Estimating Examinee Intrinsic Difficulty for Providing Greater Specificity of Feedback for Instruction

Estimating Examinee Intrinsic Difficulty for Providing Greater Specificity of Feedback for Instruction

[featured_image]
  • Version
  • Download 108
  • File Size 1.08 MB
  • File Count 1
  • Create Date August 2, 2018
  • Last Updated August 2, 2018

Estimating Examinee Intrinsic Difficulty for Providing Greater Specificity of Feedback for Instruction

Intrinsic difficulties as opposed to proportion correct scores reflect judgments of what students perceive as difficult. By estimating intrinsic difficulties of items, instructors can gain a greater sense of student misconceptions of content and provide more appropriately focused instruction. Initially, think-aloud protocols established judgment categories (or solutionstrategies) for twenty basic fraction and decimal problems. A larger scale administration to 238 high school students in general math classes yielded a one-factor solution for the actual correct or incorrect responses to the 20 items and a four-factor solution for the judgments using exploratory principal components factor analyses. BILOG was used to estimate the parameters of the 20 items using a 2PL model while TESTFACT was used to estimate a compensatory multi-dimensional IRT (MIRT) model of the eighty judgments (4 per item) for which only the slopes of the first dimension was plotted along with item response functions of the actual 20 ,items. Findings indicate that both factor-analytic results and an illustrative example of an IRT and MIRT plot for the misconception of incorrectly cross-multiplying with fractionmultiplication identifies specific intrinsic difficulties where instruction can be strengthened. The ,procedure could have great relevance for students of diverse cultural/ linguistic backgrounds.

Attached Files

FileAction
paper_30e42a58b.pdfDownload