In educational and psychological assessments, attending to item response process can be useful in understanding and improving the validity of measurement. This dissertation consists of three studies each of which proposes and applies item response theory (IRT) methods for modeling and understanding cognitive/psychological response process in assessment. The first study presents a noncompensatory multidimensional IRT model that reflects underlying components for solving passage-based items in reading comprehension tests. The second study proposes a mixture item response tree (IRTree) model that accommodates the possibility of a mixture of respondents exhibiting different response processes in responding to rating scale items. The third study examines systematic differences in response process across fast and slow item responses by attending to both content trait and response style and their possible differential influences across response types. Each study introduces different modeling approaches for incorporating item response process and discusses their practical implications. The dissertation as a whole contributes to understanding item response process in both cognitive and noncognitive assessments.