The gradeable stem rule
A practice question earns its place when a stranger could mark the answer right or wrong: "What are the causes?" invites an essay; "Which cause came first?" grades cleanly. Write stems that name the retrieval, the same marker test that governs cards, and keep one question one fact, since compound stems hide which half failed. Stems phrased as the exam phrases them, calculate, identify, justify, train the exam's verbs alongside its content.
Distractors are where questions get honest
Multiple choice lives or dies on wrong options, and the honest source of wrong options is real confusion: your own past mistakes, the lookalike term, the formula applied outside the conditions it holds under. Distractors invented at random test nothing, while distractors from your error log make every practice question a rehearsal of a trap you have actually fallen into, the personal edge no purchased bank can copy.
Match the question type to the material
Sequences want ordering questions, contrasts want which one questions, processes want what happens if, definitions want cloze per the cloze guide, and applications want small scenarios. Forcing everything into multiple choice flattens material that other shapes test naturally, and the exam's own mix, read from past papers, is the mix worth rehearsing.
Generation is the floor; your edit is the ceiling
Writing every question by hand taxes the evening; generating a base set from the chapter through the pipeline and editing toward sharper stems keeps the craft and drops the transcription. The edited set then enters practice rounds, where the production advantage, near 80 percent delayed retention against roughly a third in the retrieval experiments, pays the whole effort back with interest. Dunlosky and colleagues' 2013 review put practice testing in the top utility band, and that is what your edit protects when you repair a weak stem.
Questions as a group currency
Question writing scales socially: each member contributes a chapter's worth, the group sits the merged set, and authorship rotates, the study group format with the preparation built in. Written questions also expose your own blind spots, since the chapter section you cannot question is the section you do not yet own. And plainly: the percentages are the cited literature's, and WeSolve+ generates and drills while the exam's taste, learned from its past papers, guides your editing hand.
The question bank compounds across terms
Questions written this term drill next term's prerequisites: keep the bank organized by topic, one deck per chapter, as the deck organization guide sets out, and a course's end leaves an asset, not just a grade. Sharing banks across a study group multiplies authorship while the editing pass keeps your copy honest. Resitting old questions cold after a break also measures how much you have lost: score the same set in September that you scored in May, and the difference tells you whether the deck needs a summer maintenance session of ten cards a day.
By Mert Erarslan (Exam prep and subject routes)
Sources used on this page
- Forgetting curve
- Spaced repetition
- Testing effect
- Active recall
- Distributed practice
- Roediger and Karpicke 2006, test-enhanced learning
- Karpicke and Roediger 2008, Science
- Dunlosky et al. 2013, technique effectiveness review
- Cepeda et al. 2008, the spacing experiment
- Murre and Dros 2015, the Ebbinghaus replication
- Metacognition
- Long-term memory
- Desirable difficulty
- Study skills
- Flashcard
- Working memory
- Question
- Multiple choice
- Distractor
- Word stem
- Verb
- Currency
- Authorship
- Scotoma
- Interest
- Scenario
- Taste (sociology)
- Compound (linguistics)
What makes a good practice question?
A gradeable stem naming one retrieval, phrased in the exam's own verbs: a stranger could mark it right or wrong.
Where do good wrong answers come from?
Your error log: real confusions and license violations make distractors that rehearse actual traps.
Which question type should I use?
The material's natural shape: ordering for sequences, cloze for definitions, scenarios for application, mixed like your exam mixes.
Should I write or generate questions?
Generate the floor from your materials, edit toward sharpness: the craft survives, the transcription tax disappears.
How does this work in a study group?
Rotate authorship by chapter and sit the merged set: preparation becomes the group's currency.
Last updated: 2026-08-15 · Written with AI assistance and reviewed before publishing.
