From worked example to independent attempt: remove one support at a time

Find a small task with visible reasoning

Choose a task whose steps can be inspected: calculating a weighted average, writing a simple filter, or classifying a few examples. Start with a trustworthy worked solution that explains decisions, not just a finished answer. Keep the source and the conditions under which the method applies beside the task.

The IES practice guide recommends alternating worked solutions with problem-solving exercises. Carnegie Mellon's learning principles distinguish component skills, integration, and knowing when to apply them. Together these support examining the reasoning and practicing a task, rather than treating a reproduced demonstration as sufficient evidence of mastery. [1][2]

Annotate the decisions before hiding anything

For each step, write what information it uses and why it follows from the previous step. Mark unfamiliar terms and unresolved assumptions. If you cannot explain a step, consult the source or an appropriate teacher before removing more support. The goal is to understand the method well enough to test it, not to memorize screen positions.

Create a second version with one consequential step omitted. Solve that part and compare it against a key. Choose the missing step deliberately: a calculation, a condition, or a justification. Merely deleting a heading adds difficulty without necessarily testing the skill you intend to learn.

Remove support in response to the attempt

After a successful partial task, try a fresh problem without the full solution. Change something meaningful so you must decide how the method applies. If you get stuck, record exactly what hint you used and retry a comparable problem later. Needing a hint is useful diagnostic information, not a reason to hide the attempt.

A published study of worked examples in Cognitive Tutors compared fixed and adaptive fading in lab and classroom experiments; its abstract reports an advantage for adaptive fading on delayed transfer. That is evidence about those tutoring conditions, not proof that this manually designed sequence duplicates the intervention or its results. [3]

Worked example: an invented weighted average

The fictional example combines two groups: ten items with an average score of four and thirty items with an average score of two. The worked solution converts each group to a total: forty and sixty. Dividing their combined total of one hundred by forty items gives 2.5. The explanation identifies group size as the weight.

For the partial task, the learner supplies the denominator and explains why dividing by two would be wrong. For the independent task, use five items averaging eight and fifteen averaging four: the weighted result is five. Ask the learner to justify the calculation. These invented numbers make the answer checkable; they do not constitute a course assessment or validate a general teaching method.

Use an honest independence record

The worksheet separates worked, partial, independent, and changed-problem attempts. Record the answer, explanation, hint used, error type, and next step. Keep a copied or heavily prompted attempt labeled supported, even if its final answer is correct. Ask a knowledgeable reviewer to check reasoning when the answer key cannot resolve an ambiguity.

Our sequence is a practical suggestion, not a universal rate for removing support. Prior knowledge and task complexity may require more examples or a smaller task. Passing one arithmetic example does not demonstrate competence with real data, hidden assumptions, or unfamiliar domains. Before presenting the skill as independent, attempt another relevant problem and preserve the conditions under which you succeeded.

Sources & verification

Product details and prices can change. Check the linked provider before buying.

1. Institute of Education Sciences: Organizing Instruction and Study; September 2007 Accessed 14 Sep 2026
2. Carnegie Mellon Eberly Center: Learning Principles; undated Accessed 14 Sep 2026
3. Salden and colleagues: Worked examples and tutored problem solving; January 30, 2009; published abstract Accessed 14 Sep 2026