Known Unknowns in Judgment and Choice
- Author(s): Walters, Daniel
- Advisor(s): Fox, Craig R
- et al.
This dissertation investigates how people make inferences about missing information. Whereas most prior literature focuses on how people process known information, I show that the extent to which people make inferences about missing information impacts judgments and choices. Specifically, I investigate how (1) awareness of known unknowns affects overconfidence in judgment in Chapter 1, (2) beliefs about the knowability of unknowns impacts investment strategies in Chapter 2, and (3) inferences about forgotten unknowns influence choices from memory in Chapter 3.
Chapter 1 investigates how overconfidence can stem from neglecting to consider missing information. Most prior research has attributed overconfidence to people focusing disproportionately on evidence favoring the chosen hypothesis relative to its alternatives. In this chapter, I find that neglecting unknown evidence independently contributes to overconfidence. In a first study, respondents answered questions such as, “Which of these fast food items has more calories, a Subway sandwich, or a McDonald's cheeseburger? / How confident are you?” Using a process tracing technique, I found that participants who considered more missing evidence were less overconfident than those that thought about more known evidence. Meanwhile, participants who considered more unknown information answered the same number of questions correctly, resulting in better calibration. In two additional studies, I prompted participants to list unknowns before assessing confidence in their judgments. This “consider the unknowns” technique reduced overconfidence substantially, and was more effective than the de-biasing technique most often prescribed in the research literature (“consider the alternative”). Importantly, considering the unknowns was selective in its impact: it reduced confidence only in domains where participants were overconfident, but did not affect confidence in domains where participants were well-calibrated or under-confident.
Chapter 2 investigates how inferences about the knowability of missing information impacts investment choices. Recent research has found that people intuitively distinguish aleatory uncertainty that is inherently random or stochastic (e.g. What is the probability that a fair coin will land heads?) from epistemic uncertainty that is attributed to missing knowledge or information (e.g., Which company had a larger market capitalization at the end of 2015, Google or Apple?). In a series of surveys and experiments involving laypeople, experienced investors, and financial advisors, I found that investors who viewed stock market uncertainty as more epistemic/knowable searched for more stock information and were willing to pay more for financial advice, whereas investors who viewed stock market uncertainty as more aleatory/random diversify more. Similarly, when investors were primed to think about epistemic uncertainty they are more willing to pay for stock information whereas when they were primed to think about aleatory uncertainty they diversified more when completing an incentive-compatible investment task. Taken together, these studies point to the critical role that the perception of the nature of uncertainty can have on an investor’s judgments and choices.
Chapter 3 investigates how inferences about forgotten product features impact consumer choices from memory. Consumers must often make product judgments and choices based on information contained in memory. For instance, a consumer may learn information about the Apple Watch, then choose among smart watches later. The consumer may recall some features clearly, such as the Apple Watch’s ability to respond to text messages, but also realize that they have forgotten other features entirely. Whereas most prior research has focused on how consumers evaluate remembered information, this research examines how product judgments and choices are also guided by inferences consumers make about forgotten information. First, I found that consumers overestimated how closely the quality of forgotten features resembled remembered features. As a result, consumers tended to choose a product from memory over an equivalent product that is fully described when the remembered features are more positive and choose a product that is fully described over an equivalent product from memory when the remembered features are more negative.
Taken together, this work expands our understanding of how people make inferences about missing evidence, the nature of uncertainty, and forgotten information. Importantly, this work shows how these inferences impact critical outcomes in the field, such as overconfidence, investment strategy, and product choice.