The research we still are not doing: an agenda for the study of computer-based learning.

Media-comparative research-that is, the comparison of computer-based learning (CBL) to noncomputer instruction-is logically impossible because there are no valid comparison groups. Results from media-comparative studies are thus confounded and difficult to meaningfully interpret. In 1994, Friedman proposed that such research be supplanted by investigations into CBL designs, usage patterns, assessment methods, and integration. His proposal appears to have largely been ignored. In this article, the author updates the agenda for research in CBL (including Web-based learning). While media-comparative studies are confounded, CBL-CBL comparisons are often not. CBL instructional designs vary in configuration (e.g., discussion board or tutorial), instructional method (e.g., case-based learning, personalized feedback, or simulation), and presentation (e.g., screen layout, hyperlinks, or multimedia). Comparisons within one level (for example, comparing two instructional methods) facilitate evidence-based improvements, but comparisons between levels are confounded. Additional research questions within the CBL-CBL framework might include: Does adaptation of CBL in response to individual differences such as prior knowledge, computer experience, or learning style improve learning outcomes? Will integrating CBL with everyday clinical practice facilitate learning? How can simulations augment clinical training? And, how can CBL be integrated within and between institutions? In addressing these questions it is important to remember the most important outcome-effect on patients and practice-and outcomes specific to CBL including costs, cognitive structuring, and learning unique to the computer-based environment. CBL is not a panacea, but holds great promise. Realization of this potential requires that media-comparative studies be replaced by rigorous, theory-guided comparisons of CBL interventions.