The problem with worked examples
A worked solution shows you *what* was done. It almost never shows you *why that and not something else* — which is the only part you need. The author chose substitution over parts for a reason, and the reason is not on the page. It's the invisible curriculum, and it is the entire skill.
So students read solutions, feel that they understood, sit the exam, and discover that understanding a solution and generating one are different capacities. They are, and only one of them was practised.
What self-explanation actually is
Go through the worked example one line at a time, and for each line, say out loud: why is this step legal, and why did they choose it? If you cannot answer, you have found the gap — which is exactly the thing that reading the solution smoothly would have hidden from you forever.
- Cover the solution. Attempt the problem first — the generation effect is free and you are leaving it on the table.
- Reveal one line. Just one.
- Explain why that line follows before revealing the next.
- At the end, close it and reproduce the whole thing on a blank page. If you can't, you didn't learn it; you read it.
Where it matters most
Any subject built on worked examples: maths, physics, engineering, chemistry, statistics, and increasingly law, where a judgment is a worked example of reasoning. In all of them, the standard student behaviour — read the solution, nod, move on — produces the confident-then-devastated pattern in exams.