WeSolve+ is best for someone converting a marking scheme or a worked set they already hold into items they can sit again next week.
WeSolve+ reads the whole document and writes the questions for you
Upload your PDF, photograph your notebook, or point the camera. WeSolve+ writes questions from that material, explains why each answer is right, reads the chapter back to you as a podcast, and remembers every item you missed until you own it.
The tool below is a small browser-only tool and it is not WeSolve+: paste a few lines and text rules turn them into cards on the spot. The real app, the one that uses AI, is behind the link above.
Options to items
This is a browser-only tool, and that is all it isIt splits the text you paste by rule, and nothing else. WeSolve+ is a different thing entirely: it reads your whole PDF with AI, writes the reasoning behind every question, speaks the chapter back to you, and remembers what you missed so it can return it. Try the real app now, free!
This tool uses text rules, not AI. A question and its lettered options are treated as a single item, an "Answer:" line beneath them is read as the key, and an item with no key in your text keeps an empty back rather than an invented one. It cannot tell a good distractor from a silly one, which is what this page is for. Your text stays in the browser and is never sent anywhere.
A distractor is a hypothesis about how people go wrong
Every wrong option should be somebody's actual error, not filler. If a student confuses osmosis with diffusion, one option says diffusion. If they invert a formula, one option is the inverted result. Written this way, the pattern of wrong answers tells you which misconception a class holds, which is the only reason to run multiple choice rather than open questions in the first place. Filler options destroy that: when three of four choices are absurd, a person who understands nothing still scores by elimination, and the item measures reading rather than knowledge. Marking schemes are the least expensive source of real distractors, because the errors people actually made are written in them.
Five rules that kill most bad items
One, keep the options the same length and grammar, since the longest and most qualified option is usually the correct one and experienced candidates know it. Two, avoid none of the above and all of the above; they test logic puzzles rather than content. Three, put only one idea in the stem, because a two part stem produces an item nobody can diagnose. Four, keep negatives out of the stem where possible, and where a negative is unavoidable make it impossible to miss. Five, do not let one item give away another, which happens constantly when a set is written in one sitting from one chapter. None of these are opinions about style; each one is a way the score stops reflecting what the person knows, which is the definition of a threat to validity.
Guessing, and what a low score really means
On a four option item, blind guessing returns twenty five per cent, so on a forty item paper somebody who knows nothing scores about ten. That floor matters when you read your own result: forty per cent on a four option test is not forty per cent of the material, it is closer to twenty once the floor is removed. Some exams apply a correction for guessing that subtracts a fraction for wrong answers, and where that is in force the tactical advice changes completely, so it is worth knowing which regime you are sitting under. For self testing the cleaner fix is to cover the options and answer first, then look, since producing an answer is a different and harder act than recognising one.
What the format cannot reach
Multiple choice tests recognition and elimination well and it tests production not at all. Anything where the work is the construction, a proof, a derivation, an argument, an essay plan, is invisible to it, and a set of items that avoids those topics quietly narrows a syllabus down to its testable third. That is a real cost and worth naming on a page that supplies a tool for building them. Mixed papers exist for this reason: items for coverage and speed, open responses for the things that only appear when somebody has to build the answer themselves, marked against a rubric so the marking stays consistent.
Reading the paper afterwards, option by option
The useful number after a test is not the average, it is which wrong option won. An option that attracts a third of the group is a misconception with a name, and the ten minutes spent addressing it are worth more than the whole score sheet. An option nobody picks is dead weight and effectively turns your four option item into a three option item, raising the guessing floor from twenty five to thirty three. Checking whether strong candidates got each item right, the core of item analysis, separates a hard item from an ambiguous one, and only the second needs rewriting.
Having the options written for you
Producing three plausible distractors is the slow part, and it is exactly the part a text rule cannot do, because it requires knowing the subject well enough to predict how somebody misunderstands it. Upload a PDF, a photograph or a camera shot and WeSolve+ reads the source and writes items with reasoning attached, so you can see why an option was offered rather than trusting it. Items you get wrong are collected for the next session. At wesolveapp.com in any browser and on the App Store for iPhone and iPad, free to begin. Figures on pricing.
Sources used on this page
- Multiple choice
- Item analysis
- Item response theory
- Validity (statistics)
- Reliability (statistics)
- Rubric (academic)
- Testing effect
- Roediger and Karpicke, Test-Enhanced Learning (2006)
- Karpicke and Roediger, the critical importance of retrieval (2008)
- Cepeda et al., distributed practice meta analysis (2006)
- Dunlosky et al., improving students' learning (2013)
- What Works Clearinghouse, organizing instruction and study
- The Learning Scientists, retrieval practice
- Retrieval Practice, the research library
- Cloze test
- Active recall
- Anki manual, importing text files
- MDN, the details element
- MDN, the textarea element
- MDN, the Clipboard API
- W3C, the disclosure pattern
- WeSolve+ on the App Store
- Test (assessment)
- True/false test
- Process of elimination
- Closed-ended question
- Open-ended question
- Question
- Guessing
- Response bias
- Acquiescence bias
| Part of the item | The version that fails | The version that diagnoses |
|---|---|---|
| Stem | Which of the following is not untrue of osmosis? | Osmosis moves water across a membrane in which direction? |
| Correct option | Water moves from a dilute to a concentrated solution, always | Toward the more concentrated solution |
| First distractor | Bananas | Toward the more dilute solution |
| Second distractor | None of the above | In whichever direction the solute moves |
| Third distractor | All of the above | It does not move until equilibrium is reached |
| What you learn from a wrong pick | Nothing | Exactly which step they inverted |
How many options should an item have?
Four is conventional, but three good ones beat four where the fourth is filler. A dead option raises the guessing floor from twenty five per cent to thirty three without adding anything.
Why does the tool keep the options with the question?
Because splitting them gives you four fragments and no item. A question and its options are one unit, and an answer line under them belongs to that unit.
Should I use none of the above?
Rarely. It turns the item into a logic puzzle and, when it is the correct answer, tells you only that the candidate rejected three things, not that they know the right one.
Is anything I paste sent anywhere?
No. The page carries the rules with it and has no server to talk to. Disconnect and try it if you would rather see than be told.
What does a forty per cent score mean on a four option paper?
Less than it looks. Blind guessing returns about twenty five per cent, so forty is closer to twenty once the floor is taken out.
Can this write distractors for me?
No, and the reason is worth stating: a plausible wrong answer is a prediction about how somebody misunderstands the topic. That needs a model of the subject, which is what the app has and a text rule does not.
Last updated: 2026-08-15