Interviews in the social sciences

Here we address research design considerations and data collection issues focusing on topic guide construction and other pragmatics of the interview. We also explore issues of ethics and reflexivity that are crucial throughout the research project.

Research design

Participant selection

Participants can be selected and recruited in various ways for in-depth interview studies. The researcher must first decide what defines the people or social groups being studied. Often, this means moving from an abstract theoretical research question to a more precise empirical one. For example, the researcher might be interested in how people talk about race in contexts of diversity. Empirical settings in which this issue could be studied could include schools, workplaces or adoption agencies. The best research designs should clearly explain why the particular setting was chosen. Often there are both intrinsic and extrinsic reasons for choosing to study a particular group of people at a specific time and place³. Intrinsic motivations relate to the fact that the research is focused on an important specific social phenomenon that has been understudied. Extrinsic motivations speak to the broader theoretical research questions and explain why the case at hand is a good one through which to address them empirically.

Next, the researcher needs to decide which types of people they would like to interview. This decision amounts to delineating the inclusion and exclusion criteria for the study. The criteria might be based on demographic variables, like race or gender, but they may also be context-specific, for example, years of experience in an organization. These should be decided based on the research goals. Researchers should be clear about what characteristics would make an individual a candidate for inclusion in the study (and what would exclude them).

The next step is to identify and recruit the study’s sample. Usually, many more people fit the inclusion criteria than can be interviewed. In cases where lists of potential participants are available, the researcher might want to employ stratified sampling, dividing the list by characteristics of interest before sampling.

When there are no lists, researchers will often employ purposive sampling. Many researchers consider purposive sampling the most useful mode for interview-based research since the number of interviews to be conducted is too small to aim to be statistically representative⁴. Instead, the aim is not breadth, via representativeness, but depth via rich insights about a set of participants. In addition to purposive sampling, researchers often use snowball sampling. Both purposive and snowball sampling can be combined with quota sampling. All three types of sampling aim to ensure a variety of perspectives within the confines of a research project. A goal for in-depth interview studies can be to sample for range, being mindful of recruiting a diversity of participants fitting the inclusion criteria.

Study design

The total number of interviews depends on many factors, including the population studied, whether comparisons are to be made and the duration of interviews. Studies that rely on quota sampling where explicit comparisons are made between groups will require a larger number of interviews than studies focused on one group only. Studies where participants are interviewed over several hours, days or even repeatedly across years will tend to have fewer participants than those that entail a one-off engagement.

Researchers often stop interviewing when new interviews confirm findings from earlier interviews with no new or surprising insights (saturation)^4,5,6. As a criterion for research design, saturation assumes that data collection and analysis are happening in tandem and that researchers will stop collecting new data once there is no new information emerging from the interviews. This is not always possible. Researchers rarely have time for systematic data analysis during data collection and they often need to specify their sample in funding proposals prior to data collection. As a result, researchers often draw on existing reports of saturation to estimate a sample size prior to data collection. These suggest between 12 and 20 interviews per category of participant (although researchers have reported saturation with samples that are both smaller and larger than this)^7,8,9. The idea of saturation has been critiqued by many qualitative researchers because it assumes that meaning inheres in the data, waiting to be discovered — and confirmed — once saturation has been reached⁷. In-depth interview data are often multivalent and can give rise to different interpretations. The important consideration is, therefore, not merely how many participants are interviewed, but whether one’s research design allows for collecting rich and textured data that provide insight into participants’ understandings, accounts, perceptions and interpretations.

Sometimes, researchers will conduct interviews with more than one participant at a time. Researchers should consider the benefits and shortcomings of such an approach. Joint interviews may, for example, give researchers insight into how caregivers agree or debate childrearing decisions. At the same time, they may be less adaptive to exploring aspects of caregiving that participants may not wish to disclose to each other. In other cases, there may be more than one person interviewing each participant, such as when an interpreter is used, and so it is important to consider during the research design phase how this might shape the dynamics of the interview.

Data collection

Semi-structured interviews are typically organized around a topic guide comprised of an ordered set of broad topics (usually 3–5). Each topic includes a set of questions that form the basis of the discussion between the researcher and participant (Fig. 1). These topics are organized around key concepts that the researcher has identified (for example, through a close study of prior research, or perhaps through piloting a small, exploratory study)⁵.

Topic guide

One common way to structure a topic guide is to start with relatively easy, open-ended questions (Table 1). Opening questions should be related to the research topic but broad and easy to answer, so that they help to ease the participant into conversation.

After these broad, opening questions, the topic guide may move into topics that speak more directly to the overarching research question. The interview questions will be accompanied by probes designed to elicit concrete details and examples from the participant (see Table 1).

Abstract questions are often easier for participants to answer once they have been asked more concrete questions. In our experience, for example, questions about feelings can be difficult for some participants to answer, but when following probes concerning factual experiences these questions can become less challenging. After the main themes of the topic guide have been covered, the topic guide can move onto closing questions. At this stage, participants often repeat something they have said before, although they may sometimes introduce a new topic.

Interviews are especially well suited to gaining a deeper insight into people’s experiences. Getting these insights largely depends on the participants’ willingness to talk to the researcher. We recommend designing open-ended questions that are more likely to elicit an elaborated response and extended reflection from participants rather than questions that can be answered with yes or no.

Questions should avoid foreclosing the possibility that the participant might disagree with the premise of the question. Take for example the question: “Do you support the new family-friendly policies?” This question minimizes the possibility of the participant disagreeing with the premise of this question, which assumes that the policies are ‘family-friendly’ and asks for a yes or no answer. Instead, asking more broadly how a participant feels about the specific policy being described as ‘family-friendly’ (for example, a work-from-home policy) allows them to express agreement, disagreement or impartiality and, crucially, to explain their reasoning¹⁰.

For an uninterrupted interview that will last between 90 and 120 minutes, the topic guide should be one to two single-spaced pages with questions and probes. Ideally, the researcher will memorize the topic guide before embarking on the first interview. It is fine to carry a printed-out copy of the topic guide but memorizing the topic guide ahead of the interviews can often make the interviewer feel well prepared in guiding the participant through the interview process.

Although the topic guide helps the researcher stay on track with the broad areas they want to cover, there is no need for the researcher to feel tied down by the topic guide. For instance, if a participant brings up a theme that the researcher intended to discuss later or a point the researcher had not anticipated, the researcher may well decide to follow the lead of the participant. The researcher’s role extends beyond simply stating the questions; it entails listening and responding, making split-second decisions about what line of inquiry to pursue and allowing the interview to proceed in unexpected directions.

Optimizing the interview

The ideal place for an interview will depend on the study and what is feasible for participants. Generally, a place where the participant and researcher can both feel relaxed, where the interview can be uninterrupted and where noise or other distractions are limited is ideal. But this may not always be possible and so the researcher needs to be prepared to adapt their plans within what is feasible (and desirable for participants).

Another key tool for the interview is a recording device (assuming that permission for recording has been given). Recording can be important to capture what the participant says verbatim. Additionally, it can allow the researcher to focus on determining what probes and follow-up questions they want to pursue rather than focusing on taking notes. Sometimes, however, a participant may not allow the researcher to record, or the recording may fail. If the interview is not recorded we suggest that the researcher takes brief notes during the interview, if feasible, and then thoroughly make notes immediately after the interview and try to remember the participant’s facial expressions, gestures and tone of voice. Not having a recording of an interview need not limit the researcher from getting analytical value from it.

As soon as possible after each interview, we recommend that the researcher write a one-page interview memo comprising three key sections. The first section should identify two to three important moments from the interview. What constitutes important is up to the researcher’s discretion⁹. The researcher should note down what happened in these moments, including the participant’s facial expressions, gestures, tone of voice and maybe even the sensory details of their surroundings. This exercise is about capturing ethnographic detail from the interview. The second part of the interview memo is the analytical section with notes on how the interview fits in with previous interviews, for example, where the participant’s responses concur or diverge from other responses. The third part consists of a methodological section where the researcher notes their perception of their relationship with the participant. The interview memo allows the researcher to think critically about their positionality and practice reflexivity — key concepts for an ethical and transparent research practice in qualitative methodology^11,12.

Ethics and reflexivity

All elements of an in-depth interview can raise ethical challenges and concerns. Good ethical practice in interview studies often means going beyond the ethical procedures mandated by institutions¹³. While discussions and requirements of ethics can differ across disciplines, here we focus on the most pertinent considerations for interviews across the research process for an interdisciplinary audience.

Ethical considerations prior to interview

Before conducting interviews, researchers should consider harm minimization, informed consent, anonymity and confidentiality, and reflexivity and positionality. It is important for the researcher to develop their own ethical sensitivities and sensibilities by gaining training in interview and qualitative methods, reading methodological and field-specific texts on interviews and ethics and discussing their research plans with colleagues.

Researchers should map the potential harm to consider how this can be minimized. Primarily, researchers should consider harm from the participants’ perspective (Box 1). But, it is also important to consider and plan for potential harm to the researcher, research assistants, gatekeepers, future researchers and members of the wider community¹⁴. Even the most banal of research topics can potentially pose some form of harm to the participant, researcher and others — and the level of harm is often highly context-dependent. For example, a research project on religion in society might have very different ethical considerations in a democratic versus authoritarian research context because of how openly or not such topics can be discussed and debated¹⁵.

The researcher should consider how they will obtain and record informed consent (for example, written or oral), based on what makes the most sense for their research project and context¹⁶. Some institutions might specify how informed consent should be gained. Regardless of how consent is obtained, the participant must be made aware of the form of consent, the intentions and procedures of the interview and potential forms of harm and benefit to the participant or community before the interview commences. Moreover, the participant must agree to be interviewed before the interview commences. If, in addition to interviews, the study contains an ethnographic component, it is worth reading around this topic (see, for example, Murphy and Dingwall¹⁷). Informed consent must also be gained for how the interview will be recorded before the interview commences. These practices are important to ensure the participant is contributing on a voluntary basis. It is also important to remind participants that they can withdraw their consent at any time during the interview and for a specified period after the interview (to be decided with the participant). The researcher should indicate that participants can ask for anything shared to be off the record and/or not disseminated.

In terms of anonymity and confidentiality, it is standard practice when conducting interviews to agree not to use (or even collect) participants’ names and personal details that are not pertinent to the study. Anonymizing can often be the safer option for minimizing harm to participants as it is hard to foresee all the consequences of de-anonymizing, even if participants agree. Regardless of what a researcher decides, decisions around anonymity must be agreed with participants during the process of gaining informed consent and respected following the interview.

Although not all ethical challenges can be foreseen or planned for¹⁸, researchers should think carefully — before the interview — about power dynamics, participant vulnerability, emotional state and interactional dynamics between interviewer and participant, even when discussing low-risk topics. Researchers may then wish to plan for potential ethical issues, for example by preparing a list of relevant organizations to which participants can be signposted. A researcher interviewing a participant about debt, for instance, might prepare in advance a list of debt advice charities, organizations and helplines that could provide further support and advice. It is important to remember that the role of an interviewer is as a researcher rather than as a social worker or counsellor because researchers may not have relevant and requisite training in these other domains.

Ethical considerations post-interview

Researchers should consider how interview data are stored, analysed and disseminated. If participants have been offered anonymity and confidentiality, data should be stored in a way that does not compromise this. For example, researchers should consider removing names and any other unnecessary personal details from interview transcripts, password-protecting and encrypting files and using pseudonyms to label and store all interview data. It is also important to address where interview data are taken (for example, across borders in particular where interview data might be of interest to local authorities) and how this might affect the storage of interview data.

Examining how the researcher will represent participants is a paramount ethical consideration both in the planning stages of the interview study and after it has been conducted. Dissemination strategies also need to consider questions of anonymity and representation. In small communities, even if participants are given pseudonyms, it might be obvious who is being described. Anonymizing not only the names of those participating but also the research context is therefore a standard practice¹⁹. With particularly sensitive data or insights about the participant, it is worth considering describing participants in a more abstract way rather than as specific individuals. These practices are important both for protecting participants’ anonymity but can also affect the ability of the researcher and others to return ethically to the research context and similar contexts²⁰.

Reflexivity and positionality

Reflexivity and positionality mean considering the researcher’s role and assumptions in knowledge production¹³. A key part of reflexivity is considering the power relations between the researcher and participant within the interview setting, as well as how researchers might be perceived by participants. Further, researchers need to consider how their own identities shape the kind of knowledge and assumptions they bring to the interview, including how they approach and ask questions and their analysis of interviews (Box 2). Reflexivity is a necessary part of developing ethical sensibility as a researcher by adapting and reflecting on how one engages with participants. Participants should not feel judged, for example, when they share information that researchers might disagree with or find objectionable. How researchers deal with uncomfortable moments or information shared by participants is at their discretion, but they should consider how they will react both ahead of time and in the moment.

Researchers can develop their reflexivity by considering how they themselves would feel being asked these interview questions or represented in this way, and then adapting their practice accordingly. There might be situations where these questions are not appropriate in that they unduly centre the researchers’ experiences and worldview. Nevertheless, these prompts can provide a useful starting point for those beginning their reflexive journey and developing an ethical sensibility.

Reflexivity and ethical sensitivities require active reflection throughout the research process. For example, researchers should take care in interview memos and their notes to consider their assumptions, potential preconceptions, worldviews and own identities prior to and after interviews (Box 2). Checking in with assumptions can be a way of making sure that researchers are paying close attention to their own theoretical and analytical biases and revising them in accordance with what they learn through the interviews. Researchers should return to these notes (especially when analysing interview material), to try to unpack their own effects on the research process as well as how participants positioned and engaged with them.