Logo: Deutsches Institut für Internationale Pädagogische Forschung

Publications

Practical significance of item misfit in educational assessments

The article issues the practical significance of item misfit in educational assessments.

You can only make valid assumptions for psychometric tests backed by item response theory (IRT) measurement models, if the measurement model used is suitable for the collected data. The suitability of the items used during testing (e.g. exercises or questions) for available test data is verified by corresponding fit statistics. Mismatches generally lead to the exclusion of items. What hasn’t been considered so far is whether the modified item pool does indeed result in a modification of the outcome and thus whether the correction is of practical use for e.g. correlation analyses. The article presents a method that allows the quantification of this question.

Köhler, C. & Hartig, J. (2017). Applied Psychological Measurement, 41(5), 388–400.

Available here:

last modified Sep 28, 2018