Key ideas
Core concepts
- Novices learn better from examining worked solutions than from attempting problems independently (Atkinson et al., 2000; Sweller & Cooper, 1985).
- Structure determines effectiveness: examples must minimise extraneous cognitive load through careful design, and poorly structured examples can harm learning.
- Persist far longer than intuition suggests, continuing until complete familiarity rather than basic understanding. Typical abandonment happens far too early.
- Alternate one example with one near-identical practice problem; do not bundle several examples before practice.
Worked examples are step-by-step demonstrations of complete solution procedures with explanations, designed to reduce cognitive load and enable efficient schema construction by showing novice learners how experts solve problems. When students lack schemas for content, problem-solving consumes working memory through trial-and-error rather than schema construction. Worked examples reduce this cognitive load by presenting complete solution paths, allowing students to focus on understanding the method rather than generating solutions. Effective use requires understanding their design principles and duration of use.
Connected to
Worked-Example Effect | Cognitive Load Theory | I Do | We Do | Explicit Teaching | Scaffolding | Self-Explanation Effect | Split-Attention Effect | Redundancy Effect | Expertise Reversal Effect
The worked-example effect
The worked-example effect is a finding in educational psychology demonstrating that novices learn more efficiently from studying worked solutions than from attempting to solve problems independently. Sweller and Cooper’s seminal studies established the effect through carefully designed experiments (Sweller & Cooper, 1985). In the problem-solving condition, students received problems to solve with minimal guidance, learning through trial and error and teacher feedback. In the worked-example condition, students studied complete worked solutions showing step-by-step solution paths, then immediately practised near-identical problems. Students in the worked-example condition learned solution procedures faster, made fewer errors during learning, transferred knowledge to novel problems more successfully, and required less instructional time to reach competence. These advantages appear for novice learners; the effect reverses as expertise develops, so instruction should adapt to schema development.
Cognitive Load Theory explains why. When novices lack solution schemas, they must use means-end analysis: trying approaches, checking if they’re getting closer to the goal, backtracking when they’re not (Sweller et al., 1983). This search process consumes extensive working memory resources: holding the problem goal in mind, remembering what’s been tried, evaluating whether attempts approach the solution, deciding what to try next, and managing frustration when attempts fail. After means-end analysis fills working memory, minimal capacity remains for the actual purpose, constructing schemas for the solution method. Students might eventually solve the problem, but learning the generalisable procedure fails. By presenting complete solutions, worked examples remove means-end analysis from working memory demands. Students can devote full cognitive resources to understanding the solution method, processing each step, seeing how it connects to the overall goal, and integrating the procedure into developing schemas. The result is more efficient schema construction with less cognitive strain, faster learning, and better transfer.
Despite research clarity, most classroom implementations misunderstand the effect. Teachers typically present two or three worked examples, explain them, then ask students to solve a set of related problems. A different pattern works better: present one worked example, have students immediately solve a near-identical problem, and repeat this alternation multiple times (McLaren et al., 2008; Trafton & Reiser, 1993). Showing multiple examples before practice is the problem-solving condition: students must search through memory to recall which example applies and how to adapt it. True worked examples alternate individual examples with immediate practice, maintaining focus whilst schemas are still fresh in working memory. This misunderstanding explains why many teachers report worked examples “don’t work”: they never implemented the approach the research supports.
Structure
Worked examples must be designed to minimise every source of extraneous cognitive load, freeing working memory capacity for learning the solution method. Poorly designed worked examples can be less effective than problem-solving: when examples impose cognitive burdens through split-attention, redundancy, or transient information, the advantage diminishes (Chandler & Sweller, 1991).
Eliminate split-attention by integrating text explanations directly with the mathematical or visual content they explain (Chandler & Sweller, 1992). Putting the explanation at the top of the page with the working below forces students to hold the explanation in working memory whilst searching for the relevant part of the working. Instead, place each step’s explanation immediately adjacent to it: “We multiply both sides by 4” appears right next to the step showing this multiplication.
Avoid redundancy: don’t speak an explanation whilst it’s written for students to read (Sweller et al., 2019). This forces working memory to process identical information through two channels, creating interference rather than reinforcement. Show the written worked example for students to read at their own pace, or explain verbally whilst students listen, not both simultaneously.
Prevent transience by providing permanent written versions of examples (Leahy & Sweller, 2011). Information that disappears before processing completes, such as examples shown only on a board that gets erased or verbal explanations without written reference, forces students to hold it in working memory whilst trying to process subsequent steps. Printed worked examples can be studied repeatedly, referred back to, and reviewed later.
Chunk complex procedures by breaking solution paths into visually distinct steps, each representing one meaningful decision or operation. Present them like IKEA instructions or comic panels: discrete, digestible chunks rather than a dense paragraph explaining the entire solution in running text. Numbered or boxed steps, each showing one operation with its brief explanation, create a clear progression students can follow.
Persistence
Continue providing worked examples far longer than traditionally thought necessary, until students demonstrate complete familiarity and automaticity, not just basic understanding. Teachers abandon examples too early because expert teachers possess automated schemas that make procedures feel simple. Their intuition about when students are “ready” to practise independently reflects expert perception, not novice reality. The point where worked examples feel unnecessary to teachers often marks when students still need them most.
Persisting means aiming for complete familiarity: students should reach the point where encountering a problem type triggers automatic schema activation, not conscious recall of the method. This requires extensive exposure, far more than the 2-3 examples teachers typically provide. Base the decision on evidence: continue until success rates on practice problems exceed 80% consistently, since lower success indicates schemas aren’t yet robust and more worked examples would benefit learning. Don’t jump from worked examples directly to independent problem-solving; use completion problems (partially worked examples where students complete missing steps) to bridge the transition, maintaining support whilst building independence. Continuing worked examples when it feels like students “should be ready” requires trusting research over intuition. The discomfort teachers feel at “not challenging students enough” often indicates the right level of support for schema construction.
Content selection
When teaching procedures, use numbers that minimise calculation burden so cognitive resources focus on the method, not arithmetic. If students must consciously calculate whilst learning a procedure, working memory divides between the calculation and the method, and the procedure gets shortchanged. Teaching fraction addition, use fractions with easily found common denominators ( rather than ). Teaching algebraic manipulation, use small whole numbers ( rather than ). Teaching substitution, choose values that make arithmetic straightforward. Once procedures are automated, increase arithmetic complexity to build robustness. During initial learning, simple arithmetic is essential, not “dumbing down.”
Start with general examples, not “simple” cases. Use examples that require executing all steps of the general procedure, even if this seems more complex initially. Examples like for expanding brackets hide the general procedure because the numbers create patterns not representative of the general case. Students learn the specific pattern (middle term is double the constant) rather than the general method (multiply everything in the first bracket by everything in the second). When they later encounter , students who learnt from lack the general procedure and must learn from scratch; the “simple” example wasted time. Better to use examples like that require full execution of the general procedure (Paas & van Merriënboer, 1994): first terms , outer terms , inner terms , last terms , combining to . All steps are visible, the general procedure is clear, and there are no hidden special cases.
Avoid misleading number patterns. Patterns in worked examples powerfully influence what students learn, so if examples accidentally contain patterns not representative of the general procedure, students construct defective schemas. Teaching exponent rules with , where the base and exponent matching the result creates a coincidental pattern, may lead students to conclude the pattern holds generally. Use varied numbers that prevent pattern over-learning whilst keeping arithmetic manageable: , , , with no false patterns and the general principle visible.
Visual design
Students need both the mathematical working and explanations of what each step accomplishes, but separating them creates split-attention. A poor layout:
Steps:
1. Multiply both sides by 4
2. Simplify to get x = 12
Working:
x/4 = 3
x = 12
Students must look back and forth, losing context with each shift of attention. A better layout puts each explanation exactly where it’s needed:
x/4 = 3
x/4 × 4 = 3 × 4 ← Multiply both sides by 4
x = 12 ← Simplified
Students must also distinguish the problem they’re solving from the solution method being demonstrated. Different colours (problem in black, working in blue), different fonts or sizes, boxes or borders separating problem from solution, and clear “Question:” and “Solution:” labels all reduce the working memory burden of tracking which part they’re reading and prevent confusion between problem and solution.
Long procedures overwhelm when presented as continuous text; students lose track of progression. Chunk them like instruction manuals or comic strips, with discrete, visually separated steps students process one at a time. Number each major step, use boxes, borders, or spacing to isolate steps visually, make each chunk one decision or operation, and reveal progressively where useful (show one step, discuss, show the next):
Step 1: Expand the brackets
(2x + 3)(x - 5) = 2x² - 10x + 3x - 15
Step 2: Collect like terms
2x² - 10x + 3x - 15 = 2x² - 7x - 15
Step 3: Check by substitution
(x = 0): 2(0)² - 7(0) - 15 = -15 ✓
Each step is digestible independently and the overall progression stays clear.
Types of examples
When teaching concepts (rather than procedures), students need to discern critical features from irrelevant ones, so show what something is and what it isn’t. Teaching prime numbers: examples 2, 3, 5, 7, 11, 13 (with the explanation: exactly two factors) alongside non-examples 1 (only one factor), 4 (three factors: 1, 2, 4), and 6 (four factors). Teaching perpendicular lines: various orientations of 90° intersections alongside parallel lines, acute angle intersections, and lines that don’t meet. Contrasting examples and non-examples helps students identify the critical attributes (exactly two factors, 90° angles) and ignore irrelevant features (which numbers, line orientation).
Students require multiple examples to extract the general procedure from specific instances: at least 2-3 per problem type, but typically 5-8 for robust schema construction. Change only one critical feature between consecutive examples so students see what varies and what stays constant. Teaching solving linear equations: (basic structure), (coefficient changes), (constant changes), (operation changes). Each variation reveals which features matter and which are incidental to the general method.
Sequences can also be built so patterns become visible. Teaching subtracting negative integers: , , , . The consistent result (7) makes the pattern visible: subtracting a negative adds its absolute value. The final example (starting from 0) clarifies the effect.
General schemas must also include understanding of where procedures apply and where they don’t. Include worked examples that look routine but have unexpected features, represent boundaries of procedure applicability, or demonstrate common error situations. When teaching polynomial division, include examples where the divisor doesn’t go in evenly (remainder exists), where terms are missing in the polynomial (placeholders needed), and where the leading coefficient isn’t 1. Students learn the full scope of the procedure, not just the easy cases.
Classroom implementation
The core implementation is alternation: present one worked example, immediately have students practise one near-identical problem, and repeat. The schema is fresh in working memory during practice, immediate application strengthens encoding, success builds confidence for the next example, and working memory is never overloaded by multiple examples. A typical sequence: teacher demonstrates example 1; students practise problem 1 (near-identical to example 1); check answers and address errors; teacher demonstrates example 2 (slight variation); students practise problem 2; continue alternating until the procedure is automatic. Showing 3 examples then giving 10 practice problems is the less effective problem-solving condition.
Make examples permanent. Worked examples shown only on the board or explained only verbally create transient information that disappears before students finish processing. Provide printed booklets students can study at their own pace, refer back to during practice, review when uncertain, and keep for future reference; or worksheets with worked examples at the top and practice problems below, all on the same page. Avoid expecting students to copy worked examples whilst you explain: copying divides attention between listening and writing, so students miss the explanation whilst writing, or miss the working whilst listening.
Label subgoals explicitly to make the procedure’s structure visible:
Goal: Solve for x
Subgoal 1: Eliminate the fraction
x/4 + 3 = 7
x/4 = 4 (subtract 3 from both sides)
Subgoal 2: Isolate x
x = 16 (multiply both sides by 4)
This makes the goal hierarchy explicit, helps students organise their understanding, improves transfer to novel problems, and supports self-explanation. Subgoal-labelled examples produce better learning and transfer than unlabelled examples, especially for complex procedures (Catrambone, 1998).
After studying a worked example, have students explain to themselves (not peers) why each step works (Chi et al., 1989; Renkl, 1997): “Read through this worked example. Now explain to yourself why each step was necessary. What would happen if we skipped step 2?” For novices, explaining to peers creates additional cognitive load from social interaction, audience consideration, and managing the explanation process; self-explanation provides the benefits without the extraneous load. Use it after students have studied multiple examples and have partial understanding. Self-explanation too early, when schemas barely exist, frustrates because students lack knowledge to explain. Done at the right time, it deepens understanding of underlying principles, connects procedural knowledge to conceptual understanding, reveals gaps, and increases germane cognitive load when students are ready for it.
Fading to independence
Worked examples shouldn’t continue forever; expertise development requires transitioning to independent problem-solving, and the transition method matters. Fade gradually through completion problems, partially worked examples where students complete missing steps (Atkinson et al., 2003; Renkl & Atkinson, 2003). The progression runs from a full worked example, to completing the final step, to completing the last two steps, to solving completely:
Stage 1 (full worked example):
Solve: 3x + 5 = 17
3x = 12 (subtract 5)
x = 4 (divide by 3)
Stage 2 (complete final step):
Solve: 3x + 5 = 17
3x = 12 (subtract 5)
____ (now you complete)
Stage 3 (complete last two steps):
Solve: 3x + 5 = 17
____ (now you complete)
____
Stage 4 (complete independently):
Solve: 3x + 5 = 17
Completion problems maintain support whilst building independence (van Merriënboer, 1990). Each stage reduces scaffolding gradually based on developing competence, not arbitrary timelines.
Base fading decisions on evidence, not planned timelines. Success rates above 80% suggest reducing support: students demonstrating consistent success probably have robust enough schemas for less guidance. Below 80%, maintain or increase support. This requires systematic checking of practice problem success rates rather than assumptions about readiness. The common error is fading because “we’re on lesson 3, time to move to independent practice” rather than because of evidence of schema development.
The expertise reversal effect
Worked examples benefit novices but hinder experts; optimal instruction must adapt to developing competence (Kalyuga et al., 2003). For novices, worked examples reduce cognitive load by eliminating means-end analysis, freeing working memory for schema construction. Students with robust schemas don’t need means-end analysis: they recognise problem types and activate appropriate procedures automatically. Forcing them to study worked examples wastes time processing information they’ve automated, bores students unchallenged by material they’ve mastered, slows down their automated processing, and prevents the productive struggle that strengthens schemas.
Students in the same class often possess different levels of expertise, requiring different approaches. Flexible grouping lets novices study worked examples whilst developing experts solve problems independently. Completion problems serve the middle ground: students with partial expertise receive support where needed whilst practising where ready. Self-paced progression allows students to move from examples to practice based on demonstrated competence, not class-wide timelines. Forcing all students through identical worked examples regardless of expertise wastes advanced students’ time whilst potentially still rushing novices.
Common pitfalls
Using “simple” examples that hide the procedure. Choosing examples like because they “seem easier” fails because special cases hide the general procedure: students learn the specific pattern rather than the method, then fail when encountering general cases. Use examples requiring full execution of the general procedure, even if this seems initially harder. Students learn the complete method once rather than learning twice (special case then general case).
Redundant verbal and written explanations. Reading aloud what’s written on worksheets whilst students read along creates the Redundancy Effect, forcing working memory to process identical information through both visual and auditory channels; the redundancy creates interference, not reinforcement. Present information once in the most effective modality.
Complex arithmetic distracting from the procedure. Using numbers that make arithmetic difficult in the belief that this “challenges” students divides working memory between arithmetic and method, and the procedure gets inadequate processing. Keep arithmetic simple during initial learning and increase complexity only once the procedure is automated.
Abandoning examples too early. Providing 2-3 examples then moving to independent practice underestimates how many examples students need for robust schema construction; expert perception (“this is simple”) blinds teachers to novice reality (“this is complex and overwhelming”). Persist far longer than feels necessary, using an 80%+ success rate on practice problems as the evidence-based indicator of readiness rather than intuition or curriculum pacing guides.
Bundling examples before practice. Showing multiple examples then asking for practice, the typical classroom implementation, is the problem-solving condition, not the worked-example effect. Students must search memory for which example applies and adapt it, precisely the means-end analysis worked examples should eliminate. Alternate individual examples with immediate practice on near-identical problems: one example, one practice, repeat.
Summary
Worked examples are an evidence-based instructional strategy in which novices learn from studying worked solutions rather than attempting problems independently (Atkinson et al., 2000; Sweller & Cooper, 1985). Problem-solving consumes working memory through means-end analysis (Sweller et al., 1983), whilst worked examples eliminate this burden, allowing cognitive resources to focus on understanding methods and constructing schemas.
Effective implementation rests on two principles. Structure: examples must minimise extraneous cognitive load through integration of text and working (Chandler & Sweller, 1992), elimination of redundancy (Sweller et al., 2019), and clear visual design. Persistence: examples should continue until automation develops, not merely basic understanding. Implementation involves specific design choices: simple arithmetic that focuses attention on procedures, general examples requiring full method execution (Paas & van Merriënboer, 1994), and gradual fading through completion problems (Atkinson et al., 2003; Renkl & Atkinson, 2003). The expertise reversal effect indicates that instruction should adapt as competence develops (Kalyuga et al., 2003).
Common errors include using special-case examples that hide general procedures, creating redundant explanations, allowing complex arithmetic to distract from methods, abandoning examples prematurely, and presenting multiple examples before practice rather than alternating example and practice.
References
Atkinson, R. K., Derry, S. J., Renkl, A., & Wortham, D. W. (2000). Learning from examples: Instructional principles from the worked examples research. Review of Educational Research, 70(2), 181-214. https://doi.org/10.3102/00346543070002181
Atkinson, R. K., Renkl, A., & Merrill, M. M. (2003). Transitioning from studying examples to solving problems: Effects of self-explanation prompts and fading worked-out steps. Journal of Educational Psychology, 95(4), 774-783. https://doi.org/10.1037/0022-0663.95.4.774
Catrambone, R. (1998). The subgoal learning model: Creating better examples so that students can solve novel problems. Journal of Experimental Psychology: General, 127(4), 355-376. https://doi.org/10.1037/0096-3445.127.4.355
Chandler, P., & Sweller, J. (1991). Cognitive load theory and the format of instruction. Cognition and Instruction, 8(4), 293-332. https://doi.org/10.1207/s1532690xci0804_2
Chandler, P., & Sweller, J. (1992). The split-attention effect as a factor in the design of instruction. British Journal of Educational Psychology, 62(2), 233-246. https://doi.org/10.1111/j.2044-8279.1992.tb01017.x
Chi, M. T. H., Bassok, M., Lewis, M. W., Reimann, P., & Glaser, R. (1989). Self-explanations: How students study and use examples in learning to solve problems. Cognitive Science, 13(2), 145-182. https://doi.org/10.1207/s15516709cog1302_1
Kalyuga, S., Ayres, P., Chandler, P., & Sweller, J. (2003). The expertise reversal effect. Educational Psychologist, 38(1), 23-31. https://doi.org/10.1207/S15326985EP3801_4
Leahy, W., & Sweller, J. (2011). Cognitive load theory, modality of presentation and the transient information effect. Applied Cognitive Psychology, 25(6), 943-951. https://doi.org/10.1002/acp.1787
McLaren, B. M., Lim, S., & Koedinger, K. R. (2008). When and how often should worked examples be given to students? New results and a summary of the current state of research. In Proceedings of the 30th Annual Conference of the Cognitive Science Society (pp. 2176-2181). Cognitive Science Society.
Paas, F., & van Merriënboer, J. J. G. (1994). Variability of worked examples and transfer of geometrical problem-solving skills: A cognitive-load approach. Journal of Educational Psychology, 86(1), 122-133. https://doi.org/10.1037/0022-0663.86.1.122
Renkl, A. (1997). Learning from worked-out examples: A study on individual differences. Cognitive Science, 21(1), 1-29. https://doi.org/10.1207/s15516709cog2101_1
Renkl, A., & Atkinson, R. K. (2003). Structuring the transition from example study to problem solving in cognitive skill acquisition: A cognitive load perspective. Educational Psychologist, 38(1), 15-22. https://doi.org/10.1207/S15326985EP3801_3
Sweller, J., & Cooper, G. A. (1985). The use of worked examples as a substitute for problem solving in learning algebra. Cognition and Instruction, 2(1), 59-89. https://doi.org/10.1207/s1532690xci0201_3
Sweller, J., Mawer, R. F., & Ward, M. R. (1983). Development of expertise in mathematical problem solving. Journal of Experimental Psychology: General, 112(4), 639-661. https://doi.org/10.1037/0096-3445.112.4.639
Sweller, J., van Merriënboer, J. J. G., & Paas, F. (2019). Cognitive architecture and instructional design: 20 years later. Educational Psychology Review, 31(2), 261-292. https://doi.org/10.1007/s10648-019-09465-5
Trafton, J. G., & Reiser, B. J. (1993). Studying examples and solving problems: Contributions to skill acquisition. In Proceedings of the 15th Annual Conference of the Cognitive Science Society (pp. 1017-1022). Cognitive Science Society.
van Merriënboer, J. J. G. (1990). Strategies for programming instruction in high school: Program completion vs. program generation. Journal of Educational Computing Research, 6(3), 265-285. https://doi.org/10.2190/4NK5-17L7-TWQV-1EKN