Some recent and on-going projects in the Budescu Lab
Weighted Probability Model for High-Stakes Admission Decisions (Song and Budescu): How does one select the best applicants when there are more qualified candidates than one can accept? Is there a more reliable alternative to what the traditional method commonly used when dealing with high selection ratios? In this study, we evaluate alternative solutions to commonly used models in high- selection ratio admissions that are sensitive to the imperfect reliability of the tests and the criterion.
Decisions with Multi-Stage Lotteries (Fan, Budescu and Diecidue): We examine the factors that influence decision makers’ (DMs’) perceptions of, and actions (e.g., choices) when facing multi-stage lotteries. There have been many studies demonstrating systematic violations of the reduction of compound lottery (ROCL) axiom. . However, only few studies proposed descriptive models to characterize and quantify these violations. Based on existing literature we identified 6 relevant factors and seek to validate their effects and evaluate their importance by mathematical modeling. Our ultimate goal is to determine the optimal descriptive model of evaluation of multi-stage lotteries and explain the violations in psychological perspectives.
Overconfidence in probability intervals and the wisdom of crowds (Budescu and Park): Many empirical studies have shown that interval probability estimates on uncertain quantities are too narrow (overconfident). We show that certain aggregation methods (the ‘Wisdom of Crowds’) can mitigate the bias and improve the accuracy of the estimates by combining individual intervals. We also hypothesize that the severity of the overconfidence varies as a function of the nature of estimates – being more severe under epistemic uncertainty that for aleatory uncertainty – but predict that WoC solutions can be applied in both cases.
Empirically based linguistic probability lexicons (Ho and Budescu): How can people more effectively communicate uncertain scientific information to policymakers and laymen? Many organizations develop standardized linguistic probability lexicons (“very likely” corresponding to a probability of 90-100%) to communicate uncertainty. We study methods of increasing the efficiency of such probability lexicons in the context of climate change and intelligence analysis.
The Effects of Feedback, Scoring and Mindset on Performance and Engagement in Self-Adapted Testing (Arieli-Attali and Budescu): We examine how people adjust their choices of difficulty (reflecting the level of effort they want to invest) in a self-adapted test under various conditions of feedback and scoring, and different goal-setting. We are interested to learn how a mindset of learning vs, performance, and global feedback about performance (total scores provided after each answer) impact people motivation to exert more or less effort in the test. We hypothesize that people’s engagement is influenced by their mindset and the type of scores provided, such that learning goal accompanied with scores that indicate progress are most beneficial in eliciting effort, and consequently improving performance.
The Coupled Impact of Conflict and Imprecision from Multiple Expert Forecasts (Benjamin and Budescu): We examine how people perceive and aggregate multiple expert forecasts. We study how the relative agreement and precision alters judgements by comparing set of forecasts that are precise and conflicting to sets that are agreeing and imprecise to sets that are both imprecise and disagreeing to varying degrees. We hypothesize that people’s estimates of the most likely values and judgements on relevant characteristics are influenced by the formulation of the forecasts sets.
Sources of Uncertainty from Climate Model Projections (Budescu and Benjamin): This study examines the degree to which various sources of uncertainty influence how people credit or discount information under uncertainty, especially in climate change policy decisions. We test how sensitive people are to variations in how experts use statistical models. Are people more sensitive to uncertainty between or within experts? Between or within models? To the models’ structures or the input parameters?
The Effects of Learning Mode on Advice Giving (Benjamin and Budescu): This study examines how people communicate after learning about quantitative choices either from descriptive summaries or sequential experiences (AKA description vs. experience). How do advisers pass along their method of learning when giving advice, and to what degree do different learning modes affect the advisees’ decisions, preference, and confidence in their choices?
Does Attribute Framing Effect Perceptions of Climate Change (Benjamin, Por and Budescu): The terms “global warming” (GW) and “climate change” (CC) are often used interchangeably, but research finds GW is more polarizing resulting in less advocacy by some subgroups. We question these results based on the complexity of CC beliefs and an expectations that people with stronger ideologies have more rigid beliefs. We go beyond the one-question approach to study attribute framing by measuring CC beliefs using multiple methods across many facets of belief.