Most exam-style versions of “which of the following studies would need IRB approval?” are testing one core idea: IRB review is required only when a project is both (1) research and (2) involves human subjects (or their identifiable private data).

Core rule

A study needs IRB approval if:

  1. It is research
    • A systematic investigation, with a plan or protocol.
    • Designed to develop or contribute to generalizable knowledge (e.g., publishable findings, conference presentations, results meant to be applied beyond the specific setting).
  1. It involves human subjects
    • Living individuals.
    • Data are collected through interaction or intervention with them (surveys, interviews, experiments, tests) or
    • The investigator uses identifiable private information about them (medical records, student records, employment data, etc.).

If both are true, IRB review is required before data are collected or analyzed.

Typical examples that DO need IRB approval

These kinds of studies almost always require IRB approval (sometimes “exempt,” but still formally reviewed/registered):

  • Surveys or interviews of patients, students, employees, community members where results are meant to be published.
  • Focus groups about people’s experiences, opinions, health behaviors, or mental health.
  • Clinical studies testing devices, drugs, supplements, or procedures with living participants.
  • Pilot or feasibility studies with human subjects, even if small.
  • Master’s theses and dissertations involving human participants or identifiable private data.
  • Secondary analysis of identifiable medical or educational records.

Typical examples that usually do NOT need IRB approval

These usually do not require IRB review because they are not “research,” not “human subjects” research, or both:

  • Literature reviews, meta-analyses, and theory papers using only already-published data (no new human data collected).
  • Class projects done only for a grade, with no plan to publish or generalize, especially if data are anonymous and low risk (local rules may vary).
  • Quality improvement or program evaluation conducted solely for internal use with no intent to generalize beyond that setting (e.g., routine customer-satisfaction surveys used only for internal operations).
  • Studies using fully anonymous datasets where no one can link data back to individuals (and no indirect identifiers exist).
  • Interviews where respondents only describe a process or policy , not their own personal status, feelings, or behaviors (e.g., staff describing how a clinic workflow operates).

How to pick the correct option on an exam

When you see choices like:

  • A: Analysis of completely de‑identified, publicly available census data.
  • B: Patient survey about pain levels in a clinic, with plans to publish.
  • C: Review article summarizing existing research on depression.
  • D: Internal customer satisfaction survey for a company cafeteria, used only for internal improvement.

The one that needs IRB approval is the option that clearly:

  • Involves living people (through direct contact or identifiable data), and
  • Has a research intent to produce generalizable knowledge (e.g., publish).

In that example, option B would be the correct answer. If you paste the actual answer choices you’re looking at, I can walk through each one and tell you exactly which would require IRB approval and why.