The role of statistical numeracy in computational models of risky choice
| dc.contributor.author | Werbeloff, Merle | |
| dc.date.accessioned | 2023-08-08T12:40:37Z | |
| dc.date.available | 2023-08-08T12:40:37Z | |
| dc.date.issued | 2021 | |
| dc.description | A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy to the Faculty of Commerce, Law and Management, University of the Witwatersrand, School of Governance, 2020. | |
| dc.description.abstract | Numeracy is a strong predictor of general decision-making skill, and linked to differences in risk attitudes, such as risk aversion. However, the commonly used normative expected utility model assumes complete cognitive competence of the decision maker, and statistical numeracy is not considered directly in descriptive models of risky choice. These models are nevertheless used in policy-focused economics to assess individuals’ economic welfare, regardless of the effect of statistical numeracy. Thus, if model validity is dependent on the statistical numeracy of individual decision makers, resultant policy decisions may be biased. In an online quantitative empirical study, student respondents were categorised into numeracy groups based on latent mixture analysis of responses to statistical numeracy tests. Using the students’ risky choice responses to monetary lotteries, decision models were estimated using maximum likelihood parameter estimates on a subset of the data, followed by Markov Chain Monte Carlo Gibbs sampling methods for hierarchical Bayesian analysis. The results indicate significant differences between the numeracy groups on the utility parameter estimates, with risk aversion highest for low numeracy respondents. More complex models present identifiability problems. However, simpler models indicate successful outcomes in approximately two-thirds of in-sample estimates and out-of-sample predictions in the gain frame, based on parameter estimates specific to each numeracy group. The researcher proposes a numeracy-based modification to the models, citing the nudging and boosting policy initiatives of the behavioural economics literature as potential solutions to the presence of low numeracy and its effects on risky choice behaviour. | |
| dc.description.librarian | TL (2023) | |
| dc.faculty | Faculty of Commerce, Law and Management | |
| dc.identifier.uri | https://hdl.handle.net/10539/35804 | |
| dc.language.iso | en | |
| dc.phd.title | PhD | |
| dc.rights.holder | University of the Witswatersrand, Johannesburg | |
| dc.school | School of Governance | |
| dc.subject | UCTD | |
| dc.subject | Statistical numeracy | |
| dc.subject | Computational models | |
| dc.subject | Risky choice | |
| dc.subject.other | SDG-8: Decent work and economic growth | |
| dc.title | The role of statistical numeracy in computational models of risky choice | |
| dc.type | Thesis |