Preference inversion in discrete choice experiments

Preference inversion - where preference estimates do not align with the hierarchical order of attribute levels-remains a concern in health valuation studies. To address this, we developed a novel visual aid, the emoji scale, comprising five Apple Unicode emoji characters to represent five levels of a health-related quality-of-life measure. Using a case study of the Recovering Quality of Life - Utility index instrument, we explored whether the emoji scale could reduce preference inversion and improve respondent experience in discrete choice experiments (DCEs). We conducted a between-respondent comparison involving 2024 Australian adults, randomly allocated to one of four arms:(1) standard text-only choice tasks, (2) emoji & text (combined), (3) emoji-only (no text), and (4) text with purple colour-coding. We estimated value sets derived from each format and assessed differences in perceived task difficulty, engagement, and preference inversion using t-tests and model comparisons. Results showed that utility decrements from the emoji & text were closely aligned with those from the standard text-only format. Importantly, the emoji & text reduced preference inversion and was associated with lower reported task difficulty and higher engagement compared to other formats. These findings suggest that pairing emojis with text may enhance respondent understanding and data quality in health valuation tasks, without compromising the validity of the resulting value sets. While further work is necessary to fully understand the legitimacy of the emoji scale as an effective visual aid for conveying attribute level ordering, the emoji & text format offers a promising approach to improving DCEs

Digital Communication and Language · Language, Discourse, Communication Strategies · Multisensory perception and integration

• Dutch Tariff for the Five-Level Version of EQ-5D

  Open Access•Matthijs Versteegh, Matthijs M Versteegh et al.•Value in Health•2016

  Incluir item

• The estimation of a preference-based measure of health from the SF-36

  Open Access•John Brazier, Jennifer Roberts et al.•Journal of Health Economics•2002

  Incluir item

• A Systematic Review of Emoji

  Open Access•Qiyu Bai, QI Dan et al.•Frontiers in Psychology•2019

  Incluir item

• Internet, Phone, Mail, and Mixed‐Mode Surveys

  Don A Dillman, Jolene D Smyth et al.•Internet, Phone, Mail, and…•2014

  Incluir item

• Do health preferences contradict ordering of EQ-5D labels

  Open Access•Brandon M Craig, A Simon Pickard et al.•Quality of Life Research•2014

  Incluir item

• Further evidence on EQ-5D-5L preference inversion

  Open Access•Brandon M Craig, A MONTEIRO et al.•Quality of Life Research•2017

  Incluir item

• Valuing the SF-6Dv2 Classification System in the United Kingdom Using a Discrete-choice Experiment With Duration

  Open Access•B Mulhern, Brendan J Mulhern et al.•Medical Care•2020

  Incluir item

• A Time Trade-off-derived Value Set of the EQ-5D-5L for Canada

  Open Access•Feng Xie, Eleanor Pullenayegum et al.•Medical Care•2016

  Incluir item