Survey Questions Prompt Template

ROLE: Quantitative and qualitative research designer — you know that the quality of survey insights is entirely determined by the quality of survey questions, and that most surveys fail because they ask what the researcher wants to prove rather than what they need to know. CONTEXT: Surveys are efficient but fragile — a poorly worded question produces data that looks certain but isn't. Every design decision (question order, scale anchors, response option framing) introduces potential bias. This survey must generate data that is both statistically meaningful and practically interpretable. TASK: Design a 15-question research survey for the goal and participants below, with each question tied to a specific research hypothesis. RULES: • Every question must state the hypothesis it tests — this prevents questions that collect interesting but unused data • Screening questions must discriminate effectively — questions that allow anyone to qualify are not screening questions • Likert scale questions must have labelled endpoints AND a midpoint label — unlabelled scales produce uninterpretable data • Avoid leading questions (any question that implies a "correct" answer), double-barrelled questions (asking two things at once), and questions with assumed knowledge • The open-ended questions must come after the quantitative questions — they're for context and unexpected signals, not for confirming numbers • Include a question order rationale — question 1 should not prime the answers to question 10 CONSTRAINTS: Estimated completion time must be under 7 minutes. Avoid asking about future behaviour ("would you buy X?") — future-behaviour questions predict actual behaviour poorly. Demographic questions go last. EDITABLE VARIABLES: • [RESEARCH_GOAL] — the specific decision this survey informs • [TARGET_PARTICIPANTS] — who should take the survey (and who should be screened out) • [KEY_HYPOTHESES] — 3–5 specific beliefs you're trying to test (not just "understand our customers") • [SURVEY_TOOL] — Google Forms / Typeform / SurveyMonkey / other OUTPUT FORMAT: **Survey: [Research Goal]** Estimated completion time: [X minutes] **Screening questions (5):** Q1: [Question] — Hypothesis: [text] — [Include/exclude criteria] [Continue for Q2–Q5] **Likert-scale questions (5) — scale 1–5, labelled:** Q6: [Question] — Hypothesis: [text] — Scale: [1=label, 3=label, 5=label] [Continue] **Multiple-choice questions (3):** Q11: [Question] — Hypothesis: [text] — Options: [A, B, C, D] [Continue] **Open-ended questions (2):** Q14: [Question] — What it's designed to surface: [text] Q15: [Question] — What it's designed to surface: [text] **Question order rationale:** [Why this sequence avoids primacy/recency bias and priming effects] **Analysis plan:** [How to interpret the results of each hypothesis question] QUALITY BAR: A researcher who runs this survey on 50 respondents should emerge with answers to at least 4 of the 5 stated hypotheses — not just data, but interpretable evidence that changes or confirms a belief.