Jason Dana

We demonstrate a widespread pessimism about cost of living trends. People are badly miscalibrated about the real costs of many things they buy. They think these costs have risen, and will continue to rise, while most have fallen. While there are likely many reasons for this pessimism, we demonstrate that many people hold a "depletionist" mindset - whereby more people necessarily means less stuff to go around. Depletionism tends to overlook that people are not just mouths to feed, they are arms to work and, perhaps more importantly, minds to think that can come up with solutions to scarcity problems. Our subjects show an "efficiency neglect" in that they fail to incorporate even their own beliefs about the contributions of technological progress to reducing prices.

A consumer is a role played by a person in voluntary economic exchange. That role is typically played within the context of complex global markets that organize the production and distribution of scarce resources of every sort. Understanding consumer psychology thus entails understanding how consumers think about markets and the economy. We synthesize what is known on an under-researched topic: how lay people think about economics. We explain why lay reasoning diverges from economic science, ranging from how a reliance on only the things one sees is deceptive in the marketplace, to how our ancestral past was devoid of almost any of the elements of modern markets. We characterize lay thinking across economic areas and how understanding it could inform research in consumer psychology.

We use a classic paradigm from social psychology on attributional ambiguity (avoidance of the handicapped) and apply it to charitable giving. We find that people give less in real charitable decisions when there is attributional ambiguity. We don't find much evidence, however, that people are using the ambiguity as an excuse or self-justification. Rather, normative pressure seems to drive some giving, and norms are relaxed when there is attributional ambiguity because behaviors no longer have clear signal values. In other words, attributional ambiguity lowers donations because it relieves internalized social pressure to give.

Provides an alternate take on hypothesis testing using simulated-data experimentation. Basically, what would happen if the condition labels in your data, say, treatment and control, were acutally assigned randomly with p = .5? Could you easily distinguish such nonsense from your actual data? At sample sizes and effect sizes that are probably typical in some areas of behavioral and medical science, it would not be easy. We develop a new metric (psi) based on improvement from this random baseline, and suggest some commonsense standards based on that metric. Among the benefits of psi is that it does not rely on data modeling assumptions and thus maintains its interpretation no matter how the data are distributed.

A fun set of results from thousands of iterations of the One Bid game from The Price is Right television show. Contestants bid sequentially in an attempt to get closest to the price of an item without exceeding it. Ultimate prize winnings can be substantial, yet the final bidder does not always play the weakly dominant strategy of cutting off another player by bidding $1 more than they did, essentially leaving the other player with no chance to win. Final bidders are more likely to cut off opposite-gender opponents, even conditional on bidding higher than them. We explain why the results cannot be consistent with any explanation based on beliefs. Rather, this is an example of (likely implicit) gender favoritism.

A (tame) comment on Chater and Loewenstein's i-frame vs. s-frame. They that behavioral "nudges" have been co-opted by concerted interests that would rather avoid more effective systemic interventions, i.e., regulation. Taking a public choice perspective, we are less enthusiastic about the s-frame. Political agents are subject to the same psychological foibles as economic aganets. And in any case, they have incentives to play to the foibles of their constituencies. Further, concerted interests also manipulate the traditional regulatory process when they are not diverting things toward nudges. Indeed, some of the problems Chater and Loewenstein describe were caused at the s-frame level.HTML

Americans are miscalibrated about how attitudes have changed over time and hold inaccurate stereotypes about the past. This finding is important because their willingness to support various policies is affected by what they think the trajectory of public support for the policy is, above and beyond what they think the current level of support is.

When people have damage to the ventromedial frontal lobe of the brain, they are often left with a profound inability to make decisions. Previous research in patients with VMF damage suggested that they have intransitive preferences, i.e. A over B, B over C, but C over A, suggesting that VMF damage causes irrational choice. Using a sophisticated experimental design and analysis (see 2011 paper below), we conclude that a sample of VMF patients had fundamentally transitive preferences, though with greater variability than healthy controls. It appears VMF affects decision making because it is key in assessing value, and thus strength of preference.

Data from the Good Judgment Project in the IARPA Aggregative Contingent Estimation tournament. Crudely asking people to give a probability of something happening has informational value, even incremental upon fancier methods of elicition, such as a prediction market. This is revealed when one uses best-in-class methods of aggregating forecasts.

Is profit evil? Themes in literature, art, religion, and folklore would suggest so. We show that people tend to hold anti-profit beliefs such that profit is seen as necessarily in conflict with social good. Seven studies show the basic effect and suggest that people specifically neglect the incentive properties of profit, treating it instead as zero-sum.

A more popular press-type piece about the above.

Advice tends to be paternalistically biased toward caution when there aren't pecuniary incentives to the contrary. We consider some reasons why that might be.

We give a tractable definition of crowd wisdom: A crowd is wise if a linear aggregate, for example a mean, of its members’ judgments is closer to the target value than a randomly, but not necessarily uniformly, sampled member of the crowd. From there, we derive conditions under which crowd wisdom would hold. Crowd wisdom is very robust.

Theoretical review that distinguishes "genuine" prosocial behavior from that which is reluctant (giving in). One who gives in would have rather avoided the situation that compelled prosociality altogether. One example might be going out of one's way to avoid being asked to donate to charity. Of course it's hard to generalize, but about half of giving in experimental contexts appeared to be reluctant.

Winner of the Clifford T. Morgan Best Article Award for the journal that year. A proto version of the 2024 better accuracy paper, we introduce the idea of using a random estimator as a benchmark against which to compare estimates such as means from an experiment. An attempt to set absolute standards for estimation accuracy and move away from p-values.

Unstructured interviews are notoriously poor predictors of future performance, but people sure love them. We used experiments to identify reasons why people like them and why they might not be so effective. One reason people like them is that they tend to provide too much flexibility to sensemake; that is, to fit ad hoc hypotheses to explain a person's behavior. There is an illusory feeling of understanding. One reason they might not be so effective is dilution - the presence of too many cues of questionable validity overloads a person and weakens the impact of valid information they are getting. Here is my opinion piece in the New York Times about the research (they created the current title): HTML

Defaults have a reliable impact on choice, though probably because of several contributing factors. We demonstrate in a pretty clean way that defaults shape preferences when people are uncertain about them.

Winner of the inaugural Exeter Prize for best paper in Experimental Economics, Behavioral Economics, or Decision Theory in the previous year. Behavioral research often purports to show violations of rational choice axioms. But behavior in experiments varies. For example, in repeated choices over time, people don't always make the same choice from the same set of options. There are models of variability in choice, but the reasons why choice is probabilistic lie outside of the axiom system, making axiom testing not so simple. If we assume that preferences varying from time to time is the reason for choice variance, then we can construct parsimonious tests of transitivity. We reconsider alleged evidence of transitivity violations and conduct new experiments. We find little evidence for intransitive preferences when tests are performed correctly.

Using die-under-cup experiments where people can report the result of a die roll only they see and get paid whatever number they report. Evidence suggests people report the result of successive die rolls, rather than the first one as they are instructed. Much less over-reporting when they can't roll again. I think this demonstrates an oft-cited but rarely demonstrated self-signaling. Subjects feel more comfortable over-reporting when they are taking a mulligan on a die roll nobody else saw. Subjects rate this behavior as less of a lie than reporting a number that never appears.

Economic experiments that reconsider the nature of social preferences. A substantial portion of people who choose to give money to other anonymous subjects in a one-shot game will apparently not share once they have an excuse. The most notable finding is that people who otherwise choose generously will not click a button to reveal the other subject's payoffs because they might learn what is selfishly preferred is bad for others.

Many subjects accept $9 instead of playing a $10 dicator game in which they can allocated any portion they want to themselves in a completely anonymous and one-shot setting. The key condition is that the $9 option is preferable when it prevents the anonymous, putative recipient from knowing that a $10 dictator game would have been played.

Giving physicians a reason to dislike me. We argue that gifts can influence physician decision making in subtle ways that the physician might not be aware of. Therefore, don't take gifts.