Cluster Analysis as a Tool for Qualitative Research: The Case of Scale Construction

• In Event: Poster Session 2
  In Poster Session: Achievement, Attitudes, and Advocacy: Science Literacy for All

Abstract

This paper presents a statistical technique – cluster analysis – useful in classifying the different ways of thinking about a subject. It is a statistical technique that can complement phenomenographic research where the topic of study is measurable. We report on a cluster analysis of 94 undergraduate students’ conceptions of the size of scientific objects, using a scale construction task. A four-cluster solution yields the four distinct types of scale identified by prior research: fragmented, ordinal, linear, and logarithmic/proportion. The cluster analysis alone cannot distinguish between logarithmic and proportional approaches, as they are mathematically equivalent. However, this study shows that cluster analysis can be used to help build categories, and also extends a recently published typology of conceptions of size.

Authors

• Cesar Delgado, The University of Texas - Austin

• Noah Ledbetter