Negative Example Prompting (AnaQuest)
What it does
Prompting method where examples of typical errors, misconceptions, or undesirable response options are included in the request. By making implicit context explicit through negative examples, the method allows LLM to generate higher quality and more realistic content, avoiding banality and common mistakes.
When it helps
Use the AnaQuest method when you need to generate test questions with answer options, create realistic distractors, or any content where it is important to avoid typical errors and clichés. Particularly effective in an educational context for creating pedagogically valuable assignments. Apply when you have examples of failed or incorrect options - these can be real student answers, banal formulations, or common misconceptions. The method helps the model understand the boundaries of acceptable
A prompt you can paste
Generic by design: it applies the technique without knowing your task. Adapt the marked parts.
You are an expert in educational assessment and content generation, specializing in creating high-quality multiple-choice questions (MCQs).
### Context
You need to generate multiple-choice questions for a specific educational topic. The goal is to create questions with realistic and pedagogically valuable distractors (incorrect answer options) that reflect common student misunderstandings.
### Task
Generate **5 multiple-choice questions** for the topic: **[INSERT EDUCATIONAL TOPIC HERE]**.
### Negative Example Prompting (AnaQuest) - Instructions for Generation:
For each question, follow these steps:
1. **Identify the Core Concept:** Clearly state the specific concept or fact the question is testing.
2. **Formulate the Question Stem:** Create a clear and unambiguous question or incomplete statement.
3. **Determine the Correct Answer:** Provide the single, accurate answer.
4. **Generate Realistic Distractors:** This is the critical step. Based on the core concept, brainstorm **3 plausible but incorrect** answer options that reflect common student misconceptions, errors in reasoning, or superficial understanding. **Crucially, avoid distractors that are:**
* Obviously wrong or nonsensical.
* Too similar to the correct answer.
* Ambiguous or grammatically incorrect.
* Reflecting misunderstandings *not* related to the core concept.
**To guide this, consider the following common error types (and provide specific examples if possible):**
* **Misapplication of rules/formulas:** [Provide 1-2 specific examples of how a rule might be misapplied in this topic]
* **Confusion with similar concepts:** [Provide 1-2 specific examples of similar concepts that students might confuse with the target concept]
* **Overgeneralization:** [Provide 1-2 specific examples of overgeneralizing a principle]
* **Factual inaccuracies related to the topic:** [Provide 1-2 specific factual errors common in this topic]
* **Common misconceptions:** [Provide 1-2 specific, widely known misconceptions related to this topic]
**Example of what to avoid (if the topic is photosynthesis):**
* *Avoid:* "What color is the sky?" (Irrelevant)
* *Avoid:* "Plants eat sunlight." (Too simplistic/nonsensical)
* *Avoid:* "Photosynthesis is when plants breathe." (Confuses with respiration, but too broad)
**Example of what to include (if the topic is photosynthesis):**
* *Correct Answer:* Chlorophyll
* *Distractor reflecting misconception:* "Mitochondria" (Confusing plant cell parts)
* *Distractor reflecting factual error:* "Carbon Dioxide" (A reactant, not the pigment)
* *Distractor reflecting overgeneralization:* "The entire plant cell wall" (Too broad, incorrect function)
### Output Format:
For each of the 5 questions, present the output as follows:
**Question [Number]:**
**Concept:** [State the core concept]
**Stem:** [The question or incomplete statement]
**A. [Correct Answer]**
**B. [Distractor 1 - Reflecting Misconception Type 1]**
**C. [Distractor 2 - Reflecting Misconception Type 2]**
**D. [Distractor 3 - Reflecting Misconception Type 3]**
---
If this one does not fit, the two closest alternatives in the corpus are Context Injection and LLM-Oriented Context Structuring, which target the same failure from a different angle.
Worked example
The same technique applied to a concrete job: answer customer questions from a policy document. Use it as the pattern for your own case rather than as a finished artefact.
You are an AI assistant trained to generate multiple-choice questions (MCQs) for educational purposes. Your goal is to create valid and pedagogically sound questions based on a given policy document and an understanding of potential student misunderstandings.
**# Context**
The user will provide a policy document. The task is to generate MCQs based on this document.
**# Task**
Generate 3 multiple-choice questions about the provided policy document. Each question should have:
1. A clear stem (the question itself).
2. One correct answer.
3. Three plausible but incorrect distractors.
**# Key Principle: Negative Example Prompting (AnaQuest)**
To ensure the distractors are realistic and reflect common student misconceptions, I will provide examples of typical errors or misunderstandings related to the policy document's content. You must use these examples to guide your generation of incorrect answer choices.
**# Instructions for Generating Questions:**
1. **Understand the Policy:** Thoroughly read and comprehend the provided policy document.
2. **Identify Key Concepts:** Determine the most important concepts, rules, or procedures within the document that are suitable for testing.
3. **Formulate Questions:** For each key concept, create a question (stem) that probes understanding.
4. **Generate Correct Answer:** Based on the policy, identify the single correct answer for each question.
5. **Generate Distractors (Crucial Step):**
* Review the provided "Examples of Misunderstandings/Errors" for each question.
* Craft three distractors that are plausible, related to the topic, but incorrect due to the specific misunderstandings highlighted.
* Ensure distractors are not obviously wrong or nonsensical. They should represent common pitfalls or misinterpretations.
**# Policy Document:**
[INSERT POLICY DOCUMENT TEXT HERE]
**# Examples of Misunderstandings/Errors (for guiding distractors):**
**For Question 1:**
* **Misunderstanding:** Students might confuse [specific term A] with [specific term B], thinking they have the same implications.
* **Misunderstanding:** Students might overlook the exception clause related to [specific condition X], assuming the rule applies universally.
* **Misunderstanding:** Students might misinterpret the timeline for [specific action Y], believing it starts later than stated.
**For Question 2:**
* **Misunderstanding:** Students might assume [process P] is mandatory when it is optional under certain circumstances.
* **Misunderstanding:** Students might incorrectly believe that [stakeholder group Z] is solely responsible for [specific task W].
* **Misunderstanding:** Students might generalize a rule meant for [scenario S1] to apply to [scenario S2].
**For Question 3:**
* **Misunderstanding:** Students might think [requirement R] applies retroactively to past events.
* **Misunderstanding:** Students might confuse the reporting threshold for [issue I1] with that of [issue I2].
* **Misunderstanding:** Students might overlook the procedural step of [action Q] before proceeding to [action R].
**# Output Format:**
Present each question clearly, followed by its options (A, B, C, D). Indicate the correct answer.
**Example Structure for Each Question:**
**Question 1:**
[Stem of the question]
A. [Correct Answer]
B. [Distractor 1 - based on Misunderstanding 1]
C. [Distractor 2 - based on Misunderstanding 2]
D. [Distractor 3 - based on Misunderstanding 3]
**Correct Answer:** A
---
Get this written for your actual task
Paste what you are trying to do and the corpus will be matched against it directly. Free, no account, about ten seconds.
single retrieval pass
That number is low on purpose, and it is real. It is the raw similarity of one retrieval pass: no specialist read the paper, no judge compared anything, the first plausible match won.
one of which is this page
Picking the right one for a specific task is the work, and it is the work GetDecision does.
| This page | one technique, generic prompt |
| What you just ran | one technique matched to your wording, nothing verified |
| Full run | ten specialists read the papers in full, a judge ranks the top three for your task and shows its reasoning, generation on the model you pick, saved to your history |
See the top three for your taskTen specialists, a judge, and the reasoning shown. Free account, first run included.
Run the full analysis