Ethics in research with human participants

A behavioral study uses a person as a source of evidence. That person can be bored, frightened, misled, exposed, or left with a false belief about what the study was. Research ethics is the set of constraints on those costs, and on who is asked to bear them, so that the knowledge is not purchased by treating participants as raw material. The constraints are not a mood of kindness added after the design is finished. A design that can answer its question only by hiding a serious risk has not yet earned the right to be run.

The modern U.S. framework for this judgment is summarized in the Belmont Report (National Commission, 1979). It states three principles. Respect for persons treats participants as agents who decide whether to take part, and it adds protection where the capacity to decide is limited. Beneficence requires a weighing of benefits against risks, and an obligation not to injure. Justice requires that the people who bear the burdens of research are not a different class from the people who receive its benefits. A study can satisfy one principle and fail another. A consent form does not repair a sample that was chosen because the participants were unlikely to refuse.

Psychology’s own code, the APA Ethical Principles of Psychologists and Code of Conduct, applies these ideas to the work psychologists actually do, including research (American Psychological Association, 2017). Methods courses teach the research portions because they change the procedure: what must be said before a session, what may be withheld, what must be said after, and who must approve the plan. Cozby and Bates walk the same ground as the working rules of a methods student (Cozby & Bates, 2020).

Informed consent is a decision to participate, made by a person who has been told what the study involves in language that person can use, including the procedures, the risks and benefits that can reasonably be foreseen, the fact that withdrawal is allowed, and the limits of confidentiality. Consent is a process, not a signature as a souvenir. A form written in the dialect of a grant application can fail consent even if every legal noun is present, because the participant could not have used it to decide.

Some studies qualify for a waiver or an alteration of consent, and the waiver is itself a decision an ethics committee makes, not a convenience the researcher declares. The usual grounds are that the research could not practicably be carried out with ordinary consent, that the risk is minimal, and that the waiver does not strip a right or a welfare interest. Observing public behavior, where people have no expectation of privacy and the researcher does not intervene, is often treated differently from bringing a person into a lab and asking for an hour of their time. The difference is the intrusion and the risk, not the researcher’s curiosity.

Assent is the agreement of a person who is not in a position to give full legal consent, typically a child, alongside the permission of a parent or guardian. Assent is not a smaller signature with the same force as adult consent. It is a recognition that respect for persons still applies when legal authority sits with someone else. A child who wants to stop is a person who wants to stop. The procedure has to make stopping possible, including for participants who are unused to refusing an adult in a professional room.

Risk in this context includes physical harm, psychological distress, damage to reputation, and the loss of privacy. Minimal risk means the probability and magnitude of harm are no greater than those encountered in ordinary life or in routine tests. The phrase is a comparison, not a compliment. A procedure can be familiar and still be more than minimal for the particular people in the study. Embarrassment, a revival of a painful memory, and the revelation of illegal behavior are risks even when no one is touched.

Deception is the deliberate withholding or misrepresentation of information about the study. It is used when telling the full purpose would destroy the behavior the study needs to observe, for example when people would perform the “good participant” answer instead of the answer the question is about. Deception is not a default and it is not justified by drama. The APA code allows it when the study has significant value, when effective nondeceptive methods are not feasible, and when the deception is not about risks that would affect the decision to take part. Pain that was lied about is not a clever manipulation. It is a consent failure.

Debriefing is the explanation owed at the end, and it is obligatory when deception was used. A debriefing names the true purpose, corrects the false belief, and gives the participant a chance to ask questions and, where the rules allow, to withdraw their data once they know what the study was. A debriefing that is a hurried paragraph on the way out the door does not do that job. The historical pressure on this rule is real. Milgram’s obedience experiments led participants to believe they were delivering dangerous shocks; in the 1963 report, 26 of 40 participants continued to the final switch (Milgram, 1963). Whatever one concludes about the scientific value, the distress and the false belief are the ethical facts that changed how later studies have to be justified.

Confidentiality is the agreement that identifiable information will not be spread beyond the uses the participant was told about. Anonymity is stronger: the data cannot be connected to the person at all, not even by the researcher. A coded list that the researcher keeps in a drawer is confidentiality, not anonymity. The distinction matters when the data are sensitive. A promise of anonymity that is false, because IP addresses, student numbers, or a video still exist, is a deception about risk.

Justice shows up in recruitment. The Tuskegee syphilis study is the standard warning in U.S. training: for decades, Black men with syphilis were observed and treatment was withheld even after penicillin was known to work, so that the “natural course” could be recorded. The benefits of the knowledge, such as they were, did not flow to the men who bore the harm. A modern study that recruits only from people who need the payment, or only from students who need the course credit, has to answer a smaller version of the same question. Is this population being asked because the scientific question is about them, or because they are available and unlikely to make trouble?

Special protections attach where consent is easily bent: children, prisoners, people with impaired decision-making, and pregnant women in biomedical rules. Psychology students in a participant pool are not a protected class in that legal sense, and they are still a captive audience if the alternative to participation is a penalty. A fair pool offers an alternative way to meet the requirement. The principle is the same one as in the Belmont statement of respect. The person should be able to decline without paying for the decline with a grade, a job, or a parole decision.

An institutional review board (IRB), sometimes named a research ethics committee, reviews proposals involving human participants before the work starts. The review is prospective on purpose. A moving apology after a harmful session is not a review. Boards sort proposals by risk. Exempt and expedited paths exist for lower-risk work. Full board review is for the rest. The categories are bureaucratic and they are also the mechanism that stops a single laboratory from being the only judge in its own case.

The researcher’s write-up for the board should match the procedure that will actually be run: the recruitment words, the consent language, the deception if any, the data that will be stored, and the debriefing. A clever change made after approval, because the first sessions were “more interesting” with the change, is a new study from the board’s point of view. Students in a methods class are inside this system even when the project feels small. A course project that asks strangers about trauma, records them, or misleads them is not small. The instructor and the board are part of the method.

Ethics does not replace the scientific standards of the previous chapters. A kind study with no comparison and no falsifiable question is still not evidence. A brilliant design that humiliates people to get a clean effect has failed a different and prior test. The chapters after this one return to variables, measurement, and design. They assume the people in the procedure were treated as people. When a published study seems to have skipped that assumption, the methods reader does not borrow its procedure for the sake of the citation.