Premium
Decision heuristic or preference? Attribute non‐attendance in discrete choice problems
Author(s) -
Heidenreich Sebastian,
Watson Verity,
Ryan Mandy,
Phimister Euan
Publication year - 2018
Publication title -
health economics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.55
H-Index - 109
eISSN - 1099-1050
pISSN - 1057-9230
DOI - 10.1002/hec.3524
Subject(s) - preference , heuristic , attendance , latent class model , context (archaeology) , welfare , class (philosophy) , psychology , discrete choice , revealed preference , actuarial science , social psychology , economics , econometrics , computer science , microeconomics , artificial intelligence , machine learning , economic growth , market economy , paleontology , biology
This paper investigates if respondents' choice to not consider all characteristics of a multiattribute health service may represent preferences. Over the last decade, an increasing number of studies account for attribute non‐attendance (ANA) when using discrete choice experiments to elicit individuals' preferences. Most studies assume such behaviour is a heuristic and therefore uninformative. This assumption may result in misleading welfare estimates if ANA reflects preferences. This is the first paper to assess if ANA is a heuristic or genuine preference without relying on respondents' self‐stated motivation and the first study to explore this question within a health context. Based on findings from cognitive psychology, we expect that familiar respondents are less likely to use a decision heuristic to simplify choices than unfamiliar respondents. We employ a latent class model of discrete choice experiment data concerned with National Health Service managers' preferences for support services that assist with performance concerns. We present quantitative and qualitative evidence that in our study ANA mostly represents preferences. We also show that wrong assumptions about ANA result in inadequate welfare measures that can result in suboptimal policy advice. Future research should proceed with caution when assuming that ANA is a heuristic.