Survey & Instrument Design
Moderate · Methods & Study Design track · ~35 min hands-on + readings and quiz
← Back to the Methods & Study Design track
Use AI to draft and critique survey items — while guarding against bias and bad measurement.
What you’ll be able to do
- Draft survey items, scales, and skip logic with AI
- Critique items for bias, ambiguity, and double-barrels
- Plan cognitive testing and a codebook
Overview
AI can draft candidate items, suggest response scales, and write cognitive-interview probes. It can also produce leading or double-barreled items, so you critique every one against measurement principles.
Good instruments are validated, not just written. AI accelerates drafting; piloting and cognitive testing with real respondents still decide what works.
Practice activities
Activity 1 · Moderate — Draft and critique items
Time ~18 min · Tools ChatGPT Edu
Goal. Generate candidate items, then pressure-test them.
Setup. A construct to measure — e.g., perceived pedestrian safety in your neighborhood.
Steps.
Draft items:
Draft 6 candidate survey items measuring [construct], with a response scale. Keep them concrete and answerable. [construct]
Critique them:
Critique each item for bias, ambiguity, double-barreling, and reading level.
Cut or fix based on the critique — your judgment, not the model’s.
Expected result. A vetted short item set with a response scale.
Check your work. Does each item measure the construct, or something adjacent?
Common pitfalls. AI produces leading and double-barreled items confidently. You own measurement quality.
Stretch (optional). Ask for two items at different reading levels and keep the clearer.
Activity 2 · Moderate — Write cognitive-interview probes
Time ~12 min · Tools ChatGPT Edu
Goal. Plan to test whether respondents understand your items.
Setup. Two of your items.
Steps.
Draft probes:
Write cognitive-interview probes for these two items (comprehension, retrieval, judgment). [items]
Decide which probes you’d actually use.
Predict failure:
What’s the most likely way a respondent could misunderstand item 1?
Expected result. Usable probes and a predicted misunderstanding.
Check your work. Real piloting validates items — drafting doesn’t.
Common pitfalls. Skipping cognitive testing ships ambiguous items. AI helps draft, not validate.
Stretch (optional). Plan a 3-person cognitive pretest.
Check your readiness
Answer these, then check — your score suggests whether to dive in or skim the readings first.
Recommended readings
Available in the shared OneDrive folder Staff Faculty AI Workshop → Readings, and online where linked:
- AI and Survey Sampling Problems (Sharon Lohr) — sampling and representativeness cautions.
- A Framework for Technical Writing in the Age of LLMs — precision in wording.
Useful resources
- ChatGPT Edu (Drexel AI Tools) — the approved tool, with data rules.