AI Survey Design: Write Questions That Get Real Answers

# AI Survey Design: Write Questions That Get Real Answers

You are a research methodologist helping [ORGANIZATION_NAME] design a survey on [SURVEY_TOPIC].

## Survey Objectives

What decisions will this survey data inform? [DECISION_TO_INFORM]
Target respondents: [TARGET_RESPONDENTS]
Expected completion time: under [MAX_MINUTES] minutes
Distribution method: [DISTRIBUTION_METHOD]

## Question Design

For [SURVEY_TOPIC], draft survey questions in each category:

**Screening questions (optional):**
2–3 questions to confirm the respondent fits [TARGET_RESPONDENTS] before proceeding.

**Background questions:**
3–4 questions that provide context (role, experience level, behavior frequency).

**Core questions (the substance):**
10–12 questions that directly address [DECISION_TO_INFORM]. For each, choose the right question type:
- Likert scale (1–5 or 1–7): for measuring agreement or satisfaction
- Multiple choice: for categorical answers with known options
- Ranking: when relative preference matters
- Open text: use sparingly, for "why" and "other"

**Net Promoter Score or benchmark question:**
Include one standard benchmarkable question for [SURVEY_TOPIC].

## Bias Check

Review the drafted questions for:
- Leading language (questions that suggest the "right" answer)
- Double-barreled questions (two questions in one)
- Vague terms without clear definitions
- Social desirability bias (questions people feel pressure to answer a certain way)

## Response Scale Guidance

Recommend the best response scales and label conventions for [SURVEY_TOPIC].

## Analysis Plan

How will [ORGANIZATION_NAME] analyze responses to answer [DECISION_TO_INFORM]?