Faded worked examples are a bridge between reading a complete solution and solving a problem alone. Start with a fully explained example, remove one solution step in the next problem, then remove additional steps only after you can supply the missing work correctly and explain why it works. This method is especially useful for students learning multi-step problems in mathematics, physics, chemistry, statistics, accounting, coding, or any course with repeatable procedures.
This guide is for students who understand a solution while looking at it but freeze when the page is blank. You will build a short sequence that reduces help gradually, use a clear rule for advancing, and finish with an independent exam-style problem. The aim is not to copy more solutions; it is to make yourself generate more of each solution at a manageable pace.
The distinction matters because “read one example, then attempt a hard problem” creates a sudden jump. Fading replaces that jump with a ramp. You might first supply only the final calculation, then supply the last two steps, then choose the method and complete the entire solution. Each blank forces retrieval and reasoning while the remaining steps still keep you oriented.
The practical rule: remove help in small enough pieces that you must think, but not so much that you can only guess.
That advantage changes as knowledge grows. Guidance that was useful at the beginning can become redundant when you already know what to do. Fading responds to that change: it gives support early and deliberately removes it before practice becomes comfortable copying.
Pick a skill with a repeatable structure: solving linear equations, balancing chemical equations, calculating confidence intervals, tracing a loop, or preparing a cash-flow statement. Gather four to six problems that use the same underlying method but vary in surface details. If every problem requires a different method, you will not know whether an error came from the target procedure or from choosing among procedures.
Solve the first problem in clearly labeled steps. Put a short reason beside each step, such as “apply the distributive property,” “convert units before substitution,” or “check the loop condition before updating the variable.” Include the information an examiner expects, but remove decorative commentary. Your model should expose decisions that a competent solver makes, not merely show a chain of unexplained equations.
For many procedures, remove the last step first. The learner can follow the supplied path and finish it, then later take responsibility for earlier planning decisions. This backward-fading pattern keeps the goal visible while reducing support. If the hardest learning objective is choosing the method rather than executing it, add a separate comparison prompt at Level 3: “What feature of this problem signals this method?”
Before writing the missing line, answer one prompt: What is the goal of this step? Which rule permits it? What information changes? What mistake would produce a plausible but wrong answer? After writing, predict the next step before revealing it. These prompts make the practice diagnostic: a correct line with a weak explanation shows that recognition may be outrunning understanding.
Move forward after completing two different problems at the current level without looking and explaining the missing decisions accurately. If you need to peek, misidentify the method, or make an early structural error, return to the previous level for one problem. This two-success rule is a practical study threshold, not a universal research constant; adjust it upward for safety-critical calculations or high-stakes professional exams.
Suppose the target skill is solving x² + 5x + 6 = 0 by factoring. The complete model states the goal, finds two numbers that multiply to 6 and add to 5, writes (x + 2)(x + 3) = 0, applies the zero-product property, and concludes x = −2 or x = −3. Beside the factorization, write the reason: the constant terms multiply to 6 and the linear terms combine to 5x.
Provide the factorization and leave the two solutions blank. The learner applies the zero-product property and checks both values in the original equation. The prompt is: “Why can each factor be set equal to zero?”
Give the original equation and the target form (x + __)(x + __) = 0. The learner chooses 2 and 3, finishes the solution, and explains how multiplication and addition constrain the pair. Change the next problem to x² + 7x + 12 = 0 so the learner must use the method rather than remember the previous numbers.
At Level 3, show only a factorable quadratic and ask the learner to choose and justify a method. At Level 4, use a new equation whose leading coefficient is not 1, or mix it with a quadratic that should be solved another way. The final question is not only “Can I factor?” but “Can I recognize when factoring is appropriate without a cue?”
If you can only guess, restore the last removed step or add a cue that names the immediate goal. Productive struggle still includes a plausible route forward. Repeated random attempts are a signal that the fade was too steep, the model was unclear, or prerequisite knowledge is missing.
A sequence should keep the method stable while changing enough details to require thought. If every item differs only by one number in the same position, performance can reflect pattern matching. Vary the wording, representation, or irrelevant details gradually, and finish with a problem that requires deciding whether the learned method applies.
Examples are training wheels, not the destination. Once you can solve two varied problems at a level and explain the decisions, remove more support. Always reserve at least one untouched problem for the independent check; otherwise familiarity with the answer may feel like mastery.
No. They work best wherever a task has visible, repeatable decisions. Students can fade steps in chemical calculations, physics derivations, programming traces, accounting procedures, grammar analysis, or structured essay planning. For open-ended work, fade an outline, evidence-selection prompts, or revision checklist rather than pretending there is one correct final product.
Start by removing the last step when a procedure is new because the supplied earlier work keeps the path and goal visible. Later, remove earlier planning steps so you practice launching the solution. If selecting the method is the main exam difficulty, make method choice an explicit late-stage blank before moving to independent mixed practice.
Begin with one complete model and four to six related problems spread across several fading levels. The exact number should depend on performance, not a fixed quota. Add practice when you still need to peek or cannot explain a decision; move to an unscaffolded problem when you can complete two varied items accurately at the current level.
Compare your attempt with the model and identify whether the problem was a missing prerequisite, wrong method, faulty setup, execution slip, or weak check. Then restore one layer of help and solve a different problem. Do not erase the error immediately: write one sentence explaining what cue you missed and retest that decision later.
Yes. Copy a trustworthy model, duplicate the solution, and cover selected lines with opaque paper or replace them with blank boxes. Prepare the sequence before practice so you cannot reveal answers impulsively. Textbook solutions, instructor exemplars, and verified answer keys are safer starting points than an unverified solution generated from memory.
To study with faded worked examples, move from a complete model to an unfamiliar problem: keep one procedure stable, remove steps gradually, explain each blank, and advance only after accurate no-peek performance. The method closes the gap between “I understand this solution” and “I can produce one myself.” Start with one model and one missing final step; use Snitchnotes to schedule or generate later retrieval checks, then let independent work be the proof.
ノート、クイズ、ポッドキャスト、フラッシュカード、チャット — アップロード1回で全部。
最初のノートを無料で試す