The Reliability of Complex Item Types for Assessing Cognitive Ability

The Reliability of Complex Item Types for Assessing Cognitive Ability

[featured_image]
  • Version
  • Download 60
  • File Size 244.90 KB
  • File Count 1
  • Create Date August 2, 2018
  • Last Updated August 2, 2018

The Reliability of Complex Item Types for Assessing Cognitive Ability

The term ’,reliability’, is most often used as a statistical estimate defining error associated with the administration and scoring of parallel forms of a test, error associated with the re-administration of and scoring of the same test form over occasions, the internal consistency of scores on a form from one administration, or even the error associated with the scores from different judges. However, little has been done methodologically to investigate the reliability of two different tasks which are part of the same assessment form. The present paper is concerned with just that –,estimating the reliability of two different tasks associated with the same form. This paper presents methods of reliability for three different complex innovative item types where the actual item is one component and the other components are: (1) topics associated with the item, (2) solution strategies (intrinsic difficulties) associated with answering the question, and (3) conceptual distances between topics. Reliability was achieved for types (2) and (3) where the associated tasks were uncorrelated with responses to the actual items. Generalizability theory was used for type 1 but correct answers to the items were associated with choosing the topic selected by the instructor.

Attached Files

FileAction
paper_5bc168c1.pdfDownload