When SDL Beats a Lecture—and When It Does Not

The comparison between self-directed learning and traditional didactic teaching is useful only when the two are judged against the same learning goal. This chapter develops a conditional choice rule for health-professions education: match the method to the target outcome, the learner’s prior knowledge, the task’s complexity, and the amount of practice and feedback required. The evidence does not identify a universal winner. It shows why SDL can be a strong starting point for some knowledge goals while structured teaching or a hybrid design may be more defensible for complex performance.

Compare the outcome before the method

A fair comparison begins by naming what “better learning” means. A test score, a practical performance, a behavior in a clinical setting, and a patient outcome are different destinations. If the destination changes, the criteria for choosing a method change with it. Learner autonomy may be especially valuable for identifying gaps and navigating a large knowledge domain, while a complex procedure may require demonstration, deliberate practice, observation, and correction before the learner can safely direct more of the process.

The broad HPE review offers a useful pattern rather than a final rule. Its moderator estimates were largest for test scores, d = 0.50, smaller for behavioral outcomes, d = 0.23, and smallest for clinical performance, d = 0.08. The outcome-type moderation test was not statistically significant, so these estimates should not be treated as a proven hierarchy. They do, however, make a practical boundary visible: the evidence is more favorable and more direct for measured knowledge than for transfer into complex clinical performance.

That distinction changes the question an educator should ask. If the immediate goal is to understand a body of material, a self-directed design may allow learners to diagnose what they need and choose a productive route through explanations, cases, or retrieval practice. If the goal is to perform a procedure, a resource that explains the procedure may be necessary but insufficient. The design must also create a valid opportunity to practice, observe errors, receive feedback, and demonstrate the required standard.

Prior knowledge matters because it changes the cost of choice. A learner with a stable foundation may be able to compare resources and notice which explanation fills a gap. A novice may not know enough to recognize that the most fluent explanation is incomplete or that a chosen strategy cannot support the target performance. The relevant choice is then not between “full freedom” and “full instruction,” but between different ways of staging responsibility. Curated resources, explicit goals, and early facilitation can provide a scaffold; later choices can become more open as learners acquire the knowledge needed to use them well.

Task complexity creates a similar constraint. A single concept can be explored through self-directed reading. A clinical performance may require simultaneous coordination of knowledge, perception, physical action, communication, and judgment. Removing instruction from such a task does not remove the need for structure. It may simply hide the structure the learner needs. A hybrid design can preserve learner ownership over practice priorities while keeping demonstrations, safety standards, supervision, or feedback non-negotiable.

Modality should be treated as a means rather than a winner.