1/16 Timothy Pleskac, Psychology
Abstract: Countless everyday decisions—which candidate to support, which witness to believe, which venture to invest in, which scientific theory to endorse, when to cross a busy street—are made with very limited or no knowledge of the probability distributions over the possible outcomes. There are different ways in which people solve this problem of making a decision under uncertainty. In this talk, I will discuss two of these solutions and their implications for understanding decision making. One solution is to sample the external world for information and estimate the probabilities via these samples of information, making what we call “decisions from experience.” Understanding how people make these decisions from experience is important because when they do, rare events tend to have less impact on the decision than they deserve according to their true, objective probabilities, a property that is contrary to the common conception in behavioral decision research that people tend to overweight or overestimate rare events. A second solution is based on an ecological structure present in many of life’s gambles where risk is reward, that is the big rewards people desire are relatively unlikely to occur. We find that decision makers are keenly aware of this relationship and exploit it in the form of a heuristic—the risk-is-reward heuristic—to infer the probability of a payoff during decisions under uncertainty. I will show how this risk-is-reward heuristic can help explain several different phenomena. More broadly, these two solutions make it clear that to understand decision making, we need to not only model the decision process, but also the environment to which these processes have adapted and which they exploit.
1/23 Yesim Orhun, Marketing
Abstract: This study examines how positive reciprocity to a helpful behavior varies with the potential of strategic motives behind the helpful behavior. I elicit reciprocal choices and beliefs about kindness across scenarios where the same helpful action could be motivated by punishment-avoidance, reward-seeking, and/or pure altruism. I find that in response to the same helpful action, positive reciprocity is higher for unmotivated helpful actions than for helpful actions potentially motivated by punishment-avoidance. I show that this response difference is mediated by kindness inferences regarding the person who chose the helpful action. These results provide the first evidence that motives impact the level of positive reciprocity to the same helpful action.
Similarly, people reward motivated kindness less when existence of reward-seeking motive leads to lower kindness inferences. However, this motive does not impact kindness inferences very negatively, because rewards are not as motivating as punishment. In addition, an opposing force may be present if kindness inferences are sufficiently good: people reward kindness more when rewards were expected. Therefore, replicating previous work in this domain, I find null results regarding reciprocity towards helpful actions motivated by reward seeking. However, I differ in my conclusion regarding the role of perceived motives. By unpacking the two opposing forces across and within experiments and exposing the pivotal role of kindness inferences, I argue that motives matter for reciprocal behavior precisely because they impact kindness inferences.
The results contribute to the reciprocity literature by showing that 1) perceived motives impact reciprocal behavior, 2) this impact is due to changes in kindness inferences and may differ across motives to the extent motives shape inferences, and 3) previously inconclusive evidence regarding the role of motives can be reconciled. The findings also have implications for the positive reciprocity puzzle and labor markets.
1/30 Steven Katz, Medicine
Abstract: One of the biggest challenges clinicians and their patients with breast cancer face is to determine a treatment plan for disease with relatively favorable prognosis. Advances in treatment have improved life expectancy for many patients. But advances have come at a steep price because treatments impose substantial morbidity and burden on patients and their families. Concerns about the potential harm of treatments for patients with breast cancer have grown because population-based screening has markedly increased the number of patients diagnosed with relatively favorable prognosis. This has motivated initiatives to evaluate treatment strategies that limit morbidity and burden. However, the impact of these initiatives will be impeded by limited understanding of how decisions about treatments are made. It is increasingly challenging to incorporate patient preferences into these decisions. Indeed, the growing complexity of the process limits patient autonomy in directing treatment decisions. Physicians remain the dominant determinant of receipt of treatment and this influence will grow with advances in personalized medicine. Physician recommendations will be increasingly directed by clinical guidelines applied to individual patients. Opportunities to improve the individualizing of treatment will largely be through improving the clinical validity of tests that form the basis of the evidence that directs treatment recommendations.
2/6 Richard Lewis, Psychology
Abstract: Some kinds of preference reversals occur in multi-attribute choice settings when a preference for one option over another is reversed by the addition of further options. It has been argued that the occurrence of preference reversals violates axioms of decision theory that are the basis of normative, expected utility maximizing models of rational choice. It then follows that people’s choice behavior cannot be explained as expected utility maximization. In our new work, we assume that people rationally (optimally) integrate two sources of information about choice problems in service of maximizing expected utility: noisy estimates or calculations of value, and noisy observations of the order of feature values. We show with mathematical proof and computational simulation that for a range of types of contextual preference reversals, including attraction and compromise effects, the rational choice between existing options is one that is influenced by the addition of new choices: in other words, the rational decision maker should make preference reversals in order to maximize expected utility. We also show that the same assumptions explain some observed risk preference effects. We conclude that experiments showing that people exhibit contextual preference reversals and risk preference are not evidence that they are irrational, nor that they are risk averse; they are however evidence that they are computationally rational, and specifically that they make rational use of ordinal information about feature values.
2/20 Brian Zikmund-Fisher, Health Education
Abstract: We live in an era of health data. Patients often have direct access to test results, risk estimates, and other data related to their health. Drugs come with information about potential side effect risks, which vary in severity and likelihood. People facing particularly complex medical decisions may receive decision aids. Yet, the fact that we have those numbers does NOT mean that people can use them. Visual displays are a key approach to overcoming numeracy deficits and enabling people to make sense of the health data they have.
The Visualizing Health project was a short and highly intense (only 5 months long!) project funded by the Robert Wood Johnson Foundation designed to push the envelope both in considering visual designs for communicating health risk data and in developing iterative research approaches for testing them. The project involved a large team combining researchers and staff from both the University of Michigan’s Center for Health Communications Research and the Center for Bioethics and Social Sciences in Medicine. The UM team then worked closely on a week-by-week basis with Thomas Goetz (former editor of Wired magazine) who envisioned the project, Tim Leong (graphic designer and author of Super Graphic), and teams of graphic designers that Tim recruited.
We created 16 distinct visual data display tasks related to health risks, had teams of graphic designers develop display concepts, and iteratively tested these displays using three different online survey methodologies (Survey Sampling International, Amazon MTurk, and Google Consumer Surveys). The resulting designs and data were then assembled in a project Website that included all the images plus commentary and additional features such as a design “wizard” to help guide users to visual displays that best fit their personal needs.
The Decision Consortium session will provide an overview of the project as well as the visual designs it helped create. We will also discuss the strengths and weaknesses of this research approach—what worked well and what was sacrificed in the name of speed and creativity.
3/20 Stephanie Carpenter, Psychology
Abstract: When hurricane Katrina threatened New Orleans, many people heeded warnings and left the city while others lacked the resources and transportation to flee. Outsiders’ reactions to stranded inhabitants varied from sympathy to anger, and these appraisals affected decisions to donate. Prior research has not fully explained why such discrepancies exist in reactions to tragedy. The current research investigated how appraisals of either a human or a situation as responsible for an outcome influences helping decisions. Results indicated that participants in a high human agency mindset were more likely to donate to a helper (e.g., social worker) than to a victim (e.g., homeless person), whereas those in a situational agency mindset donated equally to helpers and to victims. Findings suggest that shifting human agency appraisals differentially motivates altruism by changing perceptions about the cause of the judged targets’ circumstances.
3/27 Erin Krupka, Information School
Abstract: Social identity describes the part of an individual’s sense of self that is derived from their perceived association with a social category. A key mechanism for social identity-driven choice stems from the normative prescriptions associated with the identities. In this paper we use a 2 (identity prime) x 2 (frame) x 2 (behavior or norms) design. Subjects have one of two identities primed and then are randomly allocated to either a negatively or positively framed decision problem or they are assigned to tell us about the prescriptive norms for either the negatively or positively framed problem. By comparing the behavior of identity-primed subjects in the negative and positive frames, we can test for the effect of identity on choice. We do so in both a laboratory experiment as well as an online experiment.
4/3 Patricia Chen, Psychology
Abstract: Steve Jobs famously quipped, “The only way to do great work is to love what you do.” This quote reflects a “find your fit” mindset, which has been increasingly promoted by our popular media. Professor Cal Newport countered this trend by offering a contradictory “develop your passion” mindset instead. He advised, “Don’t set out to discover passion. Instead, set out to develop it… it’s a path much more likely to lead you somewhere worth going.” We examine these “fit” and “develop” lay theories that are being offered as advice to people making important decisions about their careers. Our studies investigate some core questions: Do these lay theories predict expectations of passion for work? Do they predict people’s choices between different kinds of work? And how do they play out among working adults in the real world? Additionally, we offer two main propositions: first, fit and develop lay theories offer meaningful distinctions that have different motivational and behavioral outcomes; second, both lay theories can lead to passion for one’s line of work, but they do so through different means.