Research in Progress

“Preparing for Disaster, One Decision at a Time: The effect of broad bracketing on non-clustered risky choices” (with Michael Hand, Howard Kunreuther, and Rick Larrick)

People tend to frame decisions narrowly, which leads them to underappreciate the cumulative nature of probabilities. This is problematic because we make many one-period decisions repeatedly, like purchasing insurance and making investments, and they thus have implications for cumulative gains and losses. Past work has shown that presenting cumulative probabilities, or “broad bracketing,” can help people make better decisions when decisions were clustered such that participants only had to make one decision, but many real-life decisions cannot be clustered this way.  Furthermore, this research primarily looked at cases of moderate probabilities. We extend this line of work by examining the effectiveness of broad bracketing for rare, catastrophic events, when individuals must make the decision to protect themselves repeatedly (each time period). In particular, we reframe the likelihood of a random negative event occurring by extending the relevant time horizon for a single period to multiple periods. In two online experiments, we present participants with 15 rounds of a binary choice between losing a certain small amount for sure and a risky option that involves losing a large amount with a small probability—the negative event. We vary whether participants learn the one-round probability of the negative event (narrow bracketing) or the 30-round probability of the negative event happening at least one time with the risky option (broad bracketing). We conduct this comparison in an abstract context involving the choice between a safe option and a gamble (Experiment 1) and also the more externally valid context of purchasing flood insurance (Experiment 2). Participants are incentivized with real, high-stakes monetary payoffs. In both experiments, we find that broad bracketing significantly reduces interest in choosing the risky option in a way that persists at a similar level for all 15 rounds of the two experiments (see Figures 1 & 2). We also find that a significant portion of the population exhibits a negative recency effect (“gambler’s fallacy”) after experience a negative event: Immediately after the occurrence of a negative event, many people who had chosen the safe option in the round that the negative event occurred, chose the risky option in the round immediately afterwards. This behavior reflects the belief that it is less likely for two rare negative events to happen in a row than would be implied by a random process.  However, the effectiveness the broad bracket has over the narrow bracket in encouraging people to select the safe option is robust to these experience effects.


“When to promote and when play down the self? Revealing negative information to create a positive impression” (with Andras Molnar and Silvia Saccardo)

People tend to be judged on two dimensions, warmth & competence, and research suggests that when given negative information about one trait (e.g. competence), people infer positive information about the other trait (warmth).  We suggest people use this phenomenon to their advantage, engaging in a risky impression management strategy: revealing negative information on one dimension to enhance the impression they create about the other. Our results suggest that whether it is better to self-deprecate versus self-promote depends on the relative status of interactants, whether they are competing for resources, and the presence or absence of feelings of threat and envy.

“Fourth Party Punishment: When do people punish bystanders?” (with Lauren Kaufmann)

This project investigates the influence of higher-order punishment that we term “Fourth Party Punishment” on norm enforcement. Fourth Party Punishment (FPP) refers to the propensity to punish individuals who themselves have failed to punish a norm violator. We argue that FPP, a higher-order sanction, presents one solution to the second-order public good problem. In the behavioral economics tradition, we empirically test the hypothesis that in some cases, individuals may be willing to punish others because they failed to act when they were bystanders to a transgression. Through two experiments, we show suggestive evidence that individuals are willing to punish bystanders who failed to punish a norm violator, whether that punishment takes the form of an economic sanction (Study 1) or a social sanction (Study 2).