🔬 Research
Survey Questionnaire with Bias Checks
Draft a survey with item types, bias checks, skip logic, and analysis notes. Flags leading, double-barreled, and loaded items before you field.
0Reviews
Prompt
Act as a survey methodologist. Write a questionnaire a team can field. Detect bias in the items you write. Do not invent population statistics. Do not claim a sample size is powered unless Inputs include a power plan. Inputs: - Research question: [What we will decide] - Population and sample: [Who, how invited, expected n] - Channel: [Email / in-app / panel] - Length: [Minutes or item cap] - Must measure: [Constructs] - Must not ask: [Legal, medical, sensitive] - Existing items: [Paste any we already use] - Scale needs: [Likert, CSAT, NPS, open] - Language: [EN and reading level] - Decision this will inform: [Ship / kill / target] Generate: 1. Coverage map: Construct -> item IDs -> decision. Gaps. What a survey cannot answer (need interviews). 2. Bias review rules you will apply: leading, double-barreled, loaded, double negative, recall window, social desirability, order, aquiescence. Then use them. 3. Questionnaire: Numbered items. For each: text, type, options, skip logic, construct, a bias note (pass / fail+rewrite). 4. Opening and consent: why, time, voluntary, data use, in one short screen. 5. Demographics: only what Decision this will inform needs. Offer prefer-not-to-say. 6. Analysis sketch: which items are descriptive vs comparison. What you will not do (no fake p-values). How you will treat NPS if used (and whether you should). 7. Fielding: invite copy, reminder, exclude criteria, a cognitive-test of 3 items on 3 people before launch. 8. Kill list: 5 items you refused to write and why. Constraints: - No em dashes. No fake citations to survey scale papers unless the user pasted a licensed instrument. - Do not write medical diagnosis items. Do not ask immigration status, health, or income unless Inputs require and then offer skip. - If expected n is under 30, warn that crosstabs will be theater. - NPS is optional and often the wrong tool; say so if CSAT or a job-to-be-done item is better. - Short items. One idea each.
Instructions
Replace every [bracket] with your details before running. Works on ChatGPT, Claude, and Gemini.
Examples
Example Input
Research question: among trial workspaces that stalled (0-1 dashboards in 14 days), what blocked first-value, so PM can pick one onboarding fix Population and sample: email to 420 stalled-trial admins, US, expected n ~60-80 if 15-20% respond Channel: email, one reminder Length: under 6 minutes, max 12 items Must measure: where they stalled, who they needed, whether Excel was the workaround, satisfaction with setup (not brand love) Must not ask: salary, health, undocumented status, competitor NPS, customer PII Existing items: none licensed Scale needs: a 5-point setup satisfaction, one open box, no NPS this round (PM agreed) Language: US English, plain, about grade 8 Decision this will inform: pick one onboarding fix for Q4, not a pricing change
Example Output
1. Coverage Stall location -> Q3. Who they needed -> Q4. Excel workaround -> Q5. Setup satisfaction -> Q6. Open context -> Q7. Decision: one Q4 onboarding fix. Cannot answer: true prevalence in all finance teams, or why a VP wanted a dashboard (interview). n~60-80: describe this invited list, do not crosstab into tiny cells. 2. Bias rules applied No leading "did our missing guided setup hurt you." No double-barreled. Recall window: last 14 days of trial. Social desirability: "Excel is common" in the option, not a shame item. Order: behavior before satisfaction. No NPS (agreed). 3. Questionnaire S1 consent. Q1 role. Single. Admin / builder / viewer / other / prefer not. Construct: filter. Pass. Q2 did you try to create a dashboard in the last 14 days of trial. Yes / no / not sure. Skip Q3-Q5 if no. Pass. Q3 where you stopped (last point). Single: account setup / connecting data / building first view / sharing with someone / other. Construct: stall. Pass. Not "why did you fail our product." Q4 last person you needed. Single: no one / teammate / IT / vendor / VP or exec / other. Pass. Q5 last workaround you shipped a number with. Single: none / Excel or Sheets / email / another tool / other. Do not name a competitor as bad. Pass. Q6 setup satisfaction. 5-point: very dissatisfied ... very satisfied, plus N/A. Construct: setup, not brand. Pass. (Not NPS.) Q7 optional open: "If you want, what would have made first-value faster?" 400 chars. Probe, not a leading fix list. Q8 invite them to a 20-min follow-up (yes/no). Not a must. Bias fail we rewrote: "How much did the lack of guided setup block you?" (leading, pitches v1). 4. Consent "6 minutes. Voluntary. About how setup went. We will not ask for customer data. Used to pick one onboarding fix. Skip any item." 5. Demographics Role only (Q1). No salary, health, income. 6. Analysis Q3-Q5 counts and % of respondents, not of all 420. Q6 median. Q7 themes. No p-values. No NPS. If n<30, stop slicing. 7. Fielding Invite: "Help us pick one setup fix. 6 minutes. $0 required." One reminder day 3. Exclude: current employees, people who already finished onboarding. Cognitive test: Q3, Q4, Q5 on 3 stalled admins, fix wording. 8. Kill list NPS. Willingness to pay. "Would you recommend us." Immigration. Competitor smear item.