Rule-based item construction: Analysis with and comparison of linear logistic test models and cognitive diagnostic models with two item types

The dissertation focuses on demonstration and evaluation of rule-based item construction of figural reasoning items and mathematical word problems and application as well as comparison of LLTMs and CDMs as statistical analysis methods. Results show Rasch scalability of items, confirm the importance of the chosen basic parameter sets and demonstrate precise item construction and analysis processes. It is shown how LLTM and its variants can contribute substantial insights into cognitive solution processes and composition of item difficulty in relational reasoning and mathematical word problems and also for item cloning and longitudinal data. However, CDM application detects severe modeling problems and misfit. Application hints regarding test item construction as well as statistical model application and interpretation of results for practitioners and researchers are pointed out.