A picture paints a thousand numbers – and tells a story. A postpositivist approach to building substantive grounded theory using decision tree analysis

Abstract

This paper illustrates how decision tree analysis, using quantitative data normally associated with deductive methodologies, can be employed to inductively generate substantive grounded theory. Within the context of academic disadvantage and development, a practical demonstration is provided of how theoretical explanations for student performance may be generated. Specifically, the chemistry, physics and biology modules of a science foundation programme are examined using classification and regression tree analysis (CART) towards an understanding of the factors that best facilitate a pass in each of these modules. Together with research published elsewhere on performance in the mathematics module, and the foundation programme as a whole, the findings allow a number of 'conceptualisations' to emerge that may become the focus for theory generation. Such a substantive theory can contribute towards understanding, on a broader scale, of what is needed to enhance student persistence, and improve retention and throughput rates in Science, Technology, Engineering and Mathematics (STEM) in South African higher education.