"Why Blame?" Journal of Political Economy, (2013), 121(6): 1205-1246 (with Mehmet Y. Gurdal and Aldo Rustichini)
Idea (not the abstract): Will Ann hold Bob responsible for an event that Ann knows Bob cannot control? Will Ann do this even if doesn't affect her payoffs? An experiment shows that Ann will do so if the event reveals how Ann would have done had Bob chosen otherwise. We argue that this behavior can be viewed as an unconscious manifestation of the informativeness principle from contract theory. A simple delegated expertise model can account for this behavior, under the assumption that people carry over their optimal behavior from the most related environment (salient perturbation).
Intuition: (1) Are you more likely to fire your financial manager if she performs worse than the S&P 500? (2) rational superstition?
Finalist for the 2014 Exeter Prize
“Social Risk and the Dimensionality of Intentions” Management Science, (2017), 121(6): 1205-1246 (with Jeffrey V. Butler)
Idea (not the abstract): People's attitudes towards risk depend on more than consequences and their associated probabilities. People treat risk differently when it is determined by the actions of another human being, rather than a natural source of risk. This is social risk. We predict that people's attitude towards social risk will depend on their perception of intent. We experimentally manipulate the scope for intentional action and find that attitudes towards social risk are influenced by this manipulation. For example, suppose Ann knows that Bob has a conflict of interest; controlling for the probability of an unfavorable outcome, we find that: (1) Ann treats the social risk from Bob as no different than a natural source of risk if Bob is unaware of the conflict, (2) if Bob is aware of the conflict, but cannot properly foresee the consequences of his actions, then Ann views the social risk from Bob's (partially) intentional action as preferable to a natural source of risk.
“Surprised by the Hot Hand Fallacy? A Truth in the Law of Small Numbers”, Econometrica (2018), 86(6): 2019–2047 (with Adam Sanjurjo)
Idea (not the abstract): If you select a flip from a finite sequence coin flips because it is preceded by several heads, then it is more likely to be a tails. Overlooking this fact is the hot hand fallacy fallacy and it invalidates the conclusions of the original hot hand fallacy study (and its replications). When this fact is accounted for, we find that the hot hand is not a cognitive illusion.
"Subjective Beliefs and Confidence When Facts are Forgotten", Journal of Risk and Uncertainty (online edition available), (2018), (with Igor Kopylov)
Abstract: We run several experiments where people bet on propositions (facts) that they cannot recall with certainty. Forgetting is induced via interference tasks and time delays (up to one year). We use betting preferences to define subjects’ revealed beliefs and their confidence in these beliefs. Forgetting makes revealed beliefs less accurate and reduces the subjective confidence metric as well. Moreover, we find a form of comparative ignorance where subjects are more ambiguity averse when they cannot recall the truth rather than never have learnt it before. In a different vein, we identify an overconfidence pattern: on average, subjects overpay for bets on propositions that they believe in, but underpay for the opposite bets. We formulate a two-signal behavioral model of forgetting that generates all of these patterns
"A Bridge from Monty Hall to the Hot Hand: Restricted Choice, Selection Bias, and Empirical Practice", Journal of Economic Perspectives, Conditionally Accepted (2018), (with Adam Sanjurjo)
Idea (not the abstract): What does Monty Hall have to do with the hot hand? The Monty Hall problem is a puzzling brain teaser who's solution many have a difficult time accepting. The hot hand selection bias is the critical statistical error found in the original hot hand fallacy study, and it is an error that many have a difficult time accepting. These two difficulties are illuminated by the principle of restricted choice, an intuitive inferential rule from the card game contract bridge. We quantify this principle as the updating factor from the odds formulation of Bayes' rule. We show how restricted choice not only simplifies calculations, but it renders the classic conditional paradoxes intuitive. Importantly, because restricted choice highlights Monty's selection rule, it provides insights into empirically-relevant selection biases.
"Laplace's Theories of Cognitive Illusions, Heuristics, and Biases," Statistical Science, forthcoming (2018), (with Andrew Gelman)
Download Options: SSRN
Abstract: In his book from the early 1800s, Essai Philosophique sur les Probabilités, the mathematician Pierre-Simon de Laplace anticipated many ideas developed in the 1970s in cognitive psychology and behavioral economics, explaining human tendencies to deviate from norms of rationality in the presence of probability and uncertainty. A look at Laplace's theories and reasoning is striking, both in how modern they seem and in how much progress he made without the benefit of systematic experimentation. We argue that this work points to these theories being more fundamental and less contingent on recent experimental findings than we might have thought.
"Penney's Game Odds From No-Arbitrage" (2019)
Download Options: OSF
Penney's game is a two player zero-sum game in which each player chooses a three-flip pattern of heads and tails and the winner is the player whose pattern occurs first in repeated tosses of a fair coin. Because the players choose sequentially, the second mover has the advantage. In fact, for any three-flip pattern, there is another three-flip pattern that is strictly more likely to occur first. This paper provides a novel no-arbitrage argument that generates the winning odds corresponding to any pair of distinct patterns. The resulting odds formula is equivalent to that generated by Conway's ``leading number'' algorithm. The accompanying betting odds intuition adds insight into why Conway's algorithm works. The proof is simple and easy to generalize to games involving more than two outcomes, unequal probabilities, and competing patterns of various length. Additional results on the expected duration of Penney's game are presented. Code implementing and cross-validating the algorithms is included.
"Tra i Leoni: Revealing the Preferences Behind a Superstition" (2019), (with Giovanna Invernizzi, Tommaso Coen, Martin Dufwenberg, and Luiz Edgard R. Oliveira)
Download Options: OSF
Abstract: We investigate a superstition for which adherence is nearly universal. Using a combination of field interventions and a lab-style value elicitation, we measure the strength of peoples' underlying preferences, and to what extent their behavior is driven by social conformity rather than the superstition itself. Our findings indicate that both mechanisms influence behavior. While a substantial number of people are willing to incur a relatively high individual cost in order to adhere to the superstition, for many, adherence is contingent on the the behavior of others. Our findings suggest that it is the conforming nature of the majority that sustains the false beliefs of the minority.
"A Visible (Hot) Hand? Expert Players Bet on the Hot Hand and Win" (2017), (with Adam Sanjurjo)
Idea (not the abstract): While the hot hand can no longer be considered an illusion (see above), the idea that players misperceive the hot hand in basketball continues to be supported by the results from an incentivized betting study from Gilovich, Vallone, and Tversky (1985). In particular, players who believe in the hot hand were found to be unable to detect it (or any other exploitable pattern) in shooting performance. We find the analysis in this study to be severely underpowered and its measure of bettors' ability to predict to be misinterpreted. Upon obtaining the raw data and performing a more powered analysis, in contrast with the original results, we find that experts are successful at predicting shot outcomes, and that the effect sizes are considerable.
"Is the Belief in the Hot Hand a fallacy in the NBA Three Point Shootout?" (2015), (with Adam Sanjurjo)
Abstract: A prominent exhibit of the hot hand fallacy, which can be considered a quasi-replication of Gilovich, Vallone, and Tversky (1985), is the work of Koehler and Conley (2003). Koehler and Conley find no evidence of hot hand shooting in four years (1994-1997) of NBA Three-Point Contest data, despite the well-known beliefs of players, coaches, and fans alike. The NBA's Three-Point Contest has been referred to as an ideal setting to test for the hot hand (Thaler and Sunstein 2008), as it involves elite professional shooters with large incentives, yet shooting in a fairly controlled environment. We improve on the study of Koehler and Conley by instead collecting 31 years of NBA Three-Point Contest television broadcast data (1986-2017), and applying a statistical approach that: (i) is more powered, (ii) uses an improved set of statistical measures, and (iii) corrects for a substantial downward bias that is present in previous estimates of the hot hand effect. In contrast with the original results, but consistent with recent evidence found relating to the classic work Gilovich, Vallone, and Tversky (1985), we find considerable evidence of hot hand shooting in the NBA Three-Point Contest.
"A Cold Shower for the Hot Hand Fallacy" (2014), (with Adam Sanjurjo)
Idea (not the abstract): The Hot Hand Fallacy, the mistaken belief that good outcomes have a tendency to cluster has long been considered a fallacy with important economic consequences. We develop a novel empirical strategy to correct for flaws in previous studies and perform a critical test of the hot hand fallacy in its canonical domain, basketball shooting. We design a field experiment, which improves on earlier fields studies, and collect data from all extant controlled shooting studies. Our analysis reveals that the belief in the hot hand is not a fallacy: hot hand shooting exists across all extant controlled shooting studies and expert decision makers can predict which shooters have a tendency to become hot.
"How Experience Confirms the Gambler's Fallacy when Sample Size is Neglected" (2018), (with Adam Sanjurjo)
Download Options: OSF
Abstract: The Gambler's Fallacy is the mistaken belief that random sequences have a systematic tendency towards reversal, i.e. that streaks of similar outcomes are more likely to end than continue. Despite broad empirical support for gambler´s fallacy beliefs, there exists little formal explanation of why such beliefs persist. We present a simple model in which an individual formulates his beliefs about the probability of success given recent success via repeated exposure to random sequences. For each sequence he focuses on the proportion of success given recent success and then updates his beliefs, but (partially) neglects sample size. This results in probability beliefs which, in the limit, are smaller than the true (conditional) probability, i.e. gambler's fallacy beliefs. We discuss the model's novel testable predictions.
"The Weight of Personal Experience: an Experimental Measurement" (2012), (with Zacharias Maniadis)
Idea (not the abstract): We present an experiment to address the question of whether a piece of information is more influential if it comes from experience, rather than from another source. We employ a novel experimental design which controls for the value of information and other potentially important confounding factors present in related studies. Overall, our results show that an event that is personally experienced has a stronger influence on subsequent behavior than an observed event with equally valuable information content. Importantly, in early rounds when information is more valuable from a rational viewpoint, this overweighting of personal experience is not evident.
"Momentum isn’t magic – vindicating the hot hand with the mathematics of streaks" The Conversation (2017), (with Adam Sanjurjo)
This is an explanation of our hot hand fallacy work intended for the general reader.