There is a special issue of Computers in Human Behavior on learning from video and in their Editorial Fiorella and Mayer give an overview of effective and ineffective methods that are being trialed in the special issue:
What are the effective methods?
…two techniques that appear to improve learning outcomes with instructional video are segmenting—breaking the video into parts and allowing students to control the pace of the presentation—and mixed perspective—filming from both a first-person perspective and third-person.
And what isn’t worth the effort?
…some features that do not appear to be associated with improved learning outcomes with instructional video are matching the gender of the instructor to the gender of the learner, having the instructor’s face on the screen, inserting pauses throughout the video, and adding practice without feedback.
Abstract of the editorial:
In this commentary, we examine the papers in a special issue on “Developments and Trends in Learning with Instructional Video”. In particular, we focus on basic findings concerning which instructional features improve learning with instructional video (i.e., breaking the lesson into segments paced by the learner; recording from both first- and third-person perspectives) and which features or learner attributes do not (i.e., matching the instructor’s gender to the learner’s gender; having the instructor’s face on the screen; adding practice without feedback; inserting pauses throughout the video; and spatial ability). In addition, we offer recommendations for future work on designing effective video lessons.