Abstract

Using lab data on both choices and eye-movements we exam the relationship between randomness in choice and attention. We bring in 50 subjects and have each make 180 choices, involving repeated choices from the same choice sets, while tracking their eye movements. Our approach allows us to consider attention as a multi-dimensional object and link different aspects of attention to distinct patterns in choice. We show that although the monotonicity condition that underlies random utility models is frequently violated, the monotonicity condition on attention sets considered by Cattaneo et al., 2020 is satisfies by almost all observations. Despite this, attention explains at most around a third of the randomness in choice. Although randomness in choice is much larger at the aggregate compared to the individual level, attention explains randomness in choice to the same degree in both. In ongoing work we conduct revealed preference tests of both random utility and random attention models.