[quoting Morrison, 24 Oct 96.a] As for the research, I will reiterate what Steve Ross and I said in the first issue of Educational Technology Research & Development, let's focus on which strategies work best in which medium (media replication studies) rather than trying to say which medium is best. It is the instructional strategy that we can use, replicate, and modify.
Different media have different power, depending upon instructional objectives. For example, from 40-plus years (and nearly 5,000 studies) in social learning theory research, we can be fairly confident that behavioral modeling/observational learning will generally be more effective to produce the distal goal of behavior change than will a purely cognitive approach (e.g., using just text). A rich literature across a wide range of activities supports the efficacy of symbolic modeling (video, graphics, etc.) in learning new behaviors (see Bandura, 1986, Social Foundations of Thought & Action: A Social Cognitive Theory, among many others) relative to other instructional approaches.