Sequential versus one-shot decision making: Statistical guidelines and behavior in a medical watchful waiting dilemma

Many important real world risky decisions, such as deciding the size and timing of investments in a market or product, determining how to best care for a chronically ill patient, or deciding how best to set interest rates, are actually a series of risky decisions that are made over time. This type of decision is typically made by trained experts, using both statistical data, as well as memories of previous experiences. This very important class of decisions can be described as dynamic because the chain of decisions that will be made in the future changes as a result of each current decision. This dissertation looks at a particular type of dynamic decision, perhaps best described as a sequential watchful waiting decision, where the only real option is whether to perform a specific action or wait and decide again whether to perform it. The specific decision described in this work is a health care decision, about when to operate on an Abdominal Aortic Aneurysm, or AAA. Participants were given complete information about the statistically correct way to make the decision (when to wait and when to go to surgery based on AAA size) and highly significant differences were found (p < .0001) depending on whether the decisions were made as part of a watchful waiting sequence (the scenario was described to participants as following one patient over time) or as a series of one-shot decisions (the same AAA sizes were shown, only out of order, and the scenario was described to participants as seeing a number of patients in clinic). The level of riskiness of the risky wait or go decisions being balanced and the participants' comfort level with making decisions based on statistical guidelines were also found to be significant predictors, however, participants uniformly made far less statistically optimal decisions in the sequential conditions.