Improving and extending models of quantitative judgments

Izydorczyk, David

PDF
izydorczyk_dissertation_PDF14.pdf - Veröffentlichte Version
Download (6MB)

URN:	urn:nbn:de:bsz:180-madoc-639084
Dokumenttyp:	Dissertation
Erscheinungsjahr:	2022
Ort der Veröffentlichung:	Mannheim
Hochschule:	Mannheim
Gutachter:	Erdfelder, Edgar
Datum der mündl. Prüfung:	13 Dezember 2022
Sprache der Veröffentlichung:	Englisch
Einrichtung:	Fakultät für Sozialwissenschaften > Allgemeine Psychologie (Bröder 2010-)
Lizenz:	Creative Commons Namensnennung 4.0 International (CC BY 4.0)
Fachgebiet:	150 Psychologie
Freie Schlagwörter (Englisch):	judgment and decision making , mathematical modeling , psychology
Abstract:	How fast is this car approaching? What is the robability that it will rain today? How severe are the symptoms of this patient? Such quantitative judgments require nferring a continuous criterion from a number of cues or features of the judgment object (e.g., the color of the clouds). Judgments such as these are a central cognitive process which guides our decisions and behavior in our everyday life. For over half a century, researchers are investigating how people make such judgments, which information they rely on, how they combine different types of information, and how the environment or the task affect the processes underlying these judgments by using computational models of the theorized cognitive process. It is the goal of my thesis to improve and extend these models of quantitative judgments. In three articles, I implement and test improved state-of-the art versions of existing models, highlight and solve issues in the way these models are currently used, and extend the scope and possibilities of these models of quantitative judgments. In the first manuscript, I develop, test, and apply a hierarchical Bayesian version of the RulEx-J model, which is used to measure the relative contribution of rule- and exemplar-based processes in people’s judgments. The manuscript shows that the Bayesian RulEx-J model allows to estimate parameters more accurately and how it can be used to test hypotheses about latent parameters. The second manuscript shows that the current practice of not differentiating between direct retrieval of a trained exemplar and genuine judgments in the responses of participants leads to a biased estimation of parameters and reduced fit of exemplar-models. The manuscript also presents a solution to this problem by introducing a latent-mixture extended exemplar model which integrates a direct-recall process of trained exemplars. In the third manuscript, I demonstrate how to model people’s judgments of even complex and realistic stimuli by extracting the necessary cues from pairwise similarity ratings. In sum, the results of the three manuscripts described here contribute to the model-based study of the cognitive processes underlying people’s judgments. By implementing state-of-the-art methods, improving upon current practices, and broadening the scope of the existing research, the results reported in this thesis add to the development, testing, and application of theories of quantitative judgments.