What is in a packet, and who reads what
A submission is not one document. It is a small set, and different reviewers spend their attention in different parts of it.
- The protocol. Purpose, design, procedures, sample and analysis. The methodological reviewer lives here.
- The consent documents. Read closely, and often the only part a community member reviews in detail.
- Recruitment materials. Every flyer, email, script and posting, exactly as a participant will encounter them.
- Instruments. Surveys, interview guides and observation forms in full, rather than described in a sentence.
- Permissions and training. Site letters, plus certification such as CITI, current at the date you submit.
Risk is about your procedures, not your topic
New researchers describe a sensitive subject and conclude the study is high risk. Boards reason in the opposite direction. They read what you will actually do to a person: what you ask, how long it takes, what happens to someone who becomes distressed, and what could follow them out of the room.
So write risk as a list of procedures with a mitigation attached to each. A survey about workplace stress carries a risk of discomfort, mitigated by the right to skip any item and a stated stopping point. A study run inside your own workplace carries a risk of coercion, mitigated by recruiting through someone with no authority over the participants. Naming the mitigation is what turns a concern into an approved procedure.
Write the section for a reader outside your field. Boards contain members from other disciplines by design, and a procedure that seems routine inside your specialty may need one plain sentence of explanation. That sentence costs nothing and prevents a query.
Recruitment, where the returns come from
Boards ask three things about recruitment and most protocols answer only the first. Who is eligible. How will they hear about the study. And why can they say no at no cost to themselves.
The third question is where the revision letters go. Recruiting your own students, your own patients or your own staff raises a coercion problem that has to be answered in the document rather than in a follow-up email. Describe the distance you are putting between yourself and the decision, and describe what a participant sees, hears and receives at every step.
Data handling, written last and returned first
This section is usually drafted in a hurry and read carefully, which is a poor combination. Answer all five of these inside the protocol itself.
- What identifiers you collect, and whether each one is genuinely needed.
- When identifiers are separated from responses, and whether any link between them is kept.
- Where the data lives, on which device or service, and who else can reach it.
- How long it is retained, and what happens at the end of that period.
- What appears in published results, and how small subgroups are prevented from identifying anyone.
The consistency pass
Before submitting, read the packet as one document. The sample size in the protocol has to match the consent form. The time commitment in the recruitment email has to match the interview guide. The procedures described in the protocol have to match what the instrument actually asks.
Reviewers notice mismatches immediately, and a mismatch reads as carelessness in a document whose entire purpose is care. Run the pass twice, once for numbers and once for procedures. Where a packet has already come back more than once, a specialist who has sat on a review board can usually name the problem in a single reading, which is the point of IRB support.
Questions
How specific does the analysis section need to be?
Specific enough that a reviewer can see the data you collect will answer the question you asked. Name the comparisons, the tests or the coding approach, and the sample the design requires. Vague analysis language raises a subtler concern than it appears to: a study that cannot answer its own question is asking people for time without a benefit.
Can I change the study after approval?
Yes, through an amendment, and the rule is that changes are approved before they happen rather than reported afterward. Adding a survey question, changing where you recruit, or extending the sample all qualify. Keep a short log of every change with its approval date. That log makes an audit straightforward and becomes an appendix in the finished document.
Who should be listed on the protocol?
Everyone who will interact with participants or handle identifiable data, with their role and their training status. Boards check that the person conducting an interview is trained and named. Leaving an assistant off the protocol and then using one is a common finding, and it is entirely avoidable at the point of submission.
What if the board asks for something my chair disagrees with?
The board's requirement governs, and the conversation is worth having in writing. Ask the reviewer which standard the request comes from, then take that answer to your chair. Most disagreements resolve into a wording change rather than a design change. Keep the exchange and attach it to the resubmission so the reasoning travels with the packet.
How long should a protocol be?
Long enough to answer the board's questions and no longer. Padding a protocol with literature review material buries the procedures and slows the review down. Write the sections the template asks for, in the order it asks for them, and keep background to a paragraph rather than a chapter. Reviewers move faster through documents they can navigate.
