HR: 1340h
AN: ED13E-1173 [Abstracts]
TI: Mass Producing Concept Inventories
AU: * Garvin-Doxas, K
EM: garvindo@colorado.edu
AF: Center for Integrated PLasma Studies, University of Colorado, Boulder, CO 80309
AU: Klymkowsky, M
EM: klym@colorado.edu
AF: Department of Molecular Cellular and Developmental Biology, University of Colorado, Boulder, CO 80309
AU: Doxas, I
EM: doxas@colorado.edu
AF: Center for Integrated PLasma Studies, University of Colorado, Boulder, CO 80309
AB:
Concept Inventories are research based assessment instruments which derive their validity and reliability from well
researched distracters that represent students' dominant misconceptions in the field. They have formed the backbone of
research based reform efforts in Physics by providing valid, reliable common assessment instruments with which to evaluate
different teaching approaches and materials, and many disciplines are in the process of developing large numbers of Concept
Inventories for their own subject areas. Unfortunately, Concept Inventories are labour and time intensive, with instruments
taking anywhere from 2-8 years to develop, and correspondingly high price tags. The time and cost is directly related to the
fact that valid, reliable instruments require mapping the dominant misconceptions in a field, which is usually a time
consuming and labour intensive task.
This paper will describe how we use Latent Semantic Analysis (LSA) with unsupervised clustering of the LSA vectors to
identify and classify misconceptions in various science disciplines, considerably speeding up the process of misconception
discovery and classification. The paper will present results from Astronomy and Biology, and will describe current efforts to
develop a Concept Inventory for Space Physics.
DE: 0805 Elementary and secondary education
DE: 0810 Post-secondary education
DE: 0840 Evaluation and assessment
SC: Education and Human Resources [ED]
MN: Fall Meeting 2005