Bachelor/Master Theses

Perceiving distributions

The events people experience strongly shape their decisions and attitudes. Experiencing extreme weather events is related to attitudes towards climate change (Hoffman et al. 2022), disasters affect the willingness to purchase insurance (Kamiya & Yanase, 2019), and investors may overweight rare extreme positive returns (Blau et al., 2020).

What people experience typically can be described to follow probability distributions: For instance, most of the time there is little or no strong precipitation, but there is a non-zero probability of extreme downpours and flooding. People should pay attention to these extreme events. But under which conditions do they weight them accurately and when do they over- or underweight them? This raises the more general question, how people represent the information they experience from such probability distributions.
The goal of this project is to assess how people represent the information they experience. In the real world, we rarely have control over the underlying distributions. Therefore, we will rely on a combination of experiments and computational models to understand how people represent information from various distributions to make judgments and decisions. Supervisor: Nico Gradwohl

How do people perceive risk?
Risk is a central topic in research on judgment and decision making and economics alike. Conceptually, it is often quantified as the variance of an outcome distribution. But how do people perceive and assess this variance? What are meaningful ways to measure perceived variance in an experiment? Clearly, we don't expect people in real-world environments to actually calculate variance or standard deviation from a sample of numbers they observe. The master's thesis aims to experimentally compare existing measurement tools to assess perceived variance and to develop new ones. Existing tools include asking for a range instead, showing people different histograms, asking them how far apart two random samples would be on average, or asking them for the boundaries of some quantile (e.g. 50% of all values). Theoretically, all these measures should be equivalent, but there is reason to believe that people will give different answers depending on how the question is asked. The problem becomes even more interesting when the shape of the underlying outcome distribution is taken into account.
The project is of theoretical relevance (better understanding of how the mind works) but also practically relevant (for example, in finance or in a medical context). Supervisor: Prof. Dr. Benjamin Scheibehenne

How well do people know their partner?

Knowing your partner well is considered an important aspect of a relationship. Predicting each other's preferences is also highly relevant in a consumer context (think of Christmas presents and the infamous 'dead weight' of Christmas). In a longitudinal study that I conducted in collaboration with the Socioeconomic panel (N > 2.000 people), we asked a representative sample of couples to predict each other's food preferences at two points in time (3 years apart). Now the question is: What variables predict prediction accuracy in couples? How does this accuracy develop over time? Are there long-term consequences of low partner knowledge (divorce rate)? The project does not require collecting original data, but it does require conducting statistical time-series analyses. Supervisor: Prof. Dr. Benjamin Scheibehenne

Perceiving distributions

The events people experience strongly shape their decisions and attitudes. Experiencing extreme weather events is related to attitudes towards climate change (Hoffman et al. 2022), disasters affect the willingness to purchase insurance (Kamiya & Yanase, 2019), and investors may overweight rare extreme positive returns (Blau et al., 2020).

What people experience typically can be described to follow probability distributions: For instance, most of the time there is little or no strong precipitation, but there is a non-zero probability of extreme downpours and flooding. People should pay attention to these extreme events. But under which conditions do they weight them accurately and when do they over- or underweight them? This raises the more general question, how people represent the information they experience from such probability distributions.
The goal of this project is to assess how people represent the information they experience. In the real world, we rarely have control over the underlying distributions. Therefore, we will rely on a combination of experiments and computational models to understand how people represent information from various distributions to make judgments and decisions. Supervisor: Nico Gradwohl