Interviews and Focus Groups

Generally, qualitative data is collected from a smaller number of participants compared with a survey. The exact number of interviews or focus groups to conduct depends on the research question(s), the ability to identify and recruit the most pertinent participants, and when saturation is attained (i.e., no new information is obtained). However, conducting 20-30 individual interviews (or the focus group equivalent) is typically enough. It is also possible to answer research questions by collecting data from the population of interest and reach saturation in as few as 10-15 interviews. MAD consultants can provide guidance on the appropriate number of interviews for a given project.

Though qualitative methods have been implemented on significantly larger scales (e.g., 50-125 interviews), this is generally not recommended. When this scale is desired, surveys are often the better data collection tool. Alternatively, other purposes for large scale participant contact may exist, such as stakeholder engagement. While other reasons for participant contact are valid, separating them from data collection is recommended so that the integrity of both efforts is maintained and so that the budgetary impacts of large scale qualitative data collection are reduced, as such efforts are unlikely to yield new or different results compared with smaller scale qualitative data collection.

When data on potential participants is readily available, developing a sample of individuals best suited to help answer the research question(s) is recommended. The exact sample specifications will depend on the research questions at hand, but may include broad characteristics (e.g., geographic location, gender identity, educational attainment) and/or topic-specific characteristics (e.g., type of birth defect, custodial status, health insurance enrollment, incarceration or release status).

Participants may also be recruited for qualitative data collection in a variety of other ways. When clients have cultivated stakeholder relationships, participants may be sought through such partnerships (e.g., community-based organizations, other partner organizations or individuals). When relationships have not been formed, or when clients are looking to include participants beyond “the usual suspects,” MAD consultants may recruit participants in a variety of other ways, such as networking through social media groups, community-based organizations, or local government. When using mixed methods where qualitative data collection is preceded by quantitative data collection, a survey question might inquire about interest in future qualitative data collection efforts. 

When possible and consistent with relevant law and funding requirements, incentives (e.g., gift cards) should be provided to participants to both honor their time and increase the response rate. MAD consultants also utilize follow-up protocols to attempt to reach non-responders.

Though qualitative questions are generally structured differently compared with survey questions, the two question types do have some similarities. Like survey questions, qualitative questions should:

• Use plain language that a variety of participants can understand
• Avoid value-based language, particularly when exploring a sensitive topic or research question(s) 
• Use preferred language to refer to participants or other individuals, if known (e.g., baby instead of fetus, incarcerated individual instead of prisoner)
• Avoid asking questions that touch on more than one topic or issue (i.e., double-barreled question), which can complicate analysis and interpretation of findings
• Ask demographic questions at the end (if needed) 

Unlike survey questions, qualitative questions are designed to be intentionally broad. At the beginning of the interview or focus group, this question structure helps participants become comfortable with the interviewer, facilitator, or group of participants, and facilitates open conversation. As the interview or focus group progresses, broad questions allow participants to discuss a variety of answers, which may reveal unexpected findings. Qualitative questions tend to ask, “how,” and “why.” 

Qualitative question guides tend to flow from easier to answer questions to more challenging questions. It is important to note that “easy” questions may still be quite sensitive or personal, depending on the nature of the project (e.g., tell me a little bit about your son, tell me about your pregnancy). These questions are considered easier because they draw on information readily available to the participant and do not require opinion forming or complex thought. Questions that might be considered challenging are those that require participants to recall past experiences, develop opinions, or provide recommendations (e.g., what forms of assistance did you find most useful?, how has your experience been with school or child care?, what types of support have been the most valuable?, what advice would you give to other parents of children with this condition?, how do you feel about your role as an advocate?, what messages should parents be provided with?). Ideally, among more challenging questions, questions flow from personal experience to opinions or recommendations, so that participants can draw from earlier portions of the conversation. 

When asking more challenging questions, it is often helpful to have examples on-hand if participants seem to struggle to provide an answer. Examples should be helpful enough to provide an answer without leading the participant (e.g., who told you about the diagnosis? [e.g., OB/GYN, genetic counselor, maternal-fetal medicine doctor, someone else]?, who would you have like to get this information from [e.g., providers, department of health, other parents, someone else]?). 

It can be helpful to end interviews or focus groups with action or positively oriented questions (e.g., what would you recommend?, what are your hopes for your family’s future?), though simply asking participants if there is anything else they would like to discuss that has not been touched on is often also appropriate. 

As important as crafting qualitative questions is considering the length of the interview or focus group guide. For an individual interview, aiming to create a conversation that is 45-60 minutes long generally works for most participants. Focus groups may last close to 90 minutes, due to the group nature of the discussion. As previously noted, participants have limited time to devote to an interview or focus group, so the question guide should be focused on a core set of questions. For an interview, this generally translates to eight open-ended questions (including one or two opening warm-up questions and a wrap-up question), with two or three probes for each main question (which can help ensure that more specific questions are answered). This structure provides the participant with enough time to fully answer each question; it also provides the interviewer or facilitator with enough time to pursue unanticipated lines of questioning. Including too many questions risks keeping a participant longer than agreed upon, or compromising data collection (and the final product) to collect quantity over quality. 

When a significant number of questions are included in a qualitative question guide, it should be reviewed for scope creep (i.e., have the questions strayed or broadened to include more than the original intent). If the qualitative interview guide cannot be paired back, a survey or mixed methods approach should be considered instead. Remember: qualitative question guides should not become verbal surveys.

Interviews and focus groups may be conducted in-person when participants and interviewers or facilitators are similarly located. In such cases, it is recommended that the venue be centrally located for participants, be accessible by personal or mass transportation, and provide free parking. Commonly, interviews are conducted over the phone or online, and focus groups are conducted using both in-person and online methods. Qualitative data collection is often conducted using these methods because participants are rarely located close to each other, or to the interviewer or facilitator. Regardless of the method used, interviews and focus groups should be scheduled for a day and time that works well for the participant(s), which may mean accommodating evening and weekend availability, depending on the participant group. 

Interviews and focus groups should begin with a brief description of the project and its goals for context. Participants should also be provided with a Tennessen notice and given the opportunity to ask questions prior to beginning the question guide.

This is often an iterative and inductive process. This means that, rather than establishing all themes at the beginning based on what clients or MAD consultants expect participants to say and fitting the data into those themes, themes emerge gradually as answers are reviewed. 

Codebooks are essentially a dictionary for themes: it tells researchers what each code means. When developing a codebook, it is best practice to start with a few broad themes that are more likely to be mutually exclusive (i.e., not overlap). After broad themes have emerged, each theme (code) can be reexamined to identify more detailed codes within it. As responses are coded, it may be necessary to add codes to the codebook if new and important themes are identified. It may also be necessary to remove or edit codes as themes are better understood. Consultants managing larger datasets are prepared to recode previously coded responses to account for changes to the codebook.

Coding qualitative responses involves tagging the response text with a code or codes. A response may need to be tagged with one or more detailed codes within or across broad theme codes. 

After all responses have been fully coded, a lead coder analyzes the team’s work to ensure codes have been applied in the same way. Where discrepancies exist, the consulting team discusses the application to determine the most appropriate use of the code. This improves the integrity of the coded data.

After all responses are coded, MAD consultants can begin interpreting how often codes were used. When working with smaller datasets that are not formally coded in software, consultants manually review interview or session notes for themes. Generally, consultants complete a few steps for analysis of each broad theme: 

• Look for trends among sub-themes (e.g., were experiences more positive or negative)
• Look for nuances within trends (e.g., what types of negative experiences were reported, were they only experienced by certain participants)
• Determine the appropriate denominator, which may not be the same as the total number of people interviewed

Once MAD consultants have conducted preliminary analysis, they develop a narrative of the findings. In general, the following can be expected from a more detailed qualitative summary:

A sense of proportion is provided for a theme or sub-theme, rather than exact numbers. Exact numbers are generally not provided in qualitative summaries due to the conversational nature of data collection—participants may not have shared a particular thought, or interviewers or facilitators may not have had time to cover a question or probe with a participant. Interviewers or facilitators also may not have pursued an “off-book” line of questioning with all participants, either because it was not applicable, or because it was discovered later in the data collection process. So, while proportions are helpful for understanding relative differences, reporting exact numbers in qualitative research can be misleading.

The summary focuses on the themes that were most often discussed, but does not omit important topics that were discussed by only a few respondents.