Free wording checks + AI review prompts
Survey question checker
Give your AI-generated survey a closer look. Flag common wording patterns, see what needs attention, and prepare a better rewrite brief.
Your questionnaire
Start with an example or paste your own draft.
Optional. This context goes into the AI review prompt; the wording rules do not interpret it.
One question per block, with a blank line between questions. Add answer choices on the following line. Up to 20 questions and 12,000 characters. English wording checks.
Your text stays in this tab. Nothing is sent to an AI service. Copying a prompt lets you choose whether to share it elsewhere.
Review signals
Inspect the reason. Decide whether it applies.
A useful first pass
Look for assumptions, combined judgments, missing time periods and answer-option issues. Every signal includes an explanation and a next step.
Rules are transparent. No AI model or customer test is run here.
The AI prompt includes your draft, context and detected signals. It asks your assistant for proposed rewrites and false-positive checks. This site does not run that prompt.
Rules version 1.0. No account required. Text is not added to the URL, stored by this application, or included in analytics.
What this checker actually does
It applies a published set of English wording rules in your browser. It looks for patterns such as leading assumptions, evaluative language, possible combined judgments, undefined frequency periods, overlapping numeric choices and response-option gaps. A match is a reason to review the question, not proof of bias.
What happens when you copy an AI prompt?
The tool assembles your context, draft and review signals into instructions you can use with your chosen assistant. It requests neutral rewrites, checks for false positives and asks whether a proposed edit changes the intended measure. No model is connected to this page, and nothing is sent automatically.
What can it miss?
It does not understand your full study or validate questions, scales, translations, claims, representativeness or statistical design. Some good questions trigger signals; some problematic questions trigger none. Review the complete instrument, its order and the way relevant people understand it before fieldwork.
Put the review into practice
- Compare 12 leading questions with suggested rewrites
- Use AI for research: what to do and what to verify
- Adapt the message-testing questionnaire
Method references
Pew Research Center’s question-writing guidance and the Census Bureau’s instrument-development standard inform the review topics. The specific rules and triggers are MTB editorial heuristics; neither source validates or endorses this tool.