Predictors of Omitted Responses on the 2009 NAEP Mathematics Assessment

• In Event: Poster Session 4
  In Poster Session: Division D Section 1 Poster Session

Abstract

Omitted responses, when a student skips over an item, are not ignorable under 2P and 3P IRT models, and must be filled in with best guesses. However, very little empirical evidence exists regarding why students omit responses. Hierarchical multiple regressions were conducted to identify predictors of omitted responses on the 2009 NAEP Mathematics Assessment, using variables from the extensive student, teacher, and school questionnaires and item descriptions. Nearly all predictors, including student proficiency and item difficulty, did not explain substantial variance, with the exception of item format. We hypothesize that omitted responses represent differential item functioning, in which students with sufficient proficiency are unable to respond to certain items due to unfamiliarity resulting from lack of curricular coverage.

Authors

• Nathaniel J.S. Brown, Boston College

• Shenghai Dai, Indiana University - Bloomington

• Dubravka Svetina, Indiana University - Bloomington