Predicting Performance and Creating Better Student Proficiency Models by Improving Skill Codings

Interest in end-of-year accountability exams has increased dramatically since the passing of the NCLB law in 2001. This push has impacted educational research in a wide variety of ways, including a strong desire to be able to model student work in order to make conclusive statements about what students know and how this relates to how they will perform on end-of-year standardized exams. This thesis will look at using item response theory (IRT) to estimate student proficiency. This estimated proficiency will then be used to build prediction models for end-of-year exam scores. Next, methods to improve a skills model will be explored. Models that account for learning over time will then be considered. Finally, I will compare various different approaches to modeling response data.