Learning progression over multiple time points: Dynamic Bayesian Inference Network
Theories:
- learning theory and dev'l psych
- probabilty based statistical modeling framework
- longitudinal design framework
Learning Progression: National Research Council 2007, Shin et al 2009
provides diagnostic information regarding strength and weakness of student's understanding along a curriculum
Challenges
- How to dev'l assessment to elicit student performances?
- How can student performance be modeled
- How to provide feedback from tests
Design of study
Evidence centered deisgn (Mislevy 2003)
How can students inconsistent levels and patterns be explained and modeled?
Non-linear sequence of change by longitduanla accounts of student leanring beyond sors-ssectional approach
- Latent class model
- Diagnostic class model
- rule space model
- attributation hierarchial model
- hidden markov model
Bayesian Inference network
probability theory + graph theory
- Observation: categorical variable
- Level change over time = qualitative growth
- Qualitative growth is addressed by proficiency change within same student over time
