Worked Examples and Faded Scaffolding: Supporting Novices Without Creating Dependence

Ask a struggling student to solve unfamiliar problems from scratch and you will mostly teach them that they cannot. Decades of research on the worked-example effect show that novices learn more from studying complete, well-structured solutions than from attempting equivalent problems unaided. The reason is capacity: a novice solving from scratch spends nearly all their working memory on search — trying things, backtracking, guessing — and almost none on learning the underlying method. A worked example removes the search and lets them study the structure directly.

What a good worked example looks like

Not every model solution teaches. A worked example earns its keep when each step is visible, each step is justified, and the decisions are named. 'Multiply both sides by 4' is a step; 'we multiply both sides by 4 to clear the fraction, because equations are easier to handle without fractions' is teaching. When you work an example live, narrate the decision points — the moments where an expert chose between methods — because those choices are exactly what students cannot see in a finished solution.

• Keep the example clean: one method, no shortcuts, no clever tricks on the first pass.
• Pair every worked example immediately with a near-identical practice problem. Study one, attempt one is the pattern with the strongest support.
• Ask the student to explain a step back to you before moving on. Self-explanation is where the example becomes theirs.

The expertise-reversal effect: why yesterday's help becomes today's hindrance

Here is the complication that many teachers miss. The same research programme that established the worked-example effect also found its mirror image: as students gain competence, worked examples stop helping and start hurting. For a student who can already solve the problem, studying someone else's solution is redundant processing — it occupies attention without adding anything, and it denies them the retrieval practice that would actually strengthen the skill. Researchers call this the expertise-reversal effect. The practical translation is blunt: the right amount of support is a moving target, and support that is not withdrawn becomes a ceiling.

Scaffolding is only scaffolding if it comes down. Support that stays up permanently is not scaffolding — it is dependence with good intentions.

Fading: the bridge between showing and doing

The evidence-backed route from full support to independence is the faded example, sometimes called completion problems. Instead of jumping from 'watch me do it' to 'now do it alone', you remove the solution one piece at a time, usually from the end backwards, because the final steps are the easiest for a student to supply.

1. Full worked example: every step shown and explained; the student's job is to study and self-explain.
2. Completion stage one: the example stops before the final step, and the student supplies it.
3. Completion stage two: only the setup and the first step are given; the student completes the rest.
4. Guided start: the student begins alone but you provide the plan — 'this is a two-stage problem, deal with the brackets first'.
5. Independent practice: full problems, no support, with mixed variations of the method.

A weekly fading plan for tutors

In a one-to-one setting this maps naturally onto weeks. Week one on a new topic: two or three full worked examples with self-explanation, each paired with a matching attempt. Week two: completion problems, with the student supplying progressively more of the solution. Week three: independent problems with varied surface features, and worked examples used only as repair when an error pattern appears. Week four: the topic appears mixed into general practice, unannounced, which tests whether the student can recognise the method as well as execute it.

The judgement call is when to move a student along the sequence, and the honest answer is: earlier than feels comfortable. Two clean, explained completions is usually enough to advance; five is coddling. Watch for the tell-tale sign of over-support — the student who executes flawlessly when you set up the problem but freezes when the setup is theirs. That student has been left too long at the wrong stage.

The trap of the brilliant explainer

Tutors are often hired for their explanations, and parents rate lessons by how clearly the tutor explained. This creates a quiet incentive to stay at the worked-example stage forever, where lessons feel smooth and the student nods along. But a lesson where the tutor performs beautifully and the student watches is a lesson optimised for the wrong person's learning. The measure of scaffolding is never how good the support was — it is how completely, and how soon, it became unnecessary.