Assess expertise level before deciding how much guidance to provide

The same support that accelerates novices can slow down intermediates — measure first, design second.

Why it works

Cognitive schemas formed through learning compress previously separate elements into automatized units. When a learner encounters a detailed explanation of something already schematized, the brain must process both the incoming explanation and reconcile it with the existing schema — adding load rather than reducing it. What looks like redundant explanation to the designer is an interference source to the expert. The solution is to calibrate the level of guidance to the actual schema state of the learner.

How to do it

1. Before providing instruction, test what the learner can do without help on the target task.
2. For learners who fail the unsupported test: use high-guidance formats (worked examples, direct explanation).
3. For learners who partially succeed unassisted: shift toward completion problems and faded guidance.
4. For learners who fully succeed unassisted: use independent problem-solving or generative tasks.

Evidence

The expertise reversal effect has been replicated across multiple studies in mathematics, science, and other instructional domains, consistently finding that worked examples and fully guided instruction have decreasing benefits and eventually negative effects as prior knowledge increases. (rct)

Effect sizes and the exact crossover point vary by domain and by the quality of the prior-knowledge assessment; the principle is robust but the precise calibration requires judgment.

Sources

• Kalyuga et al. (2003), "The expertise reversal effect", Educational Psychologist

Common mistake

Using the same instructional format for an entire cohort, assuming all learners are at the same expertise level — producing boredom and interference in advanced learners while leaving novices unsupported.