irb-protocol

name: irb-protocol description: > Use this skill whenever a user needs help writing, revising, or evaluating an IRB or ethics protocol for qualitative or anthropological research. Triggers include: any mention of "IRB," "ethics protocol," "IRB application," "ethics review," "human subjects," "Common Rule," "informed consent form," "consent process," "how to write an IRB protocol," "ethics board," "IRB submission," "protocol narrative," "exempt vs expedited vs full board," "consent waiver," "verbal consent," "data security plan," "confidentiality," "de-identification," "deductive disclosure," "vulnerable populations," "fieldwork ethics," or "recruitment script." Also trigger when users ask about writing consent forms, recruitment materials, data management plans for IRB, risk assessment for qualitative research, digital ethnography ethics, or oral history IRB issues. Covers all qualitative and ethnographic methods. Do NOT use for upstream method selection (use methodology-selection skill), full research plan writing (use research-plan skill), or grant proposals (use grant-proposal skill). This skill handles writing IRB-ready protocol narratives and associated documents.

IRB & Ethics Protocol Writing

Write IRB and ethics protocols for qualitative anthropological research that satisfy regulatory review criteria while making legible the distinctive epistemics and practical uncertainties of ethnographic work — emergent sampling, relational consent, and shifting field conditions. The skill treats protocol writing as a translation problem: making ethnographic practice reviewable without distorting it.

Contemporary best-practice protocols do two things at once: they (a) satisfy the Common Rule's review criteria (risk minimization, risk-benefit proportionality, equitable selection, consent, privacy and confidentiality safeguards, and added protections when appropriate); and (b) anticipate reviewer questions about ambiguity, and answer them with bounded flexibility (estimated sample sizes plus an amendment plan; tiered consent strategies; contingency plans for fieldwork changes).

Quick Reference

Workflow

Step 1: Identify What the User Needs

Determine the entry point:

• Writing a protocol from scratch. The user has a research design and needs to produce a full protocol narrative. Load both reference files and work through the template section by section.
• Revising for resubmission. The user has reviewer feedback and needs to address specific concerns. Identify which protocol sections need revision and load relevant guidance.
• Writing specific components. The user needs a consent form, recruitment script, data management plan, or risk assessment — not a full protocol. Load the templates reference for modular text blocks.
• Adapting for a specific method. The user's methods trigger special protocol requirements (digital ethnography, oral history, visual methods, focus groups). Load the templates reference for method-specific guidance.

Step 2: Gather Context

Before generating any content, collect these inputs:

Required:

1. Research methods. What methods will be used? Participant observation, interviews, focus groups, oral history, visual methods, digital ethnography, archival, or mixed? Different methods trigger different protocol requirements.
2. Study population. Who are the participants? How many (estimated range)? Are there primary vs incidental participants? Are there vulnerable populations, minors, dependency relationships?
3. Risk level. Minimal risk? Sensitive topics (health, legal status, political activity)? Vulnerable populations? Cross-border data?

Important but can be inferred: 4. Institutional context. U.S. Common Rule (default assumption), Canadian TCPS 2, or other? Does the institution use a SmartForm platform or narrative-style application? 5. Review pathway. Does the user expect exempt, expedited, or full board review? If unsure, help determine the likely pathway based on methods and risk profile. 6. Consent approach. Written, verbal with waiver, layered/tiered, ongoing, community-level? If unspecified, recommend based on methods and context. 7. Recording plans. Audio, video, photo, screenshots? Separate consent for recordings is usually required. 8. Data sharing/archiving plans. Will data be archived or shared? Consent language must match sharing plans.

Helpful but not required:

• Field site(s) and whether they are public or private spaces
• Language of consent (may need translated materials)
• Whether the project involves Indigenous communities (triggers additional governance requirements: CARE, OCAP, FPIC)
• Whether the project is funded (some funders require Certificates of Confidentiality or specific data management plans)
• Prior IRB experience and career stage

Step 3: Load Appropriate References

• Always load references/irb-protocol-guide.md for the protocol template, regulatory foundations, checklist, and comparative guidance.
• Load references/irb-templates-and-examples.md when the user needs method-specific guidance, consent form language, recruitment scripts, data management plan templates, risk mitigation strategies, or annotated protocol excerpts.

Step 4: Generate Protocol Content

Follow the protocol template architecture from the guide reference. The standard protocol narrative contains these sections (adapt to institutional format):

1. Project overview and key information — rapid orientation for reviewer
2. Research questions and analytic goals — establishes systematic investigation and social value
3. Setting and sites — privacy expectations and permissions
4. Study population — equitable selection, vulnerability flags
5. Recruitment and access — coercion risk, gatekeeper roles
6. Methods and procedures — core risk assessment anchor
7. Consent processes — meets regulatory requirements while addressing ethnographic consent realities
8. Confidentiality and anonymization — deductive disclosure, group limits
9. Data management and security — auditable controls
10. Risks and mitigation — social, legal, reputational, emotional, community harms
11. Benefits, reciprocity, dissemination — beyond clinical framing
12. Vulnerable populations and power dynamics — added safeguards
13. Fieldwork contingencies and amendment plan — makes uncertainty governable

Step 5: Generate Output

Produce one or more deliverables depending on user needs:

• Full protocol narrative. Complete document following the 13-section template, with suggested word counts per section.
• Consent form or information sheet. Modular text blocks including key information summary, ongoing consent language, recording permissions, and focus group confidentiality limits.
• Oral consent script. For use with waiver of documentation.
• Recruitment script. IRB-compliant recruitment language.
• Data management plan. Storage, encryption, access control, key-file separation, retention, destruction, sharing/archiving.
• Risk mitigation plan. Deductive disclosure, emotional distress, legal risk, community harm, cross-border data.
• Individual sections. Any of the 13 protocol sections as standalone drafts.

Step 6: Quality Check

Before presenting output, verify using the dual applicant-reviewer checklist (full version in guide reference):

• Research determination and level of review are identified
• Research questions show social value and interpretive coherence
• Sites and permissions clarify public vs private, gatekeepers
• Population defines primary vs incidental participants with estimated sample size range
• Recruitment minimizes coercion and avoids role confusion
• Consent approach specifies written vs verbal, includes key information, addresses documentation waiver if used
• Ongoing consent is operational, not aspirational
• Recording plans specify separate consent and address identifiability
• Focus group limits state confidentiality cannot be guaranteed
• Anonymization addresses deductive disclosure for small-N data
• Data security describes specific controls (encryption, access, key-file separation)
• Data sharing/archiving claims do not exceed consent language
• Legal risk is framed appropriately; protections are not overstated
• Risks to participants are identified with mitigation and referrals
• Community impacts are considered where salient
• Vulnerable populations have additional protections
• Reciprocity and compensation avoid undue influence
• Fieldwork contingencies list amendment triggers and safety pivots

Parameters

• Review pathway: Exempt, expedited, full board. Determines scope and detail of protocol narrative.
• Methods in protocol: Participant observation, interviews, focus groups, oral history, visual methods, digital ethnography, archival, mixed. Different methods trigger different protocol requirements.
• Risk posture: Minimal risk, sensitive topics, vulnerable populations, high-surveillance, cross-border. Higher risk postures require more detailed safeguards.
• Consent modality: Written, verbal with waiver, layered/tiered, ongoing, community-level. Choice depends on methods, context, and risk.
• Output type: Full protocol narrative, consent form, information sheet, recruitment script, data management plan, risk assessment, individual sections.
• Institutional context: U.S. Common Rule (default), Canadian TCPS 2, other. Determines regulatory framework and required elements.

Guardrails

• Do not produce boilerplate. Every protocol section must reflect the specific project's methods, population, and risks. Generic language that could apply to any study is a failure mode.
• Do not conflate consent process with consent documentation. The Common Rule distinguishes these. A project can have a robust consent process (verbal, ongoing, relational) while requesting a waiver of written documentation. Help users frame this correctly.
• Require explicit treatment of deductive disclosure. For small-N, high-context qualitative data, "de-identification" alone is insufficient. Protocols must address how contextual details, distinctive roles, and narrative specificity can re-identify participants even without direct identifiers.
• Flag method-specific protocol requirements. When methods require special protocol attention, flag it explicitly:
  • Digital ethnography: public/private ambiguity, searchability of quotes, platform terms of service
  • Oral history: attribution vs anonymity choice, archiving options, narrator review
  • Visual methods: faces and locations as identifiers, rights and ownership
  • Focus groups: confidentiality cannot be guaranteed among participants
  • Participant observation: primary vs incidental participants
• Ethics goes beyond IRB compliance. Address AAA principles (do no harm, ongoing consent, record stewardship), community obligations, and data sovereignty (CARE, OCAP) where relevant. Do not treat the IRB as the ceiling of ethical practice.
• Do not overpromise protections. Certificates of Confidentiality have exceptions. "Confidentiality" in focus groups cannot be guaranteed. De-identification of qualitative data is always limited. Help users frame protections accurately.
• Require amendment planning. Ethnographic designs are emergent. Protocols must specify what changes require amendment (new populations, new recording types, materially increased risk) and what falls within approved flexibility (follow-up questions, emergent themes within approved domains).

Common Failure Modes

Examples

Example 1: Full protocol for participant observation + interviews

Input: "I need to write an IRB protocol for my study of street vendors negotiating municipal regulation in Lima. I'll be doing participant observation and semi-structured interviews with about 30 vendors."

Output approach:

• Load both reference files
• Set risk posture to standard (public spaces, non-sensitive population)
• Set consent modality to layered (site permission + verbal introduction + written consent for interviews + separate recording consent)
• Protocol: distinguish primary participants (30 consented interviewees) from incidental contacts during observation in public spaces
• Consent: written consent for interviews; justify observation in public spaces as not requiring individual consent; ongoing consent checkpoints
• Data security: encrypted storage, pseudonyms, deductive disclosure plan for small vendor community
• Fieldwork contingencies: amendment plan for expanded sites or populations
• Generate full 13-section protocol narrative

Example 2: Digital ethnography protocol

Input: "I'm studying how online health communities discuss alternative treatments on Reddit. I want to analyze posts and do some interviews with active members. My IRB is asking about consent for the Reddit data."

Output approach:

• Load both reference files
• Set methods to digital ethnography + interviews
• Set risk posture to elevated (health information, searchable content)
• Key protocol issues: (a) treating publicly accessible posts as human subjects data when collected and curated systematically; (b) quoting policy to prevent searchable re-identification; (c) separate consent pathway for interview participants vs observed forum content
• Consent: interviews get standard consent; forum observation uses paraphrase-only policy with no verbatim quotes, no screenshots with usernames; justify approach using context-sensitive internet research ethics
• Data security: encrypted storage, separate key-file for pseudonym-handle links, no raw data in cloud services

Example 3: Oral history with archiving

Input: "I'm collecting oral histories from civil rights activists. Some want to be named. My IRB is confused about whether this is 'research' and how to handle the naming issue."

Output approach:

• Load both reference files
• Address research determination: state whether this meets the regulatory definition of research (systematic investigation designed to contribute to generalizable knowledge) and what the institution's expected pathway is
• Set consent modality to participant choice (pseudonym vs attribution)
• Protocol: narrators choose whether to be credited or anonymized; narrators review transcripts; narrators select access conditions for archived materials (open, restricted, embargoed)
• Archiving: specify repository, access conditions, and how consent language supports the planned deposit
• Risk: attribution may create long-term risk; anonymization may be experienced as erasure; frame as informed choice with clear explanation of implications