Errors of Perception in Air Traffic Control
Steven T Shorrock investigated errors that occurred in the UK area and approach control in light of cognitive psychology research, and probed the possible implications for future air traffic management.1 He divided his investigations into two sections, one being from interviews with operational controllers, and the other from information reported in incident reports, after a loss of separation had occurred.
The table below, indicates the types of errors made by ATC with the errors of perception as the highest, as determined by the controller interviews, and Airprox reports.
|Error Types||Numbers (Interviews)||Numbers (Incident Reports)|
|Errors of Perception||66||45|
|Errors of Memory||42||20|
|Errors of Decision Making||31||72|
|Errors of Action and Speech||22||13|
The errors of perception as disclosed in the above table have been categorized below between internal error modes, and psychological error mechanisms, and have been further categorized under subheadings for each these two options.
The internal error modes show Hearback error, and No Detection (Visual) are the most common errors, in both the controller interviews and Airprox report investigations.
The psychological error mechanisms show distraction/preoccupation as the most common error in both the controller interviews and the Airprox report investigations.
|Internal error modes||Number (Interviews)||Number (Airprox)|
|No detection (auditory)||2||3|
|Late Auditory recognition||2||0|
|No detection (visual)||21||19|
|Late detection (visual)||8||8|
|Psychological Error Mechanisms||Number (Interviews)||Number (Airprox)|
|Perceptual discrimination failure||4||2|
- This was an explorative approach to research, completed in two sections, to determine the causes of errors of Air Traffic Controllers.
- Part one involved interviews with 28 validated uk area controllers. Each controller was validated on one or more sector. The controllers were from London Area and Terminal Control Centre (LATCC) at West Drayton, near London, The interviewees managers allowed their employee to participate in the interview.
- Part two involved accessing the Airprox reports filed from incidents involving losses of separation, also involving air traffic control errors. A sample of 48 incidents were used, ranging from 1995 to 1997.2
- The design of section one was always going to have flaws. Although interviews are a great way to get in depth information regarding errors made by ATC, in such a safety critical environment, where significant implications as a result of errors are possible, it is likely that some controllers may withhold information or adjust specific details. This reduces the integrity of the data, and the analysis, however, it is very difficult to improve this integrity.
- Section two also had limitations, considering their was limited information available to the researchers from the report. This enables assumptions to be made, which may reduce the integrity of the data.
- In section one, the controllers were the variables, as it was their insight which gave the only information available for the research. It is possible, that many controllers did not give complete information to the interviewer, for reasons such as differences in personalities, or protecting themselves, their colleagues, systems, or many other unknown reasons. It is also possible that the two interviewers perceived the controllers responses differently, or analysed their responses in different manners. Therefore, some information is likely to be more insightful than others, and it is probable assumptions were made in analysing the data.
- Section two involved the Airprox report as the variable. These reports had a fixed amount of data from when the reports were made and the researchers were unable to interview the controllers included. Therefore, when shortfalls of information occurred, assumptions may also have been made.
- Part one involved interviews in a closed off office located in the building in which the controllers worked. Each interviewee was interviewed by one of two interviewers.
- Part two involved the reports from Airprox, which were reports containing details from when a loss of separation had previously occurred.
- Part one involved interviews with 28 validated UK area controllers. Each controller was interviewed by one of two interviewers, in a separate, closed office, with the duration lasting between 30 to 45 minutes.
- The format of the interview was unconstructed, in an attempt to gain in-depth insites into controllers experiences, and due to the sensitivity of the data, and lack of formal reports. The interviews, although unconstructed, used three types of questions as bases, those being, Open questions, probing questions and closed questions. This enabled the interviewer to establish a line of enquiry, and subsequently follow it up. Every interview was recorded and transcribed, with the permission of the interviewees. The interviews were subsequently coded and analysed to identify the types of errors that occurred.
- Part two looked into formally reported controller errors which contributed to losses of separation, which occurred in ondon Area and Terminal Control Centre (LATCC), Scottish Area Control Centre (ScACC) and Manchester Area Control Centre (MACC) over the course of three years. This involved the analysis of 48 incidents. The information from the reports is the only information that was available for use, and this included factual backround information, a summary of the incident, a redescription of each error, details of the error detection and recovery TRACEr RETRO3
classifications, assumptions and notes
- There were no specific tests completed to determine the results of this research. However, the results were collected, categorized and analyzed by Mr Shorrick and his partner. It is unknown of their qualifications and eligibility to complete this form of research.
- Due to the limitations present when conducting interviews, it is obvious that assumptions were made, reducing the integrity of the data. This reduces the potential to generalize on the errors made.
- The small sample size of just 28 interviews means that the data and analysis provides limited potential for generalizing. Also, because each of the interviewees were UK area controllers, and worked for the same company, it is possible the culture of their company and/or sector affects their judgements or willingness to discuss openly their experiences.
- Having only area controllers may also affect the results, in comparison to if they had interviewed controllers from a range of sectors, or from a range of countries, including tower and approach controllers. This would have allowed the generalizations to be placed worldwide, as apposed to area controllers within the UK.
2. ( [CAA, 1996], [CAA, 1997a], [CAA, 1997b], [CAA, 1998], [UK Airprox Board, 1999a] and [UK Airprox Board, 1999b]).
3. ( [Shorrock, 2003] and [Shorrock and Kirwan, 2002]),