5 Consent and Trust
Where this chapter fits. Steps 4 and 6 of the research process both involve the people who take part. This chapter covers the consent text and the protection of participants during and after the interview.
5.1 A Signature Some People Cannot Give
A community health center wants to learn why some immigrant families in its area do not use a free dental program for children. The case is a composite written for this book.
A researcher plans 20 interviews with parents. She prepares the consent form that her organization uses for every study. It runs four pages. It asks for the participant’s printed name, signature, address, and date.
The first three parents she approaches decline. A staff member who knows the families explains why. Some of the parents do not have legal status, and others have relatives who do not. They have learned to be careful about any paper that carries their name and address. The form that was written to protect them is the reason they will not take part. The signed form would also be the only record that links their names to the study.
The researcher returns to her review board with a different plan. She will explain the study aloud, in the parent’s language, from a one-page script. Parents will agree by voice and sign nothing. She will record no names, and each interview will carry a number. The board approves the change, and parents begin to agree.
The researcher did not relax the ethics of the study. The second plan protects participants better than the first. This chapter covers what a researcher owes the people who take part. It also covers how to meet those duties in a way that fits the people and the setting.
5.2 Three Principles
Modern research ethics in the United States grew out of a public scandal. From 1932 to 1972, the US Public Health Service ran a study in Tuskegee, Alabama. It followed 399 Black men who had syphilis, to observe the disease without treatment. The men were not told their diagnosis. They were not offered penicillin when it became the standard treatment in the 1940s. The study ended only after a news report in 1972.
Congress responded by creating a national commission. In 1979, the commission published the Belmont Report, which states three principles.
- Respect for persons. People decide for themselves whether to take part, with full information. People who are less able to protect their own interests receive extra protection.
- Beneficence. The researcher works to reduce possible harm and to increase possible benefit.
- Justice. The burdens and benefits of research are shared fairly. A group should not carry the risks of a study because it is easy to recruit or unable to refuse.
An institutional review board applies these principles. It is a committee that reviews a study with human participants before the study begins. Universities, hospitals, and many agencies have one. A researcher who works outside such an institution is still bound by the principles.
5.3 Informed Consent
Informed consent means that a person agrees to take part after understanding what the study involves. It has three requirements.
Information. The person is told the purpose of the study, what they will be asked to do, and how long it will take. The person is also told the risks and benefits and how their information will be protected.
Understanding. The information is given in a form the person can follow. This means plain words, in the person’s own language, at a pace that allows questions.
Voluntary choice. The person is free to say no and free to stop. No one with power over the person applies pressure.
Many consent forms meet the first requirement and fail the second. A four-page form in legal language informs no one. A researcher should write the consent text for the reader with the least schooling among the participants. Table 5.1 shows three sentences from a standard form and a plain version of each.
| Standard wording | Plain wording |
|---|---|
| “Participation in this research is entirely voluntary, and you may withdraw at any time without penalty or loss of benefits to which you are otherwise entitled.” | “You do not have to take part. You can stop at any time. If you say no, your services will not change.” |
| “Audio recordings will be retained in a secure repository and destroyed upon completion of the transcription process.” | “I will keep the recording in a locked file. I will delete it after I type up what you said.” |
| “There are no anticipated risks beyond those encountered in daily life.” | “Some questions may bring up hard memories. You decide what to share.” |
The third row shows a second problem. The standard sentence says the study has no special risks. For a parent without legal status, an interview about family and health carries a risk. A consent text should name the risks that apply to these participants.
A full plain-language template is in the book’s files as consent-template.txt.
Consent is also a process. A participant agreed at the start of the interview without knowing what would come up. The researcher can check again during the conversation. “Is it all right if I ask about that?” takes five seconds. At the end, the researcher can ask whether there is anything the participant would like left out.
5.4 Written and Spoken Consent
Consent is usually recorded with a signature. Review boards can approve spoken consent, with no signature, in two situations. In the first, the signed form would be the only record that links a person to the study. The main risk to the person is that the link becomes known. In the second, the study involves little risk, and the same activity outside research would not normally require a signature.
The dental study fits the first situation. With spoken consent, the researcher reads a script, answers questions, and notes the date and that the person agreed. She keeps no name.
When participants speak another language, the consent text should be translated by a person fluent in both languages and checked by a second person. If an interpreter takes part in the interview, the interpreter is bound by the same promise of confidentiality. A family member should not interpret, and a child should never do so. A parent will not speak freely about some topics through a relative.
5.5 Confidentiality in Practice
Chapter 7 of Before the Numbers separates two terms. A study is anonymous if no one, including the researcher, can link the data to a person. A study is confidential if the researcher can make the link and promises not to reveal it. Most interview studies are confidential. The researcher met the participant and knows who said what.
A promise of confidentiality has to be kept through the whole life of the data. Five practices cover most of it.
- Separate names from data. Give each participant a code number. Keep the list that links codes to names in a different place from the transcripts, and delete it when it is no longer needed.
- Remove identifying details from transcripts. Replace names of people, employers, schools, and streets with labels in brackets.
- Protect the files. Store recordings and transcripts on an encrypted drive. Do not send them by ordinary email.
- Limit who has access. State who will hear the recordings and read the transcripts, and hold to it.
- Set a date to delete the recordings. A voice identifies a person. Once the transcript has been checked, the recording is usually no longer needed.
A pseudonym is an invented name that replaces a participant’s real one in a report. Choose pseudonyms that do not point back to the person. Do not use the same initials. In a small group, do not keep a detail that only one person fits.
5.6 When a Name Is Not the Only Clue
Removing a name does not always protect a person. Deductive disclosure occurs when a reader can work out who a participant is from the details in a report (Kaiser, 2009).
Consider this synthetic excerpt.
Synthetic “I work nights at the poultry plant. My husband was deported in March, so now it’s me and the four kids. The dentist is only open when I’m asleep.”
The excerpt has no name in it. In a town with one poultry plant, a neighbor could still identify a night-shift worker with four children whose husband was deported in March. The details that make the quotation vivid are the same details that identify her.
A researcher has three options, and each has a cost.
- Generalize the details. “The poultry plant” becomes “a factory.” “Four kids” becomes “several children.” “March” becomes “this year.” The quotation is safer and less specific.
- Remove the risky clause. The quotation can keep the point about the dentist’s hours and drop the deportation.
- Paraphrase. The researcher reports the point in her own words and does not quote.
A fourth step helps with all three. Ask the participant. At the end of the interview, the researcher can say which parts she may want to quote. She can then ask whether any of it should be changed or left out.
Any change to a quotation should be made by rule and reported. The methods section should state that identifying details were altered and should describe the kinds of changes. It should never present an altered quotation as exact.
5.7 Work With the Data: Identifying Details Checker
A quotation can identify a speaker with no name in it. The tool marks details in a quotation that could help a reader work out who said it. Do not paste real names or real quotations from a study that is under way.
Open Identifying Details Checker in its own window
- Press “Check the quotation” for the sample. The tool marks five details.
- For each marked detail, choose one of the three changes described above. Make it more general, cut the clause, or paraphrase.
- Type your revised quotation and check it again.
- Read the revision as a neighbor of the speaker would. Decide whether that reader could still tell who it is.
The tool uses fixed rules and knows nothing about the setting. In a small town, a detail that the tool leaves unmarked can identify a person. The last step is the one that protects the participant.
A one-page consent text in plain language is in the file consent-template.txt at qual.marginoferrormedia.com/corpus.
5.8 The Limits of a Promise
A researcher should promise only what she can deliver. Two limits apply in many studies.
Mandatory reporting. In most US states, members of certain professions are required to report suspected child abuse. If the researcher is a mandatory reporter, the consent text has to say so before the interview begins.
Legal demands. Research records can be subpoenaed. For studies funded by the National Institutes of Health, a Certificate of Confidentiality protects identifiable data from most legal demands. Researchers with other funding can apply for one when a study collects sensitive information. A study that collects no names, like the dental study, has a simpler protection where the design allows it.
In a focus group, the researcher cannot promise confidentiality at all. She can promise her own silence and ask the group to keep what is said in the room. She cannot guarantee that eight other people will do so. The consent text should say this.
5.9 Power, Payment, and Position
An interview takes place between two people who are not equal. The researcher chose the topic, will write the report, and often comes from an institution with standing. The participant may depend on a related institution for services. Three practices reduce the effects of that imbalance.
Keep the service provider out of recruitment. Chapter 4 described the problem of a caseworker who recruits her own clients. The health center in this case should not ask its dentists or front-desk staff to recruit.
Pay fairly and without conditions. A payment thanks the participant for their time. It becomes a problem when it is large enough to lead a person to accept a risk they would otherwise refuse. Review boards call this undue inducement. Pay an amount in line with an hour or two of local wages. Pay it whether or not the participant finishes the interview.
Examine your own position. Positionality is the researcher’s social position in relation to the participants, including language, class, race, citizenship, and role. An insider shares the participants’ background. An outsider does not. Each has advantages. Participants may speak more openly with an insider. They may also leave unsaid what “everyone knows.” They may explain more to an outsider and trust the outsider less. A researcher should state her position in the report and consider how it affected what she was told.
5.10 Giving Something Back
Participants give time and trust. A study that takes both and returns nothing has used them. Three forms of return are practical in most studies.
Send participants a short summary of the findings in plain language. Report the findings to the organization in a form it can act on. The dental study could recommend evening hours and a statement that the clinic asks no questions about legal status. Where participants are willing, invite a few of them to react to the draft findings before the report is final. Chapter 8 covers this last practice as a check on the analysis.
Read the consent text aloud to someone outside your field. Then ask three questions. What will happen if you say yes? What could go wrong? What happens if you change your mind? If the person cannot answer all three, rewrite the text.
A participant reads the final report and finds her own words. Her name is not there, and no outsider could identify her. She recognizes herself at once, and so would her sister. The quotation describes a private matter, and it sits under a heading she finds unflattering. No rule was broken, and she still feels exposed. Confidentiality protects a participant from being identified by others. It does not protect her from how it feels to see her own account in print. A researcher should choose quotations and headings as if each participant will read the report, because some of them will.
Language models can transcribe recordings, summarize interviews, and suggest codes. Each of these uses sends participants’ words to a company’s servers. That is a disclosure to a third party, and it falls under the promise the researcher made. Before using any such tool, check three things. Does the provider store the text, and for how long? Can the provider use it to train models? Did the consent text tell participants that such a service would be used? If the answer to the third question is no, remove identifying details first, or do not upload the text. Some institutions provide approved tools that keep the data inside the institution.
5.11 Putting It All Together
The Belmont Report states three principles. They are respect for persons, beneficence, and justice. Informed consent puts the first into practice. It requires information, understanding, and a voluntary choice. A consent text should be written in plain words and should name the risks that apply to the participants in the study.
Most interview studies are confidential and not anonymous. The researcher separates names from data, removes identifying details, protects the files, and deletes recordings on a stated date. A quotation with the name removed can still identify a person through its details. The researcher can generalize the details, cut a clause, or paraphrase, and she reports that she did so.
A researcher promises only what she can deliver, keeps service providers out of recruitment, pays fairly, and states her own position. She returns the findings to the people who made them possible.
5.12 Looking Ahead
The interviews are complete, and the transcripts are checked and stripped of identifying details. The researcher now has a few hundred pages of text. Chapter 6 covers the first step of analysis, which is coding. The material is the set of 12 synthetic housing interviews.
5.13 Further Reading and References
- National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research (1979). The Belmont Report. US Department of Health, Education, and Welfare.
- Jones, J. H. (1993). Bad Blood: The Tuskegee Syphilis Experiment (new ed.). Free Press. The standard history of the study.
- Kaiser, K. (2009). Protecting respondent confidentiality in qualitative research. Qualitative Health Research, 19(11), 1632 to 1641. The source of the discussion of deductive disclosure.
- Saunders, B., Kitzinger, J., and Kitzinger, C. (2015). Anonymising interview data: Challenges and compromise in practice. Qualitative Research, 15(5), 616 to 632.
- Dwyer, S. C., and Buckle, J. L. (2009). The space between: On being an insider-outsider in qualitative research. International Journal of Qualitative Methods, 8(1), 54 to 63.
- US Department of Health and Human Services. Federal Policy for the Protection of Human Subjects (45 CFR 46). The regulation that governs review boards and consent.
5.14 Key Terms
- Anonymous. No one, including the researcher, can link the data to a person.
- Beneficence. The duty to reduce possible harm and increase possible benefit.
- Certificate of Confidentiality. A federal protection that shields identifiable research data from most legal demands.
- Confidential. The researcher can link the data to a person and promises not to reveal the link.
- Deductive disclosure. Identifying a participant from the details in a report.
- Informed consent. Agreement to take part after understanding what a study involves.
- Insider. A researcher who shares the participants’ background.
- Institutional review board. A committee that reviews research with human participants before it begins.
- Justice. The fair sharing of the burdens and benefits of research.
- Outsider. A researcher who does not share the participants’ background.
- Positionality. The researcher’s social position in relation to the participants.
- Pseudonym. An invented name that replaces a participant’s real name.
- Respect for persons. The principle that people decide for themselves whether to take part.
- Undue inducement. A payment large enough to lead a person to accept a risk they would otherwise refuse.
5.15 Exercises
Check Your Understanding
- Name the three principles of the Belmont Report and give one practice that follows from each.
- Why did the review board approve spoken consent for the dental study?
- Explain deductive disclosure. Why is removing a name not always enough?
- Why can a researcher not promise confidentiality in a focus group?
Apply It
- Rewrite this sentence in plain language. “Data will be reported in aggregate, and no individually identifiable information will be disseminated.”
- Revise the synthetic excerpt about the poultry plant so that it could be published. Then state what you changed and what was lost.
- Fill in
consent-template.txtfor a study of how college students manage the cost of food. Name one risk that applies to those participants.
Think Deeper
- A participant says, during the interview, “Please don’t write this down,” and then describes something central to your research question. What do you do with what you heard?
- You are a former client of the program you are now studying. List two ways this could help the interviews and two ways it could distort them. How would you describe your position in the report?
- A colleague wants to paste full transcripts into a public AI tool to get a quick summary. The consent form did not mention any such tool. Write a short reply that explains the problem and suggests two alternatives.