A multifaceted approach to investigating the equivalence of computer-based and paper-and-pencil assessments: an example of reading diagnostics.

The purpose of this study was to present a multifaceted approach to investigating the equivalence of the computer-based and the paper-and-pencil tests and exemplify it using the Stanford Diagnostic Reading Tests. The multifaceted approach includes construct equivalence evaluation, test-level comparison and item-level comparisons, especially differential functioning of items across the two modes. The results show that less than 1% of items in grades 3 and 4 were flagged with DIF, though about 5% of the items in grade 2 were flagged with DIF. The CBT and the PPT for the same grade test shared similar test characteristics. The mode did not affect the raw score to scale score conversion relationship significantly. The underlying constructs between the two modes of the same grade test satisfied the three levels of construct equivalence, i.e., congeneric, τ-equivalent and parallel equivalent. [ABSTRACT FROM AUTHOR]