Studies of participant attention allocation in stated preference choice experiment (CE) studies have found mixed evidence that participants employ attribute non-attendance (ANA) as a decision heuristic or as a part of a fully rational search strategy. Many studies find that correcting for ANA has considerable effects on willingness-to-pay (WTP) estimates, but the right method of correction depends on the reasons for ANA. We conduct an experiment that, coupled with results from a previous survey, give us a uniquely suitable dataset to settle the heuristic-vs.-rationality debate. Our experiment presents subjects with a standard choice experiment over alternative health risk reduction strategies, with the addition of masks covering information about the alternatives. The masks can be removed by “paying” randomly drawn time costs, giving us an exogenous source of variation in ANA. Results from a “vanilla” version of this survey without masks allow us to identify the causal effects of time costs on WTP estimates. Additionally, we develop a dynamic search model for our CE context and compare its predictions with those of the directed cognition model outlined in Gabaix et al. (2006).