Using eye-tracking methods to investigate the framing of risk attributes in discrete choice experiments

Purpose: To understand if, and how, the framing of risk in a discrete choice experiment (DCE) for a national breast screening programme (NBSP) affects respondents’ decision-making heuristics and strategies. Method: A DCE was designed and piloted (n=15) to elicit the preferences of a purposive sample of female members of the public (recruited by in local cafes and from the internet) for a NBSP described by two risk attributes (probability of detecting a cancer and risk of unnecessary treatment) and an out-of-pocket cost attribute. Two survey versions were created with risk attributes framed as: (A) a percentage or (B) a percentage and icon array. Women completed the DCE in a university laboratory whilst their eye movements were tracked and monitored. After completing the last choice question, participants answered a debriefing survey comprising questions about: personal characteristics; choices made; self-reported attribute non-attendance (ANA). Respondent’s eye-movements were tracked throughout the experiment by recording a series of co-ordinates 1,000 times a second and pupil size. Eye-tracking data were analysed in terms of: direction of movement (saccades); total visual attention (dwell time) to pre-defined areas of interest; and change in pupil size (dilation).Result: In total, 40 women (20 randomised each to survey versions A and B) completed the DCE with eye-tracking. Respondents gave more visual attention, suggesting information processing, to attributes when risk was communicated with an icon array compared to a percentage only; however, these differences were not significant. All respondents made more horizontal (left-right) eye movements than vertical (up-down), although respondents who received the icon array version exhibited significantly more vertical eye-movements (49% v 44% of saccades). Average (mean) pupil size was smaller for respondents receiving the icon array version. The eye-tracking data confirmed the self-reported ANA to the risk attributes, with those reporting not attending the risk of unnecessary follow-up attribute having 25% lower dwell times for this attribute and those reporting not attending the probability of detecting a cancer attribute having a 40% shorter dwell time for this attribute. Conclusion: This study demonstrates that eye-tracking methods can be used to understand attention and attendance to information presented in a DCE, and how this compares to self-reported measures. The results suggest that risk attribute framing can have an impact on the strategies and heuristics of respondents.