Informed Consent & Ethics Committees

Informed consent is the first Belmont principle, respect for persons, turned into a practical requirement. It is not a signature you collect and file; it is a genuine agreement, made by someone who understands what they are agreeing to and is free to say no. A form can be signed and still be worthless if any of those conditions is missing. This lesson unpacks what valid consent actually requires, who cannot give it alone, and what the ethics committee that reviews your study is really looking for.

The three things valid consent requires

A consent process is only valid when all three of these hold at once:

• Information. The participant is told what the study involves in enough detail to decide: its purpose, what they will actually do, how long it takes, the foreseeable risks and discomforts, the benefits (if any), how their data will be handled, and who to contact. Omitting or burying the uncomfortable parts breaks this.
• Comprehension. The information is presented so this particular person can understand it. A dense wall of legal jargon technically "informs" but fails comprehension; the test is whether they grasp it, not whether it is on the page.
• Voluntariness. The decision is free from coercion and undue influence. A participant must be able to decline, and to withdraw later, without penalty. Pressure from someone with power over them (an instructor, an employer, a doctor), or an incentive so large it overrides judgment, both undermine this.

Capacity, vulnerability, and assent

Consent assumes capacity, the ability to understand the information and reason about the decision. Some people cannot give full consent on their own: young children, people with significant cognitive impairment, or anyone whose situation limits their free choice. Belmont singles these groups out for extra protection, not exclusion.

Vulnerable populations (prisoners, patients dependent on the researcher, employees, people in crisis) need special care because the voluntariness condition is fragile for them. A prisoner offered early release, or a patient whose doctor is running the study, may not feel truly free to refuse.

For minors, the standard splits in two: a parent or guardian gives consent, and the child gives assent — an age-appropriate agreement in their own right. A ten-year-old cannot sign a legal consent form, but they can be told, in words they understand, what will happen and be allowed to decline. Both are required: a parent's yes does not override a child's no.

What an ethics committee actually does

Before you collect a single data point, a committee reviews your plan: an Institutional Review Board (IRB) in the U.S., or a Research Ethics Committee elsewhere. They are not there to judge whether your hypothesis is interesting. They check that the risks are justified by the likely benefits, that your consent process is genuine, that recruitment isn't coercive, that data will be protected, and that the burdens fall fairly. Review comes in three intensities, scaled to risk:

• Exempt: minimal-risk studies in low-stakes categories (e.g. anonymous surveys of adults, normal educational practice). Still submitted, but exempt from full review. "Exempt" is a decision the committee makes, not one you make for yourself.
• Expedited: no more than minimal risk, reviewed by one or two members rather than the full board (e.g. non-invasive measures, voice recordings, moderate surveys).
• Full board: anything above minimal risk, involving vulnerable groups, deception, or sensitive data goes to the whole committee at a convened meeting.

When consent is altered, waived, or arrives late

If consent were absolute, whole categories of research would be impossible: no study of deception, no analysis of a hospital's existing records, no reuse of a public survey whose respondents answered a decade ago. The regulations handle this by letting the committee, and only the committee, waive or alter the consent requirement. In the United States the criteria sit in 45 CFR 46.116(f), and all of them have to hold: the research is no more than minimal risk, it could not practicably be carried out without the waiver, identifiable data is genuinely necessary if identifiable data is used, the waiver will not adversely affect participants' rights and welfare, and where appropriate participants are given the missing information afterward.

Read that list against a deception study and you can see the shape of the argument a researcher has to make. Deception is usually an alteration rather than a waiver, because participants still consent to something; one element, the true purpose, is withheld with the committee's permission, and the last criterion on the list is what turns the debriefing from a courtesy into a condition of approval.

The same clause is why analyzing data you did not collect is usually workable. A dataset that is already public and carries no identifiers often falls outside human-subjects research altogether, and your committee will tell you so in writing. Existing identifiable records, a colleague's unpublished data, a clinic's files, are a different matter and go through the waiver route. The mistake is deciding either question yourself. "It's secondary data" is not a determination a student gets to make, and it takes an email to ask.

Audit a consent form

Below is a consent form a student submitted for a study on sleep and memory. Five of its lines have a real problem: a missing or broken element of valid consent. Click any line you think is defective to see why it matters and how to fix it. The compliant lines are there too; clicking one just confirms it's fine.