Evaluating the performance of alternative blended learning designs using DEA

Abstract The extensive demand for blended learning programs imposes the problem of selecting the most appropriate instructional design from amongst a variety of alternatives that may be feasible for a particular program. The decision-making process should consider a number of qualitative factors such as the satisfaction of learning needs, educational efficiency, ease of implementation and total financial cost. In this paper, we propose that Bates’ (1995) e-learning instructional design model ACTIONS, which describes seven qualitative dimensions pertinent to selecting a design, can be used in conjunction with Data Envelopment Analysis to provide a distinct decision-making framework to aid administrators in determining which blended learning programs are the most effective. The first stage in the analysis is to explain which ACTIONS dimensions can be regarded as inputs and which can be treated as outputs for the sake of the decision process, with all seven dimensions being measurable by ordinal scores assessing the expected performance of alternative designs. In the second stage of analysis, we use Data Envelopment Analysis with ordinal data to obtain an overall expected performance index that is able to discriminate the designs most efficient and most suitable for implementation. The methodology is illustrated by an example. Discussion and Conclusions follow.