Improving the measurement and understanding of response process heterogeneity by IRTree modeling

Merhof, Viola

PDF
Dissertation_Viola_Merhof.pdf - Veröffentlichte Version
Download (3MB)

URN:	urn:nbn:de:bsz:180-madoc-672964
Dokumenttyp:	Dissertation
Erscheinungsjahr:	2024
Ort der Veröffentlichung:	Mannheim
Hochschule:	Universität Mannheim
Gutachter:	Wetzel, Eunike
Datum der mündl. Prüfung:	21 Mai 2024
Sprache der Veröffentlichung:	Englisch
Einrichtung:	Fakultät für Sozialwissenschaften > Psychologische Methodenlehre u. Diagnostik (Meiser 2009-)
Lizenz:	Creative Commons Namensnennung 4.0 International (CC BY 4.0)
Fachgebiet:	150 Psychologie
Freie Schlagwörter (Englisch):	IRTree models , item response theory , response styles , multidimensional item responding , ideal point models
Abstract:	Self-reported rating responses provide valuable information on the characteristics of a person that cannot be observed directly, and numerous psychometric approaches are available to derive individual trait estimates from the given item responses. However, the mapping of latent trait levels and manifest answers may be biased if, in addition to trait-based responding, trait-unrelated response processes such as response styles affect the category selection. This bias may further be exacerbated by additional heterogeneity of the involved response processes, such as when the relevance of latent characteristics for the respondents’ judgments differs between measurement units. Heterogeneous response processes may result in distorted trait measurements and misinterpretations of cognitive processes underlying item responding whenever the complexity of the respondents’ behavior is not adequately reflected in the analysis model. In this thesis, I demonstrate how item response tree (IRTree) modeling can be used to address the heterogeneity of response processes. In the first article, I focus on heterogeneity with regard to the person characteristics on which the response processes are based, and I show how IRTree models can be specified to effectively disentangle the influences of traits and response styles from each other. In the second article, I examine systematically changing influences of response processes throughout a questionnaire, and I develop dynamic IRTree models that incorporate such changes. In the third article, I consider that the involved response processes may adhere to heterogeneous item response functions, and I propose a general IRTree framework that can incorporate the combined influences of both dominance and ideal point response processes. The conducted research highlights the importance of modeling heterogeneous response processes to enhance the measurement of latent characteristics and to foster the interpretability of model parameters. Thereby, the flexibility of the IRTree model class for addressing heterogeneous response processes is illustrated and extended by new developments. Overall, this dissertation contributes to the field of psychometrics by providing a tool to improve the measurement and the understanding of response process heterogeneity, and thus, to increase the validity of assessments through rating scales.