DR@W Forum is an interdisciplinary discussion series which focuses on theoretical and empirical research about decision making. The usual structure of the forum is a 30 minute introduction of the topic/working paper related to decision research, followed by a discussion. The audience prefers discussing work-in-progress topics as opposed to finished papers. We meet on Thursdays between 2:30 p.m. and 3:45 p.m. in the extension on the third floor of the Library (at the Wolfson Research Exchange Area). Contact John Taylor (John.Taylor[at]wbs.ac.uk) if you would like to suggest a speaker for a future event. Notifications of upcoming DR@W Forum events along with other Behavioural Science GRP related activities can be obtained by registering with the moderated GRP BSci email list here. Abstracts for upcoming forum events can be viewed here.
- Nov302017DR@W Forum
Michela Redoano (Warwick, Economics)
- Nov232017DR@W Forum
Christina Gravert (Gothenburg) http://bsaw.ac.uk/events/drw-forum-christina-gravart-gothenburg/
Title : ‘Hidden Costs’
Abstract: We document the hidden costs of a popular nudge and show how these costs distort policy making when neglected. In a field experiment with a charity, we find reminders increasing intended behavior (donations), but also increasing avoidance behavior (unsubscriptions from the mailing list). We develop a dynamic model of donation and unsubscription behavior with limited attention. We test the model in a second field experiment which also provides evidence that the hidden costs are anticipated. The model is estimated structurally to perform a welfare analysis. Not accounting for hidden costs overstates the welfare effects for donors by factor ten and hides potential
- Nov162017DR@W Forum - David Ronayne (Oxford)
David Ronayne (Oxford) http://bsaw.ac.uk/events/drw-forum-david-ronayne-oxford/
Title of paper ‘When good advice is ignored: the role of envy and the sunk-cost fallacy’.
Abstract: In an incentivized online experiment involving 1,500 participants we find that good advice is frequently ignored. In our experiment, participants completed tasks where the higher their submitted score, the higher their chance of winning a bonus. After completing the tasks they were told their score along with the score of another participant who did especially well, aka an “advisor” or “expert”, and were asked if they would like to submit the advisor's answers in place of their own. Of those participants scoring strictly lower than the advisor, approximately a quarter chose not to switch. Participants appeared to trade-off rationality against other, behavioural factors. Specifically, we found that an individual’s level of envy, as measured by a dispositional envy scale (Smith et al., 1999) was a significant determinant of whether individuals took good advice from an advisor. Envy operated via distinct channels, dependent on the domain of the advice. When the advisor’s status had been achieved by skill, envy played a “positive” role, meaning that participants with a higher level of envy were more likely to take the advice. However, when the advisor’s status had been achieved by luck, a “negative” effect of envy emerged; the strength of which depended on how much higher the advisor’s score was, relative to the participant’s. Envy played an additional, countervailing role when the advisor was highly-remunerated. Our high-pay treatment caused the probability of accepting good advice to decrease on average by approximately 6-7 percentage points, in both domains. Furthermore, in the skill domain, high-remuneration caused the “positive” effect of envy to no longer be present. Interestingly, our second remuneration treatment, where advisors were paid a small amount for each participant that followed their advice, had no effect. Finally, we constructed a novel scale to measure proneness to the sunk-cost fallacy. We found that a participant’s proneness was negatively associated with the propensity to take good advice, regardless of the domain. In contrast, we found a more general measure of stubbornness to have no predictive power.
- Nov022017DR@W Forum
Dominik Duell (Essex)
- Oct262017DR@W Forum
Matthew Polisson (St Andrews) http://bsaw.ac.uk/events/drw-forum-matthew-polisson-st-andrews/
- Oct052017DR@W Forum
Philip Newall (Munich)
‘It Will Be Worth It In The End’ http://bsaw.ac.uk/events/drw-forum-philip-newall-munich/
- Jun292017DR@W Forum - Susan Michie (UCL Centre for Behaviour Change)
Artificial Intelligence meets behavioural science: The Human Behaviour-Change Project
To advance our understanding of behaviour change and apply that understanding to real-wold problems, evidence is needed about intervention effectiveness across contexts, and about mechanisms of action. Such evidence is currently produced on a vast but fragmented scale and more rapidly than humans can synthesise and access. The Human Behaviour Change Project brings together behavioural, computer and information scientists to build an Artificial Intelligence system to continually scan the world literature on behaviour change, extract key information and use this to build and update the scientific understanding of human behaviour to answer variants of the ‘big question’: ‘What works, compared with what, how well, for whom, in what settings, for what behaviours, for how long and why?’. A user interface will allow researchers, policy makers and practitioners to seek answers to their queries. The project will focus initially on the use-case of smoking cessation but the project team will make available resources, products and findings throughout the life of the project to encourage collaboration.
- Jun222017DR@W Forum - Paul Frijters (LSE)
Black Lives Matter – The Impact of Racial Shooting on Black Well-Being
Draft manuscript available on request.
- Jun152017DR@W Forum - Leonhard Lades (Stirling)
Self-control failures as judged by themselves (with Liam Delaney)
Behavioral economic insights about systematic decision making biases have obtained considerable influence in policy-making. All over the world, governments use these insights to nudge individuals towards making decisions that make individuals “better off, as judged by themselves”. However, behavioral economics also poses a problem for such policy applications: If choices are biased and do not always reveal individuals’ true preferences, how can policy makers identify what makes individuals better off, as judged by themselves? This question is particularly hard to answer when short-term preferences conflict with long-term preferences. In order to identify whether a policy makes individuals better off, we need a welfare standard to determine what “better off” means and we need a way to measure welfare according to that standard. This paper deals with the latter and discusses whether data from everyday life can be used to evaluate policies assuming that policy makers subscribe to a certain welfare standard (choice, subjective well-being, or rational decision making). We present data from a day reconstruction study in which we tracked self-control problems and subjective well-being of 259 participants over one day. Using this data, we first show that it is possible to identify self-acknowledged self-control failures in everyday life. In a nutshell, if individuals attempt to resist a behavior, but nonetheless enact it, a self-acknowledged self-control failure is present. Secondly, we investigate whether self-acknowledged self-control problems are costly in terms of subjective well-being. Finally, we discuss theoretical and practical issues with using everyday data in policy evaluation.
- Mar162017DR@W Forum - Eugenio Proto (Warwick Economics)
Co-operation and positive mood in the repeated prisoners dilemma.
Does mood matter for cooperation, and if so how? We address this issue in the laboratory by examining repeated play of the Prisoners’ Dilemma coupled with mood induction. We investigate not only whether mood matters but also how it interacts with ability to communicate, the role of beliefs and with both certainty and uncertainty in the number of repetitions. We find that players with an induced positive mood tend to cooperate less than players in a neutral mood setting. This difference is highest in settings with an uncertain number of repetitions and with no communication. We find that the difference is driven by both less accurate beliefs about partners’ choices and a less rational reaction to these beliefs among the players in the positive mood treatment. This interpretation of the data is corroborated by a systematic analysis of the text used during communication.
- Mar022017DR@W Forum - Tim Mullett (Psychology)
Strategies and Pro-Social Tendencies in Incentive-Equivalent Public Goods Games and Binary Dilemmas
A number of experiments have examined whether individuals display consistent tendencies towards pro-social behaviour in different tasks and situations. The results have been mixed, with most studies showing only weak correlations in prosociality across tasks. However, many of these examples compare situations where the incentives and/or outcomes are not directly equivalent. We present an experiment where subjects both completed public goods games and made choices in binary strategic dilemmas. Several linear and non-linear incentive structures are presented to participants, and these structures are designed such that the outcomes for both self and other are directly equivalent in public goods and binary choice tasks. We find a significant relationship between pro-sociality in different tasks, but that this relationship is noisy, with subjects apparently applying different strategies in different tasks. Surprisingly, the correlation in pro-social behaviour is independent of the specific incentive structure, with cross-task behaviour being predicted equally well from responses in equivalent and non-equivalent incentive structures. Finally, by using novel incentive structures we show that individuals classified as “free riders” in traditional public goods games are actually comprised of two groups: “profit maximisers” and “non-investors”, with the latter group contributing zero, or minimal amounts, even when it would be more profitable to invest more.
- Feb232017DR@W Forum: Johannes Lohse (Birmingham Business School)
DR@W Forum: Johannes Lohse (Birmingham Business School)
Is fairness really intuitive?
Economists are increasingly interested in the cognitive basis of pro-social behavior. Using response time data or time pressure, several authors have claimed that "fairness is intuitive". In light of conflicting empirical evidence, we provide theoretical arguments showing under which circumstances an increase in "fair" behavior due to time pressure provides unambiguous evidence in favor of the "fairness is intuitive" hypothesis. Drawing on recent applications of the Drift Diffusion Model (Krajbich et al., 2015a), we demonstrate how the subjective difficulty of making a choice affects choices under time pressure and time delay, thereby making an unambiguous interpretation of time pressure effects contingent on the choice situation. To explore our theoretical considerations and to retest the "fairness is intuitive" hypothesis, we analyze choices in two-person prisoner’s dilemma and binary dictator games. As in previous experiments, we exogenously manipulate response times by placing subjects under time pressure or forcing them to delay their decisions. In addition, we manipulate the subjective difficulty of choosing the fair relative to the selfish option across all choice situations. Our main finding is that time pressure does not increase the fraction of fair choices relative to time delay irrespective of the subjective difficulty of choosing the fair option. Hence, our results cast doubt on the hypothesis that "fairness is intuitive".
- Feb092017DR@W Forum - Severine Toussaert (LSE)
DR@W Forum: Séverine Toussaert (LSE)
Eliciting temptation and self-control through menu choices: a lab experiment
Unlike present-biased individuals, Gul and Pesendorfer (2001) agents may pay to restrict choice sets despite expecting to resist temptation, thus eliminating self-control costs. I design an experiment to identify these self-control types, where the temptation was to read a story during a tedious task. The identification strategy relies on a two-step procedure. First, I measure commitment demand by eliciting subjects’ preferences over menus, which did or did not allow access to the story. I then implement their preferences using a random mechanism, allowing me to observe subjects who faced the choice, yet preferred commitment. A quarter to a third of subjects can be classified as self-control types according to their preferences. Of those facing the choice, virtually all self-control types behaved as they anticipated and resisted temptation. These findings suggest that policies restricting the availability of tempting options could have much larger welfare benefits than predicted by present bias models..
- Jan262017DR@W Forum - Michalis Drouvelis (Birmingham)
Michalis Drouvelis (Birmingham)
Personality and social preferences
Theories of social preferences remain silent about the role of individuals’ personality heterogeneity in predicting economic behaviour. We conduct a laboratory experiment which sheds empirical light on the causal impact of agreeableness on two measures of social preferences: aversion to advantageous inequality and cooperative behaviour. Our findings provide robust evidence that both measures are sensitive to the personality trait of agreeableness. In particular, agreeable individuals are more averse to advantageous inequality compared to disagreeable individuals. We also find that agreeable individuals are more cooperative in relation to disagreeable individuals. Our results provide novel evidence for inspiring theory development that can account for personality effects on economic preferences
- Jan192017DR@W Forum - Benjamin Scheibehenne (Geneva)
Benjamin Scheibehenne (Geneva)
Bayesian Statistics as an Alternative for Analyzing Data and Testing Hypotheses
Empirical data in Psychology and Economics are often analyzed using null hypothesis significance testing (NHST) and the ritualized calculation of p-values. In my talk, I will point out problems of this approach and I will propose Bayesian methods as a feasible alternative for analyzing data and testing hypotheses. Based on concrete examples from the literature on descriptive social norms and from modelling decision making under risk, I will show that, in contrast to NHST, Bayesian statistics yield consistent results, it can quantify the evidence for both, the null and the alternative hypothesis based on the Bayes factor, it makes prior assumptions explicit, and it is fairly easy to use.
- Dec082016DR@W Forum - Neel Ocean (Economics, Warwick)
Do people adjust for extreme review score bias?
An important implication of the internet on modern economic life is the increasing reliance on online reviews to inform consumption decisions. Yet, extremely positive or negative reviews may be subject to a large degree of bias, as well as conflicts of interest. I introduce a model that proposes individuals weight extreme review scores to adjust for this potential bias. A randomised experiment on 501 individuals finds insufficient evidence that extreme review scores are being weighted when evaluating the quality of a good. Hence, individuals are susceptible to being influenced by deliberately falsified extreme reviews.
- Dec012016DR@W Forum - Joe Gladstone (UCL)
Joe Gladstone (UCL)
Health Risks or Financial Costs: A Randomized Controlled Trial to Improve Medication Adherence in Pharmacies
Low levels of medication adherence represent a growing problem for global health systems. We report evidence from a pre-registered randomized controlled trial, delivered through 278 UK pharmacies, aimed at increasing adherence rates. Patients (N=16,191) were asked to commit to taking their medication as prescribed by signing their name on a sticker attached to their medication packaging. In two additional trial arms, the commitment was paired with a message describing the negative consequences arising from non-adherence; either the increased risk to the patient’s own health, or the financial costs to society. Our results indicate that for participants who signed the pre-commitment without reference to the negative consequences arising from non-adherence, there was no change to their medication adherence levels in comparison to the control group. However, participants who signed a pre-commitment paired with the health warning were significantly more likely to adhere to their medication than the control group (odds ratio = 1.59, 95% CI [1.02; 2.48]). Conversely, participants who signed a pre-commitment paired with a financial cost warning were less likely to adhere to their medication (odds ratio = .64, 95% CI [0.41; 1.02]). Our results provide new insights into the psychological motivations underlying medication adherence.
- Nov252016Special Friday DR@W Forum - Jialan Wang (Illinois)
Jialan Wang (University of Illinois)
Minimum Payments and Debt Paydown in Consumer Credit Cards
Using a dataset covering one quarter of the U.S. general-purpose credit card market, we document that 29% of accounts regularly make payments at or near the minimum payment. We exploit changes in issuers’ minimum payment formulas to distinguish between liquidity constraints and anchoring as explanations for the prevalence of near-minimum payments. Nine to twenty percent of all accounts respond more to the formula changes than expected based on liquidity constraints alone, representing a lower bound on the role of anchoring. Disclosures implemented by the CARD Act, an example of one potential policy solution to anchoring, resulted in fewer than 1% of accounts adopting an alternative suggested payment. Based on back-of-envelope calculations, the disclosures led to $62 million in interest savings per year, but would have saved over $2 billion per year if all anchoring consumers had adopted the new suggested payment. Our results show that anchoring to a salient contractual term has a significant impact on household debt.
- Nov172016DR@W Forum - Jerker Denrell (WBS, Behavioural Science Group) & Adam Sanborn (Psychology, Warwick)
Jerker Denrell (Behavioural Science Group) & Adam Sanborn (Department of Psychology) - Date(s) - 17 November 2016
Implicit corrections for missing feedback: Imputation vs. statistical models
In many real-life settings feedback is only available for cases decision makers accept. How do people learn from such selective feedback? There are two approaches in statistics for this kind of missing data: imputing the missing values, and using a statistical model of the task. Elwin et al. (2007) provided evidence that people rely on a type of imputation called ‘constructivist coding’, i.e., people code rejected cases, for which no feedback is available, as failures. It is not intuitively obvious whether relying on this kind of internally generated feedback is sensible or leads to bias in an exemplar model. To examine this, we formally analyze the impact of constructivist coding on the performance of exemplar learning algorithms. Our analysis shows that constructivist coding is an adaptive strategy: it maximizes the total reward. The reason is that constructivist coding compensates for the failure of exemplar algorithms to take selection-bias into account. In two experiments we then test whether participants impute missing values through constructivist coding, or use a statistical model of the task to correct for selection bias. These experiments have a simple setup: a financial advisor is predicting the amount an investment will return, but the advisor’s predictions are noisy and have an unknown bias. Participants decide on each trial whether to invest, receiving feedback only if they do so. We find that about half of participants use an exemplar model; a large majority of these participants use constructivist coding, some of whom internally generate values that are very close to optimal. The other half of participants correct for selective feedback with a sensible task-specific strategy, the majority of whom correct for bias using a Bayesian model of the task.
- Oct272016DR@W Forum - Amelia Hunt (Aberdeen)
"Choice and consequence in eye movements and beyond" Amelia Hunt (Aberdeen)
Deciding how to allocate one’s attention when faced with multiple competing goals is a dilemma we all face in daily life. These decisions can have serious consequences — for example, in splitting your attention while driving. An important factor that should weigh into such a decision is the limitations of your own abilities. That is, if you have adequate skill and the tasks are not too demanding, you can complete multiple tasks in a given time interval. but if the tasks are difficult, you should focus all your efforts on completing one task. Using eye movements as a starting point, we observe that people fail to take the strengths and limitations of their own visual acuity into account when deciding where to look to detect a target that could appear in multiple possible locations. We extend this conclusion beyond eye movements into two other tasks (throwing and memorization), and show that the results cannot be accounted for by a lack of accurate information about one’s own probability of success given their level of skill and the set of possible decisions that person could make. We also find that experience and training have a severely limited ability to improve decision efficiency in these tasks. The results reveal surprising shortcomings in human decisions. I will speculate on the decision rules and biases that could lead to inefficient decisions in the specific situations in which we have observed them.
- Oct202016DR@W Forum - Kirill Pogorelskiy (Economics, Warwick)
Title: Media Bias and News Sharing on Social Networks: A Laboratory Study
Abstract: In this paper we use lab experiments to study the relationship between social media and voting in elections. Our treatments mimic the features of social networks (obtaining information from friends) in the presence of media bias (obtaining information from biased media outlets) in order to address concerns in both the academic and popular press literatures that voters obtaining their political news and information from social media outlets may become more polarized in their voting behavior. Our preliminary results suggest substantial effects of polarization at the expense of efficient information aggregation by voting: in all treatments voters publicly send out signals favorable to their party more often than signals unfavorable to their party, and also vote according to their private signals more often if the signal is favorable. Media bias lowers efficiency, and its negative effects are amplified when voters only exchange information with other voters with the same party preferences (our polarized social media treatment). All in all, our results provide tentative support to concerns that by filtering out unfavorable content social media may lead to polarization in voting behaviour.
- Oct132016DR@W Forum - Gerri Spassova (Monash)
"Uniformity aversion in judgements of expertise. When being right looks wrong" Gerro Spassova (Monash)
The present research documents a phenomenon we label “uniformity aversion,” whereby a critic who evaluates several options as qualitatively equal is seen as less of an expert relative to a critic whose evaluations exhibit variance. Importantly, the phenomenon is observed in the presence of strong accuracy cues indicating that the options are more likely to be of the same quality. We test uniformity aversion in a series of studies, using different product settings and different accuracy cues. The mechanism underlying uniformity aversion is discussed.
- Oct062016DR@W Forum - Nick Powdthavee (WBS Behavioural Science Group)
"Do economists care more about the average signal than total productivity in academic publications? A randomized survey experiment" Nick Powdthavee (WBS Behavioural Science Group)
It is well-known that a person’s publications are very important for judgments made in hiring, tenure, and promotions of academics. Yet, there seems to be little research on what characteristics of such lists economists consider in making these judgments. In the current study, we conduct a survey experiment on faculty members of economics departments from 44 universities around the World. By randomly assigning people to rate different hypothetical CVs, we find that economists tend to rate shorter CVs higher than longer CVs in single evaluation, even though longer CVs have everything that the shorter CVs have. However, the differences in the ratings disappear when they are asked to rate the short and long CVs together in a joint evaluation.
- Jun232016DR@W Forum: Peiran Jiao (University of Oxford)
Experience-Based Belief Distortion: When Experience is Information-Free
People overweight experience relative to descriptive and observational information, in games, portfolio choice, etc. However, little is known about how beliefs are biased by experienced payoffs. This paper offers a simple model of experience-based belief distortion, where the decision maker with good (bad) experience misinterprets bad (good) signals, and overestimates future good (bad) states. Two experiments were conducted to test the model predictions. The first experiment asked subjects to predict future prices after viewing some stock price charts, and experienced gain/loss was exogenously assigned. The second experiment further provided information about the outcome-generating processes to allow for Bayesian updating as a benchmark. Subjects who gained reported significantly more optimistic guesses than those who lost after viewing the same sequence; in belief updating, they overweighted new evidence in favor of the signal from which they gained.
- Jun162016DR@W Forum: Mahnaz Nazneen (Department of Economics)
Gender Roles and Bargaining Behaviour: A Lab Experiment in Bangladesh
There is ample evidence of gender differences in bargaining behaviour, observed in the laboratory. One possible explanation of such behaviour is Social Role Theory (SRT) which suggests that, men and women behave differently in social situations and take different roles due to expectations that society puts upon them, that almost all behavioural differences are the result of stereotypes. The aim of the study is to examine if there are gender difference in a bargaining behaviour and if these differences can be explained by SRT in a society where perceptions of gender roles are strongly formed. A standard ultimatum game was used to observe bargaining behaviour among 222 university students in a laboratory experiment in Bangladesh. Subjects were randomly assigned to a control or treatment session; where in the latter, subjects read a small vignette about how preferences of individuals are heterogeneous and depend on a number of factors including gender. The purpose of the vignette is to prime for gender differences in behaviour. The main finding is that both men and women Responders ask for a higher MAO when they are partnered with a female Proposer, in the treatment session, after controlling for personality traits, intelligence and risk preferences. Regardless of their gender, the prime influences behaviour of both men and women in a similar manner and overpowers their initial perception (if any) about gender roles. Also, consistent with the literature, I find no significant difference in the Proposer behaviour. I conclude that there is no gender bias per se. Only when subjects are nudged or provided with additional information, they adjust their belief about the other person’s behaviour and mostly this information or signal comes from the society. So, the findings are consistent with Social Role Theory.
- May262016DR@W Forum: Janina Hoffmann (University of Konstanz)
Capacity restrictions in human judgment
Making accurate judgments such as choosing a job candidate presumes an adequate weighting of more and less important aspects, say the candidate’s skills. People may find out the importance of different cues by testing rules specifying how the cues relate to the criterion. In this talk, I will present evidence from three studies suggesting that this ability to test rules is restricted by working memory capacity. In a large individual difference study, we first investigated how working memory and episodic memory affect judgment accuracy. The ability to solve rule-based tasks was predicted by working memory, whereas episodic memory predicted judgment accuracy in the exemplar-based task. Second, increasing working memory load reduced the prevalence of rule-based strategies and ultimately benefitted judgment accuracy in a task that could not be solved by rules. In a final step, we incorporated the assumption that a capacity limit restricts rule-based learning into a learning model and tested it against two alternative psychological mechanisms: a decay in learning speed and attentional learning. A capacity-restricted learning model best described and predicted the learning curve of the majority of participants. Taken together, these studies suggest that learning to accurately weigh the importance of different aspects is limited by working memory capacity.
- May192016DR@W Forum: Giorgio Coricelli (USC Dornsife)
- May122016DR@W Forum: Graham Loomes (Warwick Business School) and Lukasz Walasek (Department of Psychology)
Intrinsic and Extraneous Noise in Risky Choice Experiments
Graham Loomes (Warwick Business School) and Lukasz Walasek (Department of Psychology)
Participants' responses in decision experiments are 'noisy': when presented with exactly the same choice at different moments within the same experiment, many people are liable to answer differently from one moment to another. Some of this may be due to intrinsic variability in the way people generate their decisions; but the experimental environment may also have an impact - e.g. the complexity of the task, the workload, the (lack of) incentives. Moreover, in principle, extraneous and intrinsic factors may interact, and may operate to different degrees for different individuals, making it harder to identify core preferences. Can we identify/separate/measure such effects? We present some results which may shed light on these issues.
- May052016DR@W Forum: James Goulding (University of Nottingham)
Neo-demographics: New ways of analysing human behaviour from Big DataIn the age of mass digital communications, loyalty cards, social networking and mobile phones that never seem to leave our possession, each of us is leaving an incredible array of data in our wake. When aggregated, these mass datasets have unparalleled potential to reveal insights into individual behavioural patterns, to unearth unexpected groups and communities, and identify trends as they emerge across a population.
The EPSRC Neo-demographics project has brought together a team of computer scientists, mathematicians, business researchers and psychologists to explore that potential via multi-disciplinary methods. In collaboration with multinational industry partners, we are examining the novel behavioural patterns that can be mined from these datasets, and whether such mass analytics can be employed for social good.
In this talk, I will discuss the Neo-demographics project’s progress, presenting 1. some of the novel machine learning techniques that have been developed to analyse human behaviour; 2. applications of our work in East Africa for international development and social policy; and 3. a 3D visualization system via we visualize results, called the ‘PARM’.
- Apr072016DR@W Forum: Anca Hanea (University of Melbourne)
An IDEA of how to get the best out of experts
This talk presents a fairly novel structured expert judgement (SEJ) protocol for quantifying parameter uncertainty using multiple experts. Generally, multiple expert opinions need to be aggregated. The two main flavours of aggregation, behavioural and mathematical, define two main classes of SEJ protocols. A third one, the so-called “mixed” approach combines aspects of both behavioural and mathematical methods. The IDEA protocol proposed here mixes specific elements from the three SEJ classes of approaches mentioned above, such that their disadvantages are minimised and their respective advantages cumulate. The acronym IDEA arises from the combination of the key features of the protocol that distinguish it from other structured elicitation procedures: it encourages experts to Investigate, Discuss, and Estimate, following which judgements are combined using mathematical Aggregation.
The experts give their individual opinions in subsequent rounds of elicitations, in a remote manner. In the first round, the experts are required to answer the questions without engaging in any (virtual) discussion with the other experts. They are then given the opportunity to discuss differences of opinion and reconcile the meaning of questions and context. The debate is remote (using an online platform) rather than face-to-face. This has the advantage of promoting the wisdom of crowds, whilst avoiding the tensions associated with group discussion between dominating personalities. The second estimate is again individual and strictly anonymous. At the end of the second round the output is a set of estimates that should further be mathematically aggregated. Several aggregation rules, many of them performance based, can be used and compared. Experts’ performance may be measured in terms of accuracy, calibration and informativness.
- Mar172016DR@W Forum: Ganna Pogrebna (WMG)
Evaluation Periods and Decision Making in For-Profit versus Non-Profit Domains (with Carlo Perroni and Kimberley Scharf)
Investors who are more willing to accept risks when evaluating their investments less frequently are said to exhibit myopic loss aversion (MLA). Several experimental studies (Gneezy and Potters, 1997; Haigh and List, 2005, Langer and Weber, 2005, etc.) found that, in the “for-profit” domain, subjects bet significantly higher amounts of money on a risky lottery when they observe only a cumulative outcome of several realizations of the lottery (long evaluation period) than when they observe a series of individual realizations of this lottery (short evaluation period). We analyse decision making in the short evaluation period and the long evaluation period in the “non-profit” domain and find the reverse effect: people tend to donate more to charity when the evaluation period is short even if charitable giving involves risk. We provide a theoretical explanation of our findings which does not require an assumption of loss-aversion.
- Mar102016DR@W Forum: Ralph Hertwig (Max Planck Institute for Human Development)
- Mar032016DR@W Forum: John Fox (University of Oxford)
Will AI revolutionise the decision sciences? A medical perspective.
Headlines about medical errors and health service failures appear more and more frequently in the media. The general public and health professionals are now aware that medical practice is facing enormous challenges that affect the quality and safety of our clinical services. This has generated public and political pressure to look for solutions; the press often wants to know who to blame, while the politically minded see the need for greater privatisation, organisational change or bigger budgets. This is a global problem, not just one for the NHS.
As a cognitive scientist I see many of the underlying problems as arising from human cognitive limitations and how we bring our knowledge to bear in our reasoning, decision-making, planning etc. For example, decision-making is pivotal to everything we do individually, in groups and in organisations. Uncertainty and risk make decision making difficult - and uncertainty and risk pervade medicine so medicine is a fascinating model for cognitive research. Indeed there is now much talk about how “cognitive computing” and artificial intelligence will “revolutionise medicine”.
This talk will briefly overview some of my research on human expertise and the use of AI in medicine. I will argue that AI systems based on an understanding of human cognition and decision-making can improve quality, safety and efficiency of patient care, perhaps more than political and managerial interventions can. AI and its subfields, like knowledge engineering and machine learning, are showing us how machines can do human-level tasks well, cope with uncertainty, manage risks better, make more objective, evidence-based choices, plan and act more effectively in complex and rapidly evolving situations, while allowing people to retain control.
Will AI will revolutionise medicine? we don’t know yet. However, I am also interested in the question - whether AI and its subsidiary fields, like knowledge representation and autonomous systems, offer significant new insights into human reasoning and decision-making. The possibility that the decision sciences might themselves be revolutionised by the new wave of AI seems ripe for discussion.
- Feb252016DR@W Forum: Andis Sofianos, Neel Sagar (Department of Economics)
- Feb182016DR@W Forum: Silvia Montagna (Department of Statistics)
- Feb112016DR@W Forum: Sebastian Olschewski (University of Basel)
- Feb042016DR@W Forum: Philipp Külpmann (Department of Economics)