Validations of the Subdividing Method
As described in the previous chapter, the volume of examinees, number of raters, and examinee- and task- to-rater assignment determined the amounts and patterns of missing data and thus influence the estimation of measurement errors. How does each of these factors impact the precision of individual variance components (measurement errors for scoring performance-based assessments)? How can one use the subdividing method to alleviate the potential inaccuracy and imprecision caused by unbalanced data? Which variance components are influenced by the increase in sample size, and which variance components are influenced by the expansion of rater pool size? These questions are central to this chapter.
Unable to display preview. Download preview PDF.