reveal any statistically significant differences between any of

the experiential factors and self‐reported behaviour. That is, there was no relationship

evident between flight hours and behaviour (largest r =‐0.17, p=.27), between recent

flight time and behaviour (largest r =‐0.24, p=.14), and age and behaviour (largest r

=0.09, p=.59).Therefore, it appears that pilot experience appears to be unrelated to the

self‐reports of behaviour in this experiment.

152

7.5.1.1 Participant risk ratings of the flight (and what they relate to).

Experiment Two A provided baseline risk ratings (on a scale of 1‐100, as per those in Hunter’s risk perception scales) for each scenario. These ratings were based upon expert opinion, and were used in the design of Experiment Two B.

Ratings of the risk of each scenario were gathered in an identical fashion in

Experiment Two B, such that data was gathered from the non‐expert group both about

the intended behaviour of participants, and also the rating of risk that the participant

gave the scenario. This design allowed for the comparison of those individuals (in terms

of attitude, risk perception and experiential values) that rate the risks of the current

scenarios in a similar fashion (i.e. in the same order and magnitude) as the expert group

to those that do not.

In order to compare this, the risk rating for each scenario was examined.

However, based on the feedback provided post‐experiment, a filter needed to be

employed. Specifically, some participants remarked that they had rated the risk of the

flights they chose not to undertake as low, because they were rating the risk of their

chosen path of action, rather than the potential risk of the situation (i.e. before an option

was chosen). Since there was no practical way of contacting each individual participant

(a condition imposed by Ethics Panel) to ask how they had rated the risks in this

experiment, a rule‐based filter was placed on the data, such that any participant that

rated the medium and high risk scenarios in which they chose not to go as lower than

the other scenarios in the experiment where they had chosen to go, it was assumed that

they had rated the risk of their decision, rather than the potential risk in the situation.

153

These pilots were excluded from this analysis. In total, four pilots had rated the risks of

the scenarios in this fashion. Therefore, 34 participants remained.

From the 34 participants, 5 rated the risks in the same fashion as the expert

group (that is, the raw number was within 5 points and the order of ranking was

identical); 29 rated the risks in a different order and/or a different magnitude (i.e. the

low risk flight had a risk rating of zero, the medium risk flight a rating of 10, and the

high risk flight a risk of 40).

In order to determine whether a difference existed between pilots that rated risk

the same as the experts, and those that did not in terms of perceived riskiness of each

scenario (i.e., the rating of risk given by pilots of the scenarios), a non‐parametric equivalent to an Independent t test, namely a series of Mann‐Whitney nonparametric tests were employed (as prescribed by Tabachnick & Fidell, (2013) if sample size is small and unequal). .Alpha was set at .05. With correction for ties and z‐score conversion, there was a statistically significant difference between pilot groups (those

that rated the same as the experts, and those that did not) for one attitudinal factor, Self

Confidence, z (N = 34) = 2.29, p = .02, no other factor in either risk perception scale, or

the remainder of the attitudinal factors exhibited a significant relationship.

For this relationship, a higher score, which indicates a higher level of self‐

confidence, was evident in the group that rated risk differently to the experts.

Specifically the pilots that rated the risk in a similar fashion to the expert group scored a

mean of 2.41 (SD=0.35) for Self Confidence, as opposed to those that rated the risk

differently from the expert group, that scored a mean of 2.91 (SD=0.43).

154

7.6 Discussion

The aim of Experiment Two was to determine whether a link existed between

attitudes, risk perceptions, demographic variables and the self‐reported behaviour of

GA pilots. In order to measure the relationships, two research questions were examined in Experiment Two, these are addressed individually below.

The first research question was concerned with the relationship between

attitudes, risk perceptions and self‐reported behaviour. In the current research,

attitudinal factors were found to be unrelated to the decision to undertake the flights.

This finding is relatively unique and differs from the majority of academic literature in the area of attitude and behaviour (Ajzen, 1991; Albarracín et al., 2001; Crano & Prislin,

2006; Deery, 1999; Glasman & Albarracín, 2006; Smith & Terry, 2003) in that most studies have found that attitude affects behaviour to a greater extent than was evident in the current experiment.

For example, in previous research that is related to the relationship between

safety attitudes and behaviour in a road safety setting, a strongly negative relationship

(r=‐.79) was evident such that a safer attitude led to less risky behaviour (Ulleberg &

Rundmo, 2003). A relationship of the same direction, although not as strong, is evident

in recent aviation‐based research (Molesworth & Chang, 2009). It is noted that in this

research, the relationship was not evident utilising the ASAS, but an implicit association

test. Results of analysis of the relationship between the factors on the ASAS and

behaviour were such that no relationship was evident (Molesworth & Chang, 2009).

155

It was found however, in contrast to the lack of relationship between attitudes and behaviour, that some of the risk perception factors utilised in Experiment Two were related to self‐reported behaviour. Delayed Risk, General Flight Risk, High Risk, and finally Altitude Risk were all related to behaviour in medium risk scenarios. That is, participants that exhibited a higher level of these risk perception factors were more likely to choose not to fly. This is logical, and conforms to the accuracy hypothesis of risk perception (Brewer et al., 2004), in which the perception of a risk will lead to compensatory behaviour by an individual as an attempt to reduce the amount of risk encountered and therefore perceived.

This is also consistent with Hunter’s (2006) findings, in which he utilised his

Hazardous Events Scale (a self‐report scale of past event frequency) and identified a negative relationship (which failed to achieve statistical significance) between the number of hazardous events experienced in the past and participants’ risk perceptions on the Immediate High Risk factor (r = ‐.019). In other words, those with higher

perceptions of risk were less likely to have experienced hazardous situations in

comparison to those that rated the risks as lower.

The risk perception measures utilised in Hunter’s scales utilise a single risk

factor in the scenarios. The scenarios in the current research utilise multi‐dimensional

risk factors, such that there are at least two risk factors present for each scenario. It

appears from the current results that the perception of risks in uni‐dimensional risk

scenarios is related (albeit relatively weakly) to the perception of multi‐dimensional

risk. That is, perception of risk in relatively simple and straight‐forward situations

156

seems to be related to the perception of risk in more complex (with regards to the

hazards present) situations.

The relationship between attitude, risk perception and behaviour will be again

tested in Experiment Three to determine if these factors have a measurable effect on

pilot behaviour in a simulated flight environment, as separate from self‐reported

behaviour.

The second research question was concerned with the relationship between the experiential factors of age, flight hours, recent flight time and their self‐reports of intended behaviour. The analysis failed to reveal any statistically significant difference in behaviour based upon Participant Age, suggesting that age was not a reliable predictor of the intention to engage in conservative behaviour with regards to risk‐ taking. In plain terms, age did not appear related to the decision to undertake (or not to undertake) any of the situations. An inference from the finding is that older pilots performed no better than their younger counterparts with regards to conservative behaviour.

This finding is dissimilar to those of the road safety arena, where increasing age has been linked to more conservative behaviour (Deery, 1999). The relationship between age and behaviour will be again tested in Experiment Three to determine if age has a measurable effect on pilot behaviour in a simulated flight environment

Similar to the results for participant age, it appears that flight experience is not a

reliable predictor of conservative behaviour with regards to risk‐taking; there was no

157

statistically significant relationship evident between the experiential factors of flight

hours and recent flight hours, and risk‐taking in terms of electing to undertake (or not)

the flights. Pilots with more hours, and more hours flown in the last ninety days were

just as likely to choose to fly (or not to fly) in the scenarios presented in this experiment

as those that had few or no hours in recent months. That is, having more overall flight

experience, or flying recently appears not to affect one’s decision making with regards

to risky situations. This is an interesting finding, and it is one that is deserving of further

research.

The finding that both total flight hours, and recent flight hours are not related to

behaviour is echoed in the recent aviation literature, with studies in the Australian

aviation environment finding that these two experiential factors appear unrelated to

behaviour (Molesworth & Chang, 2009).

The current finding, that age appears unrelated to self‐reported behaviour in the

Australian GA sector is not congruent to those findings in other arenas. Specifically, the current findings are at odds to those in the road safety area, where increasing age is usually thought to decrease the level of risk that one accepts and undertakes (Deery,

1999). There are many possible causes for this finding. Aviation in Australia exhibits what could be described as a healthy safety culture, in which safety publications are distributed to all pilots, and safety management systems are utilised by most, if not all operators. In contrast, drivers in Australia operate in a system where little or no testing, checking, or other safety‐related activities like training are undertaken by authorities.

Therefore, the relative weighting given to safety in each domain is different, with

158

different safety cultures prevalent. An additional difference that may contribute to the

finding is the self‐selection bias caused by the cost and relatively rigorous requirements

of flight training, when compared to the relative economy and ease of driving in

Australia.

The relationship between total flight hours, recent flight hours and behaviour

will be further tested in Experiment Three to determine if this factor has a measurable

effect on pilot behaviour in a simulated flight environment, as opposed to the self‐report

of behaviour.

As a secondary part of the experiment, risk ratings were gathered from

participants about the overall risk of the scenarios. Inter‐group testing was carried out

to determine if any difference existed between the pilots that rated the risk as per the expert group, compared to those that did not. Pilots that rated the risks of the scenarios in a similar fashion to the expert group exhibited a lower level of self‐confidence than those that rated the risks apparent in the scenarios differently. In previous research

(Hunter, 2003), no clear relationships were evident between the expert groups and

non‐expert groups.

The finding that self‐confidence may affect relative risk rating and behaviour will

be tested in Experiment Three below, to determine what, if any, effect self‐confidence

has upon behaviour in a simulation.

159

8 Experiment Three ­ Revealed Behaviour of Pilots in

a Simulation of a High­Risk Situation

8.1 Introduction

The following experiment (Experiment Three) was undertaken as the final

part of the three‐part experimental process in the present research. The experiment

examined and explored the behaviours of pilots that are undergoing flight training.

The experiment consisted of a single session, which involved three steps:

• A flight briefing followed by,

• A simulated flight of a Cessna 172 on a computer based flight simulator, in which

pilots were asked to find a stranded, but uninjured parachutist in an aircraft with

critically low fuel, which was followed by,

• An interview about the flight to determine if participants had perceived the risks

inherent in their behaviour.

Two key dependent variables were under examination, namely; the time that pilots spent away from the airport, and the minimum altitude to which they descended during the task. It was assumed that pilots who flew for longer or lower in the task were accepting a higher level of risk, than those that completed the task early or flew higher.

This is because a greater amount of fuel remaining on‐board, or flying higher than

160

minimum legal altitude would give a pilot more options for the safe operation of the

flight in most situations including normal operations, an emergency, and/or unforeseen

event/s. The justification for this is provided below.

In terms of fuel for example, having a greater amount of fuel on board may allow

the pilot to fly a more appropriate route (rather than the most direct route) if there are

hazards like large bodies of water or large expanses of uninhabited land. It would also allow a pilot to fly to an alternate aerodrome if the first is unexpectedly closed, to fly around adverse weather should it be in the area, turn back to the home aerodrome if necessary, or even to extend the flight if there are unforeseen circumstances that arise.

In the event of an emergency like an undercarriage failure, more fuel would allow the pilot to attempt to rectify problems with the aircraft for a longer period of time, which may result in the aircraft being returned to fully operational status, and averting an emergency landing.

In terms of altitude, it was assumed that pilots who flew lower during the task

were accepting a higher level of risk, as a lower altitude would give the pilot fewer

options for the safe landing due to unforeseen circumstances or recovery of the aircraft

in the event of an emergency. For example, if a pilot experiences an engine failure, more

altitude would provide the pilot more time in which to attempt a re‐start of the engine.

If the aircraft is stalled, the pilot would have more distance from terrain, or in the case

of very low flight, the pilot would simply have sufficient altitude in which to recover the

aircraft to normal flight. Similarly, in the event of inadvertent entry into unusual

161

attitude, whether through pilot mishandling or aircraft malfunction, the pilot will have

more chance of recovering the aircraft to level flight if the aircraft is at a higher altitude.

8.2 Background

A review of the first two experiments revealed the following;

Experiment One, found that one attitudinal measure (Self Confidence) was

negatively related to age, flight experience and recency. The influence of attitude and

experience on behaviour was further tested in Experiment Two.

In Experiment Two it was identified that some of the risk perception factors

utilised were related to self‐reported behaviour. Delayed Risk, General Flight Risk, High

Risk, and Altitude Risk were all related, in a positive direction to behaviour, albeit

limited to flight situations that could be described as medium risk. That is, participants that exhibited higher ratings on the risk perception factors were more likely to choose not to fly than those that rated these risk factors lower.

Experiments One and Two found that there was no relationship evident between

the normal measures of experience and competence in the aviation field, and the

majority of risk perception, and attitudinal measures. Nor was there a relationship

evident between experience and self‐reported behaviour. In other words, it appears that during normal flying operations, older and/or more experienced pilots will not exhibit superior perception of the risks, attitudes towards safety, or behaviour in a

given operation.

162

Experiment Three (the present experiment) intends to replicate the examination of these variables (Attitude, Risk Perception and Experiential variables) with one important difference. Specifically, Experiment Three will employ actual behaviour as captured in a flight simulation, opposed to self‐reported behaviour.

Consistent with the previous experiment (Experiment Two), two questions were proposed to explore the second and third research questions (from Chapter 5).

Specifically, these are:.

1. What is the relationship between the attitudes, risk perception and the

behaviour (simulated flight behaviour) of pilots in the general aviation training

sector?

2. What is the relationship between a pilot’s experience, as measured by age, flight

hours, and recent flight time, and behaviour (simulated flight behaviour) in the

general aviation training sector?

8.3 Experimental Design

Experiment Three was designed as the final experiment in the current experimental process. It was a flight‐simulator based experiment designed to provide performance data from participants, rather than the self‐report data that the first two experiments produced.

Experiment Three is relatively unique to the field, as it utilised two dependent variables, as opposed to the traditional one dependent variable (for example, see Goh &

163

Wiegmann, 2001; Molesworth et al., 2006; O'Hare & Smitheram, 1995) in order to

gather data about the risk management behaviour of pilots. Similarly, this experience

allowed for the examination of risk management on a continuum from low to high

opposed to a forced position of completing the task/exercise.

Participants that were recruited to undertake Experiment Three had previously

undertaken Experiment One, but had not undertaken Experiment Two. In this way, the

same scenario (i.e. the Moruya Scenario) could be presented in Experiment Three as in

Experiment Two, and the chance that behaviour would be modified through previous

exposure to the scenario was removed. Participants in Experiment Three were those

that were able to attend a simulation session at a later date, after they had undertaken

Experiment One.

This strategy also allowed for a wider cohort of pilots to be tested (as it included

those that could not undertake a simulated flight), to test for the difference between self‐report and revealed performance, and for between‐groups testing with regards to those that flew the simulator, and those that completed the survey.

The experiment was designed to be a two‐part experiment. The first part of the

experiment is the simulation itself; the second part of the experiment was an interview

in which pilots were asked to give their reasoning and the factors behind the decisions

that they made during the flight. This part of the experiment also allowed for a cross‐

check that participants had actually perceived the hazards and risks present. This

allowed a check for the assumption that pilots were behaving with regard to perceived

risk, or acting with the knowledge of the risks present, rather than undertaking the

164

experiment without knowledge of possible hazards. In this way, conclusions about risk

management behaviour, and the link between risk perceptions and behaviour were

possible.

The simulation in Experiment Three was designed to be as similar as possible to

the ‘Moruya’ scenario presented in Experiment Two (given above). The simulation was

set at Moruya aerodrome (and the surrounding local area) on the south coast of New

South Wales, Australia. This location was chosen for a variety of reasons.

First, it is a real Australian location. This meant that pilots would be aware of all

relevant operational regulations. Second, Moruya aerodrome is not extensively used as

a primary way‐point by pilots located at Bankstown aerodrome for cross‐country flight

training exercises. This would reduce the chance that participants were intimately

familiar with the aerodrome, and would allow for the experiment to test the behaviour

of pilots in novel, high‐risk situations without the need for further controls (i.e., pre‐

screening of pilots to ensure that they were unfamiliar with the situation and setting).

Third, the aerodrome is such that its runway is orientated north‐south. This allowed the

simulation to be simplified, in that arrival procedures would be much simpler, as pilots

would naturally be flying in the direction required for final approach on their return

from Batemans Bay.

Moruya aerodrome is also close to Batemans Bay. This allowed the simulation

time to be kept to a minimum, so that participants were not forced to spend large

amounts of time in the experiment. It also allowed for the interview to be undertaken

165

immediately after the simulation, without participants becoming fatigued or frustrated

by the time spent in the experiment.

The scenario used was designed to be an abnormal operation, and one that was

novel to all participants. This was done so that, as far as possible, external, uncontrolled

variables, such as situational specific training, or experience with the area or search

rescue, were not a factor that had to be controlled for using other methods, such as

selective recruitment. The method therefore allowed for direct comparison between pilots based upon their choices with regards to the dependant variables chosen in the experimental design (time in flight and minimum altitude flown).

Participants were told that they were flying circuits at Moruya aerodrome. The

aircraft that they were flying had previously been used on a cross‐country exercise, and was low on fuel. On the downwind leg of their last circuit, the operator of the aircraft radioed them, and asked the pilot to fly to Batemans Bay to search for a parachutist that had landed away from the designated landing area on the shores of Batemans Bay.

The scenario started with the aircraft on the downwind leg at 1,000 feet, heading

(pointing) towards Batemans Bay. If they chose to undertake the simulation, pilots

would depart the circuit to the north, track to Batemans Bay, conduct their search, and

return to Moruya. The participants were told that the aircraft in the simulation had 18

minutes of primary flight fuel remaining, and 5 minutes of fixed reserve fuel on board.

This was based upon the fuel burn figures for normal cruise at 2,400 RPM or

approximately 100 knots. Batemans Bay is approximately 10 to 12 Nautical miles from

Moruya Aerodrome, depending upon which shore is used for the measurement, and

166

exactly which point on the bay is measured, as it is significantly wider at the mouth than

at the town of Batemans Bay.

Pilots were told that once they had sighted the parachutist, there were to report

it, fly a full orbit (360 turn), and return to Moruya. Participants were not informed as to

exactly when they should return if they did not find the skydiver. No instructions were

given as to how pilots should return to the aerodrome or what kind of approach should

be made.

Participants were provided with this information in a written brief that was

given to them before stepping into the simulator. This was done because the simulator

room is small and blackened in order that the projection screens work most effectively.

Pilots were asked to fly the simulator exactly as if they were in a real aircraft.

This simulation was designed so that pilots could complete the task legally. That is, the

task could be completed within the time period given, and there was no requirement to

break any low‐flight regulations. It was also designed to allow pilots to take a course of

action which they thought appropriate with respect to both time and altitude, without being overtly restrictive, or prescriptive.

Pilots were free to undertake any behaviour and by extension, level of risk that

they deemed appropriate in all facets of the flight. Pilots could choose the amount of fuel

used from practically nil by landing immediately (lowest risk) to flying for eighteen or

more minutes (highest risk). The altitude to which the pilots descended or ascended

167

was also completely under the control of the pilots, no limits other than those set by the aircraft’s limitations were given (the regulations notwithstanding).

The experimenter ensured that pilots all understood the written briefing, knew the location of their aircraft, and the expectations of the experimenter by providing this final oral briefing:

“The aircraft is currently at 1,000 feet, on the downwind leg of a right‐hand circuit for runway 18 at Moruya Airport. The aircraft is abeam the downwind threshold of the runway, and is therefore more than halfway down the downwind leg. The airport is below and to your right. Batemans bay is directly north of your position, which is straight ahead. If you choose to fly to Batemans, head north for approximately 5 minutes, at that time you will be able to see the bay. If you choose to land the aircraft immediately, turn base at the road that you can see in front of you. Remember that you are to treat this flight exactly as if you were in a real aircraft. Do not do anything today that you would not do in a real plane. Remember, the operation is not an emergency.”

Once pilots had completed the simulation by landing the aircraft and pressing the pause button, they were moved into the interview room adjacent to the simulator in order to complete the interview.

The goal of the interview was, on a global scale, to derive an answer to; did the pilots perceive the hazards present in the simulation, and behave in accordance (or in spite) of this knowledge?

168

As a secondary goal, the interview aimed to discover the answers to three basic questions:

• To find out why pilots made the decision that they made,

• To discover when the decision was made, and

• To find out how the decision was made.

The interview was designed as a semi‐formal interview, in that certain questions were to be asked (see Appendix 9), however the interviewer had the freedom to ask additional questions, or to ask for clarification by way of modifying the original question. The pre‐determined questions that were asked of all participants were concerned with the major decision points of the flight. There were judged to be five such major choices.

The first was the decision to leave the aerodrome area. Pilots were asked about their decision in regards to this. The second was the decision as to how to best fly between Moruya and Batemans Bay. The third decision was concerned with the search and rescue portion of the flight over Batemans Bay. The fourth was to do with the decision to return to the airport. The fifth and final decision was concerned with the individuals’ decisions whilst returning to Moruya regarding their approach and landing at the airport.

As the interview was semi‐formal, the interviewer was able to ask participants about severe or abnormal behaviour, and also able ask for clarification from pilots when

169

their answers were judged to be given in erroneous perception of the intent of the

question or were given as a result of a lack of understanding of the intent behind the

question.

Consistent with Experiment Two, it was hypothesised that attitudes (see for

example, Ajzen, 1991; Ajzen & Fishbein, 2005; Armitage & Conner, 2001) and risk

perceptions would effect behaviour (see Brewer et al., 2004).

8.4 Method

8.4.1 Participants.

Fifty‐six participants were recruited from the student population currently

enrolled in the Bachelor of Aviation (Flying) degree at the University of New South

Wales, graduates of the program, and from pilots that have trained at other flight

training schools located at Bankstown aerodrome. The participants had previously

completed part one of the experimental process (Experiment One) and had made

themselves available for this experiment.

The experimental process was designed such that pilots flew the simulator

within two weeks of completing their survey (Experiment One) as far as practical. This

ensured the accuracy of the experiential measures used, insofar as pilots would not

have had the opportunity to fly a significant number of hours between the time of

capture, and the time of the simulation. All participants were aware that the research

was focused on the perceptions, attitudes and behaviours of general aviation pilots, but were unaware of any further details of the research project’s purpose. Table 12 below

170

outlines the experiential characteristics of the participants. Because the participants were recruited from schools dealing with ab‐initio training, specialising in school‐leaver courses, the data was expected to represent the pilot demographics of this targeted cohort, rather than a population wide distribution.

Table 12. Experiment Three Participant Statistics

Participants # Age (SD) Flight Hours Total Hours in the (SD) Last 90 Days (SD) 56 20.07 (2.05) 114.85 (109.13) 17.02 (17.35)

There were nine female pilots in Experiment Three. A total of nine pilots held a

PPL, 20 held a CPL and the remaining pilots were unlicenced at the time of the research..

There were four pilots that held instructor ratings, and five with instrument ratings.

8.4.2 Materials.

The material utilised for the experiment was a PC based flight simulator. The simulator comprised two personal computers, one nineteen‐inch Liquid Crystal Display

(LCD) monitor, a Personal Computer Aviation Training Device (PCATD) and Elite

Rudder Pedals. The personal computers were used to operate the PCATD such that it was possible to have two independent visual displays for the simulator; one for exterior views, the other for interiors. The flight simulator software was X‐Plane 8.40 developed by Laminar Research Corporation. The PCATD hardware consisted of a remote instrument console and a Cirrus Two flight console with a Beech Yoke (without clock option). The remote instrument console consisted of five dials that operate the

Automatic Direction Finder (ADF), directional gyro, heading bug indicator, altimeter and Omni Bearing Selector (OBS). In addition, the remote instrument console

171

incorporated a switch that enabled the user to select one of two different

communication radios. Whilst users had access to these controls, it was made clear to

them that none of them required any manipulation in the flight, as all relevant settings

of these controls were pre‐set.

The Cirrus Two Flight Console with the Beech Yoke had the following controls

(not all features are specific to a Cessna 172):

• Rudder trim control indicator lights • Magneto switches • Rudder trim control • Engine start switch • Cowl flap switch • Alternator switches • Fuel boost pump switches • Battery switches • Fuel tank selector • Avionics master • Parking brake • Flap switch • Landing gear switch with position

Mounted on the top of the left side of the yoke was a rocker switch that operated the electric pitch trim, while a switch on the right side of the console permitted the pilot to view 360 degrees around the aircraft in 45 degree increments each time the switch was depressed. The control yoke was capable of full motion in two dimensions, as per normal aircraft controls. In addition, a single engine throttle quadrant was used, and was mounted on the right side of the flight console (as per its normal location in a

Cessna 172). The rudder pedals used in the study slid forward and aft, and incorporated a toe brake. The PCATD hardware used in the experiment was manufactured by

Precision Flight Controls.

The visual information was presented to the participants on two screens. The

internal view, which displayed the pilot’s side instrument panel of a Cessna 172,

including the airspeed indicator, altimeter, vertical speed gauge, artificial horizon, turn

172

and slip indicator, and directional gyroscope was displayed on a nineteen‐inch LCD. The exterior view, which displayed the outside view from the aircraft, was displayed using a

Hitachi CP‐X445 projector that projected onto on a two metre square screen. This was located directly in front, and outside of the simulator.

A typical view that the participant would see throughout the flight consisted of the console, on top of which the participants viewed the aircraft’s instruments on the screen inside the simulator, and the terrain was viewed by looking out of the windows of the simulator. A flight data recorder, which is a function of the X‐Plane software, was used to record the input from the pilot and the position of the aircraft. Five specific data points were saved five times per second and included:

• Time elapsed

• Throttle

• Pitch, roll, heading

• Latitude, longitude, altitude

• Distance travelled.

8.4.3 Procedure.

Participants were informed of the nature of the experimental process

verbally, and in writing from the information sheet provided (See Appendix 1).

Participants were briefed verbally that they would be required to commit to an

interview session to be conducted as soon as their simulator session was completed.

173

Participants were then issued with a flight briefing sheet (see Appendix 8), a

Visual Navigation Chart (VNC) of the area, and a ruler that was pre‐marked with both a

10 and 12 nautical‐mile reference, so that pilots could accurately measure the distance

between their position, and their target area.

The participants were all put through the same simulated flight scenario. The

flight scenario was based upon the premise that participants were flying circuits at

Moruya aerodrome. The aircraft that they were flying had previously been used on a cross‐country exercise, and because of this had very little fuel remaining. On the downwind leg of their last circuit, the owner/operator of the aircraft radioed them, and asked the pilot to fly to Batemans Bay to search for a parachutist that had landed away from the designated landing area on the shores of Batemans Bay.

The participants were told that the aircraft in the simulation had 18 minutes of

primary flight fuel, and 5 minutes of fixed reserve fuel on board. Batemans Bay is

approximately 10 to 12 Nautical miles from Moruya Aerodrome, making the flight to the

bay approximately 5‐6 minutes.

If pilots chose to undertake the flight, and they found the parachutist, there were to report it, fly a full orbit, and return to Moruya. Participants were not informed as to

exactly when they should return if they did not find the skydiver. No instructions were

given as to how pilots should return to the aerodrome or what kind of approach should

be made. Pilots were told that no radio calls were necessary for the flight, except for

when/if they found the parachutist, when they would treat the researcher as the

company, and were to inform the researcher of their discovery.

174

Pilots were also briefed on their aircraft’s exact location and orientation (using the briefing outlined above) and, once pilots had been given these briefings, they were asked if they had any questions. This served as the last check before the flight that pilots had fully understood the briefing, and that they would undertake the flight as per instructions. When participants were ready, the simulation (and data capture) was commenced by the participant pressing the pause/unpause button on the keyboard in the simulator. Once pilots finished their flight by landing at Moruya, the simulator was stopped.

The simulator was such that it provided the normal indications available to pilots in a real aircraft. That is, there was no extra information such as elapsed time, fuel burn, or the like. No GPS indication was included either, as this feature is not universal in the Cessna 172 fleet. Therefore, pilots had the aircraft clock and the fuel gauges to give them feedback on their progress with regards to fuel. Pilots also had their personal watch with which to gauge time. Indeed, pilots utilised their watches as the timekeeping equipment almost exclusively.

This simulation was designed so that pilots could complete the task legally, that is, there was no requirement to fly below the minimum altitude permissible under

Australian regulations, nor was there a need to break regulations regarding fuel minimums. This would therefore allow pilots to take a course of action that they thought appropriate with respect to both time and altitude, without being overtly restrictive, suggestive or prescriptive. It was not possible in the simulator however, to find the skydiver in the time allotted to the participants.

175

To evaluate pilots’ awareness of the hazards present, and to gain further insight into why pilots made the decisions that they made, a de‐briefing interview was conducted after each participant had landed the aircraft.

The goal of the interview was, on a global scale, to answer one basic question; how did the pilots perceive the hazards present in the simulation and did they behave in accordance (or in spite) of this knowledge?

In order to achieve this, the interview aimed to discover the answers to two questions;

1. To find out why pilots made the decision/s that they made,

2. To discover the most salient factors in this decision

8.5 Results

8.5.1 Data analysis.

The flight data from X‐Plane was transferred to SPSS version 13 for Macintosh using an import template made using the program’s import feature. Table 13 below shows the descriptive statistics of those pilots that were included in Experiment Three.

It was found that five pilots had a very high number of flight hours compared to the population (between 660 and 1,200 hours). These pilots’ flight hour data were transformed in order to obtain normality of the data set by replacing their extremely high value, with a value one greater than the next most extreme (Tabachnick & Fidell,

2013). In this case, the data was changed from 660 flight hours (one pilot reported this

176

level of experience), to 351 flight hours, and 1,200 hours (four pilots reported this level) to 352 hours.

177

Table 13. Descriptive Statistics for All Pilots in Experiment Three

Factor NM SD R ange

Demographics Participant’s Age 56 20.07 2.05 9.00

Total Flight Hours * 56 146.99 236.72 352.00

Total Hours In The Last 90 Days 56 17.01 17.35 62.00

Risk Perception Delayed Risk 56 67.81 11.09 51.88

One Nominal Risk 56 35.38 13.77 64.00 Immediate High Risk 56 78.29 9.32 45.00 (0 – 100) Risk Perception General Flight Risk 56 40.45 17.62 95.20

Two High Risk 56 51.87 15.32 86.30 Altitude Risk 56 51.53 16.94 93.71 (0 – 100) Driving Risk 56 54.71 17.64 74.67

Everyday Risk 56 33.73 15.17 75.25

ASAS Self Confidence 56 2.27 0.43 2.07

(1 – 5) Risk Orientation 56 1.43 0.50 2.00

Safety Orientation 56 2.73 0.48 2.00

* ‐ data was transformed

In the table above, a higher score for RP1 and RP2 indicated a higher rating of the risks for this risk type. A higher score for the ASAS indicated a higher level of agreement with the statements that constitute that attitude.

From the fifty‐six participants, thirty‐six (64%) elected to undertake the exercise

as briefed, while twenty decided to immediately land the aircraft.

The twenty (36%) that elected ‘not to go’ will be described immediately

following under the heading ‘No‐Go Pilots’. The thirty‐six who elected to ‘go’ will be

178

described later in this chapter under the heading ‘Go pilots’. It should be said that the caveat for all findings for the ‘go’ pilots is that they relate to the decision making and behaviour of pilots, with the knowledge that they had decided to undertake an extremely risky flight (they had made the decision to go), whereas the findings for the no‐go pilots are in the light of the fact that they chose to minimise the risk that they undertook as much as they could.

A comparison of the descriptive statistics for these two pilot groups is provided in Table 14 below.

179

Table 14. Comparison of Descriptive Statistics for “Go” and “No­Go” Pilots

Mann Whitney No‐Go Pilots Go Pilots Test results

n Mean SD n Mean SD Demographics Participant’s 20 19.90 1.94 36 20.08 2.20 0.12 Age Total Flight 20 125.20 104.38 36 107.74 113.68 0.71 Hours

Total Hours 20 13.16 12.56 36 18.76 19.47 0.47 in the Last 90 Days Risk Perception Delayed Risk 20 71.19 13.35 36 64.88 10.15 2.16

One (0 – 100) Nominal Risk 20 39.69 15.63 36 31.90 11.53 1.74 Immediate 20 81.41 10.53 36 75.55 8.89 2.46* High Risk Risk Perception General 20 41.38 22.75 36 38.38 15.27 0.39 Flight Risk Two High Risk 20 52.86 18.65 36 49.67 14.86 0.46

(0 – 100) Altitude Risk 20 54.56 21.99 36 48.11 15.05 1.63 Driving Risk 20 58.55 19.78 36 51.30 17.41 1.79 Everyday 20 35.93 16.72 36 31.31 14.74 2.34 Risk ASAS Self 20 2.25 0.42 36 2.30 0.41 0.56 Confidence (1 – 5) Risk 20 1.49 0.48 36 1.39 0.52 0.63 Orientation Safety 20 2.89 0.38 36 2.64 0.52 1.88 Orientation

In the table above, a higher score for RP1 and RP2 indicated a higher rating of the risks for this risk type. A higher score for the ASAS indicated a higher level of agreement with the statements that constitute that attitude.

* denotes p<.05

180

8.5.2 ‘No­Go’ pilots.

Approximately one‐third of pilots in this research elected not to leave the circuit area, effectively choosing not to undertake the simulation, thus reducing their risk‐ taking to the minimum possible level. This is an interesting finding in itself, and further analysis of these pilots is important to further understand the differences between these pilots, and the participants that undertook the simulation (who took a greater risk than the No‐Go pilots). The descriptive statistics of the pilots that elected not to exit the circuit in the simulated flight are shown in Table 14 above.

The pilots that chose not to go obviously behaved differently to their peers that did. The important question out of this difference is what causes the difference to occur?

Given that the current research is particularly concerned with the role of attitude, risk perceptions and experiential values, these will be used to test if the no‐go pilots are different on these factors, than the pilots that went. In other words, do the pilots that did not fly have a better (or worse) attitude? Do they possess a better level of risk perception? Are they more or less experienced than the pilots that flew? These questions are posed and explored below.

8.5.2.1 Difference in risk perception of go and no go pilots

In order to determine whether a difference existed between pilots (Go and No‐

Go) in terms of risk perception (as measured by Risk Perception Scales One and Two – reliabilities of Hunter’s scales can be found in section 6.2.1 above), a series of Mann‐

Whitney nonparametric tests were employed. This test was used, as the numbers within groups varied by a considerable amount (Tabachnick & Fidell, 2013). Alpha was set at

181

.025 after Bonferroni correction (0.05/2). With correction for ties and z‐score

conversion, there was a statistically significant difference between pilots (Go and No‐

Go) for one risk perception factor, Immediate High Risk, z (N = 56) = 2.46, p = .01. For this relationship, a higher score, which indicates that the individual perceived the risk evident within a given situation as higher, was evident in the group that did not go.

This result indicates that only one of the factors constituting Risk Perception

(Immediate High Risk) appears related to pilots’ decision to minimise the risk that they

are exposed to by electing to land immediately. This factor, Immediate High Risk, relates

to Hunter’s Risk Perception Scale – Other, where respondents rate the level of perceived

risk for a third person, in this case another pilot GA pilot.

8.5.2.2 Difference in attitude between Go and No­Go pilots

Consistent with the previous analysis, a series of Mann‐Whitney nonparametric

tests were employed in order to determine whether a difference existed between pilots

(Go and No‐Go) in terms of attitude (as measured by the ASAS). Alpha was set at .025

after Bonferroni correction (0.05/2). With correction for ties and z‐score conversion, there was no statistically significant difference found, largest z, (Safety Orientation) z(1,

N = 56) = 1.88, p = .06. Therefore, it appears that Go pilots did not differ from No‐Go pilots on any of the aviation attitude variables. This suggests that the construct of attitude as measured with the ASAS, is unrelated to the decision to land immediately, and therefore minimise risk exposure.

182

8.5.2.3 Difference in participant experience and age Between Go and No­Go Pilots

Consistent with the two previous analyses, a series of Mann‐Whitney nonparametric tests were employed in order to determine whether a difference existed between pilots (Go and No‐Go) in terms of experience, recency, and age. Alpha was set at .025 after Bonferroni correction (0.05/2). With correction for ties and z‐score conversion, there was no statistically significant difference found, largest z, (Total Flight

Hours) z (1, N = 57) = 0.71, p = .48. This result suggests that the typical markers of competence, and metrics used for advancement within the industry had little affect on the decision to undertake a potentially risky flight in the studied pilots.

8.5.3 Go pilots.

Descriptive statistics for the go pilots are provided in Table 14 above. The behavioural statistics from the simulated flight are provided in Table 15 below.

Table 15. “Go” Pilot Behavioural Scores

Variables Mean SD Total Time in Flight (minutes) 17.80 4.84 Minimum Altitude in Flight (feet) 690.20 223.76

183

Table 16. Correlation Analysis Between the Fourteen Predictor Variables and the Two Dependent Variables

(Criterions).

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

1. Age —– 2. Total Hours .615** —– 3. Recent Hours .438** .819* —–

4. Delayed Risk ‐.083 .110 .184 —– 5. Nominal Risk ‐.063 .290 .328 .456** —– 6. Immediate High Risk .155 .293 .199 .271 .351* —–

7. General Flight Risk .018 .339* .048 .384* .402* .190 —–

8. High Risk ‐.057 .216 ‐.076 .376* .343* .278 .861** —– 9. Altitude Risk .167 .341* .048 .372* .283 .424* .732* .779* —– 10. Driving Risk .039 .174 ‐.027 ‐.020 .083 .404* .296 .376* .360* —– 11. Everyday Risk .269 .521* .291 .143 .350* .341* .662** .501* .506** .453** —– 12. Self Confidence ‐.243 ‐.380* ‐.290 ‐.288 ‐.298 ‐.308 ‐.075 ‐.171 ‐.165 ‐.225 ‐.134 —– 13. Risk Orientation ‐.066 .062 .069 .048 ‐.027 ‐.227 ‐.209 ‐.253 ‐.285 ‐.322 ‐.273 ‐.217 —– 14. Safety Orientation ‐.111 .232 .310 .215 ‐.522**.378* .168 .191 .345* ‐.036 .373* ‐.104 ‐.009 —–

15. Total Time in Flight .128 ‐.208 ‐.158 ‐.164 ‐.262 ‐.238 ‐.056 ‐.262 ‐.135 ‐.228 ‐.010 ‐.382* .036 .003 —–

16. Minimum Altitude in Flight ‐.330* ‐.196 .061 .122 ‐.068 ‐.228 ‐.098 ‐.036 ‐.056 ‐.058 ‐.156 ‐.189 ‐.021 ‐.033 ‐.173 —–

* p < 0.05 ** p < 0.01

184

In order to investigate the relationship between risk perception, attitude,

experiential variables, and behaviour, a series of multiple linear regressions were to

be employed. These regressions were to use the variables of Total Flight Time (time

to complete the scenario) and Minimum Altitude During Flight as the criteria in two

separate analyses. The predictor variables were to be gathered from the experiential

(Participant Age, Total Flight Hours, Total Hours in the Last 90 Days) attitudinal

(Self‐Confidence, Risk Orientation and Safety Orientation), and risk perception

variables (Delayed Risk, Nominal Risk, Immediate High Risk, General Flight Risk,

High Risk, Altitude Risk, Driving Risk, and Everyday Risk). The predictor variables

used in the regression were to be finalised based upon the outcome of a correlation

analysis to determine which variables were appropriate.

Prior to this it was important to establish that the two risky flight

behavioural measures (Total Flight Time and Minimum Altitude in Flight) were

independent of one another, hence the relationship between the two variables were

examined. A Pearson product‐moment correlation failed to reveal a statistically

significant relationship between Total Flight Time and Minimum Altitude in Flight, r

(34) = ‐ .173, p = .314. This result suggests that the two risky behavioural measures are not significantly related, and therefore are independent of one another.

185

8.5.3.1 Risk perception and behaviour of Go Pilots

In order to determine which variables would feature as predictor variables in the multiple regression, a correlation analysis was performed (Tabachnick & Fidell,

2013). This analysis was performed with Total Flight Time and Minimum Altitude

During Flight as the dependant variables, and the 8 variables from Risk Perception

Scales One and Two as the independent variables.

The correlation analysis failed to reveal any significant relationships

between the factors in Risk Perception Scale One, Risk Perception Scale Two, and

the behavioural variables of Total Time in Flight and Minimum Altitude in flight, largest r (34) = ‐ .262, p = .122. Therefore, no regression analysis was performed, as it would add nothing further to the analysis.

In summary, the results suggest that the risk perception factors measured in

Risk Perception Scale One and Two had no relationship to the risk taking behaviour

of pilots with regard to minimum altitude or time flown during the flight. Therefore,

risk perception factors were found to have little affect on pilots’ behaviour, once

they had decided to undertake the flight.

8.5.3.2 Attitudes and behaviour of Go Pilots

Consistent with the previous analysis, a correlation analysis was first

performed in order to determine which variables would feature as predictor

variables in the multiple regression for attitudes and behaviour, (Tabachnick &

186

Fidell, 2013). This analysis was performed with Total Flight Time and Minimum

Altitude During Flight as the dependant variables, and the three attitudinal variables

mentioned above as the independent variables.

There was one significant correlation found, which was between ‘Self

Confidence’ and Total Time in Flight, r (34) = .38, p = .02. Therefore, no multiple

regression was performed as the regression analysis would add nothing further to the analysis. The size and direction of the relationship suggested that pilots that flew

for a longer period of time in the simulation had a higher level of self‐confidence

compared to their counterparts who flew for a shorter period in the simulation.

In terms of the second dependent variable, minimum altitude, the results

failed to reveal any significant relationship between the attitude factors from the

ASAS and the minimum altitude to which pilots descended during the simulation,

largest r (34) = ‐ .189, p = .269.

In summary, the results revealed that there was little evidence that the

attitude of pilots towards aviation safety has an effect upon behaviour, with only one attitudinal variable showing a statistically significant relationship with behaviour. This will be discussed in further detail below.

8.5.3.3 Participant experience and behaviour of Go Pilots

Consistent with the previous two analyses, a correlational analysis prior to a

multiple regression was performed to examine the relationship between the

187

experiential variables (flight hours, recency and age), and the two behavioural

measures (Tabachnick & Fidell, 2013).

The correlation analysis revealed one significant negative relationship between the factor of Participant Age and the behavioural variable of Minimum

Altitude in flight, r (34) = ‐.330, p = .05. Therefore, as above, no multiple regression was performed. The size and direction of the relationship suggests that older pilots flew lower in the simulation in comparison to their younger counterparts.

In summary, the results suggest that the experiential variables had little

effect on the risk‐taking behaviour of pilots with regard to time flown during the

flight. Only one variable, Participant Age was significantly related to a risk‐taking

behaviour variable of Minimum Altitude During Flight. Therefore, experiential

factors were found to have relatively little effect on pilots’ behaviour, once they had decided to undertake the flight. This will be discussed in further detail below.

8.5.4 Interview results.

Pilots were interviewed after they had landed the aircraft. The goal of the

interview was, as stated above, to determine whether pilots had perceived the

hazards present in the simulation and behaved in accordance (or in spite) of this

knowledge? Secondarily, the interview aimed to discover the answers to two basic

questions;

1. To find out why pilots made the decision/s that they made,

188

2. To discover the most salient factors in this decision

Table 17. Comparison of the Factors that Contributed to the Go/No Go

decision in the Simulator

Decision to Fly/Not Fly Go No Go Fuel State 63% 90% Safety of 16% 0% Skydiver Navigation 19% 0%

Safety 2% 0% Lack of 10% Appropriate Skill Level

Yes/No, Was the Scenario Se en as a Hazard? Go No Go 84% Yes 100% Yes

The first decision made by all pilots in the experiment was the initial choice made by the pilot as to whether they would fly the simulation or not. In more direct terms, what are the reasons that pilots elected to fly?

The decision to fly to Batemans Bay was made by all participants before the flight began, as indicated by the fact that no pilots reported undertaking the flight

189

without prior thought. The decision to fly (or not to fly) was, however, made in varying ways.

Most often, pilots were aware of the lack of fuel in the simulation, 79% of Go pilots reported it as an important factor in their decision, whilst 90% of No‐Go pilots cited the lack of fuel as the reason for their decision. The pilots that flew appeared to treat it as a limitation to their flight, rather than a factor that made the flight untenable. A quote from a Go‐Pilot sums up the general theme of the justification for undertaking the simulation,

“The weather was fine, you did have enough fuel to get there and back, there’s a chance you could be of some help, so you might as well give it a go.”

The general indication given by No‐Go participants was that the lack of fuel was an unacceptable risk (90% of pilots cited this factor). A quote from one of the pilots that did not go provides a succinct counterpoint to the attitude of the pilots that went;

“The idea of flying with stuff all fuel on board . . . just nuh. You’d have to be an absolute (expletive removed) to do anything worse really.”

19% of Go pilots cited navigation as either a primary factor (explicitly saying

“primary”, “the first thing” or equivalent) in their decision‐making, while 3% cited this as a secondary factor (by saying “another thing” or equivalent) in their decision‐

190

making (22% combined). These pilots had therefore already made their decision to

go, and were already deciding upon the detail of how to get out and back most efficiently, rather than making more basic decisions, like whether or not to fly the simulation.

Of the Go pilots, 16% cited the safety of the skydiver as the primary factor in

their decision to fly the simulation. These pilots seemed to place the safety of the

skydiver, or the fact that they were the closest available person to render aerial

assistance, above the immediate safety concerns for themselves or their aircraft. Of

these pilots, only one seemed to have been primarily swayed in their decision to fly

based on the fact that somebody had asked for their assistance. The following quote

is an excerpt from their interview when asked exactly why they had chosen to fly.

“Cause the guy just landed, and I’m only five or six minutes away and I can

make it back so. . . There’s a runway there, so why not. . . If I had ten minutes of fuel I

would have landed and gone back, ’cause I knew that the first few minutes are the

most important. Probably out of necessity, like you’ve gotta do it. You know?”

Interestingly, this notion of feeling obliged to help was not limited to those pilots that mentioned it as a primary factor in their decision to fly. One pilot, who had previously spoken of fuel being the primary concern in their decision to fly, said later in the interview,

191

“Cause I was the only aircraft in the area. If I heard that there was someone

that had gone missing or not landed where they had supposed to, you want to try

and help out. Obviously if I didn’t have enough fuel, I wouldn’t go, but the time it was

going to take to go out there and back was about 5 minutes ’cause time to search, so

it’s what I’d do if it actually happened.”

10% of the No‐Go pilots cited their lack of flying skills as their primary

reason not to undertake the flight. That is, they had not completed navigation training and were not confident of being able to find the airport again, once they had

left the circuit area. Given that these pilots had not been exposed to unsupervised

flight outside of their local training area, this decision was appropriate and based upon sound reasoning.

All of the No‐Go pilots identified the simulation as a significant risk. In

contrast to this, only 84% of Go‐Pilots indicated that they though that the flight was

a significant risk (the remaining 16% indicated that it was not). Given that the

majority of these pilots had identified fuel as a major factor in their decision to go, it

is argued that they had correctly identified the hazard of low fuel, but incorrectly

assessed and managed the risk that was a result of that hazard. This gives further

validity to the finding above that risk perception has the greatest influence of any

tested variable on the decision to go or not in the simulation.

192

Again, the majority of the go‐pilots were aware of the fuel state of the

aircraft, yet they typically treated it as a limitation to their performance, rather than

as an unacceptable risk factor for the flight. One quote from an interview sums up

the attitude of a large number of the pilots that went, insofar as pilots appeared to

feel that having five minutes of fuel as a buffer was an acceptable risk in the light of

the circumstances ‐ that another person requires help, which they can provide.

“It was low, but it wasn’t low enough. If it was thirteen minutes, six and a half there, and six and a half back and there was thirteen minutes of fuel on board, obviously I wouldn’t have said ‘no, I’m not going, it’s cutting it too fine’. But I thought

eighteen minutes would be enough, plus the five‐minute reserve, would be enough

to get there, get back, and have at least a minute or two to search for the skydiver.

And also another reason I guess was I guess the feeling of selfishness ‐ saying that

just because I have a certain amount of fuel, there could be someone out there

injured, so I thought I might as well take a bit of a risk, in the event that someone

was lying injured in the middle of the bush from a skydiving accident. So that came

into my thinking as well”

There was some evidence that pilots went further in their assessment of the

situation, in that they believed variously that they had sufficient control of the

situation, or that their fuel management skills were advanced and therefore capable

of managing this slight, or more generically, that the lack of fuel was not a cause for

193

serious concern as it was well within their ability to conduct a flight of this type. An

example of this thinking is as follows,

“I didn’t think it was a risk on safety at all. I believed that there was enough

fuel to get there and back, and I couldn’t see any problems with it”

Given that some of the Go‐Pilots cited a perceived obligation as a factor in

their decision to undertake the simulation, it is prudent to discuss the possibility of

the Hawthorne effect being at play. All pilots that cited an obligation to go, cited an obligation to the skydiver, or to their flight school (the school in the scenario), not to

the experimenter. Also, many pilots chose not to go, and did not identify any need to please, or undertake the simulation because the experimenter had asked them to.

Because of these factors, it is surmised that pilots likely felt little or no pressure to undertake the simulation based on the fact that the experimenter was present.

However, despite this inference, it is not acceptable to discount a possible

Hawthorne effect. Hence, there is a chance that pilot behaviour was modified by the presence of the experimenter.

8.6 Discussion

Experiment Three was designed to determine whether there was a

relationship/s present between attitude and behaviour in high‐risk situations. From

this, Experiment Three was therefore an attempt to provide some evidence for the

194

fundamental question of whether an individual’s attitude is related to their

behaviour, (as is suggested by many theorists – see Ajzen, 1991; Ajzen & Fishbein,

2005; Crano & Prislin, 2006; Glasman & Albarracín, 2006) in the Australian GA

setting.

More specifically, the aim of Experiment Three was to investigate the

revealed risk‐taking behaviour of GA pilots in a flight simulation, and to determine

whether a link existed between attitudes, risk perceptions, experiential variables

and behaviour.

The first research question in Experiment Three was concerned with the

relationship between attitudes, risk perceptions and behaviour.

The first decision made by pilots in Experiment Three was the decision as to whether or not they would choose to undertake the flight as briefed. Approximately one‐third of pilots elected not to leave the circuit in the simulation, effectively minimising the risk that they undertook as far as they could. These pilots differed from those pilots that went on one risk perception factor. Specifically, pilots who

chose not to fly the simulation exhibited a higher level of risk perception for the

Immediate High Risk factor from Hunter’s Risk Perception Scale One. The ‘No‐Go’

pilots rated the risk involved in ‘immediate high‐risk’ situations, on average, eight

percent higher than did the ‘Go Pilots’.

195

This is consistent with Hunter’s (2006) findings, in which he identified a negative relationship (which failed to achieve statistical significance) between the number of hazardous events experienced in the past (which implies that pilots were

willing to behave in a more risky fashion in the past) and participants’ risk

perceptions on the Immediate High Risk factor (r = ‐.019).

Two thirds of pilots chose to undertake the flight as described. Once pilots

had chosen to fly, their behaviour appeared to be related to the attitudinal factor of

Self Confidence in a positive direction; the higher an individual’s self‐confidence, the

longer they would fly in the simulation. This finding appears to be in agreement

with findings in road safety literature, where individuals that exhibited a higher

level of self‐confidence were more likely to engage in risky behaviours (Brown &

Groeger, 1988; Deery, 1999). However, the two findings are not directly analogous;

as the pilots that chose to go in this experiment did in‐fact choose to undertake

more risk than those that chose to land immediately. Therefore, the finding is such

that once an individual has made a decision to take a higher than necessary risk,

self‐confidence appears to be related to behaving in a more risky fashion. In plain

terms, the pilots that flew the simulation were taking more risk than those that did not. Of the pilots that went, those with a higher level of self‐confidence were more likely than their peers to take further risk, by flying longer or lower than their peers.

There was no other significant relationship found between behaviour and the

attitudinal measures in the study. Therefore, regardless of whether a pilot had

196

committed him or herself to taking a higher level of risk by leaving the circuit area, or if they chose to land the aircraft immediately to reduce their risk exposure, their orientation to risk and their orientation towards safety (the other two factors in the

ASAS) appeared to have little effect on their behaviour.

One possible explanation for the wider lack of relationships between attitude and behaviour is that the risk taken in this experiment is taken for the benefit of another person (the skydiver), and is not likely to be linked to a pilot’s normal behavioural motivators (which may include attitudes), their sensation seeking, or other intrinsic rewards for taking risks (Horvath & Zuckerman, 1993). The context that pilots were faced with that another person might possibly need their help may also have circumvented the normal influence of the attitudes of pilots such that they were ignored for the benefit of the skydiver. Indeed, 16% of pilots reported that they were primarily concerned with the skydiver in the debriefing interview, rather than the more pressing personal risks of a lack of fuel on board the aircraft.

Overall, the influence of attitude on behaviour was, at best, found to be weak.

This result appears to differ with the majority of academic literature in the area of attitude and behaviour (Ajzen, 1991; Albarracín et al., 2001; Crano & Prislin, 2006;

Deery, 1999; Glasman & Albarracín, 2006; Smith & Terry, 2003) in that most studies have found that attitude affects behaviour to a greater extent than was evident in the current experiment (and indeed the current research as a whole). This is a potentially significant finding, as it appears that in the GA training sector, in relation

197

to very risky situations, attitudes of pilots appear to have little affect on risk‐taking

behaviour. It is acknowledged however, that this study measured attitude with a

single scale, which did not perfectly satisfy the principle of compatibility. This may

have affected the results somewhat, and given an outcome in which the

relationship/s evident were lower than reality. In brief, the current experiment, and the experimental process as a whole found little evidence that aviation safety

related attitudes are related to the risk minimisation behaviour of pilots.

In the current research it was initially thought that the relationship between risk perception and behaviour was such that those that perceived greater risk would

likely modify their behaviour so as to minimise their exposure to this risk (see

Brewer et al., 2004). Similar to the attitudinal factors above, the risk perception

factors used in the experiment were found to be unrelated to behaviour, once the

decision to ‘go’ had been made. However, the decision to go appeared to be related

to risk perception. Remember that the best risk minimisation behaviour in the current experiment was the choice not to fly at all. Pilots with a higher level of perceived risk regarding the flight were more likely to choose not to fly. This finding conforms, in part, to the theory of risk perception given by Brewer et al., (2004) in which a higher level of perceived risk was likely to lead to risk minimising behaviour. However, the theory did not explain the varying risk behaviour of pilots that chose to fly (i.e., it could not explain why some Go‐Pilots chose to fly for longer, or flew lower than others).

198

A possible explanation of the apparent incongruity of the findings in the

current research (for both attitudes, and risk perceptions) with the existing

literature may indeed be that the findings are reflective of the behaviour/s of a

unique subset of the pilot population, which is as yet unidentified or poorly understood in the literature; those people that are willing to undertake highly risky

behaviours, within a relatively risk averse culture. That is, the findings may be

correct and relevant for pilots that are willing to take extreme risks in an extremely

risk‐averse industry. Also of note is that the current participant group is relatively

restricted in demographic values (i.e., relatively small age ranger). The findings may

be indicative only of pilots that belong to this subset of the population.

Another alternative to the apparent incongruity between the current results

and those in the literature is that it is a possibility that pilots in Australia, flying in

Australian conditions, and under Australian legislative requirements are indeed

affected by different factors than American (or other nationality) pilots. These may

be the prevailing national or industry norms, different air‐law and airspace

requirements, or simply the comparative lack of facilities for GA in Australia

(meaning that pilots are potentially more comfortable flying without alternates

available). Alternatively, pilots in Australia may behave based upon different factors

(e.g., attitude, risk perception), or behave differently given these factors when

compared to their counterparts in the United States (or other countries).

199

The second research question for Experiment Three was concerned with the relationship between experience and behaviour. In terms of the relationship/s between the experiential variables of age, total flight hours and recent flight hours, the current research found very few significant relationships.

The analysis from Experiment Three indicated that age was not a reliable predictor of conservative behaviour with regards to the decision to undertake the flight in the current experiment. This echoes the findings of Experiment Two, where age of pilots appeared unrelated to the choice made by pilots between flying or not.

In other words, in the current research, the initial decision made by pilots to reduce their risk taking to the lowest possible levels was apparently unaffected by age. It can be inferred from this therefore, that the minimisation of risk (by not undertaking the proposed flights) is unrelated to age.

Age was, however, weakly related to the minimum altitude to which pilots descended in the simulation. In this case, the relationship was negative, such that the older the participant was, the lower they were likely to fly during the simulated flight. Given that there was no relationship between age and the decision to fly, age and the length of flight, and the relationship present between age and minimum altitude was weak, age appeared to have a relatively minor effect upon the behaviour of pilots.

200

The finding that age appears mostly unrelated to behaviour, is not congruent

to those findings in the road safety area, where increasing age is usually thought to decrease the level of risk that one accepts and undertakes (Deery, 1999).

In terms of flight hours, the common measure of experience in the industry, it

appears from the results that they are not a reliable predictor of conservative

behaviour with regards to risk taking; there was no statistically significant

relationship between flight hours and risk taking in the current experiment. Based

on this finding, the exclusion of pilots from employment within GA, or from holding

particular licence types based on experience (and presumably the assumption that

with experience comes superior behaviour) becomes somewhat questionable. This

is tempered with the acknowledgment that this result applies specifically to

situations in which there is a very significant risk present. Therefore, future

research in the area may utilise situations that are more representative of a normal

operational environment, as opposed to that used in the current experiment.

The validity of the current measures of ability and competence in the

aviation industry appears to be an area that is deserving of more research in future,

as there is little by way of empirical evidence available regarding the efficacy of the

current measures. Future research in the area might further focus on pilots with a

higher number of flight hours than in the current research, such as those about to

enter, or already in a commercial airline. This research may better show the

interaction between experience and behaviour in situations where the

201

consequences are potentially greater for risky behaviour (such as in a commercial

context).

This kind of research would also allow for testing of the relationship between attitudes, risk perceptions, experience and behaviour in a commercial and/or multi‐ crew environment, rather than a single pilot GA environment as used in this methodology. It would also allow a more comprehensive study of the validity of the

use of flight hours for a discriminator for entry into airline employment, as pilots on

the verge of, or already in airline service were not included in this study.

Similar to flight hours, recent flight time failed to show a statistically

significant relationship with behaviour. This was regardless of whether the

behaviour was self‐reported (as in Experiment Two), or revealed in the simulation of the current experiment. Pilots with more hours flown in the last ninety days were just as likely to choose to fly in the current experiment as those that had few or no hours in recent months. Similarly, recency apparently played no role in the decision/s made by pilots in the current experiment after they had made the decision to fly. That is, recency did not appear to be related to either the time spent flying, or the minimum altitude flown by the pilots in the Moruya scenario. This echoes the findings of Molesworth and Chang (2009) in which recent flight (also hours in the last 90 days) was not related to behaviour in a flight simulation.

202

In sum, it appears that experience, as measured by age, flight hours and

recency were unrelated to pilots’ decision to fly the simulation. That is, making the

most appropriate decision, given the context, appears to be dependent upon risk

perception, not experience or attitudinal factors; pilots will behave most

appropriately in a given risky situation if they have a relatively higher perception of

the risk in situations that involve both a high impact in terms of possible

consequence and significant time pressure.

Further, there was no relationship evident between the normal measures of

experience and competency in the aviation field, total flight hours, age and recency,

and behaviour in the current research. This finding brings with it the prospect that experience, as measured by the metrics in use currently, is not related to risk‐ management behaviour in aviation.

In simple terms, the results from Experiment Three provide some evidence

that pilots will behave most appropriately in a given risky situation if they have a

higher level of risk perception, not if they are older, have more hours, or fly more

regularly.

203

9 General Discussion

9.1 Introduction

The current study was designed to determine the relationships between attitude, risk perception, experience and behaviour in the context of the Australian

General Aviation industry. The methodology employed three experiments in order to achieve this.

Experiment One was designed to examine relationships between the safety‐ related attitudes, risk perceptions, and experience of pilots in the Australian GA sector. Experiment Two was designed to examine the relationship between the variables measured in Experiment One to the rating of self reported behaviour and the rating of overall risk for three aviation scenarios that featured varying levels of risk associated with them, and various hazards within them. Experiment Three was designed to examine the relationship between the variables captured in Experiment

One to the revealed behaviour of pilots in a high‐risk flight simulation.

In layperson's terms, the methodology was designed as a three‐part process in which participants would state their thoughts about safety, then reveal what they thought that they would do in a risky situation, or alternately, illustrate what they would do in a risky situation.

204

The current research utilised two forms of data capture to gather data about

participants’ behaviour: a survey in which participants indicated their preferred

options for behaviour in a given scenario, and actual behaviour captured through a flight simulator. This is similar in design to Paris and Van den Broucke’s (2008) experiment in the road environment, in which attitudes were related to actual behaviours as measured through GPS tracking devices fitted to cars.

The use of actual behaviour (as opposed to self‐reported behaviour) as a

variable in peer‐reviewed papers in the attitude‐behaviour field is surprisingly rare

(Armitage & Conner, 2001; Paris & Van Den Broucke, 2008), making this

methodology relatively unique in the attitude‐behaviour research. Also unique in

the aviation risk management literature, and indeed the wider operational risk

management literature was the use of scenarios that varied in overall risk level, and

containing multiple risk factors. This allowed the current research to test whether

risk perception based upon uni‐dimensional situations (single hazard) were

predictive of behaviour in situations that featured multidimensional risks. Further,

it allowed the current research to test whether risk ratings of multidimensional

risks are related to similar factors as the uni‐dimensional risks, or if there are

unique markers or predictors of superior risk perception for multidimensional

risks.

The current study employed three experiments in order to answer the research aims, which were to investigate if:

205

1. There is a relationship between attitudes, risk perceptions, and experiential

values (flight hours, recency, and age)

2. There is a relationship between the attitudes, risk perception and the

behaviour of pilots in the General Aviation training sector

3. There is a relationship between a pilot’s experience, as measured by age,

flight hours, and recent flight time, and behaviour in the general aviation

training sector

In summary, the current study utilised the strength of multiple

methodologies that have been used previously in academic research, and built upon

these to gain further insight into the predictors of, and contributory factors that are

part of the causal chain of risk management behaviour.

9.2 The Relationships Between Attitudes, Risk

Perceptions, and Experience

It appears, from the preceding research, that there are very few links

between experience and attitude, experience and risk perception and attitude and risk perception in Australian General Aviation.

Attitudes and risk perceptions did not appear to be related to each other on a

global level; there was only a relationship evident between attitudes and the

206

perception of risk in highly hazardous situations. Perception of risk in other

situations, such as normal operational situations, or normal everyday life situations

did not appear to be related to attitude. Therefore, it cannot be stated categorically

that a stronger or ‘better’ attitude will generically lead to a superior level of risk

perception by a pilot.

Pilots with fewer flight hours displayed a higher level of self‐confidence than their less experienced counterparts. A similar finding was evident for recency, where flying in the last 90 days was related to reduced self‐confidence. As stated above, this disagrees with the findings of Molesworth and Chang (2009) where a weakly positive relationship was observed between attitudes and flight hours. This may be explained somewhat by the different cohorts used in the studies. The current cohort utilised pilots from a range of experience levels, and a range of flight schools in the Sydney basin, whereas Molesworth and Chang utilised only trainee

pilots from a single flight training school (Molesworth & Chang, 2009).

Interestingly, pilots that rated the risks of the scenarios in Experiment Two

in a different fashion to the expert group exhibited a higher level of self‐confidence

than those that rated the risks in the scenarios similarly to the expert group. In

simple terms, those that were expert‐like in their risk perception displayed lower

self‐confidence than those that were not.

207

The finding that the level of self‐confidence of individuals reduced as experience increased, is however in agreement with other findings in the Australian

Aviation literature (Drinkwater & Molesworth, 2010), and with findings in the road safety literature, were experience is linked to confidence that the individual will not be involved in an accident (Deery, 1999), as opposed to confidence regarding their innate or learned ability.

It might be argued that the increase in self‐confidence in the current cohort was an appropriate attitude, as the attitude was characterised by a greater agreement with statements like “I am a very capable pilot” and “I am very skilful on controls”, which may be empirically true of pilots with more experience. Relative skill levels and capabilities of the pilots with differing experience level (i.e., testing whether pilots with more experience were empirically superior), were not tested however, so the nature of such a suggestion is at best, speculative.

Flight hours appeared to be unrelated to risk perception, as measured in

Hunter’s Risk Perception Scales 1 and 2 (Hunter, 2002). This finding is in agreement with that of Molesworth and Chang (2009), where no relationship was uncovered also. However, this is in direct contrast to Hunter’s (2006), findings, where a weakly negative relationship was evident between hours and risk perception. Possible explanations of this difference are that the studies were undertaken in different countries (Hunter’s in the U.S.), and that Hunter’s study featured only private pilots licence holders. From the findings of the current research therefore, it appears that

208

the perception of risk evident in situations ranging from those risks found in everyday life situations, to extremely high‐risk aviation situations, is not influenced by flight experience.

The literature of previous research in the area of the effect/s of age upon safety attitude is relatively sparse (Siu et al., 2003). Previous safety‐based research into the relationship between age and attitudes has found generally consistent results; older age is generally thought to be correlated to a more conservative or appropriate attitude (Deery, 1999; Popa et al., 2005; Siu et al., 2003).

In the current research there was a weak negative relationship between age and Self Confidence. This finding suggests that older pilots exhibit a lower self‐ confidence than their younger counterparts. This finding is in agreement with the findings of the majority of the research outlined above (Page, 1969; Popa et al.,

2005; Siu et al., 2003). This is tempered by the knowledge that the relationship was weak.

None of the eight factors on Hunter’s risk perception scales were related to a pilot’s age in the current research. This result disagrees with Hunter’s (2002) findings, where risk perception was negatively related to age. Hunter’s finding was however weak (r = ‐.08), and was found in a cohort that differed from that of the current research on nationality (U.S. compared to Australian), and also on

209

experience (Hunter’s cohort contained airline pilots as well as the types of pilots in

the current research)

An implication of this finding is that it suggests that during normal flying

operations, which will (hopefully) represent the majority of operational situations

that a pilot will encounter, and based upon the findings of this research, older pilots will not exhibit superior perception of the risks in a given operation. In basic terms, older pilots will be just as likely as younger pilots to perceive or misperceive risks in normal operations.

Perhaps the most interesting and unexpected finding was that there was no

relationship evident between risk perception and the normal measures of

experience and competence in the aviation field, total flight hours, and recency. This

finding disagrees with research by Hunter (2006) that utilised the same risk

perception measures used in the current research in which a relationship, albeit

relatively weak (largest r = ‐.185), was found between risk perception and

experience in GA pilots from the United States (Hunter, 2006). This is a significant

finding, in that this sample of Australian GA pilots appears to be different to their US counterparts with respect to risk perceptions. However, this is tempered with the acknowledgement that the current research utilised a smaller number of pilots, and pilots that held different licence types from one another (as opposed to all pilots in the cohort holding single licence type), which is dissimilar to Hunter’s experiment.

210

This may mean that risk perception is indeed influenced by experience, but this was not evident in the cohort utilised in this study.

Alternatively, it is also possible that the effect is quite weak, or simply not present in Australian GA pilots. This is given more credence from the fact that other research in the Australian GA industry has failed to find a link between risk perception and experience (Drinkwater & Molesworth, 2010; Molesworth & Chang,

2009).

In sum, it appears that older or more experienced pilots will not exhibit superior perception of the risks or hold more conservative attitudes in a given operation than their less experienced colleagues, older pilots will be just as likely as younger pilots to perceive or misperceive risks, or to hold potentially inappropriate attitudes.

9.3 Relationships between attitude, risk perception,

experience and behaviour

In previous research in the Australian GA industry, Molesworth and Chang

(2009) found no statistically significant relationship between attitudes and behaviours. This appears to be replicated to some degree in the current research.

Overall, the current experimental process found relatively little evidence that

211

aviation safety related attitudes are related to the risk minimisation behaviour of pilots.

In terms of the relationship between attitudes and self‐reports of intended behaviour, attitudes were related to the risk ratings given by pilots regarding medium‐risk multi‐dimensional scenarios.

In terms of the relationship between attitudes and revealed behaviour (as opposed to self‐report of intention), only self‐confidence was related to risk taking behaviour in the simulation. This was a negative relationship, such that a higher level of self‐confidence led to a safer behaviour. This contrasts with results obtained by Goh and Wiegmann (2001), where pilots that flew into potentially dangerous meteorological conditions reported a higher level of self‐rated skill and judgement

(t=2.05, p< .05) than those pilots who chose not to fly into these dangerous conditions. It also contrasts with result found by Hunter (2005), where a weak positive correlation between scores on the Hazardous Events Scale and ASAS – Self

Confidence (r = .208) was evident.

Further, this result appears to differ with the majority of academic literature in the area of attitude and behaviour (Ajzen, 1991; Albarracín et al., 2001; Crano &

Prislin, 2006; Deery, 1999; Glasman & Albarracín, 2006; Smith & Terry, 2003) in that most studies have found that attitude affects behaviour to a greater extent. This is a significant finding, as it appears that in the GA training sector, in relation to very

212

risky behaviours, attitudes of pilots appear to have little affect on risk‐taking

behaviour.

In the current research it was expected that that those that perceived greater

risk (higher level of risk perception) would likely modify their behaviour so as to

minimise their exposure to this risk (see Brewer et al., 2004). The analysis carried

out showed that the decision to undertake a risky flight was influenced only by risk

perceptions. This was such that the pilots that flew (or chose to fly) in the

experiment exhibited a lower level of risk perception for the Immediate High Risk

factor than those that chose not to fly. From the current evidence, risk perception‐

based scales would likely give a more reliable prediction of future risk behaviour

than the current attitudinal scales.

Experience, as has been stated above, is used within the aviation industry as

a differentiator and discriminator between pilots for licensing and employment. The

current experimental methodology tested the relationship between experience

(measured in age, flight hours, and recent flight time) and the behaviour of pilots.

The analysis indicated that the initial decisions made by pilots to reduce their

risk taking by electing not to undertake the flight/s was apparently unaffected by age. It can be inferred from this therefore, that the minimisation of risk to the lowest possible levels in Australian GA is unrelated to age. Age was, however related to the minimum altitude to which pilots descended in the simulation, such that the older

213

the participant was, the lower they were likely to fly during the simulated flight. Age

therefore appeared to have a relatively minor effect upon the behaviour of pilots.

This finding is again, not congruent to those findings in the road safety area, where

increasing age is usually thought to decrease the level of risk that one accepts and

undertakes (Deery, 1999). As noted above though, this might be attributed to safety

culture differences prevalent between aviation and the road environment and self‐

selective biases introduced by financial costs and the mental and physical

requirements of aviation.

It appears, from the analysis undertaken, that flight hours are not a reliable

predictor of conservative behaviour with regards to risk taking. As has been stated

above, the exclusion of pilots from employment within GA, or from holding particular licence types based on experience is potentially questionable.

The remaining experiential factor of recency also failed to show a statistically

significant relationship with behaviour in any of the scenarios presented. This was

regardless of whether the behaviour was self‐reported, or revealed in the

simulation. This finding is also in agreement with recent research in the Australian

GA industry (Drinkwater & Molesworth, 2010; Molesworth & Chang, 2009), and in

the broader aviation field (Hunter, 2006).

In sum, there was no relationship evident between the normal measures of

experience and competence in the aviation field ‐ total flight hours, age and recency

214

‐ and behaviour in the current research. The results provide some evidence that pilots will behave most appropriately in a given risky situation if they have a higher level of risk perception, not if they are older, have more hours, or fly more regularly.

In other words, this finding brings with it the prospect that experience, as measured by the metrics in use currently, is not related to risk‐management behaviour in aviation. This is deserving of further research to determine if this effect is specific to this cohort, or if it is indicative of the wider population.

Remember that the Australian GA industry has recently introduced Threat

and Error Management training to the training syllabus. The introduction of TEM to

GA has also introduced the concept of risk management into the GA sector, which up

until recently, had not explicitly trained pilots in risk management principles

(Molesworth et al., 2006), and therefore relied upon other factors, like experiential training, or future training from operators (like airlines) to further pilots risk management skill level. The findings of the current research, that the common metrics of experience and competence appeared not to be related to risk management gives support to the need for increased levels of risk management training in the Australian GA industry.

215

9.4 Synergy (or Otherwise) of the Two Behavioural

Methodologies Used

As has been documented above, very similar results were reported between the two methodologies.

Experiment Two utilised self‐reports of intended behaviour to gather data on behaviour. Experiment Three on the other hand, utilised a flight simulation to gather information regarding participant behaviour and decision‐making. Both methodologies found that attitudes were unrelated to the decision to undertake (or otherwise) risky behaviours, but rather that risk perceptions appeared to be responsible for the decision/s. Both experiments also found that experience, as measured by the common metrics in aviation, was unrelated to behaviour in the current scenarios.

This result reflects favourably on pen and paper measures to determine pre‐ cursors to risky flight behaviour. Moreover, as noted above, despite the different tools used to determine factors that predict risk management in aviation (e.g., experiment two ‐ pen and paper and experiment three ‐ simulation), there were little differences in the results. It could be concluded that the use of pen and paper measures is a more time efficient method in examining key predictors of risky flight behaviour compared to simulation.

216

An implication of this is that self‐reports may be especially useful as a

method for aviation safety research where the focus in on predictors of risk taking

behaviour (i.e., decision to undertake a risky flight). This is opposed to a project in

which the variable of interest was tactical, such as discovering how a pilot

undertakes a precautionary landing. To be clear, this is not to say that the method

cannot, or should not be used for tactical research, rather, that in the current

research, there was only evidence available that self‐report was useful for strategic,

decision‐based research.

In the current research it was found that the self‐reports of behaviour were

likely to present more conservative behaviour than seen in the simulations. If this

finding generalises to the population then the use of simulation, which is relatively

expensive and time consuming when compared to simulator‐like pen‐and‐paper

studies (especially if data transcription is abolished by utilising computer based

survey techniques), may be unnecessary for studies that are attempting to

determine the pre‐cursors, or causal factors of risk management behaviour in an

aviation specific setting.

As an addition to this, organisations that wish to determine the existing risk management behaviour of their personnel may be able to use survey‐based measures (which may include semi‐formal face‐to‐face interviews), rather than the use of simulation measures in order to obtain information. Alternately, these

217

measures could be used as augmentations to the current use of simulation for

training and competence checks in larger organisations.

9.5 Limitations of the Research

Whilst the results outlined above are interesting and may have serious

implications for Australian General Aviation, the results must be interpreted with a

level of reservation. The results obtained in this research are based upon a relatively

small cohort of pilots that are limited in terms of their experience and vocation.

Specifically, no airline pilots were included in the study, only General Aviation pilots

were utilised. The generalisability of these results within the wider aviation

industry and other domains remains untested.

The experiment examined pilot performance in a controlled and simulated

environment, where normal operational conditions such as movement, noise,

temperature, radio calls, traffic and the threat of injury or death were absent. While

the manipulation or removal of these variables is beneficial from an experimental perspective, their overall impact on pilots’ performance remains unclear.

One methodological limitation of the study, from an overall perspective, is

the relatively small and limited sample used. This was used because of the financial,

resource, and time limitations of the current research project. The methodology,

however, is easily adaptable to use within a commercial environment, and is

218

relatively expedient and inexpensive to conduct. One repercussion of the small sample size, however, is the participants were generally residents of Sydney and had trained at Bankstown or Camden airports, both General Aviation Aerodrome

Procedures (GAAP) airports at the time of the experiment. GAAP airports are an

Australian oddity in that the rest of the world uses a different airspace system (see

CASA, 2010a). This operating environment may have affected the participants’ behaviours and decision‐making. Further still, the majority of pilots that undertook the experiment/s were sourced from four schools around Sydney, which adds the possibility of an effect from the prevalent safety culture of these schools (for a review of safety culture see Wiegmann, Zhang, Von Thaden, Sharma, & Gibbons,

2004).

Another methodological limitation related to the number of participants in the study relates to the reliability of the ASAS and Risk Perception Scales. Subscales of these two measures have not achieved the generally accepted Cronbach’s Alpha of 0.7 when measured for reliability. Because of this, the results of this study must be interpreted with due caution. Factor analysis could not be undertaken on the current sample, as it is not a large enough group for the tests to be adequately reliable (Tabachnick & Fidell, 2013).

In terms of the limitation of sample size and type used in this thesis, further research that is focused on broader populations, both within GA, and in the commercial aviation environments would be especially useful in gaining an accurate

219

picture of the factors that guide behaviour within the greater aviation industry, as opposed to that within the general aviation sector only.

Similar to this point, the current methodology was designed to discover the causal factors behind risk management of pilots in a range of situations, from relatively low‐risk situations, to those that involve extreme risks, with a focus upon the risk management of pilots in riskier situations. Research into the factors that guide pilot risk management in a wider range of ‘normal’ situations would be of use to the industry, as to date; this information appears not to be available in the literature. This research could further provide evidence for the appropriate use, or modification of the standards in use for licensing and employment. Additional to this, the current methodology was related only to single pilot operations within GA in Australia. The generalisability of this to the Australian commercial sector, the worldwide GA industry, and even further afield, the worldwide commercial sector is unknown. The methodology is easily modified however, and research in these jurisdictions would be helpful in gaining an accurate picture of exactly how attitudes, risk perceptions and experiential factors affect behaviour of pilots

Another limitation related to sample is that the populations that undertook

Experiment Two and Three were notably different from one another, namely in age.

This is despite the fact that they were drawn from the same master sample. As was noted above, participation in Experiment Two or Three was dependant upon each

220

individual’s availability to undertake a simulator session. Therefore the results of

the research need to be interpreted with this limitation in mind.

As has been stated above, the use of self‐report to gather information

regarding behaviour has limitations. These have been acknowledged, and the use of

a methodology that utilised revealed behaviour as an augmentation to the self‐

reports of Experiment Two was utilised to address this limitation somewhat. In the

current research, each of these methodologies revealed similar, and importantly,

consistent results.

A limiting factor that may potentially impact the results of the current study

relates to the perceived reality of simulated flights for the pilots studied. Although

the pilots were instructed on multiple occasions to treat the simulated flight “as

though they were in a real aircraft”, and to “do exactly what you would do in real life”, the fact that it was a simulation, and that the potential negative consequences of a high level of risk taking, like loss of licence, injury, or death, are not present in a simulation, may have impacted the results. This problem is a common limitation for

simulation‐based experiments. This limitation is tempered though, with the

knowledge that some pilots did indeed choose not to go. From this it can be gleaned

that the pilots in the simulation did not behave as though the experiment was

without consequence. That is, some pilots appear to have behaved as though they

were actually at risk in the simulation (as in a real flight), and chose not to

undertake the flight.

221

In the current research there is no objective data (as it was not captured as part of the methodology) as to whether participants behave in a real operational environment in the same way that they have behaved in the simulation. This is an obvious area for future research. A longitudinal study that followed individuals for a period of their career would be one effective way to capture such data. In this way, data regarding real‐world behaviour, incidents, accidents and longitudinal attitudinal and risk perception data could be captured.

Another possible limitation of the study is the lack of fidelity of the simulator

in control sensitivity or ‘feel’. The simulator software used was at the time of the

experiment, one of the most advanced and realistic simulators available for use on

the available equipment. Similarly, the physical controls are high fidelity in terms of

range of movement, physical design, dimension and positioning. It was evident from

pre‐testing of the simulator that the simulator was not a perfect simulation of a

Cessna 172, which the author is familiar with, but that it was an acceptable model.

The same simulator has also been used in other experiments (Drinkwater &

Molesworth, 2010).

Regardless of the reality of the model, from an empirical perspective, it can

be argued that the use of the simulator was consistent across the groups, and all

were exposed to the same conditions in terms of the control of the aircraft.

Therefore, it can be argued that the effect of the novel handling characteristics of the

simulator was universal to the studied population. It can also be argued that the

222

decision‐making that pilots were undertaking was somewhat (or possibly it was

almost completely) independent of the physical act of flying the simulation. That is,

pilots that chose to fly chose to do so before they had flown the simulator, and

therefore it is only subsequent decisions that may have been affected by the

simulator in any case.

Another limitation of the research concerns the limited visibility in the

simulator. A general aviation aircraft, dependant on the aircraft type and model, will

have between 300 to 360 degrees of visibility. In normal circuit operations, pilots

will use 270 to 300 degrees of visibility for spatial positioning and traffic separation.

In search operations, like the one used in the simulation, a pilot requires as much

visibility as is possible. The simulator restricted pilots’ field of view to 160 degrees.

While this field of view is adequate for the operation of the flight, general aviation pilots are not used to a restricted field of view. It can be argued that all pilots experienced the same limitations relating to the field of view, however, it may have been the case that some pilots had a preference for flying with reference to visual cues outside this arc and, therefore, modified their behaviour somewhat to overcome this limitation. As above though, the overarching argument can be made that the simulator’s shortcomings were universal, and that the decision‐making of pilots was, at least initially, independent of the simulator’s fidelity.

Apart from the technical limitations of the equipment utilised, it has been

found in previous research that attitudes better correlate with behaviours if the

223

principle of compatibility is satisfied, that is, when both the attitude measured and behaviour revealed are concerned with the same action, target, context and time based elements (Ajzen & Fishbein, 2005). The attitudinal measures used in this experiment, it could be argued, did not fully satisfy the requirements of the principal of compatibility, because they are not attitudes regarding a flight in which pilots are asked to choose between two options, one with extreme risk, and one with relatively low risks. Instead, they are measures that are concerned with a pilot’s attitude towards flight in a range of conditions, from risky through to normal operations, with their attitudes towards risk in aviation, and their orientation towards safety in aviation.

To counter this argument though is the fact that these attitudes have been previously found to be related to reported behaviour (Hunter, 2005), which may give weight to the argument that the measures were, in the very least, explanative and part of the causal chain in aviation behaviour in past experiments, and were therefore expected to relate to behaviour here.

9.5.1 Implications and future research

The current research revealed no relationship between both attitude and experience, and risk perception and experience in Experiment One. Similarly, there was no relationship evident between the normal measures of experience in the aviation field (e.g., total flight hours) and behaviour. If effective risk management is central to the reasoning behind the use of screening measures, the results of the

224

present research bring into question the validity of the current measures. Using

flight hour minimums as a discriminator for licensing and employment is not supported from the evidence available in the current research, assuming that these minimas are in place to ensure appropriate behaviour of pilots in terms of ‘risk taking’. Further research in this area would be of particular utility to the industry, as pilot shortages due to lack of appropriately experienced pilots have the potential to cause significant disruption, both economic and schedule wise, to airlines worldwide.

One area that warrants further investigation is the viability and accuracy of

the measures used to determine attitudes in Australian General Aviation. The

reliabilities of the currently available scales, like the ASAS used in this research, are

sub‐optimal at best. The creation of reliable and robust scales that measure

constructs like attitude would be most helpful in advancing the understanding of the

role of psychological traits in pilot behaviour.

As a result of the fact that there were so few significant findings with regards

to a contributory relationship of experience, attitude and risk perception with

behaviour, a logical area of future research is to investigate the other factors that

may contribute to the decision making and risk management behaviour of pilots.

Road safety research has found many other factors that are modifiers of behaviour.

Dahlen, Martin, Ragan and Kuhlman (2005 ) found that sensation seeking ‐ the

degree to which an individual desires new and intense stimuli was related to

225

aggressive driving, and inappropriate expression of anger, predicted the periodic

loss of concentration whilst driving, loss of control of the vehicle, aggressive and

risky driving, both physically and verbally expressing anger, and the use of their

vehicle to show anger (Dahlen et al., 2005 ).

Another interesting area for future research is the role of feedback on

attitude, risk perception and behaviour. Recent aviation examples (see for instance

Molesworth, Wiggins, & O'Hare, 2006) have found that feedback has an effect on

behaviour. Further research into this may yield useful performance‐enhancing

techniques for safety critical industries.

In a different study, Scott‐Parker, Watson and King (2009) found that the

anticipated punishment for a behaviour, the expected rewards for a behaviour, and

imitation of parents and peers was related to risky driving behaviour. There are

many other factors that have also been found to modify risk‐related behaviours.

These include (but are not limited to) aggression (Davey, Wishart, Freeman, &

Watson, 2007), normlessness (Ulleberg & Rundmo, 2003), implicit attitude (Hatfield et al., 2008), and alienation (Gulliver & Begg, 2007 ).

Explicit testing of these factors in the Australian GA sector may yield a more

comprehensive understanding of the modifiers of safety related behaviour.

226

9.6 Conclusion

This thesis examined and explored the risk perceptions, attitudes and

experiential metrics of pilots that are undergoing flight training, are private pilots, or are instructors in the Australian GA industry and compared these factors to

behavioural measures. The study employed three experiments in order to explore

these relationships.

The multi‐experimental process was designed to provide data from

participants in three different manners and for three different primary purposes.

First, factual experiential data, attitudinal data and risk perception data were collected using pen and paper surveys; these were gathered to enable the comparison between these factors and also to compare these factors to behaviour in later experiments.

Second, self‐reported behaviour in low to high‐risk situations was gathered

utilising a specifically constructed survey. These data were compared to the data

from the first experiment to determine if self‐reports of intended behaviour were

related to any of the metrics used in aviation as discriminators.

Thirdly, revealed flight behaviour was observed utilising a flight simulator.

Much like the previous method, this was compared to the metrics in the first

experiment to determine if revealed behaviour was related to these factors.

227

The first stage of the experimental process found there was no relationship evident between the risk perception of pilots and the normal measures of competence and experience in the aviation field: total flight hours, age and recency.

It therefore appears that during normal flying operations in Australia, which represent the majority of operational situations that a pilot will encounter, older pilots, those with more hours, or those that fly regularly will not exhibit superior perception of the risks in a given operation.

Older pilots will be just as likely as younger pilots to perceive or misperceive risks in normal operations. Similarly, attitudes were found to be largely unrelated to the measures of experience used in the experiment. There was however, a relationship between the experiential factors measured and the attitudinal factor of

Self Confidence, in which more experienced pilots displayed lower levels of self‐ confidence.

This finding, that the majority of attitudinal measures used, and all of the risk perception measures utilised are not related to experience, differs from aviation research undertaken in the United States (Hunter, 2005), where risk perception and attitude were both related to experience. This is a significant finding, in that

Australian GA pilots appear to be different to their U.S .counterparts with respect to the effects of experience.

228

The second and third stages of the experimental process were concerned with the exploration of the causal factors behind behaviour. One of the most interesting outcomes of the experimental process is that the influence of attitude on behaviour was, at best, found to be weak. This result differs with the majority of academic literature in the area of attitude and behaviour (see Albarracín et al., 2001;

Azjen, 1991; Crano & Prislin, 2006; Deery, 1999; Glasman & Albarracín, 2006; Smith

& Terry, 2003) in which authors have generally claimed a causal and reasonably robust relationship between attitudes, and the corresponding behaviours.

This may be a significant finding, as it appears that in the Australian GA sector, in relation to the management of risk in normal through to extremely risky situations, attitudes of pilots appear to have little effect on risk‐taking behaviour.

The preceding experiments revealed that the decision to undertake (or more importantly, not to undertake) a risky flight was influenced only by risk perception

– a higher level of risk perception led to a lower likelihood of a pilot choosing to undertake the flight. Therefore, holding a ‘good’ attitude, or having a higher level of experience appeared to have no influence on the decisions by pilots as to whether they would undertake the flights.

When coupled with the findings above, this means that although older and more experienced pilots appeared to be less self‐confident, they did not behave any differently to younger, or less experienced, or less self‐confident pilots with regards

229

to effective risk management. As has been stated this brings into the question the validity of using arbitrary flight hour minimums, age minima, and flight hour requirements as a discriminator for licensing and employment, if risk management is seen as a large component of flying competence. Further testing of the validity of experiential variables as behavioural predictors in aviation is required.

The experimental process undertaken has therefore found that it appears

possible that attitude, age and experience do not significantly affect the risk

management outcome of pilots decisions in Australian GA. Making the most

appropriate decision, given the context, appears to be most dependent upon risk

perception, not experiential or attitudinal factors. In other words, pilots with

superior risk perception are more likely to perform more appropriately in high‐risk

situations than those with lower perceptions of the prevalent risks.

230

References

Åberg, L. (1999). The Role of Attitudes in Transportation Studies. Lund, Sweden:

Borlänge and the Department of Psychology, Uppsala University.

Ajzen, I. (1988). Attitudes, Personality and Behaviour. Stony Stratford, Milton Kenes:

Open University Press.

Ajzen, I. (1991). The Theory of Planned Behaviour. Organizational Behaviour and

Human Decision Processes, 50, 179-211.

Ajzen, I. (2001). Nature and Operation of Attitudes. Annual Review of Psychology, 52,

27-58.

Ajzen, I., & Fishbein, M. (1980). Understanding Attitudes and Predicting Social

Behaviour. Englewood Cliffs, NJ: Prentice Hall.

Ajzen, I., & Fishbein, M. (2005). The Influence of Attitudes on Behaviour. In D.

Albarracín, B. T. Johnson & M. P. Zanna (Eds.), The Handbook of Attitudes. Mahwah,

NJ: Erlbaum.

231

Albarracín, D., Fishbein, M., Johnson, B. T., & Muellerleile, P. A. (2001). Theories of

Reasoned Action and Planned Behaviour as Models of Condom Use: A Meta-

Analysis. Psychological Bulletin, 127(1), 142-161.

Albarracín, D., Wallace, H. M., & Glasman, L. R. (2004). Survival and Change in

Judgments: A Model of Activation and Comparison Advances in Experimental Social

Psychology (Vol. 36, pp. 251-315). San Diego: Academic Press Inc.

Archibald, R. B., & Reece, W. S. (1977). The Impact of the Energy Crisis on the Demand

for Fuel Efficiency: The Case of General Aviation. Transport Research, 11, 161-165.

Arezes, P. M., & Miguel, A. S. (2008). Risk Perception and Safety Behaviour: A Study in

an Occupational Environment. Safety Science, 46(6), 900-907.

Arkes, H. R., & Tetlock, P. E. (2004). Attributions of Implicit Prejudice, Or "Would

Jesse Jackson 'Fail' the Implicit Association Test?". Psychological Inquiry, 15(4), 257-

278.

Armitage, C. J., & Conner, M. (2001). Efficacy of the Theory of Planned Behaviour: A

Meta-Analytic Review. British Journal of Social Psychology, 40, 471-499.

ATSB. (2005). Aviation Safety Indicators: A Report on Safety Indicators Relating to

Australian Aviation (No. 1 921092 07 6). Canberra, ACT: Australian Transport Safety

Bureau.

232

ATSB. (2006). Analysis of Fatality Trends Involving Civil Aviation Aircraft in Australian

Airspace between 1990 and 2005. Canberra, ACT: Australian Transport Safety

Bureau.

ATSB. (2009). Avoidable Accidents No.1: Low Level Flying. Canberra, ACT: Australian

Transport Safety Bureau.

Aven, T. (2003). Foundations of Risk Analysis: A Knowledge and Decision-Oriented

Perspective: John Wiley & Sons, Ltd.

Aven, T., & Kristensen, V. (2005). Perspectives on Risk: Review and Discussion of the

Basis for Establishing a Unified and Holistic Approach. Reliability Engineering and

System Safety, 90, 1-14.

Azjen, I. (1991). The Theory of Planned Behaviour. Organizational Behaviour and

Human Decision Processes, 50, 179-211.

Bagozzi, R. P. (1992). The Self-Regulation of Attitudes, Intentions, and Behavior. Social

Psychology Quarterly, 55(2), 178-204.

Bandura, A. (1977). Self-Efficacy: Toward a Unifying Theory of Behavioral Change.

Psychological Review, 84(2), 191-215.

233

Barnes, P. H. (1996). Life as a Coiled Spring: Hazard and Risk Perception in the

Queensland Fire Service. Unpublished PhD Thesis, Griffith University, Brisbane.

Barowsky, A., Shinar, D., & Oron-Gilad, T. (2010). Age, Skill and Hazard Perception in

Driving. Accident Analysis and Prevention, 42(4), 1240-1249.

Beckwith, J. A. E. (1996). Judgement Strategies in Determining Risk Acceptability.

Unpublished PhD Thesis, Curtin University of Technology, Perth.

Bentler, P. M., & Speckart, G. (1981). Attitudes Cause Behaviors - a Structural Equation

Analysis. Journal of Personality and Social Psychology, 40(2), 226-238.

Blanton, H., Jaccard, J., Gonzales, P. M., & Christe, C. (2006). Decoding the Implicit

Association Test: Implications for Criterion Prediction. Journal of Experimental

Social Psychology, 42(2), 192-212.

Boeing. (2005). Statistical Summary of Commercial Jet Airplane Accidents (Worldwide

Operations 1959-2004). Seattle, WA: Boeing.

Bohnenblust, H., & Slovic, P. (1998). Integrating Technical Analysis and Public Values

in Risk-Based Decision Making. Reliability Engineering and System Safety, 59(1),

151-159.

234

Bohner, G., Rank, S., Reinhard, M. A., Einwiller, S., & Erb, H. P. (1998). Motivational

Determinants of Systematic Processing: Expectancy Moderates Effects of Desired

Confidence on Processing Effort. European Journal of Social Psychology, 28(2), 185-

206.

Brenot, J., Bonnefous, S., & Mays, C. (1996). Cultural Theory and Risk Perception:

Validity and Utility Explored in the French Context. Radiation Protection Dosimetry,

68(3/4), 239-243.

Brewer, N. T., Weinstein, N. D., Cuite, C. L., & Herrington Jr., J. E. (2004). Risk

Perceptions and Their Relation to Risk Behaviour. Annals of Behavioural Medicine,

27(2), 125-130.

Bronfman, N. C., & Cifuentes, L. A. (2003). Risk Perception in a Developing Country:

The Case of Chile. Risk Analysis, 23(6), 1271-1285.

Brown, I. D., & Groeger, J. A. (1988). Risk Perception and Decision Taking During the

Transition between Novice and Experienced Driver Status. Ergonomics, 31(4), 585-

597.

BTRE. (2005). General Aviation: An Industry Overview. Report 111 Canberra, ACT:

Bureau of Transport and Regional Economics.

235

BTRE. (2006). Aviation Statistics General Aviation 2004. Canberra, ACT: Bureau of

Transport and Regional Economics.

Button, K., & Drexler, J. (2006). Are Measures of Air-Misses a Useful Guide to Air

Transport Safety Policy? Journal of Air Transport Management, 12(4), 168-174.

Cacioppo, J. T. (1984). The Elaboration Likelihood Model of Persuasion. Advances in

Consumer Research, 11, 673-675.

Cacioppo, J. T., Kao, C. F., Petty, R. E., & Rodriguez, R. (1986). Central and Peripheral

Routes to Persuasion - an Individual Difference Perspective. Journal of Personality

and Social Psychology, 51(5), 1032-1043.

CASA. (2008). Flight Crew Licensing Procedures Part 3. Canberra, ACT: Civil Aviation

Safety Authority.

CASA. (2010a). Class D. Retrieved 10, 06/2010, from

http://www.casa.gov.au/scripts/nc.dll?WCMS:STANDARD::pc=PC_93379

CASA. (2010b). Stages in Becoming a Pilot. Learning to Fly Retrieved 11/06/2010,

from http://www.casa.gov.au/scripts/nc.dll?WCMS:STANDARD::pc=PC_90020

236

Chaiken, S. (1980). Heuristic Versus Systematic Information Processing and the Use of

Source Versus Message Cues in Persuasion. Journal of Personality and Social

Psychology, 39(5), 752-766.

Chaiken, S., & Stangor, C. (1987). Attitudes and Attitude Change. Annual Review of

Psychology, 38, 575-630.

Chang, B. P. I., & Mitchell, C. J. (2009). Processing Fluency as a Predictor of Salience

Asymmetries in the Implicit Association Test. Quarterly Journal of Experimental

Psychology, 62(10), 2030-2054.

Chapman, P. R., & Underwood, G. (1998). Visual Search of Driving Situations: Danger

and Experience. Perception, 27, 951-964.

Cohen, J. B., & Reed II, A. (2006). A Multiple Pathway Anchoring and Adjustment

(Mpaa) Model of Attitude Generation and Recruitment. Journal of Consumer

Research, 33(1), 1-15.

Conner, M., Kirk, S. F. L., Cade, J. E., & Barrett, J. H. (2001). Why Do Women Use

Dietary Supplements? The Use of the Theory of Planned Behaviour to Explore Beliefs

About Their Use. Social Science & Medicine, 52(4), 621-633.

237

Conner, M., Povey, R., Sparks, P., James, R., & Shepherd, R. (2003). Moderating Role of

Attitudinal Ambivalence within the Theory of Planned Behaviour. British Journal of

Social Psychology, 42, 75-94.

Conner, M., Sparks, P., Povey, R., James, R., Shepherd, R., & Armitage, C. J. (2002).

Moderator Effects of Attitudinal Ambivalence on Attitude-Behaviour Relationships.

European Journal of Social Psychology, 32(5), 705-718.

Conrey, F. R., & Smith, E. R. (2007). Attitude Representation: Attitudes as Patterns in a

Distributed, Connectionist Representational System. Social Cognition, 25(5), 718-735.

Cook, A. J., Moore, K., & Steel, G. D. (2005). Taking a Position: A Reinterpretation of

the Theory of Planned Behaviour. Journal for the Theory of Social Behaviour, 35(2),

143-154.

Crano, W. D., & Prislin, R. (2006). Attitudes and Persuasion. Annual Review of

Psychology, 57, 345-374.

Crespo, A. H., & del Bosque, I. R. (2008). The Effect of Innovativeness on the Adoption

of B2c E-Commerce: A Model Based on the Theory of Planned Behaviour. Computers

in Human Behavior, 24(6), 2830-2847.

Cunningham, W. A., & Zelazo, P. D. (2007). Attitudes and Evaluations: A Social

Cognitive Neuroscience Perspective. Trends in Cognitive Sciences, 11(3), 97-104.

238

Cunningham, W. A., Zelazo, P. D., Packer, D. J., & Van Bavel, J. J. (2007). The Iterative

Reprocessing Model: A Multilevel Framework for Attitudes and Evaluation. Social

Cognition, 25(5), 736-760.

Dahlen, E. R., Martin, R. C., Ragan, K., & Kuhlman, M. M. (2005 ). Driving Anger,

Sensation Seeking, Impulsiveness, and Boredom Proneness in the Prediction of Unsafe

Driving. . Accident Analysis and Prevention, 37, 341- 348.

Davey, J., Wishart, D., Freeman, J., & Watson, B. (2007). An Application of the Driver

Behavior Questionnaire in an Australian Organisational Fleet Setting. Transportation

Research Part F, 10, 11-21.

Deaux, K., Dane, F. C., Wrightsman, L. S., & Sigelman, C. K. (1993). Social Psychology

in the 90's. Pacific Grove, California: Brooks/Cole publishing company.

Deery, H. A. (1999). Hazard and Risk Perception Among Novice Drivers. Journal of

Safety Research, 30(4), 225-236.

DeJoy, D. M. (1989). The Optimism Bias and Traffic Accident Risk Perception. Accident

Analysis and Prevention, 21(4), 333-340.

DeJoy, D. M. (1992). An Examination of Gender Differences in Traffic Accident Risk

Perception. Accident Analysis and Prevention, 24(3), 237-246.

239

Delhomme, P. (1991). Comparing One's Driving with Others': Assessment of Abilities

and Frequency of Offences, Evidence for a Superior Conformity of Self-Bias?

Accident Analysis and Prevention, 23(6), 493-508.

Dockery, T. M., & Bedeian, A. G. (1989). Attitudes Versus Actions - Lapiere (1934)

Classic Study Revisited. Social Behavior and Personality, 17(1), 9-16.

Douglas, M. (1978). Cultural Bias., Occasional Paper no. 35: Royal Anthropological

Institute of Great Britain and Ireland.

Douglas, M., & Wildavsky, A. (1982). Risk and Culture. Berkley, CA: University of

California Press.

Drinkwater, J. L., & Molesworth, B. R. C. (2010). Pilot See, Pilot Do: Examining the

Predictors of Pilots' Risk Management. Safety Science, 48, 1445-1451.

Driskill, W. E., Weismuller, J. J., Quebe, J., Hand, D. K., Dittmar, M. J., & Hunter, D. R.

(1997). The Use of Weather Information in Aeronautical Decision-Making.

Washington, D.C.: Federal Aviation Administration.

Eagly, A. H., & Chaiken, S. (1993). The Psychology of Attitudes. Belmont, CA:

Thompson Wadsworth.

240

Eagly, A. H., & Chaiken, S. (2007). The Advantages of an Inclusive Definition of

Attitude. Social Cognition, 25(5), 582-602.

Eggers, V. H. K. (2000). Customers and Actors of ATM: General Aviation. Air and

Space Europe, 2, 50-52.

Faber, M. H., & Stewart, M. G. (2003). Risk Assessment for Civil Engineering Facilities:

Critical Overview and Discussion. Reliability Engineering and System Safety, 80, 173-

184.

Fabrigar, L. R., Priester, J. R., Petty, R. E., & Wegener, D. T. (1998). The Impact of

Attitude Accessibility on Elaboration of Persuasive Messages. Personality and Social

Psychology Bulletin, 24(4), 339-352.

Farrand, P., & McKenna, F. P. (2001). Risk Perception in Novice Drivers: The

Relationship between Questionnaire Measures and Response Latency. Transportation

Research Part F, 4, 201-212.

Fazio, R. H. (2007). Attitudes as Object-Evaluation Associations of Varying Strength.

Social Cognition, 25(5), 603-637.

Fazio, R. H., Chen, J., McDonel, E. C., & Sherman, S. J. (1982). Attitude Accessibility,

Attitude Behavior Consistency, and the Strength of the Object Evaluation Association.

Journal of Experimental Social Psychology, 18(4), 339-357.

241

Fazio, R. H., Powell, M. C., & Williams, C. J. (1989). The Role of Attitude Accessibility

in the Attitude-to-Behavior Process. Journal of Consumer Research, 16(3), 280-288.

Fazio, R. H., & Williams, C. J. (1986). Attitude Accessibility as a Moderator of the

Attitude-Perception and Attitude-Behavior Relations - an Investigation of the 1984

Presidential-Election. Journal of Personality and Social Psychology, 51(3), 505-514.

Fazio, R. H., & Zanna, M. P. (1978). Predictive-Validity of Attitudes - Roles of Direct

Experience and Confidence. Journal of Personality, 46(2), 228-243.

Finucane, M. L., Alhakami, A., Slovic, P., & Johnson, S. M. (2000). The Affect Heuristic

in Judgments of Risks and Benefits. Journal of Behavioral Decision Making, 13(1), 1-

17.

Fischoff, B., Slovic, P., & Lichtenstein, S. (1982). Lay Foibles and Expert Fables in

Judgements About Risk. The American Statistician, 36(3), 240 - 255.

Fischoff, B., Slovic, P., Lichtenstein, S., Read, S., & Combs, B. (1978). How Safe Is Safe

Enough? A Psychometric Study of Attitudes Towards Technological Risks and

Benefits. Policy Sciences, 9(2), 127.

Fischoff, B., Watson, S. R., & Hope, C. (1984). Defining Risk. Policy Sciences, 17(2).

242

Fishbein, M., & Ajzen, I. (1974). Attitudes Towards Objects as Predictors of Single and

Multiple Behavioral Criteria. Psychological Review, 81(1), 59-74.

Forward, S. E. (2009). The Theory of Planned Behaviour: The Role of Descriptive

Norms and Past Behaviour in the Prediction of Drivers' Intentions to Violate.

Transportation Research Part F-Traffic Psychology and Behaviour, 12(3), 198-207.

Freudenburg, W. R., Coleman, C., Gonzales, J., & Helgeland, C. (1996). Media Coverage

of Hazard Events: Analyzing the Assumptions. Risk Analysis, 16(1), 31-42.

Gawronski, B., & Bodenhausen, G. V. (2007). Unravelling the Process Underlying

Evaluation: Attitudes from the Perspective of the APE Model. Social Cognition, 25(5),

687-717.

Gawronski, B., Lebel, E. P., & Peters, K. R. (2007). What Do Implicit Measures Tell Us?

Scrutinizing the Validity of Three Common Assumptions. Perspectives on

Psychological Science, 2(2), 181-193.

Glasman, L. R., & Albarracín, D. (2006). Forming Attitudes That Predict Behaviour: A

Meta-Analysis of the Attitude-Behaviour Relation. Psychological Bulletin, 132(5),

778-882.

243

Goh, J., & Wiegmann, D. A. (2001). Visual Flight Rules Flight into Instrument

Meteorological Conditions: An Empirical Investigation of the Possible Causes. The

International Journal of Aviation Psychology, 11(4), 359-379.

Goldvarg, E., & Johnson-Laird, P. N. (2001). Naive Causality: A Mental Model Theory

of Causal Meaning and Reasoning. Cognitive Science, 25(4), 565-610.

Green, M. F. (2001). Aviation System Safety and Pilot Risk Perception: Implications for

Enhancing Decision-Making Skills. Journal of Air Transportation World Wide, 6(1).

Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998). Measuring Individual

Differences in Implicit Cognition: The Implicit Association Test. Journal of

Personality and Social Psychology, 74(6), 1464-1480.

Greve, W. (2001). Traps and Gaps in Action Explanation: Theoretical Problems of a

Psychology of Human Action. Psychological Review, 108(2), 435-451.

Groeger, J. A., & Brown, I. D. (1989). Assessing One's Own and Others' Driving Ability:

Influences of Sex, Age, and Experience. Accident Analysis and Prevention, 21(2),

155-168.

Gulliver, P., & Begg, D. (2007 ). Personality Factors as Predictors of Persistent Risky

Driving Behavior and Crash Involvement among Young Adults. Injury Prevention, 13,

376-381.

244

Guppy, A. (1993). Subjective Probability of Accident and Apprehension in Relation to

Self-Other Bias, Age, and Reported Behavior. Accident Analysis and Prevention,

25(4), 375-382.

Hafer, C. L., Reynolds, K. L., & Obertynski, M. A. (1996). Message Comprehensibility

and Persuasion: Effects of Complex Language in Counter Attitudinal Appeals to

Laypeople. Social Cognition, 14(4), 317-337.

Han, H., Hsu, L.-T., & Sheu, C. (2010). Application of the Theory of Planned Behavior

to Green Hotel Choice: Testing the Effect of Environmental Friendly Activities.

Tourism Management, 31(3), 325-334.

Hatfield, J., Fernandes, R., Fuance, G., & Job, R. F. S. (2008). An Implicit Non-Self-

Report Measure of Attitudes to Speeding: Development and Validation. Accident

Analysis and Prevention, 40, 616-627.

Hawkins, F. H. (1993). Human Factors in Flight. Burlington, VT USA.: Ashgate,

Aldershot.

Hini, D., Gendall, P., & Kearns, Z. (1995). The Link between Environmental Attitudes

and Behaviour. Marketing Bulletin, 6, 22-31.

Horvath, P., & Zuckerman, M. (1993). Sensation Seeking, Risk Appraisal, and Risky

Behaviour. Personality and Individual Differences, 14(1), 41-52.

245

Hunter, D. R. (1995). Airman Research Questionnaire: Methodology and Overall

Results. Springfield, VA: Federal Aviation Administration.

Hunter, D. R. (2002). Risk Perception and Risk Tolerance in Aircraft Pilots. Washington,

D.C.: Office of Aerospace Medicine - Federal Aviation Authority.

Hunter, D. R. (2003). Measuring General Aviation Pilot Judgment Using a Situational

Judgment Technique. The International Journal of Aviation Psychology, 13(4), 373-

386.

Hunter, D. R. (2005). Measurement of Hazardous Attitudes among Pilots. The

International Journal of Aviation Psychology, 15(1), 23-43.

Hunter, D. R. (2006). Risk Perception Among General Aviation Pilots. The International

Journal of Aviation Psychology, 16(2), 135-144.

Hunter, D. R., Martinussen, M., & Wiggins, M. (2003). Understanding How Pilots Make

Weather-Related Decisions. The International Journal of Aviation Psychology, 13(1),

73-87.

Hutchins, E., Holder, B. E., & Pérez, R. A. (2002). Culture and Flight Deck Operations.

San Diego: University of San Diego.

246

Infrastructure, D. o. (2009). Aviation White Paper - Flight Path to the Future. Canberra,

ACT: Department of Infrastructure Transport Regional Development and Local

Government.

Iverson, H. (2004). Risk Taking Attitudes and Risky Driving Behaviour. Transportation

Research Part F, 7, 135-150.

Janic, M. (2000). An Assessment of Risk and Safety in Civil Aviation. Journal of Air

Transport Management, 6, 43-50.

Job, R. F. (1990). The Application of Learning Theory to Driving Confidence: The Effect

of Age and the Impact of Random Breath Testing. Accident Analysis and Prevention,

22(2), 97-107.

Johnson, S. E., & Hall, A. (2005). The Prediction of Safe Lifting Behavior: An

Application of the Theory of Planned Behavior. Journal of Safety Research, 36(1), 63-

73.

Kahle, L. R., & Berman, J. J. (1979). Attitudes Cause Behaviors - Cross-Lagged Panel

Analysis. Journal of Personality and Social Psychology, 37(3), 315-321.

Kakoko, D. C., Astrom, A. N., Lugoe, W. L., & Lie, G. T. (2006). Predicting Intended

Use of Voluntary HIV Counselling and Testing Services among Tanzanian Teachers

Using the Theory of Planned Behaviour. Social Science and Medicine, 63(4), 991-999.

247

Kaplan, S., & Garrick, J. B. (1981). On the Quantitative Definition of Risk. Risk

Analysis, 1(1).

Kasperson, R. E., Renn, O., Slovic, P., Brown, H. S., Emel, J., Goble, R., Kasperson, J.

X., & Ratick, S. (1988). The Social Amplification of Risk: A Conceptual Framework.

Risk Analysis, 8(2), 177-187.

Katz, D. (1960). The Functional Approach to the Study of Attitudes. Public Opinion

Quarterly, 24(2), 163-204.

Klinke, A., & Renn, O. (2002). A New Approach to Risk Evaluation and Management:

Risk-Based, Precaution-Based, and Discourse-Based Strategies. Risk Analysis, 22(6),

1071 - 1094.

Kos, J. M., & Clarke, V. A. (2001). Is Optimistic Bias Influenced by Control or Delay?

Health Education Research. Theory and Practice, 16(5), 533-540.

Kothandapani, V. (1971). Validation of Feeling, Belief, and Intention to Act as Three

Components of Attitude and Their Contribution to Prediction of Contraceptive

Behaviour. Journal of Personality and Social Psychology, 19, 321-333.

Kraus, S. J. (1995). Attitudes and the Prediction of Behavior - a Meta Analysis of the

Empirical Literature. Personality and Social Psychology Bulletin, 21(1), 58-75.

248

Krishnan, H. S., & Smith, R. E. (1998). The Relative Endurance of Attitudes, Confidence

and Attitude-Behavior Consistency: The Role of Information Source and Delay.

Journal of Consumer Psychology, 7(3), 273-298.

Kristensen, V., Aven, T., & Ford, D. (2006). A New Perspective on Renn and Klinke's

Approach to Risk Evaluation and Management. Reliability Engineering and System

Safety, 91, 421-432.

Kruglanski, A. W., & Thompson, E. P. (1999). Persuasion by a Single Route: A View

from the Unimodel. Psychological Inquiry, 10(2), 83-109.

Leippe, M. R., & Elkin, R. A. (1987). When Motives Clash - Issue Involvement and

Response Involvement as Determinants of Persuasion. Journal of Personality and

Social Psychology, 52(2), 269-278.

Lenne, M. G., Ashby, K., & Fitzharris, M. (2008). Analysis of General Aviation Crashes

in Australia Using the Human Factors Analysis and Classification System.

International Journal of Aviation Psychology, 18(4), 340-352.

Lichtenstein, S., Slovic, P., Fischoff, B., Layman, M., & Combs, B. (1978). Judged

Frequency of Lethal Events. Journal of Experimental Psychology: Human Learning

and Memory, 4(6), 551-578.

249

Lord, C. G., Paulson, R. M., Sia, T. L., Thomas, J. C., & Lepper, M. R. (2004). Houses

Built on Sand: Effects of Exemplar Stability on Susceptibility to Attitude Change.

Journal of Personality and Social Psychology, 87(6), 733-749.

MacGregor, D. G., Slovic, P., & Granger Morgan, M. (1994). Perception of Risks from

Electromagnetic Fields: A Psychometric Evaluation of a Risk-Communication

Approach. Risk Analysis, 14(5), 815-828.

Machin, M. A., & Sankey, K. S. (2008). Relationships between Young Drivers'

Personality Characteristics, Risk Perceptions and Driving Behaviour. Accident

Analysis and Prevention, 40, 541.

Machlis, G. E., & Rosa, E. A. (1990). Desired Risk: Broadening the Social Amplification

of Risk Framework. Risk Analysis, 10(1), 161.

Maio, G. R., Bell, D. W., & Esses, V. M. (1996). Ambivalence and Persuasion: The

Processing of Messages About Immigrant Groups. Journal of Experimental Social

Psychology, 32(6), 513-536.

Maio, G. R., & Olson, J. M. (1995). Relations between Values, Attitudes and Behavioural

Intentions: The Moderating Role of Attitude Formation. Journal of Experimental

Social Psychology, 31, 266-285.

250

Manstead, A. S. R., Proffitt, C., & Smart, J. L. (1983). Predicting and Understanding

Mothers Infant-Feeding Intentions and Behavior - Testing the Theory of Reasoned

Action. Journal of Personality and Social Psychology, 44(4), 657-671.

Marris, C., Langford, I. H., & O'Riordan, T. (1998). A Quantitative Test of the Cultural

Theory of Risk Perceptions: Comparison with the Psychometric Paradigm. Risk

Analysis, 18(5), 635-647.

Matthews, M. L., & Moran, A. R. (1986). Age Differences in Male Drivers' Perception of

Accident Risk: The Role of Perceived Driving Ability. Accident Analysis and

Prevention, 18(4), 292-303.

McKenna, F. P. (1993). It Won't Happen to Me: Unrealistic Optimism or Illusion of

Control? British Journal of Psychology, 84, 39-50.

McMillan, B., & Conner, M. (2003). Applying an Extended Version of the Theory of

Planned Behavior to Illicit Drug Use Among Students. Journal of Applied Social

Psychology, 33(8), 1662-1683.

Miesen, H. (2003). Predicting and Explaining Literary Reading: An Application of the

Theory of Planned Behavior. Poetics, 31(3-4), 189-212.

Millar, M. G., & Millar, K. U. (1998). Effects of Prior Experience and Thought on the

Attitude-Behavior Relation. Social Behavior and Personality, 26(2), 105-113.

251

Millar, M. G., & Tesser, A. (1986). Effects of Affective and Cognitive Focus on the

Attitude-Behaviour Relation. Journal of Personality and Social Psychology, 51(2),

270-276.

Molesworth, B. R. C. (2005). Experiential Training and Risk Management Behaviour

Amongst Pilots. Unpublished PhD Thesis, University of Western Sydney, Sydney.

Molesworth, B. R. C., & Chang, B. (2009). Predicting Pilots' Risk-Taking Behavior

through an Implicit Association Test. Human Factors, 51(6), 845-857.

Molesworth, B. R. C., Wiggins, M. W., & O'Hare, D. (2006). Improving Pilots’ Risk

Assessment Skills in Low-Flying Operations: The Role of Feedback and Experience.

Accident Analysis and Prevention, 38, 954-960.

Moyano Dìaz, E. (2002). Theory of Planned Behavior and Pedestrians' Intentions to

Violate Traffic Regulations. Transportation Research Part F: Traffic Psychology and

Behaviour, 5(3), 169-175.

Nighswonger Kraus, N., & Slovic, P. (1988). Taxonomic Analysis of Perceived Risk:

Modelling Individual and Group Perceptions within Homogenous Hazard Domains.

Risk Analysis, 8(3), 435-455.

252

O'Hare, D. (2003). Aeronautical Decision Making: Metaphors, Models, and Methods. In

P. S. Tsang & M. A. Vidulich (Eds.), Principles and Practices of Aviation (pp. 201-

237). Mahwah, NJ: Lawrence Erlbaum.

O'Hare, D., & Smitheram, T. (1995). ''Pressing on'' into Deteriorating Conditions: An

Application of Behavioral Decision Theory to Pilot Decision Making. International

Journal of Aviation Psychology, 5(4), 351-370.

Oltedal, S., Moen, B., Klempe, H., & Rundmo, T. (2004). Explaining Risk Perception. An

Evaluation of Cultural Theory (Report): Norwegian University of Science and

Technology, Department of Psychology, Norway.

Page, M. M. (1969). Social Psychology of a Classical Conditioning of Attitudes

Experiment. Journal of Personality and Social Psychology, 11(2), 177-186.

Paris, H., & Van Den Broucke, S. (2008). Measuring Cognitive Determinants of

Speeding: An Application of the Theory of Planned Behaviour. Transportation

Research Part F, 11, 168-180.

Petty, R. E. (2006). A Metacognitive Model of Attitudes. Journal of Consumer Research,

33, 22-24.

253

Petty, R. E., Briñol, P., & DeMarree, K. G. (2007). The Meta Cognitive Model (MCM) of

Attitudes: Implications for Attitude Measurement, Change, and Strength. Social

Cognition, 25(4), 657-686.

Petty, R. E., Wegener, D. T., & Fabrigar, L. R. (1997). Attitudes and Attitude Change.

Annual Review of Psychology, 48, 609-647.

Petty, R. E., Wheeler, C., & Bizer, G. Y. (1999). Is There One Persuasion Process or

More? Lumping Versus Splitting in Attitude Change Theories. Psychological Inquiry,

10(2), 156-163.

Pidgeon, N., Hood, C., Jones, D., Turner, B., & Gibson, R. (1992). Risk Perception.

London: The Royal Society Study Group.

Popa, M., Rotaru, C., & Oprescu, I. (2005). A Tri-Modal Risk Attitude Indicator for

Aviation Personnel: Design and Diagnostic Validity. Europe's Journal of Psychology.

Poulter, D. R., Chapman, P., Bibby, P. A., Clarke, D. D., & Crundall, D. (2008). An

Application of the Theory of Planned Behaviour to Truck Driving Behaviour and

Compliance with Regulations. Accident Analysis and Prevention, 40(6), 2058-2064.

Prislin, R., Wood, W., & Pool, G. J. (1998). Structural Consistency and the Deduction of

Novel from Existing Attitudes. Journal of Experimental Social Psychology, 34(1), 66-

89.

254

Quick, B. L., Stephenson, M. T., Witte, K., Vaught, C., Booth-Butterfield, S., & Patel, D.

(2008). An Examination of Antecedents to Coal Miners' Hearing Protection

Behaviors: A Test of the Theory of Planned Behavior. Journal of Safety Research,

39(3), 329-338.

Razavi, T. (2001). Self-Report Measures: An Overview of Concerns and Limitations of

Questionnaire Use in Occupational Stress Research, Discussion Papers in Accounting

and Management Science. Southampton, UK: University of Southampton.

Reason, J. (1990). Human Error. New York, NY: Cambridge University Press.

Renn, O. (1998). The Role of Risk Perception for Risk Management. Reliability

Engineering and System Safety, 59, 49-62.

Rothengatter, T. (2002). Drivers' Illusions - No More Risk. Tranportation Research Part

F, 5, 249-258.

Rundmo, T. (1996). Associations between Risk Perception and Safety. Safety Science,

24(3), 197-209.

Rydell, R. J., & McConnell, A. R. (2006). Understanding Implicit and Explicit Attitude

Change: A Systems of Reasoning Analysis. Attitudes and Social Cognition, 91(6),

995-1008.

255

Ryeng, E. R. (2012). The Effect of Sanctions and Police Enforcement on Drivers' Choice

of Speed. Accident Analysis and Prevention,45, 446-454.

Schwarz, N. (2007). Attitude Construction: Evaluation in Context. Social Cognition,

25(5), 638-656.

Scott-Parker, B., Watson, B., & King, M. (2009). Understanding the Psychosocial Factors

Influencing the Risky Behavior of Young Drivers. Transportation Research Part F, 12

470-482.

Shappell, S. A., & Wiegmann, D. A. (2000). The Human Factors Analysis and

Classification System - HFACS. Springfield, VA: Federal Aviation Administration.

Shappell, S. A., & Wiegmann, D. A. (2003). A Human Error Analysis of General

Aviation Controlled Flight into Terrain Accidents Occurring between 1990-1998.

Springfield, VA: Federal Aviation Administration.

Siegrist, M., Keller, C., & Kiers, H. A. L. (2005). A New Look at the Psychometric

Paradigm of Perception of Hazards. Risk Analysis, 25(1), 211-222.

Simpson, P. A. (2001). Naturalistic Decision Making in Aviation Environments.

Melbourne, Vic.: DSTO Aeronautical and Maritime Research Laboratory.

256

Siu, O. L., Phillips, D. R., & Leung, T. W. (2003). Age Differences in Safety Attitudes

and Safety Performance in Hong Kong Construction Workers. Journal of Safety

Research, 34(2), 199-205.

Sjöberg, L. (1996). A Discussion of the Limitations of the Psychometric and Cultural

Theory Approached to Risk Perception. Radiation Protection Dosimetry, 68(3/4), 219-

225.

Sjöberg, L. (2000). Factors in Risk Perception. Risk Analysis, 20(1), 1-11.

Sjöberg, L. (2002). Are Received Risk Perception Models Alive and Well? Risk Analysis,

22(4), 665-669.

Sjöberg, L., Moen, B., & Rundmo, T. (2004). Explaining Risk Perception. An Evaluation

of the Psychometric Paradigm in Risk Perception Research Trondheim: Norwegian

University of Science and Technology, Department of Psychology.

Slovic, P., & Weber, E. U. (2002, April 12-13, 2002). Perception of Risk Posed by

Extreme Events. Paper presented at the Risk Management Strategies in an Uncertain

World, Palisades, New York.

Smith, J. R., & Terry, D. J. (2003). Attitude-Behaviour Consistency: The Role of Group

Norms, Attitude Accessibility, and Mode of Behavioural Decision-Making. European

Journal of Social Psychology, 33, 591-608.

257

Smith, S. M., & Petty, R. E. (1996). Message Framing and Persuasion: A Message

Processing Analysis. Personality and Social Psychology Bulletin, 22(3), 257-268.

Sparks, P., & Shepherd, R. (1994). Public Perceptions of the Potential Hazards

Associated with Food Production and Food Consumption: An Empirical Study. Risk

Analysis, 14(5), 799-806.

Stahlberg, D., & Frey, D. (1996). Introduction to Social Psychology. Oxford, UK:

Blackwell Publishers.

Steg, L., & Sievers, I. (2000). Cultural Theory and Individual Perceptions of

Environmental Risks. Environment and Behaviour, 32(2), 250-269.

Svenson, O., Fischoff, B., & MacGregor, D. (1985). Perceived Driving Safety and

Seatbelt Usage. Accident Analysis and Prevention, 17(2), 119-113.

Tabachnick, B. G., & Fidell, L. S. (2013). Using Multivariate Statistics(5th ed.): Pearson.

Terry, D. J., & O'Leary, J. E. (1995). The Theory of Planned Behavior - the Effects of

Perceived Behavioral-Control and Self-Efficacy. British Journal of Social Psychology,

34, 199-220.

Tesser, A., & Shaffer, D. R. (1990). Attitudes and Attitude Change. Annual Review of

Psychology, 41, 479-523.

258

Thomas, M. J. W. (2004). Predictors of Threat and Error Management: Identification of

Core Nontechnical Skills and Implications for Training Systems Design. International

Journal of Aviation Psychology, 14(2), 207-231.

Tonglet, M., Phillips, P. S., & Read, A. D. (2004). Using the Theory of Planned

Behaviour to Investigate the Determinants of Recycling Behaviour: A Case Study

from Brixworth, UK. Resources Conservation and Recycling, 41(3), 191-214.

Turco, R. M. (1996). Self-Referencing, Quality of Argument, and Persuasion. Current

Psychological Research and Reviews, 15(3), 258-276.

Tversky, A., & Kahneman, D. (1973). Availability: A Heuristic for Judging Frequency

and Probability. Cognitive Psychology, 5, 207-232.

Ulleberg, P., & Rundmo, T. (2003). Personality, Attitudes and Risk Perception as

Predictors of Risky Driving Behaviour among Young Drivers. Safety Science, 41, 427-

443.

van Overwalle, F., & Siebler, F. (2005). A Connectionist Model of Attitude Formation

and Change. Personality and Social Psychology Review, 9(3), 231-274.

Virgin Blue. (2010). Virgin Blue Flight Crew - Domestic. Retrieved 01/06/10/2010,

from http://www.bfound.net/detail.aspx?jobId=78873&CoId=43&rq=7

259

Vriljing, J. K., van Hengel, R. J., & Houben, R. J. (1995). A Framework for Risk

Evaluation. Journal of Hazardous Materials, 43, 245-261.

Wahlberg, A. A. F., & Sjöberg, L. (2000). Risk Perception and the Media. Journal of

Risk Research, 3(1), 31-50.

Wallis, T. S. A., & Horswill, M. S. (2007). Using Fuzzy Signal Detection Theory to

Determine Why Experienced and Trained Drivers Respond Faster Than Novices in a

Hazard Perception Test. Accident Analysis and Prevention, 39, 1177-1185.

Weigel, R. H., & Newman, L. S. (1976). Increasing Attitude-Behavior Correspondence

by Broadening Scope of Behavioral Measure. Journal of Personality and Social

Psychology, 33(6), 793-802.

Weinstein, N. D. (1980). Unrealistic Optimism About Future Life Events. Journal of

Personality and Social Psychology, 39(5), 806-820.

Werner, P. D. (1978). Personality and Attitude-Activism Correspondence. Journal of

Personality and Social Psychology, 36(12), 1375-1390.

Wetton, M. A., Horswill, M. S., Hatherly, C., Wood, J. M., Pachana, N. A., & Anstey, K.

J. (2010). The Development and Validation of Two Complimentary Measures of

Driving Hazard Perception Ability. Accident Analysis and Prevention, 42(4), 1232-

1239.

260

Wicker, A. W. (1969). Attitudes Versus Actions - Relationship of Verbal and Overt

Behavioral Responses to Attitude Objects. Journal of Social Issues, 25(4), 41-78.

Wiegmann, D. A., Zhang, H., Von Thaden, T. L., Sharma, G., & Gibbons, A. M. (2004).

Safety Culture: An Integrative Review. The International Journal of Aviation

Psychology, 14(2), 117-134.

Wiener, J. B., & Rogers, M. D. (2002). Comparing Precaution in the United States and

Europe. Journal of Risk Research, 5(4), 317-349.

Wildavsky, A., & Dake, K. (1990). Theories of Risk Perception: Who Fears What and

Why? Daedalus, 119(4), 41.

Wilkinson, I. (2001). Social Theories of Risk Perception: At Once Indispensable and

Insufficient. Current Sociology, 49(1), 1-22.

Wilson, D. R., & Fallshore, M. (2001). Optimistic and Ability Biases in Pilots' Decisions

and Perceptions of Risk Regarding Vfr Flight into Imc. Paper presented at the the 11th

International Biennial Symposium on Aviation Psychology, Columbus, Ohio,.

Wilson, T. D., Lindsey, S., & Schooler, T. Y. (2000). A Model of Dual Attitudes.

Psychological Review, 107(1), 101-126.

261

Wogalter, M. S., Brelsford, J. W., Desaulniers, D. R., & Laughery, K. R. (1991).

Consumer Product Warnings: The Role of Hazard Perception. Journal of Safety

Research, 22, 71-82.

Wood, W. (2000). Attitude Change: Persuasion and Social Influence. Annual Review of

Psychology, 51, 539-570.

262

Appendix 1. Participant Consent Form

(UNSW Letterhead Here)

Approval No 08/2009/33

THE UNIVERSITY OF NEW SOUTH WALES

PARTICIPANT INFORMATION STATEMENT AND CONSENT FORM

She’ll Be Right: Pilot’s Assessment of Risk in Australian General Aviation I, Justin Drinkwater would like to invite you to participate in a study on perceptions, attitudes, and the associated behaviour in flight training. You were selected as a possible participant in this study because you are a pilot in the Australian General Aviation sector.

If you decide to participate, you will be asked to complete a series of questionnaires, namely; demographics questionnaire, Risk Perception scale, Aviation Safety Attitude questionnaire in addition to a questionnaire relating to flight performance. In total it is anticipated that the research will take no longer than one hour to complete.

You may benefit from the study by gaining an insight into the perceptions and attitudes of both yourself and others, as well as your behaviour. You may also benefit from this experiment through the publication of findings that are related to flight safety and relevant to the wider aviation community.

Any information that is obtained in connection with this study and that can be identified with you will remain confidential and will be disclosed only with your permission, except as required by law. If you give your permission by signing this document, I plan to publish the results. A report will be written for UNSW, and the results may be published in conference proceedings, books or journals. In any publication, information will be provided in such a way that you cannot be identified.

Complaints may be directed to the Ethics Secretariat, The University of New South Wales, SYDNEY 2052, AUSTRALIA (phone 9385 4234, fax 9385 6648, email [email protected]). Any complaint you make will be treated in confidence and investigated, and you will be informed of the outcome.

263

Your decision whether or not to participate will not prejudice your future relations with The University of New South Wales. If you decide to participate, you are free to withdraw your consent and to discontinue participation at any time without prejudice.

If you have any questions, please feel free to ask me. If you have any additional questions later, I (Tel: 9385 6767), will be happy to answer them.

You will be given a copy of this form to keep.

Justin Drinkwater SYDNEY 2052 AUSTRALIA SYDNEY 2052 AUSTRALIA Chief Investigator Telephone: +61 (0)2 9385 6767 Email: [email protected] Location: Room 208, Old Main Building

264

THE UNIVERSITY OF NEW SOUTH WALES

PARTICIPANT INFORMATION STATEMENT AND CONSENT FORM

(continued)

“She’ll Be Right”: Pilot’s Assessment of Risk in Australian General

Aviation

You are making a decision whether or not to participate. Your signature indicates

that, having read the Participant Information Statement, you have decided to take

part in the study.

Signature of Research Participant .…………………………………………………….

(Please PRINT name) …………………………………………………… Date

Signature of Witness.…………………………………………………….

(Please PRINT name) .…………………………………………………….

Nature of Witness .……………………………………………………

Signature(s) of Investigator(s) ……………………………………………………

Please PRINT Name .…………………………………………………….

265

REVOCATION OF CONSENT

“She’ll Be Right”: Assessment of Risk in Australian Flight Training

I hereby wish to WITHDRAW my consent to participate in the research proposal described above and understand that such withdrawal WILL NOT jeopardise any treatment or my relationship with The University of New South Wales.

Signature …………………………………………………… Date ……………………………………………………

Please PRINT Name ……………………………………………………

The section for Revocation of Consent should be forwarded to Justin Drinkwater,

Department of Aviation, UNSW, Sydney, NSW 2052.

266

Appendix 2. Demographic Survey

Demographic Information

Age: ____ Gender: Male: F Female: F

Please i ndicate w hich of the following licences that you hold:

GFPT F Private F

Commercial F ATPL F

Please indicate w hich of th e following ratings that you hold:

Instructor F Instrument F

Have you e ver con ducte d any low‐level flight training?

Yes: F No: F

Each of the following questions is related to your flying experience. Please estimate these figures as accurately as possible.

Number of hours (total) experience:

Number of hours (total) as pilot in command:

Number of hours (total) actual IFR experience:

Number of cross‐country hours experience (excluding training):

Number of hours (total) during the previous 90 days:

267

Number of cross‐country hours during the previous 90 days:

Number of times you have been forced to f l y belo w 500f t AGL:

Number of hours (total) low‐level flying:

Have you ever been involved in an aircraft accident or incident involving low‐level flying?

Yes: No:

Do you know of anyone who has been involved in an accident or incident involving low‐level flying?

Yes: No:

Over the previous six months, how often have you used GPS as a primary source of navigation?

Never Rarely So meti m es Fre qu en tly Al wa y s

268

Appendix 3. Hunter’s Risk Perception Scale One

Aviation Research HRP 1

In this exercise, you will see several descriptions of other pilots who are involved in aviation situations. Your task will be to decide how risky each situation is. Unless the description says otherwise, you may assume that the pilot involved in the situation is an average general aviation pilot, with about 300 hours of total experience, who has flown about 30 hours over the last 12 months.

You will rate the risk in each of the situations on a scale of 1 to 100.

The 1 to 100 risk scale is defined as follows:

1 ‐‐ Virtually zero risk involved in this situation. It is about as safe as sitting on

the couch watching TV.

50 ‐‐ The same amount of risk as driving your car on a freeway in moderate traffic

and good weather conditions during the day.

100 ‐‐ Extremely high risk of a serious, probably fatal accident. The pilot will be

very fortunate to escape from this situation alive and with the aircraft undamaged.

269

Item Rating Scenario 1 On short final a pilot drops his microphone on the floor. He looks down while bending over trying to reach it. He inadvertently moves the control yoke and the aircraft banks sharply. 2 The pilot is in a hurry to get going and does not carefully check his seat, seat belt, and shoulder harness. When he rotates, the seat moves backward on its tracks. As it slides backward, the pilot pulls back on the control yoke, sending the nose of the aircraft upward. As the airspeed begins to decay, he strains forward to push the yoke back to a neutral position. 3 A line of thunderstorms block the route of flight, but a pilot sees that there is a space of about 10 miles between two of the cells. He can see all the way to clear skies on the other side of the thunderstorm line, and there does not seem to be any precipitation along the route, although it does go under the extended anvil of one of the cells. As he tries to go between the storms, he suddenly encounters severe turbulence and the aircraft begins to be pelted with hail. 4 Low ceilings obscure the tops of the mountains, but the pilot thinks that he can see through the pass to clear sky on the other side of the mountain ridges. He starts up the wide valley that gradually gets narrower. As he approaches the pass he notices that he occasionally loses sight of the blue sky on the other side. He drops down closer to the road leading through the pass and presses on. As he goes through the pass, the ceiling continues to drop and he finds himself suddenly in the clouds. He holds his heading and altitude and hopes for the best. 5 Just after takeoff a pilot hears a banging noise on the passenger side of the aircraft. He looks over at the passenger seat and finds that he can't locate one end of the seatbelt. He trims the aircraft for level flight, releases the controls, and tries to open the door to retrieve the seatbelt. 6 During the planning for a 2 hour cross‐country flight, a pilot makes a mistake in computing the fuel consumption. He believes that he will have over an hour of fuel remaining upon arrival, but he will really only have about 15 minutes of fuel left.

270

7 After working a full day, a businesswoman drives out to the airport for her three hour flight home. She is tired, and the sun is setting, but the weather forecast is for clear sky and good visibility. About an hour after takeoff, she begins to feel very tired and sleepy. She regrets not bringing any coffee along, and opens the cockpit air vent to get some fresh, cool air. 8 It is late afternoon and the VFR pilot is flying west into the setting sun. For the last hour, the visibility has been steadily decreasing, however his arrival airport remains VFR, with 4 miles visibility and haze. This is a busy uncontrolled airfield with a single East‐West runway. He decides to do a straight‐in approach. 9 When he took off about an hour earlier, there was a quartering headwind of about 15 knots. He made it into the air, but it was a rocky takeoff, and one he hoped none of the other pilots at the small airport noticed. Now as he entered the downwind leg for landing, he noticed that the windsock was indicating almost a direct crosswind of about the same strength. On final he is holding a large crab to keep from drifting away from the centreline, and as he starts the flare he begins to drift toward the side of the runway. 10 While on a local sightseeing flight, the pilot notices that the weather is deteriorating to the west. A line of clouds is moving in his direction, but they are still over 20 miles away. He decides to cut his flight short and turns to return to his home airfield about 25 miles east of his present position. 11 The instructor pilot had been suffering from a cold and when he arose in the morning, he took an over‐the‐counter antihistamine to try and control his runny nose. After a morning of giving instruction in the flight simulator, he had a lesson scheduled after lunch with a pilot working on his commercial licence. He felt a little drowsy, but the weather was good and they were going to be working on short‐ field landings, so he did not cancel the lesson. 12 A pilot is cruising in good weather to a destination airport about an hour away. It is midday, and there are three hours of fuel on board. 13 An experienced pilot with a rated passenger are taxiing out for takeoff. They are at a controlled airfield, on the ground‐control radio frequency. They have been cleared to "taxi to and hold short of Runway 31 " and are now approaching the hold‐short line.

271

14 An instrument‐rated pilot on an IFR flight plan has just climbed through a 4000 foot thick layer of clouds. Although icing was not forecast, he notices a trace of ice on the edges of the windscreen. The aircraft is not equipped for flight into known or forecast icing conditions. As he approaches his destination airport, air traffic control issues a clearance that will require him to hold for approximately 15 minutes in the cloud layer. 15 For the first part of this late night flight, the low‐time VFR pilot has enjoyed a spectacular view of the stars as he cruised at 8,500 feet with over 25 miles visibility. As he nears his destination airport, which sits on the far side of a large lake, he notices that the visibility is decreasing because of haze nearer the surface. As he starts across the lake at about 2,500 feet he loses sight of the lights on the shore, and the dim lights scattered far apart on the ground seem to be indistinguishable from the stars. 16 It is time for an oil change and the pilot/owner decides to do it himself. He consults with his local A&P mechanic and then follows his instructions. He does not have the work inspected afterwards and makes the appropriate log book notation himself. 17 While cruising at 4,500 feet AGL, the engine on the single‐engine aircraft sputters and quits. The pilot checks the fuel settings and tries to restart the engine but is unsuccessful. He sees a level field within gliding distance and turns toward it. He will be landing into the wind.

272

Appendix 4. Hunter’s Risk Perception Scale Two

Aviation Research HRP 2

In this exercise, you will be given descriptions of common aviation and everyday situations. After you have read the description of each situation, you will decide how risky the situation would be if YOU were in that situation tomorrow. Base your rating on your personal training and experiences, and use the 1 to 100 risk rating scale shown below.

The 1 to 100 risk rating scale is defined as follows:

1 ‐‐ Virtually zero risk involved in this situation. It is about as safe as sitting on the couch watching TV.

50 ‐‐ The same amount of risk as driving your car on a freeway in moderate traffic and good weather conditions during the day.

100 ‐‐ Extremely high risk of a serious, probably fatal accident. The pilot will be very fortunate to escape from this situation alive and with the aircraft undamaged.

Item Rating Scenario 1 During the daytime, fly from your local airport to another airport about 150 miles away, in clear weather, in a well‐maintained aircraft. 2 Jaywalk (cross in the middle of the block) across a busy downtown street. 3 Make a two‐hour cross‐country flight with friends after checking your weight and balance. 4 Fly across a large lake or inlet at 500 feet above ground level. 5 At night, take a cross‐country flight in which you land with over an hour of fuel remaining. 6 Climb up a 10‐foot ladder to replace an outside light bulb.

273

7 Fly in clear air at 6,500 feet between two thunderstorms about 25 miles part. 8 Take a two‐hour sightseeing flight over an area of wooded valleys and hills, at 3,000 above ground level. 9 During the daytime, take a cross‐country flight in which you land with 30 minutes of fuel remaining. 10 Make a traffic pattern so that you end up turning for final with about a 45 degree bank. 11 Drive your car on a freeway near your home at night, at 110 KPH in moderate traffic. 12 Take a two‐hour flight in a jet aircraft on a major US air carrier. 13 During the daytime, take a cross‐country flight in which you land with over an hour of fuel remaining. 14 During the daytime, fly from your local airport to another airport about 150 miles away, in a well‐maintained aircraft, when the weather is marginal VFR (3 miles visibility and 2,000 foot overcast). 15 Fly across a large lake or inlet at 1,500 feel above ground level. 16 Make a traffic pattern so that you end up turning for final with about a 30 degree bank. 17 Drive your car on a freeway near your home, during the day, at 110 KPH in moderate traffic, during heavy rain. 18 Start a light aircraft with a dead battery by hand‐propping it. 19 Make a two‐hour cross country flight with friends, without checking your weight and balance. 20 Drive your car on a freeway near your home during the day, at 110 KPH in moderate traffic. 21 At night, take a cross‐country flight in which you land with 30 minutes of fuel remaining. 22 Take a two‐hour sightseeing flight over an area of wooded valleys and hills, at 1,000 above ground level. 23 At night, fly from your local airport to another airport about 150 miles away, in clear weather, in a well‐maintained aircraft. 24 Fly across a large lake or inlet at 3,500 feet above ground level. 25 Ride an elevator from the ground floor to the 25th floor of an office building 26 At night, fly from your local airport to another airport about 150 miles away, in a well‐maintained aircraft, when the weather is marginal VFR (3 miles visibility and 2,000 foot overcast.

274

Appendix 5. The Aviation Safety Attitude Scale

Show your response to the questions 1 2 3 4 5 below using the following scale Strongly Strongly Agree Disagree I would duck below minimas to get home I am capable of instrument flight I am a very careful pilot I never feel stressed when I am flying The rules controlling flying are much too strict I am a very capable pilot I am so careful that I will never have an accident I am very skilful on controls I know aviation procedures very well I deal with stress very well It is riskier to fly at night than during the day Most of the time accidents are caused by things beyond the pilot's control I have a thorough knowledge of my aircraft Aviation weather forecasts are usually accurate I am a very cautious pilot The pilot should have more control over how he/she flies Usually, your first response is your best I find it easy to understand the weather information I get before flights You should decide quickly and then make judgements later It is very unlikely that a pilot of my ability would have an accident

275

I fly enough to maintain my proficiency I know how to get help from ATC if I get into trouble There are few situations I couldn't get out of If you don't push yourself and the aircraft a little, you'll never know what you could do I often feel stressed when flying in or near weather Sometimes you'll just have to depend on luck to get you through Speed is more important that accuracy during an emergency

276

Appendix 6. Experiment Two A Scenarios

Cross­Country Scenario

You are planning on flying a normal cross‐country flight for fun. You have had significant previous experience both with the aircraft, and with the route you are flying.

The expected time of arrival for this flight is 20 minutes before last light. You have a night rating, but haven’t used it in the last month, and have flown at night very rarely in the last six months.

You have the chance to leave an hour earlier than planned if you wish by taking an alternate aircraft of the same type. This aircraft is slightly slower than the original aircraft, as it is an older aircraft with a slightly less powerful engine. This will add about 10 – 20 minutes to your flight time. The alternate aircraft does not have the more modern avionics suite that is found in the original aircraft. The aircraft specifically lacks a GPS, and has an older radio fitted.

Your expected landing at the airport will be into the west.

1) What would you do in this situation?

______

277

2) Which aircraft would you choose?

Original (newer aircraft) Alternate (older aircraft) N/A

2a) Why did you choose this option?

______

3) On a scale of 1 to 100, 1 being almost zero risk (sitting at home in front of the television) and 100 being extreme risk (almost certain chance of death), how would you rate this scenario?

______/100

4) I would describe this flight as containing the following amount of risk (please circle) low medium high

5) What would best describe your decision making for this scenario? (please circle)

1. Recall past examples from your experience

2. Consider the pros and cons of the alternatives

3. Recall and follow a specific rule

4. Recall past examples that you’ve heard or read about

5. Base your decision on a model in which decisions should be made

6. Know immediately whether it was the right decision

Proceed to the next scenario.

278

Ferry to LAME Scenario

You have been asked by a friend to ferry their aircraft to the nearest LAME for its 100 hourly service. You fly together relatively frequently, and he allows you the use of his plane at the cost of the fuel used because of your friendship.

The airport that you are to ferry the aircraft to is situated at sea level, but it is bordered by 2000 foot hills. The cloud level is estimated at 3000 feet AMSL at the airport at present. The prevailing wind is from the south, and is currently at about 10 to 15 knots. The runway is 34/16.

The terrain between your home airport and the LAME’s port is generally at 2000 feet AMSL, and is approximately 1.5 hours flying time from your home airport. There are very few airports that are close to the route you will fly. The best available is a well maintained private strip, however you have not used this strip before, as the owner reportedly has a habit of leaving machinery on it to discourage use.

There is a storm front to the south of the airport, and the forecast is for the cloud base to drop throughout the day. This has been reported by sites to the south of the airport and is due to reach the destination about 15 minutes after your ETA. You plan is to drop the plane at the LAME, and quickly depart in another aircraft that the LAME has just completed a 100 hourly check on.

1) What would you do in this situation?

______

279

2) Why did you choose this option?

______

3) On a scale of 1 to 100, 1 being almost zero risk (sitting at home in front of the television) and 100 being extreme risk (almost certain chance of death), how would you rate this scenario?

______/100

4) I would describe this flight as containing the following amount of risk (please circle) low medium high

5) What would best describe your decision making for this scenario? (please circle)

1. Recall past examples from your experience

2. Consider the pros and cons of the alternatives

3. Recall and follow a specific rule

4. Recall past examples that you’ve heard or read about

5. Base your decision on a model in which decisions should be made

6. Know immediately whether it was the right decision

Proceed to the next scenario.

280

Sightseeing Scenario

You are about to take your friends on a local sightseeing flight over the local mountains. You have conducted this flight many times in your career, as it is very close to your flight school, and was often used in your low‐level training.

You expect to fly at around 500 feet AGL during parts of the flight, although this will only be in areas that have cleared land in which an emergency landing can be conducted if you experience power failure.

Your friends are running a little bit late for the flight because they’ve been at a lunch at the local winery where they’ve been drinking moderate amounts of alcohol.

1) What would you do in this situation?

______

2) Why did you choose this option?

______

If you chose NOT to go, proceed to Question 4 in this Scenario

281

3) As you fly towards the mountains you find that one of your friends is getting motion sickness. He tells you that he'd prefer to return to the airport and pay the excess in cost to return to the airport. This will cost him approximately $120. You have no issues with duty time or fuel on board.

Describe what you would do in this situation.

______

4) On a scale of 1 to 100, 1 being almost zero risk (sitting at home in front of the television) and 100 being extreme risk (almost certain chance of death), how would you rate this scenario?

______/100

5) I would describe this flight as containing the following amount of risk (please circle) low medium high

6) What would best describe your decision making for this scenario? (please circle)

1. Recall past examples from your experience

2. Consider the pros and cons of the alternatives

3. Recall and follow a specific rule

4. Recall past examples that you’ve heard or read about

5. Base your decision on a model in which decisions should be made

6. Know immediately whether it was the right decision

Proceed to the next scenario.

282

Sydney – Melbourne Scenario

You are asked to ferry a Piper Navajo on a flight from Sydney to Melbourne to deliver it to its new owner. The current owner asked you to fly the aircraft as you were going to Melbourne to see family for Christmas and your brother’s birthday, and this would “kill two birds with one stone”.

The maintenance release checks out and the owner assures you that the plane is fully serviceable. He goes so far to say that he’d be “happy to have his children in it”.

You’ve had a fair amount of time on type, and are confidant that you can undertake the flight, especially because you’ve just passed your ATPL theory exams on the first attempt.

Your brother is the only other person on board for the flight. He is due to be married on New Year’s Eve. Your parents call the night before the flight, and tell you that they’ve organised for a family get together for the next day and that they’ll meet you at the airport to pick you both up.

1) What would you do in this situation?

______

2) Why did you choose this option?

______

283

______

If you chose NOT to go, proceed to Question 4 in this Scenario

3) During your flight, the weather in Melbourne begins to deteriorate. Your brother has his mobile on board, and he checks the current weather on it using the internet. There appears to be a large storm approaching the city, which is likely to be at Melbourne at about the same time as your ETA. The storm looks like a typical Melbourne summer storm, and should pass relatively quickly. Upon reaching Melbourne, your endurance will be about 2 hours, excluding your reserves. You have planned to fly near to Benalla airport, as this takes you away from high terrain. Because of your planned track, you have the option of diverting to Benalla to wait for the weather to pass. The airport has food and fuel available, and is less than an hour’s flight from Melbourne. Would you choose to land at Benalla, or to continue to Melbourne?

Land at Benalla Continue to Melbourne? Other?

3b) Why did you choose this option?

______

4) On a scale of 1 to 100, 1 being almost zero risk (sitting at home in front of the television) and 100 being extreme risk (almost certain chance of death), how would you rate this scenario? ______/100

5) I would describe this flight as containing the following amount of risk (please circle) low medium high

284

6) What would best describe your decision making for this scenario? (please circle)

1. Recall past examples from your experience

2. Consider the pros and cons of the alternatives

3. Recall and follow a specific rule

4. Recall past examples that you’ve heard or read about

5. Base your decision on a model in which decisions should be made

6. Know immediately whether it was the right decision

Proceed to the next scenario.

285

Hunter Valley Scenario

You offer to take your best friend flying for his 30th birthday. You know that he loves the Hunter valley, and seeing as though you both live at Camden, it makes sense to fly to the Hunter for the weekend. Your partners will drive and meet you both at Cessnock airport.

The aircraft that you are going to fly is familiar to you, as you fly it most weekends. You also fly to Cessnock quite regularly as part of your job as a flight instructor.

The weather for the weekend is forecast to be fine with little wind. As a backup plan, you have organised to have your partner drive both couples in case the weather turns bad or any other unforseen problems arise.

1) What would you do in this situation?

______

2) Why did you choose this option?

______p.t.o

286

3) On a scale of 1 to 100, 1 being almost zero risk (sitting at home in front of the television) and 100 being extreme risk (almost certain chance of death), how would you rate this scenario?

______/100

4) I would describe this flight as containing the following amount of risk (please circle) low medium high

5) What would best describe your decision making for this scenario? (please circle)

1. Recall past examples from your experience

2. Consider the pros and cons of the alternatives

3. Recall and follow a specific rule

4. Recall past examples that you’ve heard or read about

5. Base your decision on a model in which decisions should be made

6. Know immediately whether it was the right decision

Proceed to the next scenario.

287

Victor One Scenario

You have planned to fly Victor One with your friends who are in Sydney for a summer holiday.

You have had a big day on the day before your planned flight, and you don’t get to sleep until 03:00. You also feel a cold coming on. You wake at about 07:00 in order to work on the tax documentation for your business. This work is relatively complex, and from past experience you know that it has a tendency to tire you.

Your ETD for the flight is 17:00, and you arrive at the airport at 16:00 to ensure that the aircraft is ready, and to complete the paperwork for the flight.

You are flying in a Cessna 210, with which you’ve had a reasonable amount of flight time, but haven’t flown this type extensively for about 2 years, since you flew mail runs.

The weather is changing fairly rapidly in the local area and there is a gusty wind blowing. Pilots at the flight school indicate that it’s relatively smooth flying, but that landing is a little difficult because of the crosswinds. You haven’t completed many cross‐wind landings recently, because you’ve been concentrating on your business and have been flying a lot less as a result.

Your friends arrive for the flight at 16:50. At about the same time a rain shower sets in at Bankstown, and the weather radar shows that this is like to last until about 17:30.

At about 17:00 your partner rings you to ask you to bring home some take‐away food for dinner, because the car has broken down, and they couldn’t get to the supermarket to do the shopping.

You tell your friends that the flight will be a little late because of the weather. They indicate that they are very keen on flying today ‐ one of them has travelled down from Byron Bay especially for the flight, so they don’t mind waiting.

288

You have been drinking coffee throughout the day, although you notice that you are feeling tired. The instructor at the flight school comes to you with your paper work – you’ve made some very basic errors in addition, and you forgot to sign your name.

1) What would you do in this situation?

______

2) Why did you choose this option?

______

3) On a scale of 1 to 100, 1 being almost zero risk (sitting at home in front of the television) and 100 being extreme risk (almost certain chance of death), how would you rate this scenario?

______/100

4) I would describe this flight as containing the following amount of risk (please circle) low medium high

5) What would best describe your decision making for this scenario? (please circle)

1. Recall past examples from your experience

2. Consider the pros and cons of the alternatives

3. Recall and follow a specific rule

289

4. Recall past examples that you’ve heard or read about

5. Base your decision on a model in which decisions should be made

6. Know immediately whether it was the right decision

Proceed to the next scenario.

290

Moruya Scenario

Your brother has recently started a job at Moruya, in which he is a flight instructor and part‐time drop pilot for the local skydiving operation. You have gone down to Moruya to visit him, and to hopefully go for a flight, as it is one of your favourite areas in the state to fly.

A plane has returned from a local flight early, and your brother tells you that you can have the plane for a quick flight. As a result of the time pressure, you don’t re‐ fuel the plane. The aircraft has about 1 hour of fuel remaining before the flight, and you decide to fly for about half an hour.

Your brother’s flight school has obtained a special exemption on their AOC to carry only 5 minutes of fixed reserve fuel.

You are currently on the downwind leg of a circuit for runway 18 at Moruya aerodrome. The owner of the flight school radios you with a request for help. He wants you to fly up to Batemans Bay to search for a parachutist that has landed away from the designated landing area for the parachuting operation, which is on the shores of Batemans Bay. The owner tells you that skydiver should be easy to find, as his parachute is bright white, and he has been trained to make the parachute as visible as possible in the even of missing the landing area.

You are currently at 1000 feet AGL. Batemans Bay is directly in front of you at the present time. The bay is about 10‐12 nautical miles from your position, and will require approximately 5‐6 minutes of flight time to reach at 100 knots.

At 100 knots (2400 RPM), the aircraft has approximately 18 minutes of primary fuel available, with the 5 minutes of fixed reserves also on board.

The owner has asked that once you have found the parachutist to report it to him on the radio, fly a full orbit, and then return to Moruya. p.t.o.

291

1) What would you do in this situation?

______

2) Why did you choose this option?

______

(if you chose not to go, proceed to Question 6 in this scenario)

3) Once overhead Batemans Bay, it becomes apparent that finding the skydiver is unlikely to be easy, as there is a significant amount of area to search. The bay has a north and south coastline, with Batemans Bay township on the south‐western side of the bay. The northern coastline is approximately 1 to 2 nautical miles to the north of the southern coastline. Your flight to the area has taken about 5 minutes so far. The skydiver is not immediately visible, although you can see some white areas on the northern shore and on the far eastern edge of the southern shoreline.

Describe what you would do in this situation

______

292

4) Describe your plan for when your turn back to Moruya if you do not find the skydiver.

______

5) Describe your plan for approach and landing at Moruya.

______

6) On a scale of 1 to 100, 1 being almost zero risk (sitting at home in front of the television) and 100 being extreme risk (almost certain chance of death), how would you rate this scenario?

______/100

7) I would describe this flight as containing the following amount of risk (please circle) low medium high

8) What would best describe your decision making for this scenario? (please circle)

1. Recall past examples from your experience

2. Consider the pros and cons of the alternatives

3. Recall and follow a specific rule

4. Recall past examples that you’ve heard or read about

293

5. Base your decision on a model in which decisions should be made

6. Know immediately whether it was the right decision

Proceed to the next scenario.

294

USA to Australia Scenario

You are asked to ferry a new Cirrus SR‐22 from the USA to Australia. The flight is usually undertaken by the most experienced ferry pilot, but he is sick with the flu, and he cannot come along for this flight.

You’ve had a fair amount of time on type, and are confidant that you can undertake the flight

Your calculations show that the longest leg over water will require nearly your entire maximum fuel payload. You expect to land with less than 5 minutes of primary fuel available. This is exclusive of your 45 minutes of reserves.

There are no islands with runways as an alternate between the two islands. The single alternate available is to the north of the destination island, but the island airport is very rudimentary, with a dirt runway, no re‐fuelling capabilities and no nav‐aids.

The company that is undertaking this ferry service is also ferrying a Cessna Caravan at the same time. This means that any additional weight can be carried on that aircraft. This may make a difference to your fuel burn, but you will not know the exact effect of this change until you reach the USA and fly the aircraft.

1) What would you do in this situation?

______

295

2) Why did you choose this option?

______

3) On a scale of 1 to 100, 1 being almost zero risk (sitting at home in front of the television) and 100 being extreme risk (almost certain chance of death), how would you rate this scenario?

______/100

4) I would describe this flight as containing the following amount of risk (please circle) low medium high

5) What would best describe your decision making for this scenario? (please circle)

1. Recall past examples from your experience

2. Consider the pros and cons of the alternatives

3. Recall and follow a specific rule

4. Recall past examples that you’ve heard or read about

5. Base your decision on a model in which decisions should be made

6. Know immediately whether it was the right decision

Proceed to the next scenario.

296

Mount Hotham Scenario

You flew to Mount Hotham in the middle of summer to watch a mountain biking competition being run at the nearby resort.

Your best friend is competing in the competition, and you are likely to give him a lift back to Sydney, as his team is staying two extra nights to train, whilst he needs to get to Sydney airport to fly to New Zealand to compete in another competition.

The day that you are due to leave is 30 degrees at Mt. Hotham Airport, which is at 4,200 feet AMSL. You, your friend, and full fuel will take you very close to MTOW. You can afford to fly with approximately half of a full fuel load, and make it to Bankstown.

The team’s boss comes to you with an offer. He will give you $5,000 to fly his ‘package’ back to Sydney. When questioned, he won’t divulge its contents, and he says it won’t be at the airport until about 10 minutes before your ETD. He says that it’s extremely important that it’s in Sydney as soon as possible, and that driving it will be too slow.

When the ‘package’ arrives, it is actually three items. They are packed in large metal cases that are padlocked, and are very heavy and bulky. It will be a struggle to fit them into the aircraft. It is also impossible to weigh the items now, as there is not sufficient time to take them into the resort to weigh them. You estimate that they weigh about 250kg, which puts you very close to MTOW with half fuel on board.

1) What would you do in this situation?

______

297

2) Why did you choose this option?

______

3) On a scale of 1 to 100, 1 being almost zero risk (sitting at home in front of the television) and 100 being extreme risk (almost certain chance of death), how would you rate this scenario?

______/100

4) I would describe this flight as containing the following amount of risk (please circle) low medium high

5) What would best describe your decision making for this scenario? (please circle)

1. Recall past examples from your experience

2. Consider the pros and cons of the alternatives

3. Recall and follow a specific rule

4. Recall past examples that you’ve heard or read about

5. Base your decision on a model in which decisions should be made

6. Know immediately whether it was the right decision

Thank you – You have completed the survey. Please make sure you have answered all the applicable questions, and return the survey to the researcher.

298

Appendix 7. Experiment Two B Scenarios

Ferry to LAME Scenario

You have been asked by a friend to ferry their aircraft to the nearest LAME for its 100 hourly service. You fly together relatively frequently, and he allows you the use of his plane at the cost of the fuel used because of your friendship.

The airport that you are to ferry the aircraft to is situated at sea level, but it is bordered by 2000 foot hills. The cloud level is estimated at 3000 feet AMSL at the airport at present. The prevailing wind is from the south, and is currently at about 10 to 15 knots. The runway is 34/16.

The terrain between your home airport and the LAME’s port is generally at 2000 feet AMSL, and is approximately 1.5 hours flying time from your home airport. There are very few airports that are close to the route you will fly. The best available is a well maintained private strip, however you have not used this strip before, as the owner reportedly has a habit of leaving machinery on it to discourage use.

There is a storm front to the south of the airport, and the forecast is for the cloud base to drop throughout the day. This has been reported by sites to the south of the airport and is due to reach the destination about 15 minutes after your ETA. You plan is to drop the plane at the LAME, and quickly depart in another aircraft that the LAME has just completed a 100 hourly check on.

299

1) What would you do in this situation?

______

2) Why did you choose this option?

______

3) On a scale of 1 to 100, 1 being almost zero risk (sitting at home in front of the television) and 100 being extreme risk (almost certain chance of death), how would you rate this scenario?

______/100

4) I would describe this flight as containing the following amount of risk (please circle) low medium high

5) What would best describe your decision making for this scenario? (please circle)

1. Recall past examples from your experience

2. Consider the pros and cons of the alternatives

3. Recall and follow a specific rule

4. Recall past examples that you’ve heard or read about

5. Base your decision on a model in which decisions should be made

6. Know immediately whether it was the right decision

Proceed to the next scenario.

300

Hunter Valley Scenario

You offer to take your best friend flying for his 30th birthday. You know that he loves the Hunter valley, and seeing as though you both live at Camden, it makes sense to fly to the Hunter for the weekend. Your partners will drive and meet you both at Cessnock airport.

The aircraft that you are going to fly is familiar to you, as you fly it most weekends. You also fly to Cessnock quite regularly as part of your job as a flight instructor.

The weather for the weekend is forecast to be fine with little wind. As a backup plan, you have organised to have your partner drive both couples in case the weather turns bad or any other unforseen problems arise.

1) What would you do in this situation?

______

2) Why did you choose this option?

______p.t.o

301

3) On a scale of 1 to 100, 1 being almost zero risk (sitting at home in front of the television) and 100 being extreme risk (almost certain chance of death), how would you rate this scenario?

______/100

4) I would describe this flight as containing the following amount of risk (please circle) low medium high

5) What would best describe your decision making for this scenario? (please circle)

1. Recall past examples from your experience

2. Consider the pros and cons of the alternatives

3. Recall and follow a specific rule

4. Recall past examples that you’ve heard or read about

5. Base your decision on a model in which decisions should be made

6. Know immediately whether it was the right decision

Proceed to the next scenario.

302

Moruya Scenario

Your brother has recently started a job at Moruya, in which he is a flight instructor and part‐time drop pilot for the local skydiving operation. You have gone down to Moruya to visit him, and to hopefully go for a flight, as it is one of your favourite areas in the state to fly.

A plane has returned from a local flight early, and your brother tells you that you can have the plane for a quick flight. As a result of the time pressure, you don’t re‐ fuel the plane. The aircraft has about 1 hour of fuel remaining before the flight, and you decide to fly for about half an hour.

Your brother’s flight school has obtained a special exemption on their AOC to carry only 5 minutes of fixed reserve fuel.

You are currently on the downwind leg of a circuit for runway 18 at Moruya aerodrome. The owner of the flight school radios you with a request for help. He wants you to fly up to Batemans Bay to search for a parachutist that has landed away from the designated landing area for the parachuting operation, which is on the shores of Batemans Bay. The owner tells you that skydiver should be easy to find, as his parachute is bright white, and he has been trained to make the parachute as visible as possible in the even of missing the landing area.

You are currently at 1000 feet AGL. Batemans Bay is directly in front of you at the present time. The bay is about 10‐12 nautical miles from your position, and will require approximately 5‐6 minutes of flight time to reach at 100 knots.

At 100 knots (2400 RPM), the aircraft has approximately 18 minutes of primary fuel available, with the 5 minutes of fixed reserves also on board.

The owner has asked that once you have found the parachutist to report it to him on the radio, fly a full orbit, and then return to Moruya. p.t.o.

303

1) What would you do in this situation?

______

2) Why did you choose this option?

______

(if you chose not to go, proceed to Question 6 in this scenario)

3) Once overhead Batemans Bay, it becomes apparent that finding the skydiver is unlikely to be easy, as there is a significant amount of area to search. The bay has a north and south coastline, with Batemans Bay township on the south‐western side of the bay. The northern coastline is approximately 1 to 2 nautical miles to the north of the southern coastline. Your flight to the area has taken about 5 minutes so far. The skydiver is not immediately visible, although you can see some white areas on the northern shore and on the far eastern edge of the southern shoreline.

Describe what you would do in this situation

4) Describe your plan for when your turn back to Moruya if you do not find the skydiver.

______

304

5) Describe your plan for approach and landing at Moruya.

______4

6) On a scale of 1 to 100, 1 being almost zero risk (sitting at home in front of the television) and 100 being extreme risk (almost certain chance of death), how would you rate this scenario?

______/100

7) I would describe this flight as containing the following amount of risk (please circle) low medium high

8) What would best describe your decision making for this scenario? (please circle)

1. Recall past examples from your experience

2. Consider the pros and cons of the alternatives

3. Recall and follow a specific rule

4. Recall past examples that you’ve heard or read about

5. Base your decision on a model in which decisions should be made

6. Know immediately whether it was the right decision

Proceed to the next scenario.

305

Appendix 8. Flight Briefing

Flight Briefing

Background: You are a private pilot on your day off from work. You've decided to spend some time at Moruya airport as your brother works as an instructor at that aerodrome. You are hoping to get the chance to fly a few circuits today, as your work schedule has been busy and you haven't flown in a couple of weeks. After a short time at the aerodrome an aircraft becomes available for an hour, as the pilot decided to return early from a cross‐country trip.

Flight: You are flying circuits at Moruya aerodrome on runway 18 in a Cessna

172. On the downwind leg of your final circuit for this flight, the company that owns the aircraft you are flying calls you and asks you to perform an abnormal operation for them.

They have just received a call from the local skydiving operator who was informed that one of their skydivers has landed outside of the normal drop zone. The caller gave a good description of the drop location, and from this, it is estimated that the landing site of the skydiver is somewhere on the shore of the Clyde river, east of the bridge. Based on previous experience this is between 10 and 12 NM from the aerodrome on a heading of approximately North. You have been asked by your company to perform a preliminary search for the skydiver as your aircraft is the closest to the landing site.

306

From all indications, the skydiver is uninjured, but the operator needs to find the man so that a car can be sent to get him back to the skydiving centre. The aircraft you are flying has been previously used on a cross‐country exercise and has approximately 18 minutes of fuel remaining with an extra 5 minutes of reserves.

This is based on fuel burn for a 110 knot cruise at 2500 RPM for this specific aircraft. The flight to the site will take between 5.5 and 6.5 minutes at 110 knots.

The only aerodrome within 15 minutes flight time of your current position is the aerodrome you are currently flying at.

Your company informs you that the site should be visible, as the skydiver's parachute is white and is quite large ‐ the skydiver knows to keep his chute open if he has missed the landing zone so as to be visible. Once you have identified the site by sighting the parachute, your company has asked that you inform them, fly an orbit around the site and return to the aerodrome.

There are no specific clearances necessary to fly to the site. Your company also informs you that this is not an emergency, and you are still under normal flight rules for this flight. Your company makes it clear that they are only asking for assistance, for which they would be most appreciative.

307

Appendix 9. Interview Questions

Sector 1. Before you started the flight, can you explain to me what you were

thinking, or what decisions you thought needed to be made?

Sector 2. En‐route to the site, what were you thinking about?

Sector 3. When you were over or approaching the target area, what was going

through your mind with regards to the outcomes or purposes of the flight?

Sector 4. What factors did you consider in you decision to turn back to the

airport?

Sector 5. On the way back to Moruya, were there any other things going

through your mind?

308