Open AccessA GENERATIVE APPROACH TO THE DEVELOPMENT OF HIDDEN‐FIGURE ITEMS
Author(s) -
Bejar Isaac I.,
Yocom Peter
Publication year - 1986
Publication title -
ets research report series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.235
H-Index - 5
ISSN - 2330-8516
DOI - 10.1002/j.2330-8516.1986.tb00175.x
Subject(s) - dimension (graph theory) , item response theory , process (computing) , psychology , artificial intelligence , computer science , test (biology) , generative grammar , cognitive psychology , item analysis , machine learning , psychometrics , mathematics , developmental psychology , paleontology , pure mathematics , biology , operating system
This report explores an approach to item development and psychometric modeling which explicitly incorporates knowledge about the mental models used by examinees in the solution of items into a psychometric model that characterize performances on a test, as well as incorporating that knowledge into the item development process. The paper focuses on the hidden figure item type. Although there is an extensive literature on the correlates of performance for this type of item little is known about the mental models that may explain performance on the item. The approach taken in this paper is to search for a complexity dimension that accounts for the difficulty of hidden figures. Although several complexity dimensions can be postulated we chose one inspired by artificial intelligence research on vision. A computer‐based system was developed to analyze as well as generate items based on this framework. To empirically determine the validity of the chosen framework two experiments were conducted. The results suggest that this approach to psychometric modeling is viable. The practical and theoretical implications of the research are discussed.