This study looked at how three different analysis methods could help us to understand rater effects on exam reliability. The techniques we looked at were:
- generalizability theory (G-theory)
- item response theory (IRT): in particular the Many-Facets Partial Credit Rasch Model (MFRM)
- multilevel modelling (MLM)
We used data from AS component papers in geography and psychology for 2009, 2010 and 2011 from Edexcel.