My research lies broadly in the area of judgment and decision making, with a particular focus on ethical decision making. Often through the use of experimental economic methods, I have investigated such topics as how people avoid their own generosity and justify dishonesty. A second, separate focus of my research is the development of accuracy benchmarks for estimating parameters in psychological experiments that can allow us to measure replicability and set standards for acceptable measurement error. Some working papers below. Comments are always welcome.

Abstract: We derive standards for acceptable measurement error when inferring treatment effects from sample means. The standards arise from the amount of measurement accuracy required for sample means to beat a naive benchmark estimator that randomizes the direction and magnitude of treatment effects. We show that at sample and effect sizes common to many areas of psychology, particularly in human subjects research, measurement error is too substantial to make meaningful claims about treatment effects. Alternatively, given some amount of measurement error, we provide minimum sample size recommendations for conducting meaningful experiments.

Abstract: We report evidence for taste-based discrimination on the part of contestants in a bidding game on The Price Is Right television show. The game is well-suited for studying gender discrimination because it combines high stakes for making correct decisions in a field setting with arbitrary assignment of players to groups and roles with respect to gender. Analysis of over 3,900 games reveals that contestants are 19% more likely to use an efficient but aggressive strategy against opposite-gender opponents. The bias in female bidders appears to be driven by incorrect beliefs that male bidders are more skilled. The bias in male bidders appears to be driven by increased competitiveness against bidders who are perceived to be skilled when they are female. Favorable treatment of same-gender opponents costs players an average of $193 in prize winnings across all games, increasing to $425 in the last game of the day.

Abstract: Prior research has suggested that people prefer to remain uncertain about the possible negative social consequences of their actions and that this uncertainty facilitates selfish behavior. Our participants played an economic game where they were uncertain about how selfish actions would affect other players; we offered participants various incentives to "look" at the potential consequences of their actions, and this reduced selfishness. Contrary to the predictions of both "crowding out" and adverse selection, participants who were paid to look were more generous than participants who looked without payment. We also find that these payments can be cost-effective: small payments can lead to social welfare gains that are larger than the total cost of subsidies. Our results suggest an efficient way of changing behavior because it may be cheaper to pay someone to look at information about their social footprint - thus activating their social preferences - than it would be to directly monitor/reward prosocial behavior.

Abstract: In opposition to economic first principles, four studies show that people appear to view profit as necessarily socially harmful. Studies 1 and 2 find a strong negative correlation between profit and perceived social value across both real firms and entire industries. This relationship holds for both perceived profit and actual profit information for public firms. Study 3 confirms that this effect holds when profit motive is manipulated. Otherwise identically-described organizations are seen as providing less value and doing more harm when described as "for-profit" rather than nonprofit. Study 4 demonstrates that people hold a zero-sum conception of profit that neglects the disciplining effects of competition. People see harmful business practices as profitable, even after an intervention encouraging the consideration of long-term consequences under competition. This tendency is not significantly related to self-reported political ideology. Even in one of the most market-oriented cultures in the world's history, people doubt the ability of profit-seeking business to benefit society.

Abstract: We develop a general measure of estimation accuracy for fundamental research designs called v. The v measure compares the estimation accuracy of the ubiquitous Ordinary Least Squares estimator, which includes sample means as a special case, to a benchmark estimator that randomizes the direction of treatment effects. For sample and effect sizes common to experimental psychology, v suggests that OLS produces estimates that are insuciently accurate for the type of hypotheses being tested. We demonstrate how v can be used to determine sample sizes to obtain minimum acceptable estimation accuracy.

Abstract: Unstructured interviews are a ubiquitous tool for making screening decisions despite vast evidence of their invalidity. In three studies, we investigated the propensity for "sensemaking" - the ability for interviewers to make sense of virtually anything the interviewee says - and "dilution" - the tendency for non-diagnostic information to weaken the predictive value of quality information. In study 1, participants predicted two fellow students' semester GPAs from background information and, for one of them, an unstructured interview. In one condition, the interviewee secretly answered questions according to a random system. Consistent with sensemaking, random interviews did not perturb predictions or diminish perceptions of the quality of information that the interview yielded. Consistent with dilution, participants made better predictions about students whom they did not interview. Study 2 showed that merely watching a random interview, rather than conducting it, did little to mitigate sensemaking. Study 3 showed that participants believe unstructured interviews will help accuracy, so much so that they would rather have random interviews than no interview. Impressions formed from unstructured interviews can seem valid and inspire confidence even when interviews are useless. Our simple recommendation for those making screening decisions is not to use them.

Jason Dana [C.V.]
Assistant Professor
Department of Psychology
University of Pennsylvania
danajd@sas.upenn.edu
Solomon Labs C17A
3720 Walnut Street
Philadelphia, PA 19104

Other affiliations:
Philosophy, Politics, and Economics program