<XML><RECORDS><RECORD><REFERENCE_TYPE>3</REFERENCE_TYPE><REFNUM>7811</REFNUM><AUTHORS><AUTHOR>Johnson,C.W.</AUTHOR></AUTHORS><YEAR>2004</YEAR><TITLE>Looking Beyond the Cockpit: Human Computer Interaction in the Causal Complex of Aviation Accidents</TITLE><PLACE_PUBLISHED>In A. Pritchett and A. Jackson (eds), Human-Computer Interaction in Aerospace, EURISCO, Toulouse. </PLACE_PUBLISHED><PUBLISHER>N/A</PUBLISHER><PAGES>17-24</PAGES><LABEL>Johnson:2004:7811</LABEL><KEYWORDS><KEYWORD>Accident analysis</KEYWORD></KEYWORDS<ABSTRACT>Aviation mishaps stem from complex combinations of human error, managerial failure, design flaws, environmental and meteorological factors. Mackie [1] uses ‘causal fields’ to denote a subjective frame of reference that guides our search for particular causes. This helps to explain why many investigators focus on ‘primary’ forms of human computer interaction. There is a tendency to focus on the interaction between aircrew and onboard systems or between air traffic managers and their ground based software. In contrast, we argue that the causal field must be extended to cover secondary and tertiary problems in human computer interaction. Secondary failures indirectly contribute to adverse events and increase the likelihood of primary forms of error’. For instance, recent mishaps have been attributed to secondary problems in the software that general pilots use to plan their flights prior to departure. Tertiary usability problems stem from the difficulty of applying computer software to understand the events leading to an accident or incident. Tertiary software is not directly involved in the adverse event but is critical to reconstruct previous interaction. For instance, a number of recent investigations have been unable to probe flight data records to accurately identify the sequence of aircrew actions leading to an accident. This creates uncertainty about the role that human ‘error’ plays in the course of an adverse event. A detailed analysis of NTSB accident and incident reports involving human computer interaction failures between 1999 and 2004 is used to illustrate our argument. </ABSTRACT></RECORD></RECORDS></XML>