For Research Analysts ·
What you'll accomplish
By the end of this guide, you'll use ChatGPT as a methodology partner for survey design — drafting question batteries, checking for bias, writing screeners, and validating logic — cutting your survey design time from 3–4 hours to under 90 minutes for a typical 20-question survey.
What you'll need
Start a new conversation in ChatGPT. Give it the full context in one message — the more context, the better the output.
Template:
I'm designing a survey and need your help. Here's the project context:
Research objective: [what we're trying to find out]
Client: [industry/type of company]
Target respondent: [who will take the survey — demographics, behaviors, experience]
Sample size: [N]
Survey platform: [Qualtrics / SurveyMonkey / etc.]
Topics to cover: [list the content areas]
I'll ask you to help with: (1) writing the screener, (2) drafting question batteries for each topic, and (3) reviewing the full draft for bias and logic issues. Ready?
What you should see: ChatGPT confirms it understands the project and asks to proceed or clarifies anything unclear.
Ask ChatGPT to draft the respondent screener — the qualification questions that determine who takes the survey and who gets terminated.
Example:
Write a screener for this survey. Target respondent: U.S. adults 25-54 who have purchased pet food in the past 3 months. Quotas: 50% dog owners, 50% cat owners; balance by age and gender. Include termination logic for anyone outside these criteria.
What you should see: A screener with intro questions, qualifier questions with termination logic, and quota routing.
Work through the survey section by section. For each content area, give ChatGPT:
Example:
Now draft the awareness and consideration section. We need: unaided brand awareness (open-ended), aided brand awareness (multi-select from a list), brand consideration (single select), and reasons for non-consideration (multi-select). The brand list is: [Brand A, Brand B, Brand C, Brand D].
Once you have all sections drafted, ask ChatGPT to audit the complete survey:
Here is the complete survey draft: [paste all questions]. Please review for:
1. Leading or biased language
2. Double-barreled questions (asking two things at once)
3. Acquiescence bias (questions where "yes" or "agree" is the obvious polite answer)
4. Order effects (questions that may bias responses to later questions)
5. Response option issues (missing options, overlapping ranges, unbalanced scales)
Flag each issue with the question number and suggest a rewrite.
What you should see: A list of flagged questions with specific issues and suggested revisions.
Ask for a final logic review:
Based on this survey structure, check the flow: (1) Is there a logical order from general to specific? (2) Are any sensitive questions placed too early? (3) Is the survey too long for our target respondent (aim for 12-15 minutes)? Suggest reordering if needed.
Full project briefing:
I'm designing a survey. Objective: [objective]. Target: [respondent profile]. Platform: [Qualtrics/etc.]. Topics: [list]. Help me draft the screener and question battery section by section.
Screener writing:
Write a screener for: target = [description], quotas = [requirements]. Include: intro, qualifier questions, termination logic for each disqualifying answer.
Question battery for one topic:
Write a question battery for [topic]. I need: [question types — awareness, satisfaction, etc.]. Use [scale type]. Brand list: [brands]. Keep response options balanced and unbiased.
Bias audit:
Review this survey draft for bias, double-barreled questions, order effects, and response option issues. Flag each problem by question number with a suggested fix. [paste draft]