PCORI funded the Pilot Projects to explore how to conduct and use patient-centered outcomes research in ways that can better serve patients and the healthcare community. Learn more.

Background

Focus groups and individual interviews are two common methods that healthcare researchers use to learn about people’s experiences and opinions. There is little research comparing the usefulness of focus groups and individual interviews.

Project Purpose

The goal of this study was to find out which method—focus groups or individual interviews—is a better way to answer research questions. This study looked at (1) how many focus groups or interviews are needed to get a full list of topics that people raise, and (2) whether focus groups yield different topics than interviews. The researchers studied this by asking questions in focus groups and interviews about how people in the African-American community in Durham, North Carolina, get health information.

Research Methods

The researchers invited African-American men 18 to 64 years old who lived in Durham and spoke English to take part in the study. A total of 364 men joined the study, and 333 of them completed it.

The researchers randomly assigned people to either focus groups or individual interviews. Each focus group had eight people. The researchers asked the same questions of all participants. The questions were about how African-American men get health information from people they know.

The research team created written transcripts of the focus groups and individual interviews. Two people on the research team then reviewed the transcripts individually. The research team compared their analyses and looked at how much and what types of information came from each focus group and individual interview.

The researchers also noted how many focus groups or individual interviews they had to review before they heard the same ideas repeated by participants without any new ideas emerging. The researchers determined how many transcripts they would need to read until they would see 80 percent or 90 percent of the same ideas. They also noted how often participants talked about sensitive personal information.

The researchers compared the cost of getting the same amount of information from focus groups versus individual interviews. Finally, the research team summarized and organized what participants said about how they get health information.

Findings

To identify 80 percent of the ideas that participants mentioned, it took eight individual interviews or three focus groups with a total of 24 participants. To identify 90 percent of the ideas, it took 16 individual interviews or five focus groups with a total of 40 people. To reach each level, interviews took more time to complete but cost less overall. To identify all of the ideas that participants mentioned, individual interviews cost 64 to 80 percent as much as focus groups. But people talked more about sensitive personal information in the focus groups than in the individual interviews.

The answers to the questions posed in the interviews and focus groups showed that social relationships influence how African-American men get health information. Different types of social relationships have varying kinds of influence. Some social relationships offer direct support, while others provide indirect messages about health.

Limitations

The results of the comparison between focus groups and interviews might be different if the study had used a different topic, a different set of questions, or different researchers to review the transcripts, or if the study included people from different backgrounds.

Importance of Findings

Other researchers can use these findings to help them decide whether focus groups or individual interviews would work better for gathering information in a study they are planning. The findings might also help researchers determine how many people they need to include. The findings about the differences in cost between focus groups and individual interviews are different than what people might expect.

Sharing the Results

The ideas mentioned in the focus groups and individual interviews were presented in Durham through a live theater performance in October 2014. Nearly 400 residents and healthcare workers attended. The research team has also published manuscripts in medical journals. (see below)

PCORI funded the Pilot Projects to explore how to conduct and use patient-centered outcomes research in ways that can better serve patients and the healthcare community. Learn more.

Project Purpose

Despite the widespread use of focus groups in qualitative and mixed-methods research, most of the recommended parameters for focus group methodology lack an evidence base. Two conspicuous gaps in need of empirical study relate to sampling and the relative value of focus groups compared with in-depth interviews.

The methodological research objectives of the study were (1) to empirically determine the number of focus groups required to reach data saturation; and (2) to evaluate the differences between the data produced in focus groups and data gathered during in-depth interviews. Addressing these objectives led to the final objective: (3) to explore health-seeking behavior among the African-American community in Durham, North Carolina.

Study Design

In this comparative qualitative study, participants were randomized to individual interviews (IDIs) or focus groups (FGs). To minimize externally introduced variability, researchers attempted to include eight men per focus group (the modal recommendation for group size in the literature). A single interviewer conducted all of the FGs and IDIs, with the same guide, and the same office space for all events. A team of two analysts coded all of the data. 

Participants, Interventions, Settings, and Outcomes

Participants were men ages 18–64 who self-identified as African American; were current residents of the city of Durham, North Carolina; spoke English; and were willing and able to provide informed consent. The setting of this research was the general community, both for recruitment and conduct of the qualitative research.

The comparative element of this research was the mode of data collection: IDI or FG. Participants were randomized at the time of enrollment to one of these two modes. Researchers compared IDI and FG data on the following “outcome” measures: (1) amount and type of thematic content generated; (2) amount of free-listing data generated (per event and per person); (3) number of data collection events required to reach data saturation; and (4) cost-effectiveness to reach thematic saturation.

Data Analysis

All FG and IDI recordings were transcribed using a transcription protocol developed for thematic analyses. After verifying the accuracy of the transcripts, two members of the study team (including the data collector) independently and sequentially reviewed each focus group transcript, identified new themes, and created new codes corresponding to those themes.

Researchers used a well-known template to define and agree on all codes. Analysts used NVivo 10 to apply codes to the transcripts. The research team calculated thematic saturation for FGs and IDIs on the original data set and a randomly ordered data set (using cumulative number of new codes per transcript divided by total number of codes, until 80 percent and 90 percent of codes had been discovered). Using a bootstrap simulation, 10,000 random samples were generated from the original data set and calculated the median thematic saturation levels and the range around them. These saturation levels were then used as the basis of comparing cost-effectiveness. Researchers also compared the number of free-list items generated in FGs and IDIs and the voluntary discussion of sensitive personal topics in each. Finally, the research team organized and summarized the thematic content about health-seeking behaviors among African-American men.

Findings

The median number of data collection events (in the bootstrap samples) required to reach 80 percent and 90 percent thematic saturation was eight and 16, respectively, for IDIs, and three and five for FGs. Interviews took longer but were more cost-effective at both levels. At the median, IDIs cost 20 percent to 36 percent less to reach thematic saturation. At the event level, each type generated similar numbers of unique items in a free-listing task. When compared on a per-person basis, however, individual interviews were more effective at generating a broad range of items. Focus groups were more effective than individual interviews for eliciting disclosure of sensitive personal information in this study. With regard to the substantive topic of the research, findings confirmed the importance of social relationships in influencing the health-seeking behaviors of African-American men. These findings also helped to characterize the nature of influence across different types of relationships, according to the direct support or indirect messages they provide.

Limitations

A potential limitation of the methodological findings is that they come from one relatively homogenous population responding to questions on a semi-structured guide. Generalizability may be limited because of degree of instrument structure, sample homogeneity, complexity of the study topic, study purpose, or analyst categorization style.

Conclusions

These findings build the evidence base about adequate sample sizes for focus group and interview research and also establish the ability of each approach to generate different types of content. Findings also challenge some prevailing assumptions about the cost-efficiency of the two methods.

Dissemination

Dissemination of the substantive topic area findings was completed in October 2014 in the form of a live performance of an “ethnodrama” based on the data, held at a theater in Durham and attended by nearly 400 community members and healthcare providers. A total of five manuscripts have been published or are under review.