Using cognitive science to assign test weights

✍ Scribed by R. Bhaskar; Jesse F. Dillard

• Book ID: 104645573
• Publisher: Springer Netherlands
• Year: 1983
• Tongue: English
• Weight: 375 KB
• Volume: 12
• Category: Article
• ISSN: 0020-4277
• DOI: 10.1007/bf00154128

✦ Synopsis

This article attempts to develop an objective method for assigning weights to questions on examinations. Departing from the statistical discrimination approach of classical test theory, we use a domain-dependent approach.

Our method is developed using cognitive science. Previously formulated methods of task analysis are used to break up a domain into segments that are independent pieces of knowledge. A Pareto distribution is used to assign relative weights. This distribution is selected simply because it is ubiquitous in the social sciences. Word frequency rankings, business firm sizes and ranking scientists by number of publications are all distributions that are skewed, or Pareto distributions. Optimistically, we are hoping that ranking knowledge elements by importance will follow the same distribution.