Response Validity of a Reading Motivation Adaptive Assessment

• In Event: Roundtable Session 14
  In Roundtable Session: Effects of Aberrant Response Patterns on Test Scores

Abstract

Generally, it is recommended to conduct data screening to identify invalid responses. The data screening will improve the quality of data and can lead to accurate results. This paper employs three data screening methods to identify invalid responses on a low-stakes assessment data. First two methods identified the number of invalid participants. The third method identified the invalid response, hence the effort-moderated IRT model was applied for comparing the model fit. The data after applying the third method was with lower model fit statistics and with similar but slightly better information than original data. It is clear that more research is needed on data screening method, but the normative threshold method can improve quality of data and model fit statistics.

Authors

• Suk Keun Im, The University of Kansas

• Wenhao Wang, The University of Kansas