Influence of confirmation bias in aviation lost situations

Gilbey and Hill (2012¹) conducted a study to examine the confirmation bias empirically. The main purpose of this study was to find out how confirmation bias might influence the pilots when they try to reorient themselves. Navigational and flying experiences and the effectiveness of an easy debiasing interference on the task were also examined. The term “confirmation bias” is most directly linked to failure in practical human factors framework and has also been recognized as a possible error of interpretation and judgment responsible for a number errors, incidents and accidents in aviation navigation (Gilbey & Hill, 2012).
Only orienteers made the right choice significantly more than by chance. All other groups completed the task either at or worse than probable by chance. The efforts to enhance the rate of disconfirming choices were mainly ineffective.

The table shows the extent to which disconfirmation bias influence the participants to reorient.
Study	Participants	Percentage of disconfirming choices of the time	Percentage of all participants who made disconfirming choice	Mean of disconfirming choice	Significance compared to "by chance"
1	Pilots (No)	18	42.4	0.53	lower
2	Psych students (Basic lecture)	19	52.8	0.56	much lower
3	Orienteers (No)	67	76	2.0	much higher
4	Pilots (Lecture)	26	44	0.78	lower
5	Psych students (Lecture)	22	55	0.69	much lower

In short, findings imply that when lost, both pilots and psychology students, but not orienteers, take up a confirmatory approach to determine their location, which might bring pilots who are already lost, an even greater risk.

Methods

Research approach

This is an exploratory study to find out how confirmation bias might influence the pilots when they try to reorient themselves.

Sample

Study 1: 66 student pilots with experiences of flying solo, separated in four groups of flown hours of 20, 55, 100 and 160.
Study 2: 36 second year cognitive psychology students after attending a simple lecture on confirmation bias.
Study 3: 21 orienteers
Study 4: 18 student pilots with experience of flying solo and flying around 55 hours after attending detailed lecture on confirmation bias.
Study 5: 29 second year cognitive psychology students after attending detailed lecture on confirmation bias.
Total of 170 persons.

Design

It was a field experimental study.

Variables

The dependent variables:
• the total number (between 0-3) of disconfirming choices made for all three scenarios
The independent variable:
• studies 1-5
• Experience of flying with hours of 20, 55, 100 and 160 respectively
• Cognitive psychology knowledge
• Simple or detailed lecture on confirmation bias
• Orienteering experience

Materials

Studies 1-5: The participants (student pilots, psychology students or orienteers respectively) were asked to reorient themselves as fast as possible, pretending to be lost as a passenger in motorcycle, yacht and light aircraft. Three features were given for them to make the right choice (the unique feature) in maps. The three scenarios were written in detail and in easy language.
Study 6: The findings of Studies 1-5 and Scenarios 1-3 were used to reveal the difference between them.

Procedure

In studies 1-3, the extent to which the respective participants choose confirmatory evidence instead of disconfirmatory evidence to determine their position when lost in three hypothetical scenarios was explored. The task in study was completed within 45 minutes.
In studies 4-5, an effort to increase the disconfirmatory choices for student pilots and psychology students while completing the same tasks was made by giving detailed lecture and worked example on confirmation bias 10 days before they completed the task.
In Study 6, the difference between the findings of Studies 1-5 and Scenarios 1-3 were revealed.

Data analysis

• Three pair-wise comparisons (McNemar) and its extention Cochran’s Q test were used.
• Kruskal- Wallis’ H test explored the mean total number of disconfirming choices among Studies 1–5.
• Mann–Whitney U-test, post hoc pair-wise analysis was also done.

Generalization potential

• Flight navigation and map reading experiences do not prevent from biased responding. It is also possible that some participants made their choices randomly. The detailed written scenarios may be interpreted and understood differently by each participant. These limit the external validity of the findings of this research study.
• The research results are generalizable to orienteers to certain amount.
• More experiments should be done in order to further validate the results.

References

1. GILBEY Andrew and HILL Stephen (2012). Confirmation bias in General Aviation lost procedures. Applied Cognitive Psychology. Sep/Oct2012, Vol. 26 Issue 5, p785-795. 11p. DOI: 10.1002/acp.2860.