minimum number of participants in qualitative research

How many participants do I need for qualitative research?

• Participant recruitment
• Qualitative research

6 min read David Renwick

For those new to the qualitative research space, there’s one question that’s usually pretty tough to figure out, and that’s the question of how many participants to include in a study. Regardless of whether it’s research as part of the discovery phase for a new product, or perhaps an in-depth canvas of the users of an existing service, researchers can often find it difficult to agree on the numbers. So is there an easy answer? Let’s find out.

Here, we’ll look into the right number of participants for qualitative research studies. If you want to know about participants for quantitative research, read Nielsen Norman Group’s article .

Getting the numbers right

So you need to run a series of user interviews or usability tests and aren’t sure exactly how many people you should reach out to. It can be a tricky situation – especially for those without much experience. Do you test a small selection of 1 or 2 people to make the recruitment process easier? Or, do you go big and test with a series of 10 people over the course of a month? The answer lies somewhere in between.

It’s often a good idea (for qualitative research methods like interviews and usability tests) to start with 5 participants and then scale up by a further 5 based on how complicated the subject matter is. You may also find it helpful to add additional participants if you’re new to user research or you’re working in a new area.

What you’re actually looking for here is what’s known as saturation.

Understanding saturation

Whether it’s qualitative research as part of a master’s thesis or as research for a new online dating app, saturation is the best metric you can use to identify when you’ve hit the right number of participants.

In a nutshell, saturation is when you’ve reached the point where adding further participants doesn’t give you any further insights. It’s true that you may still pick up on the occasional interesting detail, but all of your big revelations and learnings have come and gone. A good measure is to sit down after each session with a participant and analyze the number of new insights you’ve noted down.

Interestingly, in a paper titled How Many Interviews Are Enough? , authors Greg Guest, Arwen Bunce and Laura Johnson noted that saturation usually occurs with around 12 participants in homogeneous groups (meaning people in the same role at an organization, for example). However, carrying out ethnographic research on a larger domain with a diverse set of participants will almost certainly require a larger sample.

Ensuring you’ve hit the right number of participants

How do you know when you’ve reached saturation point? You have to keep conducting interviews or usability tests until you’re no longer uncovering new insights or concepts.

While this may seem to run counter to the idea of just gathering as much data from as many people as possible, there’s a strong case for focusing on a smaller group of participants. In The logic of small samples in interview-based , authors Mira Crouch and Heather McKenzie note that using fewer than 20 participants during a qualitative research study will result in better data. Why? With a smaller group, it’s easier for you (the researcher) to build strong close relationships with your participants, which in turn leads to more natural conversations and better data.

There’s also a school of thought that you should interview 5 or so people per persona. For example, if you’re working in a company that has well-defined personas, you might want to use those as a basis for your study, and then you would interview 5 people based on each persona. This maybe worth considering or particularly important when you have a product that has very distinct user groups (e.g. students and staff, teachers and parents etc).

How your domain affects sample size

The scope of the topic you’re researching will change the amount of information you’ll need to gather before you’ve hit the saturation point. Your topic is also commonly referred to as the domain.

If you’re working in quite a confined domain, for example, a single screen of a mobile app or a very specific scenario, you’ll likely find interviews with 5 participants to be perfectly fine. Moving into more complicated domains, like the entire checkout process for an online shopping app, will push up your sample size.

As Mitchel Seaman notes : “Exploring a big issue like young peoples’ opinions about healthcare coverage, a broad emotional issue like postmarital sexuality, or a poorly-understood domain for your team like mobile device use in another country can drastically increase the number of interviews you’ll want to conduct.”

In-person or remote

Does the location of your participants change the number you need for qualitative user research? Well, not really – but there are other factors to consider.

• Budget: If you choose to conduct remote interviews/usability tests, you’ll likely find you’ve got lower costs as you won’t need to travel to your participants or have them travel to you. This also affects…
• Participant access: Remote qualitative research can be a lifesaver when it comes to participant access. No longer are you confined to the people you have physical access to — instead you can reach out to anyone you’d like.
• Quality: On the other hand, remote research does have its downsides. For one, you’ll likely find you’re not able to build the same kinds of relationships over the internet or phone as those in person, which in turn means you never quite get the same level of insights.

Is there value in outsourcing recruitment?

Recruitment is understandably an intensive logistical exercise with many moving parts. If you’ve ever had to recruit people for a study before, you’ll understand the need for long lead times (to ensure you have enough participants for the project) and the countless long email chains as you discuss suitable times.

Outsourcing your participant recruitment is just one way to lighten the logistical load during your research. Instead of having to go out and look for participants, you have them essentially delivered to you in the right number and with the right attributes.

We’ve got one such service at Optimal Workshop, which means it’s the perfect accompaniment if you’re also using our platform of UX tools. Read more about that here .

So that’s really most of what there is to know about participant recruitment in a qualitative research context. As we said at the start, while it can appear quite tricky to figure out exactly how many people you need to recruit, it’s actually not all that difficult in reality.

Overall, the number of participants you need for your qualitative research can depend on your project among other factors. It’s important to keep saturation in mind, as well as the locale of participants. You also need to get the most you can out of what’s available to you. Remember: Some research is better than none!

Capture, analyze and visualize your qualitative data.

Try our qualitative research tool for usability testing, interviewing and note-taking. Reframer by Optimal Workshop.

Published on August 8, 2019

David Renwick

David is Optimal Workshop's Content Strategist and Editor of CRUX. You can usually find him alongside one of the office dogs 🐕 (Bella, Bowie, Frida, Tana or Steezy). Connect with him on LinkedIn.

Recommended for you

Product Update - August 2023

We’re excited to share some new features and product updates that we’ve rolled out lately and what’s coming up next. What’s new… Speeding up task analysis in Treejack 🚤 Customer feedback and research have shown that users analyse results in Treejack in a task-by-task manner. To better support this way of working, we’ve updated Treejack... View Article

How we created a new Optimal Workshop blog

We’ve just redesigned and relaunched the Optimal Workshop blog. Here’s a little bit about how (and why) we did it.

7 ways to get other departments into user research

Anna Lee Anda shares some ways to get other departments into user research.

Try Optimal Workshop tools for free

What are you looking for.

Explore all tags

Discover more from Optimal Workshop

Subscribe now to keep reading and get access to the full archive.

Type your email…

Continue reading

Sample Size Policy for Qualitative Studies Using In-Depth Interviews

• Published: 12 September 2012
• Volume 41 , pages 1319–1320, ( 2012 )

Cite this article

• Shari L. Dworkin 1  

298k Accesses

566 Citations

28 Altmetric

Explore all metrics

Avoid common mistakes on your manuscript.

In recent years, there has been an increase in submissions to the Journal that draw on qualitative research methods. This increase is welcome and indicates not only the interdisciplinarity embraced by the Journal (Zucker, 2002 ) but also its commitment to a wide array of methodologies.

For those who do select qualitative methods and use grounded theory and in-depth interviews in particular, there appear to be a lot of questions that authors have had recently about how to write a rigorous Method section. This topic will be addressed in a subsequent Editorial. At this time, however, the most common question we receive is: “How large does my sample size have to be?” and hence I would like to take this opportunity to answer this question by discussing relevant debates and then the policy of the Archives of Sexual Behavior . Footnote 1

The sample size used in qualitative research methods is often smaller than that used in quantitative research methods. This is because qualitative research methods are often concerned with garnering an in-depth understanding of a phenomenon or are focused on meaning (and heterogeneities in meaning )—which are often centered on the how and why of a particular issue, process, situation, subculture, scene or set of social interactions. In-depth interview work is not as concerned with making generalizations to a larger population of interest and does not tend to rely on hypothesis testing but rather is more inductive and emergent in its process. As such, the aim of grounded theory and in-depth interviews is to create “categories from the data and then to analyze relationships between categories” while attending to how the “lived experience” of research participants can be understood (Charmaz, 1990 , p. 1162).

There are several debates concerning what sample size is the right size for such endeavors. Most scholars argue that the concept of saturation is the most important factor to think about when mulling over sample size decisions in qualitative research (Mason, 2010 ). Saturation is defined by many as the point at which the data collection process no longer offers any new or relevant data. Another way to state this is that conceptual categories in a research project can be considered saturated “when gathering fresh data no longer sparks new theoretical insights, nor reveals new properties of your core theoretical categories” (Charmaz, 2006 , p. 113). Saturation depends on many factors and not all of them are under the researcher’s control. Some of these include: How homogenous or heterogeneous is the population being studied? What are the selection criteria? How much money is in the budget to carry out the study? Are there key stratifiers (e.g., conceptual, demographic) that are critical for an in-depth understanding of the topic being examined? What is the timeline that the researcher faces? How experienced is the researcher in being able to even determine when she or he has actually reached saturation (Charmaz, 2006 )? Is the author carrying out theoretical sampling and is, therefore, concerned with ensuring depth on relevant concepts and examining a range of concepts and characteristics that are deemed critical for emergent findings (Glaser & Strauss, 1967 ; Strauss & Corbin, 1994 , 2007 )?

While some experts in qualitative research avoid the topic of “how many” interviews “are enough,” there is indeed variability in what is suggested as a minimum. An extremely large number of articles, book chapters, and books recommend guidance and suggest anywhere from 5 to 50 participants as adequate. All of these pieces of work engage in nuanced debates when responding to the question of “how many” and frequently respond with a vague (and, actually, reasonable) “it depends.” Numerous factors are said to be important, including “the quality of data, the scope of the study, the nature of the topic, the amount of useful information obtained from each participant, the use of shadowed data, and the qualitative method and study designed used” (Morse, 2000 , p. 1). Others argue that the “how many” question can be the wrong question and that the rigor of the method “depends upon developing the range of relevant conceptual categories, saturating (filling, supporting, and providing repeated evidence for) those categories,” and fully explaining the data (Charmaz, 1990 ). Indeed, there have been countless conferences and conference sessions on these debates, reports written, and myriad publications are available as well (for a compilation of debates, see Baker & Edwards, 2012 ).

Taking all of these perspectives into account, the Archives of Sexual Behavior is putting forward a policy for authors in order to have more clarity on what is expected in terms of sample size for studies drawing on grounded theory and in-depth interviews. The policy of the Archives of Sexual Behavior will be that it adheres to the recommendation that 25–30 participants is the minimum sample size required to reach saturation and redundancy in grounded theory studies that use in-depth interviews. This number is considered adequate for publications in journals because it (1) may allow for thorough examination of the characteristics that address the research questions and to distinguish conceptual categories of interest, (2) maximizes the possibility that enough data have been collected to clarify relationships between conceptual categories and identify variation in processes, and (3) maximizes the chances that negative cases and hypothetical negative cases have been explored in the data (Charmaz, 2006 ; Morse, 1994 , 1995 ).

The Journal does not want to paradoxically and rigidly quantify sample size when the endeavor at hand is qualitative in nature and the debates on this matter are complex. However, we are providing this practical guidance. We want to ensure that more of our submissions have an adequate sample size so as to get closer to reaching the goal of saturation and redundancy across relevant characteristics and concepts. The current recommendation that is being put forward does not include any comment on other qualitative methodologies, such as content and textual analysis, participant observation, focus groups, case studies, clinical cases or mixed quantitative–qualitative methods. The current recommendation also does not apply to phenomenological studies or life history approaches. The current guidance is intended to offer one clear and consistent standard for research projects that use grounded theory and draw on in-depth interviews.

Editor’s note: Dr. Dworkin is an Associate Editor of the Journal and is responsible for qualitative submissions.

Baker, S. E., & Edwards, R. (2012). How many qualitative interviews is enough? National Center for Research Methods. Available at: http://eprints.ncrm.ac.uk/2273/ .

Charmaz, K. (1990). ‘Discovering’ chronic illness: Using grounded theory. Social Science and Medicine, 30 , 1161–1172.

Article   PubMed   Google Scholar  

Charmaz, K. (2006). Constructing grounded theory: A practical guide through qualitative analysis . London: Sage Publications.

Google Scholar  

Glaser, B. G., & Strauss, A. L. (1967). The discovery of grounded theory: Strategies for qualitative research . Chicago: Aldine Publishing Co.

Mason, M. (2010). Sample size and saturation in PhD studies using qualitative interviews. Forum: Qualitative Social Research, 11 (3) [Article No. 8].

Morse, J. M. (1994). Designing funded qualitative research. In N. Denzin & Y. Lincoln (Eds.), Handbook of qualitative research (pp. 220–235). Thousand Oaks, CA: Sage Publications.

Morse, J. M. (1995). The significance of saturation. Qualitative Health Research, 5 , 147–149.

Article   Google Scholar  

Morse, J. M. (2000). Determining sample size. Qualitative Health Research, 10 , 3–5.

Strauss, A. L., & Corbin, J. M. (1994). Grounded theory methodology. In N. K. Denzin & Y. S. Lincoln (Eds.), Handbook of qualitative research (pp. 273–285). Thousand Oaks, CA: Sage Publications.

Strauss, A. L., & Corbin, J. M. (2007). Basics of qualitative research: Techniques and procedures for developing grounded theory . Thousand Oaks, CA: Sage Publications.

Zucker, K. J. (2002). From the Editor’s desk: Receiving the torch in the era of sexology’s renaissance. Archives of Sexual Behavior, 31 , 1–6.

Download references

Author information

Authors and affiliations.

Department of Social and Behavioral Sciences, University of California at San Francisco, 3333 California St., LHTS #455, San Francisco, CA, 94118, USA

Shari L. Dworkin

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Shari L. Dworkin .

Rights and permissions

Reprints and permissions

About this article

Dworkin, S.L. Sample Size Policy for Qualitative Studies Using In-Depth Interviews. Arch Sex Behav 41 , 1319–1320 (2012). https://doi.org/10.1007/s10508-012-0016-6

Download citation

Published : 12 September 2012

Issue Date : December 2012

DOI : https://doi.org/10.1007/s10508-012-0016-6

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Find a journal
• Publish with us
• Track your research

• Print this article

Review Article

How many participants do we have to include in properly powered experiments a tutorial of power analysis with reference tables.

Given that an effect size of d = .4 is a good first estimate of the smallest effect size of interest in psychological research, we already need over 50 participants for a simple comparison of two within-participants conditions if we want to run a study with 80% power. This is more than current practice. In addition, as soon as a between-groups variable or an interaction is involved, numbers of 100, 200, and even more participants are needed. As long as we do not accept these facts, we will keep on running underpowered studies with unclear results. Addressing the issue requires a change in the way research is evaluated by supervisors, examiners, reviewers, and editors. The present paper describes reference numbers needed for the designs most often used by psychologists, including single-variable between-groups and repeated-measures designs with two and three levels, two-factor designs involving two repeated-measures variables or one between-groups variable and one repeated-measures variable (split-plot design). The numbers are given for the traditional, frequentist analysis with p < .05 and Bayesian analysis with BF > 10. These numbers provide researchers with a standard to determine (and justify) the sample size of an upcoming study. The article also describes how researchers can improve the power of their study by including multiple observations per condition per participant.

• Page/Article: 16
• DOI: 10.5334/joc.72
• Accepted on 22 May 2019
• Published on 19 Jul 2019
• Peer Reviewed

• Research article
• Open access
• Published: 11 March 2011

What about N? A methodological study of sample-size reporting in focus group studies

• Benedicte Carlsen 1 &
• Claire Glenton 2 , 3  

BMC Medical Research Methodology volume  11 , Article number:  26 ( 2011 ) Cite this article

68k Accesses

394 Citations

4 Altmetric

Metrics details

Focus group studies are increasingly published in health related journals, but we know little about how researchers use this method, particularly how they determine the number of focus groups to conduct. The methodological literature commonly advises researchers to follow principles of data saturation, although practical advise on how to do this is lacking. Our objectives were firstly, to describe the current status of sample size in focus group studies reported in health journals. Secondly, to assess whether and how researchers explain the number of focus groups they carry out.

We searched PubMed for studies that had used focus groups and that had been published in open access journals during 2008, and extracted data on the number of focus groups and on any explanation authors gave for this number. We also did a qualitative assessment of the papers with regard to how number of groups was explained and discussed.

We identified 220 papers published in 117 journals. In these papers insufficient reporting of sample sizes was common. The number of focus groups conducted varied greatly (mean 8.4, median 5, range 1 to 96). Thirty seven (17%) studies attempted to explain the number of groups. Six studies referred to rules of thumb in the literature, three stated that they were unable to organize more groups for practical reasons, while 28 studies stated that they had reached a point of saturation. Among those stating that they had reached a point of saturation, several appeared not to have followed principles from grounded theory where data collection and analysis is an iterative process until saturation is reached. Studies with high numbers of focus groups did not offer explanations for number of groups. Too much data as a study weakness was not an issue discussed in any of the reviewed papers.

Conclusions

Based on these findings we suggest that journals adopt more stringent requirements for focus group method reporting. The often poor and inconsistent reporting seen in these studies may also reflect the lack of clear, evidence-based guidance about deciding on sample size. More empirical research is needed to develop focus group methodology.

Peer Review reports

Transparency and accountability are key elements in any research report, not least in qualitative studies. Thorough reporting of methods allows readers to assess the quality and relevance of research findings. In addition, for qualitative research methodology to advance, information about how these methods are used and how they work best is needed.

A focus group or focus interview is commonly defined as a method of collecting research data through moderated group discussion based on the participants' perceptions and experience of a topic decided by the researcher [ 1 – 6 ]. Focus groups differ from group interviews in that the emphasis is on the interaction between the participants rather than between the moderator or researcher and the participants [ 4 ]. In recent years, focus groups have become increasingly popular within health science research; in a Medline search in 1999, Twohig and Putnam [ 7 ] found no focus group studies before 1985 but more than 1000 studies between 1985 and 1999. Focus groups are well suited to explore people's subjective experiences and attitudes, and health researchers are repeatedly encouraged to use focus groups to evaluate health services, to elicit the views of key stakeholders or decision makers or to explore the views of marginalised groups that typically would not respond to a postal survey or would be intimidated by a conventional interview situation [ 2 , 5 , 7 , 8 ]. Focus group interviews are also recommended as a pre- or post-study to prepare or interpret data from surveys or trial studies.

Despite this apparent sharp increase in focus group studies in health sciences journals since the 1980 s, there is surprisingly little knowledge about how this method is used and why [ 1 , 9 , 10 ]. In 1993, one of the "founding fathers" of focus group research methodology, Richard Krueger, complained that the field is burdened with "poor design and shoddy reporting" [ 11 ]. After reviewing the field in 1996, Morgan concluded that "At present the reporting of focus group procedures is a haphazard affair at best." Twohig and Putnam's review [ 7 ] of focus group studies carried out in the 1990 s showed a "tremendous variation" in how the method was used and concluded that reporting is often inadequate. We were interested in discovering whether there is more reason for optimism a decade later.

There is an extensive literature, ranging from academic classics in grounded theory to pragmatic "how to do" books, that teaches novices how focus groups are usually conducted and indicates how to organise focus groups and how to analyse the data, often within a restricted budget and time span [ 2 , 4 – 6 , 12 – 17 ]. Within this literature, advice on how to decide on the number of groups to conduct is often meagre compared to advice on other aspects of the method. In fact, some teaching books claim that there are no existing guidelines for deciding number of groups [ 13 , 16 , 18 ].

Most of this guidance recommends that the focus group should be the unit of analysis in focus group-studies. In line with this, sample size should refer to number of groups and not the total number of participants in a study. However, there is of course an intrinsic relationship between these two aspects of sample size. Thus, some authors suggest that if the number of participants in a study is small, it is possible to increase the number of groups by reducing the size of the groups [ 4 ]. Guidance on group size is common and seldom goes beyond a minimum of 4 and a maximum of 12 participants per group [ 2 , 5 , 6 , 13 ]. Here, it is interesting to note that Fern's experimental study indicated that more information is obtained by conducting two groups of four participants then one group of eight participants [ 19 ].

The strength of qualitative methods is their ability to explore the depth and complexity of phenomena. Thus, Sandelowski, in her discussion of sample size in qualitative research, emphasizes that both too few and too many groups can lower the quality of focus group studies [ 20 ]. Quantity must be balanced against quality, and the more hours of taped interviews or pages of transcribed material, the less depth and richness the authors will be able to extract from the material [ 21 ]. While this knowledge does not directly offer advice regarding how to achieve the optimal number of focus groups, it does highlight an issue that authors should be aware of when determining number of groups.

Most advice about sample size in focus group studies refers to "point of saturation" or "theoretical saturation", The term "theoretical saturation" was introduced by Glaser and Strauss [ 22 ] in 1967 in their outline of "grounded theory" and has since spread to most qualitative research environments, albeit in a simplified version, usually referred to as "data saturation". In this approach, interviews should be conducted with different categories of informants following a line of "theoretical sampling" to continuously construct and refine theory or hypotheses. The method requires that data collection, i.e. recruiting, interviewing and analysis, is conducted as an iterative process for each interview, thus representing quite a different approach from the traditional quantitative design of successively calculating sample size beforehand and analysing all data collected [ 23 , 24 ].

The concept of point of saturation has been criticised for being too vague to operationalise [ 25 ]. Glaser and Strauss suggest that researchers should: "sample until theoretical saturation of each category is reached" [ 22 ](1967:61-62, 11-112) while Strauss and Corbin state that researchers should collect data until "no new or relevant data seem to emerge regarding a category." [ 15 ] (1990:188). However, the authors present no definition of "new or relevant data", and give no advice regarding the number of interviews with no new information that is required before the researcher can be reasonably certain that saturation has been reached.

More pragmatic authorities on focus group methodology recognise that many researchers work on assignment and their funders may require that the number of groups are decided pre-study [ 2 , 4 , 12 ]. The recommendations and references to what is common regarding number of groups vary within these text books. In general they recommend from two to five groups per category of participants. However, the authors usually underline that these are only rules of thumb and that the number of focus groups depends on the complexity of the research question and the composition of the groups. Often, the advice given is to follow the rules of thumb but to suggest a slightly higher number to be "on the safe side". Some authors suggest that the researchers then use point of saturation as they go along to decide the final number of groups needed [ 2 , 4 ].

While some of these text books still refer to Glaser and Strauss, others do not. In these cases the concept of theoretical sampling and theoretical saturation and the theory behind it seem to have been lost on the way and replaced with the less theoretically sustained terms of "purposive sampling" and "data saturation" or merely "saturation". However, the basic procedure of selecting informants and deciding on the number of groups through a constant process of analysing data and obtaining new data until no new essential information is found, remains.

It has been claimed by critics within the field of qualitative methodology that the concept of saturation is often misused in research reports [ 23 , 26 ]. Charmaz [ 25 ] (2005:528) claims that "Often, researchers invoke the criterion of saturation to justify small samples - very small samples with thin data." In reality, these small sample sizes may, in fact, be a result of lack of time or lack of funds. Whether "saturation" has thus become the magic word which obscures faulty designs in focus group reports, as Charmaz insinuates, has yet to be investigated.

In general, empirical research into how focus group methods are used and work is scarce. We have found only two review studies that document and discuss sample size in focus group studies [ 7 , 8 ] and to our knowledge no study has yet assessed how decisions about sample size in focus group studies are reported. Nor does the effectiveness of different sample sizes appear to have been evaluated. This lack of empirical evidence suggests that advice offered with regard to sample size is, as a rule, based on common assumptions or personal experience with the method.

Our objective is twofold. First, we aim to survey sample size in current focus group studies in health science journals. Second, we aim to survey and assess the extent to which researchers describe and justify the number of focus groups they carry out.

We carried out a structured search for papers that included focus group methods, and extracted and analysed quantitative and qualitative information about the authors' use of these methods.

Eligibility criteria

We searched PubMed Central in 2009 for primary studies that used focus groups and were published in 2008. We chose PubMed Central because it offers good coverage of medical journals and allowed us to apply a "Free Full Text filter" to limit searches to publications from open-access journals. This approach gave us a manageable sample of studies and easy and immediate access to papers, and allows others to easily access publications cited in this review and check our conclusions.

As mentioned, definitions of focus group interviews usually refer to researcher-initiated gathering of a small group of people with the aim of facilitating discussion about a given topic [ 1 , 4 – 6 ]. However, we included any study where the authors themselves used the term "focus group" to refer to their method of data collection. We included studies using focus groups only and studies using mixed methods. We chose studies published in 2008 as we started working with the review in 2009 and wanted to review the state of the art in the field. In addition we repeated the same search for 2003 and 1998, i.e. five and ten years earlier as a crude check on the expansion of the field.

We excluded papers that described planned studies and papers that described internet-based focus groups as we assumed that recruiting participants for such groups possibly invokes other challenges than recruiting participants to face-to-face groups.

Search methods for identification of studies

We searched Pubmed Central using the following search string:

"2008"[Publication Date] AND ("focus group"[Title/Abstract] OR "focus groups"[Title/Abstract] OR "focus interviews"[Title/Abstract]) AND (English[lang] AND pubmed pmc local[sb]). The search was conducted on December 21 st 2009.

Data selection, extraction and analysis

One review author (BC) independently assessed the relevance of full text versions of all papers identified from the electronic searches. Decisions to exclude papers were checked by the other review author (CG).

Both authors independently extracted data from the first ten included studies using a standard form based on the variables we aimed at analysing. The data were then checked against each other, and, if necessary, reference was made to the original paper. Any discrepancies between the two data extraction sheets were discussed by the two reviewers and resolved by consensus. For the remaining studies, one reviewer (BC) extracted the data alone and checked with the other author when in doubt.

We extracted data about the following aspects as numerical data:

Number of focus groups in each study

Maximum and minimum number of participants in the focus groups in each study

Total number of participants

Whether any explanation for number of groups was given

Whether this explanation was tied to:

◦ Practical issues (such as convenience when recruiting or limited resources to conduct interviews)

◦ Recommendations in the literature (pragmatic guidelines to number of focus groups)

◦ Saturation

◦ Capacity when analyzing, i.e. balancing between depth and breadth of data

We analysed this data descriptively (with SPSS 15.0).

In addition, we carried out a matrix-based qualitative analysis of the texts. Here, we explored the study reports adopting the same approach that we would have used for transcribed interview material, using a cross-case thematic analysis described by Miles and Huberman, among others [ 27 ]. Specifically, we extracted and condensed relevant passages from all included papers in a separate document with our comments regarding level and style of reporting and the authors' explanations for sample size. We especially focussed on:

How clear and accessible information regarding sample size was in each study

Whether explanations for number of focus groups was clear, credible and consistently reported

Whether and how the number of groups were referred to by the authors when discussing study strengths and limitations

The current status of sample size in focus group studies

Through our electronic database search we identified 240 papers published in 2008. In comparison, our search for 1998 and 2003 resulted in 23 and 62 studies respectively.

We considered the full text version of each of the 240 papers from 2008 for inclusion in the review. Twenty papers were excluded after reading the full text version. Of these, six were excluded because they were planned studies, four were excluded because the focus groups were Internet-based, three were excluded because they were not primary studies and seven were excluded because they did not, in fact, report the results of focus group studies even though the term "focus groups" appeared in their abstracts. (Some of these referred to focus groups as part of an intervention that was evaluated in the reported study).

Two hundred and twenty papers met our inclusion criteria. (See Additional file 1 for a list of included studies.) These were published in 117 different journals. Sixty-one percent had used mixed methods of data collection, while 39% had used focus groups only.

Many of the 220 papers were characterised by an insufficient reporting of sample size:

22 (10%) did not report number of focus groups

42 (19%) did not report total number of participants

102 (46%) did not report minimum and maximum number of participants in the focus groups

In addition, it was sometimes difficult for us to find focus group and participant numbers as this information was sometimes reported in different sections of the papers (See for example [ 28 , 29 ]. For mixed method studies, it was sometimes difficult to separate between sample size information for different data collection methods (See for example [ 30 ]).

Those papers that did report numbers of groups and participants showed a great range in these numbers, but data distribution was positively skewed, i.e. there were many studies with a few focus groups and few studies with high numbers of focus groups (Table 1 ).

Authors' explanations for number of focus groups

Authors' explanations of how they had decided on or ended up with the number of focus groups carried out varied, but were often unclear or completely lacking (e.g. [ 31 , 32 ]). (See Table 2 .) When authors used mixed methods, explanations of sample size for the quantitative part of the study were often meticulous, while the sample size of the focus interviews in contrast was often unclear and superficial, as in this example:

- purposive sampling was undertaken till the necessary number was attained [ 33 ].

Of the 220 studies included, 183 (83,2%) gave no explanation for number of focus groups (i.e. 37 did give some type of explanation). Often, authors explained the number of participants rather than the number of focus groups [ 34 , 35 ]. Typical for these cases were situations where focus group studies were carried out alongside clinical intervention studies and where the authors had invited all participants from the intervention study to participate in the focus group interviews [ 34 , 36 ]. As these accounts explain the number of participants rather than the number of focus groups, this was categorised as "no explanation".

From Table 2 we also see that no study justified the number of groups by referring to the need to balance data quality and quantity (capacity to analyse data). The table also shows that all the explanations for sample size were found in studies that had between two and 13 focus groups. We tested if there was any correlation between the presence of an explanation for number of focus groups and the number of groups conducted (Pearson correlation). We found no linear relationship. None of the eleven single-group studies attempted to justify why one group was sufficient. These studies all use mixed methods and the qualitative assessment showed that they typically used the focus group study as a pilot for developing questionnaires etc.

As noted, none of the studies that included more than 13 groups gave any explanations for sample size and the qualitative assessment showed that none of the studies referred to a high number of groups as a study limitation. On the contrary, when authors considered their sample size to be relatively high, this was seen as advantageous:

By conducting a large number of focus groups with a significant cross-section of the facility's employees, we reduced the possibility that results would be dramatically affected by a single focus group or methodological choice during coding [ 37 ].

On the other hand, when authors had included only a small number of focus groups, they frequently described this as a limitation. However, in several of these studies, authors also claimed the number of focus groups had been determined by data saturation. This should, in theory, imply that the number of focus groups was, in fact, appropriate. (See for example [ 38 – 43 ]). But while authors often described a small number of focus groups as a weakness, several studies justified their choice of method with reference to the possibilities that focus groups gave to go in depth and provide a thick description of the issue.

Among the 37 studies that did give an explanation for the number of focus groups, 28 claimed that they had stopped once they had reached a point of saturation. However, more than half of these explanations (15 of 28) did not report convincingly that an iterative process, involving data collection and analysis that ended with saturation, had taken place. The reason for this was mainly inconsistencies in the description of the methodological procedures. One common example was that, despite their reference to point of saturation, their number of focus groups was pre-determined, as the extraction below shows (See also [ 44 , 45 ]):

Three focus groups were conducted with residents of a specific community. It was a small community, and it was determined that after three focus groups, there would be a saturation point regarding their barriers and suggested solutions [ 46 ].

Other common examples where claims of saturation appeared unsubstantiated were studies that used convenience sampling, and included everyone who volunteered to participate, instead of purposive or theoretical sampling, and where, in addition all data seemed to have been gathered before the analysis (See for example [ 41 , 42 , 47 – 49 ]).

We also noted that no author reported or discussed the number of focus groups that had been conducted with no new relevant information before it was decided that a point of saturation had been reached. On the contrary, authors normally gave superficial and vague references to the process of reaching point of saturation, many of them referring to a "feeling" of having reached this point:

Focus groups were [...] conducted until we felt "saturation" (no new themes identified) was reached [ 41 ].

At the end of the fifth focus group, we felt that theme saturation had been met, so we stopped recruitment... [ 50 ]

We also found some examples of adequate reporting regarding point of saturation. For example Barimani et al [ 51 ] explained their use of grounded theory and described how a constant comparison between empirical findings and theory had guided their sampling procedure until theoretical saturation was met. This paper was, however, unusually long, around 9000 words. Another example is Shuval et al [ 52 ]. In their paper of around 3000 words, the authors have not specifically referred to the term "saturation" but have offered a clear description of the sampling procedure and how saturation was met:

We used purposeful sampling [with reference to Patton 1990] to select key informants (on the basis of researchers' acquaintance), promote group interaction, and capture the diverse characteristics of participants (i.e., sex, age, living environment, ethnicity/religion, and self-reported income). All students whom we approached agreed to participate in focus groups .

Focus groups were held until no new themes emerged (9). [...] Researchers reviewed the transcripts to reflect on each session before conducting the next, thereby enabling newly identified concepts to be examined in subsequent sessions [ 52 ].

Recommendations in the literature

Six of the 37 studies that gave an explicit explanation for the number of focus groups referred to rules of thumb in the literature. These were mostly references to pragmatic guidelines of how many groups are necessary to reach a point of saturation (e.g. [ 53 ]). As the example below illustrates, authors reported that the field lacks consistent guidelines.

Recommendations for both the number of focus groups and sample size vary. The number of recommended sessions depends on the complexity of the study design and the target sample's level of distinctiveness [ 18 – 21 ] . Stewart et al. (2007) observed that rarely are more than 3-4 focus groups conducted in the social sciences. We felt that two groups at each site would limit bias that might be seen in a single group or site and allow us to examine themes common across groups [ 54 ].

However, when we checked with the literature, referrals to recommendations were not always accurate. For example Gutterling et al, conducted three focus groups and explain the number thus:

For good results, just a few focus groups are sufficient, as data become saturated and little new information emerges after the first few groups [Morgan 96] [ 55 ].

Looking up Morgan [ 1 ], we found that he claims that most studies use four to six groups because they then reach saturation, but he also underlines that the more categories of participants and less standardisation of questions, the higher number of focus groups.

Practical reasons

Three of the 37 studies reported practical reasons for the number of focus groups conducted. The information offered regarding recruitment constraints was incomplete as, for example, in this explanation:

The number of available participants was limited and the number of focus groups was therefore few [ 56 ].

Two of the explanations appeared to be tied to difficulties in recruiting participants to additional groups [ 56 , 57 ], while one mentioned limited resources ("budgetary and staffing constraints") which led the researchers to decide pre-study to conduct five focus groups [ 38 ]. This last study also claimed to have reached saturation (without stating this as the procedure for deciding number of groups):

However, data analysis indicates that all themes reached saturation, meaning additional participants would likely not have added to the depth or breadth of parent responses [ 38 ]

An additional eight studies also described recruitment limitations, but only as an explanation for the total number of participants, not for how many groups the participants were divided into. In these studies, the size of the groups seems to have been decided beforehand, due to text book recommendations, and thus the number of groups was given by the total number of already recruited participants divided by the number of participants per group.

The results from our searches from 1998, 2003 and 2008 support the claims that there has been an increase in focus group studies over the last ten years. The wide range of health journals publishing focus group studies in 2008 indicates that this method is now widely accepted. At the same time, the fact that many journals publish only one or two focus group studies a year could also mean that the methodological competence among editors and reviewers to assess focus group studies is lacking.

The great variation in the number of focus groups that we registered was surprising, and was wider than authors of teaching materials and text books assume. For example Stewart et al [ 13 ] (2007:58) claim: "Most focus group applications involve more than one group, but seldom more than three or four groups." Twohig and Putnam also found a much narrower variation in their review of focus group studies in primary care research, with a range of two to eight groups per study [ 7 ].

Overall, reporting of sample size and explanations for this size was poor. Where such explanations were given, our study confirms the dominant role of the concept of data saturation.

We also discovered that all explanations were found in studies of between two and 13 groups. Some of these studies refer to existing pragmatic guidelines to justify their numbers, although the two to five focus groups per category recommended in these guidelines sometimes appear to have become two to five groups in total in the studies. We could speculate that studies using only one focus group, a number that goes against the rules of thumb offered by these guidelines, is simply too hard to justify and explanations are therefore evaded. Also, all the single group studies were mixed methods studies, where the focus group typically was used as a pilot to develop or test a questionnaire for the survey part of the study. In these examples, it is understandable that the focus group is offered less attention than the main part of the study. At the other end of the scale, one could also speculate that when the sample size reaches two-digit numbers, a "quantitative study logic" kicks in where a big N is seen as a positive asset and therefore less important to justify.

Roughly half of the studies that referred to data saturation as an explanation for number of focus groups did not appear to be consistent in their use of approach. These findings support earlier reviews of the field. Twohig and Putnam [ 7 ], were also startled by the variation in procedures and reporting of focus group studies, and Webb and Kevern [ 58 ], who reviewed focus group studies in nursing research, found that authors used terms such as "Grounded theory" in non-rigorous ways. These authors also conclude that researchers, on the whole, did not follow basic premises for reaching saturation such as concurrent data generation and analysis, or an "iterative process" [ 15 ]. Our study shows the same tendencies, and suggests that the increased use of focus groups in health care studies has not led to an improvement in the quality of reporting.

We were also struck by what we did not find. In our own experience with focus group research, recruitment problems are much more common than this review indicates. In addition, a number of practical limitations arise that can limit the number of focus groups conducted, including limited money and time. Excepting one study, where resource constraints were brought up, the only practical limitations mentioned were difficulties in recruiting more participants. Another non-finding was that none of the studies discussed a large number of focus groups as a potential limitation of the study. Given frequent references in these studies to the advantages of qualitative methodology for eliciting richness and depth of the data, it is not evident why the authors never used the argument that data from a large number of focus groups is difficult to analyse thoroughly.

How can poor reporting be explained?

The inadequate reporting indicated in our study could reflect the fact that most health science journals do not require specific standards of reporting from contributors presenting qualitative research. However, the poor reporting among these authors also seem to indicate confusion about when and how to decide the number of focus groups, which may reflect a lack of properly described, consistent advice to researchers wishing to carry out focus group-based data collection.

The lack of attention to sample size in the teaching material could easily be perceived as an indication that sample size is unimportant in such studies. In addition, the advice that is offered is confusing and sometimes conflicting. While the Glaser and Strauss's procedure of theoretical saturation instructs authors to use theoretical sampling and to analyse and collect data iteratively until saturation is achieved, it does not offer a detailed interpretation of how to operationalise this approach. The "how to do" literature on focus group methodology, on the other hand, offers pragmatic advice regarding the number of groups that researchers should expect to conduct before point of saturation is reached, but do not clarify how to decide about point of saturation in practice. This advice may thus tempt researchers to follow their suggestions for number of groups and do the analysis after collecting all data. Then, as they expect data to be saturated, their critical sense could be undermined when drawing conclusions about saturation.

Despite problems associated with the practical application of the concept of data saturation, it seems to have become something of an ideal in qualitative health research. This could be due to the fact that it is the only theory that offers advice, albeit, poorly operationalised, about the exact number of interviews needed. It is plausible that editors in the traditionally positivistic realm of health research are inclined to prefer explanations for exact number of groups. Practical limitations might not be as acceptable, especially not explicit references to economic or resource limitations.

In the methodology discussions of these studies, authors often point out that small sample sizes are legitimate in qualitative studies. At the same time, they often feel the need to justify small sample sizes which they invariably see as a study limitation. This may be a consequence of the fact that qualitative studies are still in a minority in health science journals. Here, more positivistic traditions may make it difficult to argue that a qualitative study can have too many groups. Nevertheless, the quality of qualitative studies does depend on the depth and richness of the data and its analysis. Reference to the trade-off between number of focus groups and the thickness of our description should therefore be an acceptable explanation for (a limited) sample size. There is also an ethical side to sample size: an excessive number of interviews means placing a burden on patients or health workers that is not legitimised by added scientific value and can thereby be seen as unethical.

Study strengths and limitations

A limitation of our study was that our sample was taken from open-access journals. While this gave us easy and immediate access to a manageable sample of studies and also allows our readers to easily check our results, a quick search without the open access filter indicates that our sample represents less than 20% of all published focus group studies in 2008. We know relatively little about how our sample might differ from the remaining 80% of available studies. Current research does suggest that articles published in open-access journals are more often cited than other articles [ 59 ]. It is therefore possible that the articles we evaluated are of a higher profile than other non-open-access articles and may be more likely to serve as examples for other researchers. While we emphasise that our findings are primarily valid for open access studies, it therefore seems all the more important to secure the quality of these reports.

We decided to include all studies claiming to be focus group studies. We have therefore also included mixed method studies, where the focus group interviews are often part of a predominantly quantitative design. In such studies the authors may not aim to adhere to standards for reporting qualitative studies. On the other hand, it could be argued that researchers who report that they have used focus groups should adhere to the methodological standards for such studies.

There are some uncertainties in our findings due to the poor reporting in the material. For instance, it was sometimes difficult to decide whether an explanation for number of focus groups had been given and what this explanation was. The number of studies that give practical reasons for their number of groups or that refers to data saturation is therefore slightly uncertain. Because of meagre and unclear reporting, it was also difficult to tell whether the studies that claimed to have decided on sample size through an iterative research process and point of saturation did, in fact, analyse their data after data collection had ended. Usually, several confusing aspects of the reporting appeared in the same studies.

As our study looked at focus groups only, future research should consider whether sample size reporting of individual interviews shows similar problems.

While researchers should always provide correct and detailed information about the methods used, our study shows poor and inconsistent reporting of focus group sample size. Editorial teams should be encouraged to use guidelines for reporting of methods for qualitative studies, such as RATS http://www.biomedcentral.com/info/ifora/rats .

Our study also indicates that poor reporting could reflect a lack of clear, evidence-based guidance about how to achieve optimal sample size. To amend this situation, text books and teaching material based on empirical studies into the use of focus group methodology and applicable and precise recommendations are needed. Ironically, one barrier to high-quality methodological studies is the current lack of proper reporting by authors of primary studies.

Authors' information

BC and CG are social anthropologists with PhDs in health services research. Both authors have conducted multimethod research, including focus group studies, partly on assignment and have mainly published in health science journals.

Morgan DL: Focus Groups. Annual Review of Sociology. 1996, 22: 129-152. 10.1146/annurev.soc.22.1.129.

Article   Google Scholar  

Krueger RA, Casey MA: Focus groups: a practical guide for applied research. 2009, Thousand Oaks, California: Sage, 4

Google Scholar  

Powell RA, Single HM: Focus groups. Int J Qual Health Care. 1996, 8: 499-504.

CAS   PubMed   Google Scholar  

Morgan DL: Focus Groups as Qualitative Research. 1997, Thousand Oaks: Sage Publications, 2

Book   Google Scholar  

Kitzinger J: Qualitative Research: Introducing focus groups. BMJ. 1995, 311: 299-302.

Article   CAS   PubMed   PubMed Central   Google Scholar  

Bender DE, Ewbank D: The focus group as a tool for health research: issues in design and analysis. Health Transition Review. 1994, 4: 63-79.

Twohig PL, Putnam W: Group interviews in primary care research: advancing the state of the art or ritualized research?. Fam Pract. 2002, 19: 278-284. 10.1093/fampra/19.3.278.

Article   PubMed   Google Scholar  

Halcomb EJ, Gholizadeh L, DiGiacomo M, Phillips J, Davidson PM: Literature review: considerations in undertaking focus group research with culturally and linguistically diverse groups. Journal of Clinical Nursing. 2007, 16: 1000-1011. 10.1111/j.1365-2702.2006.01760.x.

Dyas J, Apekey T, Tilling M, Siriwardena AN: Strategies for improving patient recruitment to focus groups in primary care: a case study reflective paper using an analytical framework. BMC Medical Research Methodology. 2009, 9: 65-10.1186/1471-2288-9-65.

Article   PubMed   PubMed Central   Google Scholar  

Fern EF: Focus groups: A review of some contradictory evidence, implications, and suggestions for future research. Advances in consumer research. 1983, 10: 121-126.

Krueger RA: Quality control in focus group research. Successful Focus Groups: Advancing the State of the Art. Edited by: Morgan DL. 1993, Newbury Park: Sage, 65-85.

Chapter   Google Scholar  

Hughes D, DuMont K: Using focus groups to facilitate culturally anchored research. American Journal of Community Psychology. 1993, 21: 775-806. 10.1007/BF00942247.

Stewart DW, Shamdasani PN, Rook DW: Focus Groups. Theory and Practice. 2007, Thousand Oaks: Sage Publications

Bryant A, Charmaz K: The SAGE Handbook of Grounded Theory. 2007, London: SAGE

Strauss A, Corbin J: Basics of Qualitative Research. Grounded Theory Procedures and Techniques. 1990, Newbury Park, California: SAGE Publications

Bowling A: Research methods in health. 2002, Berkshire, UK: Open University Press, 2

Morgan DL: The Focus Group Guide Book. 1998, Thousand Oaks: SAGE Publications

Patton MQ: Qualitative Research and Evaluation Methods. 2002, London: Sage Publications

Fern EF: The Use of Focus Groups for Idea Generation: The Effects of Group Size, Acquaintanceship, and Moderator on Response Quantity and Quality. Journal of Marketing Research. 1982, 19: 1-13. 10.2307/3151525.

Sandelowski M: Sample-size in qualitative research. Research in Nursing & Health. 1995, 18: 179-183.

Article   CAS   Google Scholar  

Morse JM: Sampling in Grounded Theory. The SAGE Handbook of Grounded Theory. Edited by: Bryant A, Charmaz K. 2007, London: SAGE Publications

Glaser BG, Strauss AL: The Discovery of Grounded Theory: Strategies for Qualitative Research. 1967, Chicago: Aldine

Denzin NK, Lincoln YS, (Eds): The SAGE Handbook of Qualitative Research. 2005, Thousand Oaks, Third

Kennedy TJT, Lingard LA: Making sense of grounded theory in medical education. Medical Education. 2006, 40: 101-108. 10.1111/j.1365-2929.2005.02378.x.

Charmaz K: Grounded Theory in the 21st Century: Applications for Advancing Social Justice Studies. The SAGE Handbook of Qualitative Research. Edited by: Denzin NK, Lincoln YS. 2005, London: SAGE Publications, 507-535. Third

Charmaz K, Mitchell RG: Grounded Theory in Ethnography. Handbook of Ethnography. Edited by: Atkinson P, Coffey A, Delamont S, Lofland J, Lofland L. 2001, Los Angeles: SAGE

Miles M, Huberman A: Qualitative data analysis: an expanded sourcebook. 1994, London: Sage

Constantine R, Brownstein JN, Hoover S, Wordlaw-Stinson L, Orenstein D, Jones P, Farris R: Strategies for controlling blood pressure among low-income populations in Georgia. Prev Chronic Dis. 2008, 5: A52-

PubMed   PubMed Central   Google Scholar  

Cabassa LJ, Hansen MC, Palinkas LA, Ell K: Azucar y nervios: explanatory models and treatment experiences of Hispanics with diabetes and depression. Soc Sci Med. 2008, 66: 2413-2424. 10.1016/j.socscimed.2008.01.054.

Kothari A, Driedger SM, Bickford J, Morrison J, Sawada M, Graham ID, Crighton E: Mapping as a knowledge translation tool for Ontario Early Years Centres: views from data analysts and managers. Implement Sci. 2008, 3: 4-10.1186/1748-5908-3-4.

Winickoff JP, Park ER, Hipple BJ, Berkowitz A, Vieira C, Friebely J, Healey EA, Rigotti NA: Clinical effort against secondhand smoke exposure: development of framework and intervention. Pediatrics. 2008, 122: e363-375. 10.1542/peds.2008-0478.

Lloyd CE, Johnson MR, Mughal S, Sturt JA, Collins GS, Roy T, Bibi R, Barnett AH: Securing recruitment and obtaining informed consent in minority ethnic groups in the UK. BMC Health Serv Res. 2008, 8: 68-10.1186/1472-6963-8-68.

Perez F, Aung KD, Ndoro T, Engelsmann B, Dabis F: Participation of traditional birth attendants in prevention of mother-to-child transmission of HIV services in two rural districts in Zimbabwe: a feasibility study. BMC Public Health. 2008, 8: 401-10.1186/1471-2458-8-401.

Wild K, Boise L, Lundell J, Foucek A: Unobtrusive In-Home Monitoring of Cognitive and Physical Health: Reactions and Perceptions of Older Adults. J Appl Gerontol. 2008, 27: 181-200. 10.1177/0733464807311435.

Tentler A, Silberman J, Paterniti DA, Kravitz RL, Epstein RM: Factors affecting physicians' responses to patients' requests for antidepressants: focus group study. J Gen Intern Med. 2008, 23: 51-57. 10.1007/s11606-007-0441-8.

Kerrison S, Laws S, Cane M, Thompson A: The patient's experience of being a human subject. J R Soc Med. 2008, 101: 416-422. 10.1258/jrsm.2007.070288.

Rosen T, Lachs MS, Bharucha AJ, Stevens SM, Teresi JA, Nebres F, Pillemer K: Resident-to-resident aggression in long-term care facilities: insights from focus groups of nursing home residents and staff. J Am Geriatr Soc. 2008, 56: 1398-1408. 10.1111/j.1532-5415.2008.01808.x.

Davis AM, James RL, Curtis MR, Felts SM, Daley CM: Pediatric obesity attitudes, services, and information among rural parents: a qualitative study. Obesity (Silver Spring). 2008, 16: 2133-2140. 10.1038/oby.2008.312.

Hatchette JE, McGrath PJ, Murray M, Finley GA: The role of peer communication in the socialization of adolescents' pain experiences: a qualitative investigation. BMC Pediatr. 2008, 8: 2-10.1186/1471-2431-8-2.

Burnet DL, Plaut AJ, Ossowski K, Ahmad A, Quinn MT, Radovick S, Gorawara-Bhat R, Chin MH: Community and family perspectives on addressing overweight in urban, African-American youth. J Gen Intern Med. 2008, 23: 175-179. 10.1007/s11606-007-0469-9.

Power JD, Badley EM, French MR, Wall AJ, Hawker GA: Fatigue in osteoarthritis: a qualitative study. BMC Musculoskelet Disord. 2008, 9: 63-10.1186/1471-2474-9-63.

Hsu C, Phillips WR, Sherman KJ, Hawkes R, Cherkin DC: Healing in primary care: a vision shared by patients, physicians, nurses, and clinical staff. Ann Fam Med. 2008, 6: 307-314. 10.1370/afm.838.

Agampodi SB, Agampodi TC, Ukd P: Adolescents perception of reproductive health care services in Sri Lanka. BMC Health Serv Res. 2008, 8: 98-10.1186/1472-6963-8-98.

Goldstein NE, Mehta D, Siddiqui S, Teitelbaum E, Zeidman J, Singson M, Pe E, Bradley EH, Morrison RS: "That's like an act of suicide" patients' attitudes toward deactivation of implantable defibrillators. J Gen Intern Med. 2008, 23 (Suppl 1): 7-12. 10.1007/s11606-007-0239-8.

Mohler MJ, Coons SJ, Hornbrook MC, Herrinton LJ, Wendel CS, Grant M, Krouse RS: The health-related quality of life in long-term colorectal cancer survivors study: objectives, methods and patient sample. Curr Med Res Opin. 2008, 24: 2059-2070. 10.1185/03007990802118360.

Griffin SF, Wilson DK, Wilcox S, Buck J, Ainsworth BE: Physical activity influences in a disadvantaged African American community and the communities' proposed solutions. Health Promot Pract. 2008, 9: 180-190. 10.1177/1524839906296011.

O'Donnell CA, Higgins M, Chauhan R, Mullen K: Asylum seekers' expectations of and trust in general practice: a qualitative study. Br J Gen Pract. 2008, 58: e1-11.

Joos S, Musselmann B, Miksch A, Rosemann T, Szecsenyi J: The role of complementary and alternative medicine (CAM) in Germany - A focus group study of GPs. BMC Health Services Research. 2008, 8: 127-10.1186/1472-6963-8-127.

Neergaard M, Olesen F, Jensen A, Sondergaard J: Palliative care for cancer patients in a primary health care setting: Bereaved relatives' experience, a qualitative group interview study. BMC Palliative Care. 2008, 7: 1-10.1186/1472-684X-7-1.

Nicolaidis C, Gregg J, Galian H, McFarland B, Curry M, Gerrity M: "You always end up feeling like you're some hypochondriac": intimate partner violence survivors' experiences addressing depression and pain. J Gen Intern Med. 2008, 23: 1157-1163. 10.1007/s11606-008-0606-0.

Bariman M, Hylander I: Linkage in the chain of care: a grounded theory of professional cooperation between antenatal care, postpartum care and child health care. Int J Integr Care. 2008, 8:

Shuval K, Weissblueth E, Brezis M, Araida A, Faridi Z, Ali A, Katz DL: The Role of Culture, Environment, and Religion in the Promotion of Physical Activity Among Arab Israelis. Prev Chronic Dis. 2008, 8:

Blixen CE, Webster NJ, Hund AJ, Perzynski AT, Kanuch SW, Stoller EP, McCormick RA, Dawson NV: Communicating about alcohol consumption to nonharmful drinkers with hepatitis C: patient and provider perspectives. J Gen Intern Med. 2008, 23: 242-247. 10.1007/s11606-007-0483-y.

Castel LD, Williams KA, Bosworth HB, Eisen SV, Hahn EA, Irwin DE, Kelly MA, Morse J, Stover A, DeWalt DA, DeVellis RF: Content validity in the PROMIS social-health domain: a qualitative analysis of focus-group data. Qual Life Res. 2008, 17: 737-749. 10.1007/s11136-008-9352-3.

Gutteling JJ, de Man RA, Busschbach JJ, Darlington AS: Quality of health care and patient satisfaction in liver disease: the development and preliminary results of the QUOTE-Liver questionnaire. BMC Gastroenterol. 2008, 8: 25-10.1186/1471-230X-8-25.

Nilsson C, Skar L, Soderberg S: Swedish district nurses' attitudes to implement information and communication technology in home nursing. Open Nurs J. 2008, 2: 68-72. 10.2174/1874434600802010068.

Watson DB, Thomson RG, Murtagh MJ: Professional centred shared decision making: patient decision aids in practice in primary care. BMC Health Serv Res. 2008, 8: 5-10.1186/1472-6963-8-5.

Webb C, Kevern J: Focus groups as a research method: a critique of some aspects of their use in nursing research. Journal of Advanced Nursing. 2001, 33: 798-805. 10.1046/j.1365-2648.2001.01720.x.

Article   CAS   PubMed   Google Scholar  

Wagner AB: Open Access Citation Advantage: An Annotated Bibliography. Issues in Science and Technology Librarianship. 2010, 60: [ http://www.istl.org/10-winter/article2.html ]

Pre-publication history

The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-2288/11/26/prepub

Download references

Acknowledgements

The study was funded through BC's position at the University of Bergen. The authors would like to thank the referees for very helpful input to approve our reporting and Regina Küfner Lein, at the Medical and Dental library, Bergen University Library for help with the literature search.

Author information

Authors and affiliations.

Uni Rokkan Centre, Nygårdsgt 5, N-5015, Bergen, Norway

Benedicte Carlsen

SINTEF Society and Technology, P.O. Box 124, Blindern, Oslo, 0314, Norway

Claire Glenton

The Norwegian Knowledge Centre for the Health Services, P.O. Box 7004, St. Olavs plass, N-0130, Oslo, Norway

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Benedicte Carlsen .

Additional information

Competing interests.

The authors declare that they have no competing interests.

Authors' contributions

BC designed the study, participated in developing the search strategy, conducted the search, extracted relevant data for analysis, participated in analysing the data and drafted the manuscript. CG participated in developing the search strategy, checked extracted data, participated in analysing the data and revised the manuscript AB. Both authors read and rewrote the manuscript and have approved the final manuscript.

Electronic supplementary material

12874_2010_539_moesm1_esm.doc.

Additional file 1: The table shows author, journal, number of focus groups and explanation for number of focus groups for all 220 included studies. (DOC 236 KB)

Rights and permissions

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License ( http://creativecommons.org/licenses/by/2.0 ), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Reprints and permissions

About this article

Cite this article.

Carlsen, B., Glenton, C. What about N? A methodological study of sample-size reporting in focus group studies. BMC Med Res Methodol 11 , 26 (2011). https://doi.org/10.1186/1471-2288-11-26

Download citation

Received : 18 August 2010

Accepted : 11 March 2011

Published : 11 March 2011

DOI : https://doi.org/10.1186/1471-2288-11-26

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Focus Group
• Focus Group Interview
• Data Saturation
• Focus Group Study

BMC Medical Research Methodology

ISSN: 1471-2288

• General enquiries: [email protected]

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• HHS Author Manuscripts

What Influences Saturation? Estimating Sample Sizes in Focus Group Research

Monique m. hennink.

Hubert Department of Global Health, Rollins School of Public Health, Emory University, 1518 Clifton Road, Atlanta, GA 30322, USA

Bonnie N. Kaiser

Department of Anthropology, University of California, San Diego (previous)

Duke University (current)

Mary Beth Weber

Hubert Department of Global Health, Rollins School of Public Health, Emory University

Saturation is commonly used to determine sample sizes in qualitative research, yet there is little guidance on what influences saturation. We aimed to assess saturation and identify parameters to estimate sample sizes for focus group studies in advance of data collection. We used two approaches to assess saturation in data from 10 focus group discussions. Four focus groups were sufficient to identify a range of new issues (code saturation), but more groups were needed to fully understand these issues (meaning saturation). Group stratification influenced meaning saturation, whereby one focus group per stratum was needed to identify issues; two groups per stratum provided a more comprehensive understanding of issues, but more groups per stratum provided little additional benefit. We identify six parameters influencing saturation in focus group data: study purpose, type of codes, group stratification, number of groups per stratum, and type and degree of saturation.

Introduction

Selecting an appropriate sample size for qualitative research remains challenging. Since the goal is to select a sample that will yield rich data to understand the phenomenon studied, sample sizes may vary significantly depending on the characteristics of each study. The concept of saturation is the most common guiding principle to assess the adequacy of data for a purposive sample ( Morse 1995 ; 2015 ). Saturation was developed by Glaser & Straus (1967) in their grounded theory approach to qualitative research, which focuses on developing conceptual or explanatory models from textual data. Within this context, theoretical saturation (also called theoretical sampling) is used. Theoretical sampling is both continuous and data-driven, involving an iterative process of concurrent sampling, data collection, and data analysis to determine further data sources ( Charmaz 2014 ) and continuing until all constructs of a phenomenon are explored and exhausted to support an emerging theory ( Glaser and Strauss 1967 ). In grounded theory, theoretical saturation therefore focuses on the adequacy of a sample to provide conceptual depth and richness to support an emerging theory, rather than on sample size per se ( Corbin and Strauss 2015 ; Birks and Mills 2011 ).

However, the term saturation has become part of the broader lexicon of qualitative research (O’Reilly and Parker 2012). It is now widely used outside of its grounded theory origins and has become “the most frequently touted guarantee of qualitative rigor” ( Morse 2015 , p.587). In this broader application, it is often called data saturation or thematic saturation and refers to the point in data collection when issues begin to be repeated and further data collection becomes redundant i . This broader use of saturation focuses more directly on assessing sample size rather than the adequacy of data to develop theory (as in theoretical saturation). Yet, it remains unclear what saturation means when used outside of grounded theory, how it can be assessed, and what influences saturation for different qualitative methods, types of data, or research objectives (Nelson 2017; O’Reilly and Parker 2012; Kerr et al 2010 ). Identifying influences on saturation in this broader context can be used by researchers, reviewers, ethical review boards, and funding agencies to determine effective sample sizes in qualitative research proposals.

Determining the number of focus group discussions needed in a study is a key part of research design, as sample size influences multiple study components (e.g., budget, personnel, timetable). However, saturation (and therefore sample size) cannot be determined in advance, as it requires the review of study data. There is limited methodological research on parameters that influence saturation to assist researchers in selecting an effective sample size in advance of data collection ( Morse 1995 ; Guest et al 2006 ; Kerr et al 2010 ; Carlsen & Glenton 2011 ; Bryman 2012 ). Guidance is needed on how to estimate saturation prior to data collection to identify an effective sample size for qualitative research proposals. In this study, we aim to assess saturation in data from focus group discussions and identify parameters to estimate sample sizes for focus group studies in advance of data collection. This goal contributes to continual calls for an evidence base of empirical research on saturation ( Morse 1995 ; Guest et al 2006 ; Kerr et al 2010 ; Carlsen & Glenton 2011 ).

Assessing Saturation in Focus Group Research

There is a growing concern in qualitative research over researchers claiming to have reached saturation in qualitative studies without providing adequate explanation or justification on how saturation was assessed or achieved ( Hennink, Kaiser & Marconi 2016 ; Malterud et al 2015; Bowen 2008 ; Guest, Bunce & Johnson 2006 ; Morse, 1995 , 2000 , 2015 ). Carlsen and Glenton (2011) conducted a systematic review of 220 studies using focus group discussions to identify how authors justified their sample size. They found that of those studies that explained the sample size, 83% used saturation to justify the number of focus groups in the study, but few authors stated clearly how saturation was assessed or achieved, and many included unsubstantiated claims of saturation and reported reaching saturation while still using the predetermined number of focus groups. Guest et al (2016) similarly reviewed 62 textbooks on qualitative research or focus group methodology and found that 42 provided no guidance at all on the number of focus group discussions needed for a study, 6 recommended saturation, 10 gave a numeric recommendation, and 4 mentioned saturation and provided a suggested number of focus groups. The sample size recommendations ranged widely from 2 to 40 groups, with a commonly cited guideline to conduct at least two focus groups for each defining demographic characteristic. A critical outcome of this review is that none of the recommendations on sample sizes were supported by empirical data demonstrating when saturation is achieved in focus group research. These reviews highlight a significant gap in the methodological literature on empirical research assessing saturation and sample size guidelines for focus group research.

There is a small but emerging body of methodological literature assessing saturation in studies using in-depth interviews ( Hennink et al 2016 ; Francis et al 2010 ; Guest et al, 2006 ). However, there are few methodological studies where the authors assess saturation and provide guidance on sample sizes for focus group research. Guest et al (2016) used data from a study with 40 focus group discussions to develop empirically based recommendations on sample sizes for focus group studies. In this study, the authors documented the process of identifying codes to ascertain when each code was developed and then determined the number of focus groups needed to identify 80% and 90% of thematic codes across the study. They assessed code frequency across data as a proxy for salience of themes and accounted for any temporal bias in identifying codes by randomizing the order of focus groups and replicating their analyses of saturation. Results showed that 64% of themes were generated from the first focus group, 84% by the third focus group and 90% by the sixth group. This pattern remained regardless of the order in which focus groups were reviewed. Three focus groups were also enough to identify the most prevalent themes across these data. The authors conclude that when averaging the sequential and randomized order of focus groups, two to three focus groups are sufficient to capture 80% of themes, including the most prevalent themes, and three to six groups for 90% of themes in a homogenous study population using a semi-structured discussion guide ( Guest et al 2016 ).

In an earlier study, Coenen et al (2012) assessed saturation in focus group discussions by different approaches to developing codes: an inductive approach of identifying themes from the data itself and a deductive approach of applying themes to data from an existing theoretical framework. The authors used maximum variation sampling to create a diverse sample of participants, which differs from the largely homogenous sample used by Guest et al (2016) . Saturation was defined as the point at which linking concepts of two consecutive focus groups revealed no additional second-level categories. The authors deemed that saturation occurred at five focus groups, regardless of the approach to code development.

It is difficult to compare results of these two studies given that they operationalize saturation differently – percentage of codes identified across data ( Guest et al 2016 ) and absence of new category development in consecutive focus groups ( Coenen et al 2012 ). Nonetheless, across both studies, saturation is achieved by six focus groups. The findings of these studies are significant, as they demonstrate that saturation is achieved at a relatively small number of focus groups, compared with typical guidance given in methodology textbooks that is not empirically based. Taken together, these studies begin to contribute an understanding of saturation in focus group research using homogenous and diverse samples, amongst inductive and deductively derived codes, and saturation in codes versus categories. Despite the differences in the type of sample, type of codes assessed, and operationalization of saturation, both studies reached saturation at a relatively similar number of focus groups.

However, there are two important limitations of these studies. First, the assessment of saturation is based on identifying the occurrences of new themes, without also assessing the understanding of these themes across the data. Identifying the presence of themes is only the first step in reaching saturation. The first time a theme is identified in data may not be detailed or insightful; therefore, additional data may be required to fully capture the meaning of the issue and to understand the depth, breadth, and nuance of the issue ( Kerr et al 2010 ; Hennink et al 2016 ). Thus, the authors of these studies provide no guidance on sample sizes needed to reach saturation in the meaning of issues in data. Second, they hardly acknowledge the group format of data collection in focus groups and how this may influence saturation. Focus group discussions involve non-directive interviewing whereby group participants engage in discussion, which generates a different type of data than interviews with a single participant due to the interaction and spontaneity of the group dialogue ( Morgan, 1997 ; Krueger and Casey 2015 ). The group format has potential to generate a broad range of issues and perspectives, but it may also sacrifice narrative depth and understanding of issues. It is unclear how these elements of focus group discussions influence saturation. Finally, there is no examination of how group composition or demographic stratification of focus groups influence saturation. Assessing how these design elements of focus group research influence sample size and saturation is critically important yet is omitted from current literature.

In this study, we aimed to assess saturation in focus group data and identify parameters to estimate sample sizes for focus group studies in advance of data collection. We utilize the broader application of saturation used outside the grounded theory approach, as described above. This focus is important given that saturation is commonly applied to a wide range of research approaches without adequate description or justification of how it was applied or achieved.

We use two approaches to assess saturation that we developed in an earlier study – code saturation and meaning saturation ( Hennink et al 2016 ) – to assess saturation in focus group data. First, we assessed the sample size needed to reach code saturation, which we define as the point when no additional issues are identified in data and the codebook has stabilized. Second, we assessed the sample size needed to reach meaning saturation, which we define as the point at which we fully understand the issues identified and when no further insights or nuances are found. We then examined code and group characteristics to identify parameters that influence saturation in focus group data. We sought to answer the following specific research questions in this study:

• How many focus group discussions are needed to reach code saturation?
• How many focus group discussions are needed to reach meaning saturation?
• How do code characteristics and focus group composition influence saturation?
• What parameters can be used to assess saturation a priori in focus group research?

Study Background

We used data from the South Asian Health and Prevention Education (SHAPE) study for our analyses ( clinicaltrials.gov # {"type":"clinical-trial","attrs":{"text":"NCT01084928","term_id":"NCT01084928"}} NCT01084928 ). Below we provide an overview of the data collection and analysis of the broader SHAPE study as context for our analyses of saturation in these data. The SHAPE study was a planning and feasibility study to inform the design of a diabetes prevention program for South Asian Americans. South Asians (individuals from the Indian subcontinent) are at a high risk for developing diabetes, often presenting with the condition at younger ages and lower body mass indices than other race-ethnic groups ( Gujral et al 2013 ). Although there is strong evidence from randomized controlled trials showing that lifestyle education interventions can prevent or delay type 2 diabetes in high-risk populations ( Crandall et al 2008 ), there is a need to translate these programs to the South Asian population. SHAPE included a formative phase of qualitative research to inform the development and tailoring of the intervention to the needs of the South Asian community and to ensure its cultural acceptability.

Data Collection and Analysis

SHAPE data comprised focus group discussions with self-identifying South Asians adults living in Atlanta, Georgia. Participants were purposively recruited through advertisements in local South Asian magazines, health fairs, and screenings targeting South Asians, and community locations such as South Asian shopping centers and community organizations. Sixteen focus group discussions were conducted in community locations. Focus groups were stratified by age (18–39 years and 40 years or older) and sex, comprising four groups in each stratum. Focus groups lasted 60–90 minutes and were conducted in English by a trained moderator matched for sex, but not ethnicity, to the participant group. Participants were asked open-ended questions on their views of diabetes and obesity, diet and physical activity behaviors, and barriers and facilitators for a healthy lifestyle, as well as providing feedback on specific design elements of the intervention. Participants were given refreshments, travel reimbursement, and a gift bag. We used data from the first ten focus group discussions in this study, as the final 6 groups focused only on some of the discussion topics and therefore, were not suitable for our analyses of saturation. Data were collected between November 2009 and March 2010. The Emory Institutional Review Board (IRB00019630) approved the study. Individual informed consent was sought from participants before each focus group discussion. Participants were informed of the study procedures, risks and benefits and provided written consent to take part in the focus group discussions and for the audio recording.

All focus groups were digitally recorded, transcribed verbatim, de-identified, and entered into the MaxQDA program (Verbi Software, Germany) for analysis. We conducted a close reading of transcripts to identify issues raised by participants. Each issue was verified by two analysts before its inclusion in a codebook, comprising a codename and a description of each issue. A total of 50 codes were developed including both inductive codes derived from the transcripts and deductive codes originating from the discussion guide. Inter-coder agreement was assessed between two coders to compare the consistency of code use and rectify discrepancies before the whole data set was coded.

To assess saturation, we used a similar process developed in an earlier study ( Hennink et al 2016 ). For our analyses on saturation, we documented the process of code development and conducted separate analyses of these procedural data, as described in the sections below.

Assessing Code Saturation

To assess code saturation, we reviewed each focus group discussion transcript in the order in which groups were conducted and documented the development of codes. We recorded all new codes developed and their characteristics, including the code name, code definition, type of code (inductive or deductive), notes about issues with new codes (e.g., clarity of the issue captured, completeness of the code definition), and whether any previously developed codes were present in the transcript. Code definitions included a description of the issue captured, instructions for code application, and an example of text relevant to the code. To document the evolution of code development, we also recorded changes made to codes or code definitions as we proceeded, including the type of change and the focus group at which the change was made. We continued to document code development and the iterative evolution of codes for each focus group discussion until the final codebook was complete.

We then categorized codes for analysis, using the same categorizations as we developed in our earlier work on saturation ( Hennink et al 2016 ) as follows. First, codes were categorized as inductive or deductive. Inductive codes were content driven and raised by participants themselves, whereas deductive codes originated from the discussion guide and were then verified with data. Second, changes in code development were categorized as a change in code name and change in code definition (e.g. code expanded, inclusion criteria or examples added). Third, codes were also categorized as concrete or conceptual. ‘Concrete’ codes were those capturing explicit, definitive issues in data; for example, the code ‘food taste’ captured concrete discussion about the taste of food. Similarly, the code ‘family time’ captured any discussion about exercise time competing with family responsibilities. ‘Conceptual’ codes were those capturing abstract constructs such as perceptions, emotions, judgements, or feelings. For example, the conceptual code ‘denial’ captured comments about failure to recognize symptoms of diabetes, refusing testing, or rejecting a diagnosis of diabetes, for example “They just don’t want to admit that okay we have this disease.” These categorizations of codes were used in our analyses to quantify the types of codes developed, types of changes to code development, and timing of code development. Finally, codes were categorized as high or low prevalence. Code prevalence was defined by the number of focus group discussions in which a code was present. On average, codes were present in 7 focus group discussions; therefore, we defined high prevalence codes as those present in more than 7 focus group discussions and low prevalence codes as those present in equal to or fewer than 7 focus groups. In total, there were 27 high-prevalence codes and 23 low-prevalence codes.

To assess whether code saturation was influenced by the order in which focus groups were conducted, we randomized the order of groups and mapped hypothetical code development onto the random order. To do this, we randomized the focus groups using a random number generator, but we did not repeat the process of reviewing transcripts given the bias this would have introduced, as this process had already been done with the same transcripts in their actual order. Instead, we assumed that codes would be developed after the same number of repetitions of that issue across the focus groups. For example, in the actual code development, the code ‘cultural expectations’ was created in focus group 3, after the issue was mentioned in focus groups 1 and 2. Thus, in the random order, we assumed that the same code would likewise be developed after 3 groups in which the issue was raised. Our aim was to reflect researchers’ style of code development in the random order as in the actual order, so that we could more directly assess the effect of the order of focus groups on code development. We replicated the pattern of code development in the randomized order of groups by calculating the number of times a code was present (as indicated by the number of focus groups in which the code was applied to the data) before the focus group in which the code was created. We then used these numbers to map hypothetical code development in the randomized order of groups. We then compared hypothetical code development with that from the actual order in which focus groups were conducted.

Assessing Meaning Saturation

We followed the same process to assess meaning saturation (described below) that we used in our previous study on saturation with in-depth interview data ( Hennink et al 2016 ), with the addition of two components to reflect the use of focus group data in this study.

To assess meaning saturation, we selected 19 codes that were central to the aims of the original study on diet, exercise and diabetes and included different types of codes. These codes comprised a mix of concrete (13 codes) and conceptual codes (6 codes) and high prevalence (10 codes) and low prevalence (9 codes) codes (as defined above). This selection reflected the nature of codes developed in this study, whereby there were more concrete than conceptual codes. To assess meaning saturation, we traced these 19 codes to identify what we learned about the code in each successive focus group discussion. This involved using the coded data to search for the code in the first focus group discussion and noting what we learned about this issue from this focus group, then searching for the code in the next focus group and noting any new aspects or nuances of the code from that group, and continuing until all 10 focus groups had been reviewed. This process was repeated for all 19 codes that were traced. For each code, we noted at which focus group there were no new aspects of a code raised and no further understanding of the code, only the repetition of earlier aspects. We deemed this as the point of meaning saturation for that code. We then compared the number of focus group discussions needed to reach meaning saturation with the number needed to reach code saturation determined in our earlier analyses.

To assess whether meaning saturation is influenced by the type of code, we compared the timing of saturation for concrete and conceptual codes. Concrete codes included: ‘family time’, ‘homeopathy’, ‘exercise instructor’, ‘exercise measures’, ‘exercise gender’, ‘exercise venues’, ‘physical appearance’, ‘ingredient cost’, ‘food taste’, ‘diabetes cause’, ‘US-Indian food’, ‘exercise barriers’, and ‘exercise perception’. Conceptual codes included: ‘denial’, ‘exercise pleasure’, ‘work success’, ‘women’s responsibility’, ‘mood’, and ‘cultural expectations’. To assess whether meaning saturation is influenced by the prevalence of a code, we compared saturation by high and low prevalence codes.

To assess whether meaning saturation is influenced by the number of participants who discussed a code, we noted the number of participants contributing to the discussion of each code across all focus groups. If 4 people had discussed a code in the first focus group, 2 in the second, and 6 in the third, we determined that a total of 12 participants had discussed this code across the data. We then identified whether there was any pattern in saturation by the number of participants discussing a code. Finally, to assess how saturation is influenced by the demographic stratification of the focus groups (described earlier), we noted the age and sex composition of each group on the trajectories and identified any patterns in saturation by these strata.

Part I: Code Saturation

Code development.

Figure 1 shows the timing of code development across all focus groups in the study. The figure shows the focus group discussions in the order in which they were conducted, the number of new codes developed in each successive focus group, the type of code developed (inductive, deductive), and the demographic stratum of each focus group. A total of 50 codes were developed in the study comprising 58% inductive and 42% deductive codes. Deductive codes were developed only from focus groups 1 and 2, with only inductive codes added thereafter. The vast majority of codes (60%) were identified in the first focus group discussion, with a sharp decline in new codes after this. From the second focus group, an additional 12 codes were developed, with 84% of codes developed at this point. Focus groups 3 to 6 added 8 new codes, with only a few new codes per focus group; most of these new codes (5/8) were of low prevalence across the data. After focus group 6, no further new codes were developed.

Timing of code development and code saturation.

Note. Two deductive codes were developed in Focus Group 2. These codes were derived from questions in the discussion guide that were not probed in Focus Group 1 but were probed in Focus Group 2. FGD = focus group discussion; Y = Younger; M = Male; O = Older; F = Female.

Given that the majority of new codes (60%) were identified in the first focus group discussion, we assessed whether the order in which the focus groups were conducted influenced the pattern of new code development and code saturation, particularly given the demographic stratification of the focus groups in this study. To assess this, we compared the pattern of code development in the actual order in which focus groups were conducted with a randomized order of focus groups. Figure 2 shows the same pattern of code development in both the actual and the randomized order of focus groups, with approximately 60% of codes developed in the first focus group discussion and a strong decline in new codes identified in subsequent focus groups. We also find that both scenarios reach saturation with over 90% of codes developed at focus group 4 (94% and 92% in the actual and random order, respectively). Therefore, the order in which focus groups are conducted has little influence on the pattern of new code development or on code saturation.

Timing of code development—Actual versus randomized order of FGDs.

Note. FGD = focus group discussion.

Codebook development

We recorded codebook development by documenting the timing of changes to codes and code definitions ( Table 1 ). Most changes to code definitions occurred in focus groups 2 and 3, with no changes to the codebook occurring after focus group 6. The two most common types of code definition changes were adding examples and conceptually expanding a code definition, which consisted of adding a new dimension of the issue to the code definition. For example, the original code definition of the code ‘exercise with friends’ was “Discussion about whom to exercise with in the intervention (e.g., friends, other South Asians).” Following focus group 2, this code definition was expanded to include “social support to exercise”, and following focus group 5, it was expanded again to include the concept of “group accountability as a motivator to exercise.” While over half of the codes (58%) were inductive, most of the code definition changes (84%) were made to deductive codes, to ensure that these codes, which were derived externally from the data, effectively reflected the issue raised in the data. Some codes were refined multiple times, with over one-third of the code definition changes made to only three codes (‘exercise perception’, ‘exercise barriers’, and ‘healthy diet barriers’).

Changes to codebook

Code prevalence

To identify when more or less prevalent codes were developed, we examined code development by code prevalence and type of code. Figure 3 depicts each code as a separate bar: the location of a code on the x-axis indicates the focus group in which the code was developed, and the height of the bar shows the number of focus groups in which the code was used. For example, the first 13 bars show that these codes were all developed in the first focus group discussion and were all high prevalence codes, present in all 10 focus groups. The dashed line indicates the average number of focus group in which a code was used – about 7 focus groups. Figure 3 shows that 27 codes were of high prevalence (above the line), and 23 were of low prevalence (below the line) across all data. The majority of high-prevalence codes (81%, 22/27) were identified in the first focus group discussion, and by the third focus group, 96% (26/27) of all high-prevalence codes were identified. Thus, the vast majority of high-prevalence codes were identified in early focus group discussions. Most low-prevalence codes (65%, 15/23) were developed after the first focus group, with a clustering at focus group 2. This shows that more focus groups are needed to identify low-prevalence codes.

Timing of code development by code prevalence and type.

a Dashedline indicates average code prevalence across all data at 7.2 FGDs.

Figure 3 also shows when the different types of codes (concrete or conceptual) were developed and their prevalence across the data. The first focus group almost exclusively generated concrete codes (97%, 29/30), most of which were also of high prevalence (76%, 22/29). In contrast, half of the conceptual codes were low prevalence (50%, 3/6) and developed later, in focus groups 2, 3, and 6. Overall, codes developed in early focus groups were high prevalence concrete codes, while those developed in later focus groups were mostly low prevalence and included more conceptual codes.

Code saturation

We did not use a set threshold to determine code saturation but were guided by the results of our analyses. We determined that code saturation was reached after four focus group discussions, based on code identification (94% of codes had been identified), code prevalence (96% of high-prevalence codes were identified) and codebook stability (90% of codebook changes were made). Therefore, four focus group discussions were sufficient to identify the range of issue present in these data.

Part II: Meaning Saturation

Having established that code saturation was reached at four focus group discussions, we then explored whether four groups are also sufficient to reach meaning saturation, whereby we gain a comprehensive understanding of the issues raised. To assess meaning saturation, we traced a range of codes across all focus groups and noted what we learned about each code from successive focus groups until no more new dimensions of the code were uncovered. Meaning saturation was deemed to be reached at the last focus group discussion in which a new dimension of the code was identified. For example, the code ‘food taste’ was identified in the first focus group discussion and reached meaning saturation at the second focus group, as no more new aspects of this code were identified (after this point, there was only repetition of dimensions already identified). The code ‘work success’ was also identified in the first focus group discussion, but new aspects of the code were identified in the fourth and sixth focus group but none thereafter, so this code reached meaning saturation by focus group six. We traced 17 codes to assess meaning saturation, comprising a mix of concrete and conceptual codes, and high- and low-prevalence codes. Below we assess the influence of these factors on meaning saturation of these codes.

Figure 4 shows the results of the code tracing, indicating the focus group at which each code was first identified in data and the focus group at which it reached meaning saturation. Most codes were identified in the first focus group discussion, but they did not reach meaning saturation until later focus groups. This shows that data from multiple focus groups are needed to understand many of the issues, with successive focus groups adding different dimensions of a code until a more complete understanding of the issue is reached. Codes also reached meaning saturation at different points in the data, some requiring more data to fully understand the issue. Both concrete and conceptual codes needed data from a range of focus groups to reach meaning saturation. For example, the concrete code ‘exercise gender’ and the conceptual code ‘work success’ needed data from 6 and 7 focus groups, respectively, to capture the various perspectives on each of these issues. This shows that reaching saturation needs to go beyond code saturation (whereby codes are identified in data) towards meaning saturation to fully understand the issues raised, and capture the different dimensions, context, and nuances of the issues.

Timing of first use of codes and their meaning saturation.

* concrete code.

Figure 4 showed that codes reached meaning saturation at different points in the data. Below, we examine a range of influences on reaching meaning saturation, including the type of code (concrete or conceptual), code prevalence, demographic strata of focus groups, and the number of participants who discussed an issue across data.

We explored when different types of codes (concrete or conceptual) reached meaning saturation. Concrete codes reached meaning saturation at different points. A few concrete codes reached meaning saturation after one or two focus groups, for example the codes ‘exercise instructor’ and ‘food taste’ ( Figure 4 ). Several concrete codes clustered to reach meaning saturation at focus group 5. The remaining concrete codes reached meaning saturation later, at focus groups 6 to 11. Conceptual codes showed a more consistent pattern in reaching meaning saturation. All conceptual codes were first identified in focus group 1 or 2, but they did not reach meaning saturation until focus group 5 or later, with one code not reaching saturation (‘cultural expectations’). No conceptual codes reached meaning saturation in fewer than five focus groups, which shows that more data are needed to reach meaning saturation for conceptual codes than for some concrete codes. Overall, with the exception of three concrete codes, all other codes reached meaning saturation at focus group 5 or later.

We examined how code prevalence (high or low prevalence) influences meaning saturation. Figure 4 shows that high-prevalence codes reached meaning saturation between focus groups 4 and 10, with a clustering at focus groups 5 and 6. High-prevalence conceptual codes needed more focus groups to reach meaning saturation (6–10 focus groups) compared with high-prevalence concrete codes (4–5 focus groups). Low-prevalence codes reached meaning saturation between focus groups 2 and 8 and also showed a clustering at focus groups 5 and 6. While all low-prevalence conceptual codes reached saturation at focus group 5 or 6, low-prevalence concrete codes showed a much more mixed pattern, reaching saturation between focus groups 2 and 8. These results show that even high-prevalence codes require a range of focus groups to fully understand issues, with high-prevalence conceptual codes requiring more data than high prevalence concrete codes.

The strongest pattern in meaning saturation was found by demographic strata of the focus groups. Each focus group was stratified by age (younger/older) and sex (men/women), with two to three focus groups conducted within each stratum across the study. Figure 4 indicates the order in which focus group discussions were conducted and the characteristics of each stratum. At focus group 5, each stratum had been included once, and by focus group 10, each stratum had been included two to three times. Results show that regardless of the type of code or code prevalence (described above), meaning saturation clusters at focus group 5 once all strata have been included once. This shows that for many codes, meaning saturation is reached once the perspectives of all strata have been included, and therefore the diversity within each code has been captured. This is shown in Table 2 , which exemplifies the different dimensions of a code that are captured in each demographic stratum. For example, with the concrete code ‘exercise barriers’, the first focus group identified issues specific to this stratum of young men (e.g., priority for education over exercise and physical appearance), in addition to other issues. The second focus group adds the perspective of older women about the lack of awareness of exercise benefits beyond weight loss. The third and fourth focus group add a range of issues related to older men, such as barriers of weather and managing an exercise routine. The fifth focus group adds the perspective of young women, regarding the need to be accompanied to exercise classes and the difficulty of exercising at home. Once the perspectives of all strata are included, this code reached meaning saturation, and thereafter only repeated issues were found in later focus groups. Similarly, the conceptual code ‘mood’, which identified emotions related to diet and exercise, captures novel aspects of the code from each of the different strata before it reached meaning saturation.

Examples of code dimensions identified across demographic strata of focus group discussions

There are four codes (in Figure 4 ) that reached meaning saturation before all strata were completed. These were all concrete codes (‘family time’, ‘food taste’, ‘exercise instructor’ and ‘diabetes cause’), with little variation in the way the issue was discussed across all data. For example, the code ‘food taste’ repeatedly focused on the issue that healthy food was not tasty, with no variation across focus groups, while the code ‘exercise instructor’ highlighted that participants prioritized the experience of an instructor over their cultural background, with no further nuances of this issue across focus groups. These results show that codes that are nuanced by the characteristics of demographic strata (e.g., age or sex) reach saturation only after all strata are included, while codes that are not influenced by demographic strata will reach saturation without all strata included.

Given our results on the influence of demographic strata on meaning saturation, we assessed whether more than one focus group per stratum was needed to reach meaning saturation. Results show that some codes needed data from multiple focus groups per stratum to reach meaning saturation. These included both concrete codes (‘exercise venues’, ‘physical appearance’) and conceptual codes (‘cultural expectations’, ‘work success’). The clustering of codes reaching meaning saturation at focus group 6 also shows that a range of codes needed more than one focus group from the strata with young men and older men to reach meaning saturation. For the strata that included three focus group discussions in each – older men and older women – only one code (‘exercise gender’) identified new elements from all three focus groups with older women, suggesting this issue is nuanced only for older women. Other codes showed no new issues in the third focus group in these strata. These results show that conducting 1–2 focus groups per stratum is likely to contribute to a more comprehensive understanding of a code but suggests limited value in conducting three groups per stratum.

We also examined whether the total number of participants discussing a code influenced meaning saturation but found no clear patterns. Codes discussed by a high number of participants across data (e.g., 51) and those discussed with a low number of participants (e.g., 9) both reached meaning saturation at the same point (focus group 5). More participants discussed high-prevalence codes than low-prevalence codes and concrete codes versus conceptual codes; however, there were no patterns by meaning saturation.

Through this study, we contribute to a small but growing evidence base of empirical research on saturation. We sought to assess saturation using two approaches - code saturation and meaning saturation - and to develop parameters of saturation to estimate and justify sample sizes for focus group research in advance of data collection.

Our results show that code saturation was reached at four focus group discussions, whereby 94% of all codes and 96% of high-prevalence codes had been identified, and the codebook had stabilized. The first focus group generated 60% of all new codes with a sharp decline thereafter, regardless of the order in which focus groups were conducted. Most codes developed in early focus groups were high-prevalence, concrete codes. Comparing these results with our earlier study on saturation in in-depth interviews ( Hennink et al 2016 ), we also found the first in-depth interview generated the majority of new codes (53%), most of which were also concrete and high-prevalence codes. While the first focus group generated more new codes than the first in-depth interview, it is not remarkably higher considering issues are generated by a group of participants. These results also reflect the findings of other studies examining saturation in focus group data, whereby code saturation was reached at 5 focus groups ( Coenen et al 2012 ) and between 3 to 6 focus groups ( Guest et al 2016 ). These collective findings provide important evidence that relatively few focus groups are needed to generate a majority of new issues in a study. This contradicts general guidelines provided in academic literature (albeit not based on empirical research) recommending much higher numbers of focus groups in a study (e.g., 10, 20, or 40 focus groups). However, it is important to remember that code saturation identifies the presence of issues in data, in particular high-prevalence concrete issues, but may not provide a full understanding of all issues, their diversity, or nuances. This goal may be suitable for some research objectives, particularly for designing research instruments or interventions; however, these limitations of code saturation should be borne in mind if using this strategy.

We found that reaching meaning saturation requires more data than code saturation. While four focus groups were sufficient to identify the majority of issues across the data, more data were needed to fully understand these issues. Our results showed that even issues identified in the first focus group discussion needed more data to fully understand the issue, regardless of the type of code (concrete/conceptual; high/low prevalence). In addition, codes reached meaning saturation at different points in the data; some codes required much more data than others to reach meaning saturation. Even low-prevalence and conceptual issues contributed to building a comprehensive understanding of a phenomenon; therefore, the prevalence of issues in data does not indicate their significance in understanding the study phenomenon. As indicated by Morse (1995 , p148), “it is often the infrequent gem that puts other data into perspective, that becomes the central key to understanding the data…it is the implicit that is interesting”. Reaching meaning saturation thus relates to the “informational power” (Malterud et al 2015) of the sample to provide depth of understanding of the issues. It goes beyond identifying the presence of issues and moves towards gaining “conceptual depth” to capture the range, complexity, subtlety, resonance, and thereby the validity of issues in data ( Nelson 2016 ). Identifying a sample size adequate to meet these characteristics is critical to maximize the benefits of conducting qualitative research.

The most consistent influence on meaning saturation was the demographic strata of the focus groups. Meaning saturation clustered at focus group five for most concrete codes – this represents the point at which at least one focus group from each demographic stratum was included. This indicates that once the perspectives of each demographic stratum have been captured, meaning saturation was reached on most codes. This finding is compelling because it shows that codes that are nuanced by the characteristics of demographic strata will reach saturation only after all strata are included, while codes not influenced by demographic strata will reach saturation without all strata included, as these issues have less diversity by these characteristics. In our data, the topics diet and exercise are highly nuanced by both sex and age for the South Asian study population; therefore, we continued to identify new insights across codes with each demographic stratum until all strata were included, with few new insights thereafter. Most conceptual codes needed more data, beyond one focus group per stratum to reach meaning saturation.

Many researchers stratify focus groups by demographic characteristics precisely because they anticipate different nuances to emerge from the various strata and to enable analytic comparisons to distinguish patterns in data. The most useful strata to use are often guided by research literature and built into the study design. Overall, it is not the number of groups per se that determines meaning saturation but the point at which all strata are included in the study – in our study this was at five focus groups, but for other studies this point may be different. A common guideline for focus group research is to conduct at least two focus groups for each demographic stratum in the study ( Krueger and Casey 2015 ; Barbour 2007 ; Fern 2001 ; Greenbaum 2000 ; Morgan 1997 ). Our results support conducting two groups per stratum to provide a more comprehensive understanding of issues in particular to fully capture nuances of conceptual codes. However, we found little additional benefit in conducting more than two groups per stratum.

These results have important implications for estimating sample sizes a priori for focus group studies. Based on our findings, we recommend using both the number of strata and number of groups per stratum as key criteria to identify an adequate sample size to reach meaning saturation. For example, researchers doing a focus group study stratified by one characteristic (e.g., sex) would need to conduct two groups to include both strata but should ideally conduct two groups for each of these strata - thereby making a sample size of 4 focus groups. Researchers doing a study using two strata (e.g., sex and age) would need to conduct four groups to include all strata (e.g. younger women, older women, younger men, and older men) and ideally should conduct two groups for each of these strata, for a total sample size of eight focus groups. While this strategy may not be new to seasoned qualitative researchers, our study provides the empirical evidence that was previously lacking to support this approach and gives clear justification for why more groups per stratum are not necessarily better. For focus group studies where groups are not stratified by demographic characteristics, authors of other empirical studies have provided guidance on reaching saturation - Guest et al (2016) show that a homogenous study population where focus groups are not stratified can reach saturation in three to six focus groups ( Guest et al 2016 ), and even with a more diverse study population, saturation may be reached at five focus groups ( Coenen et al 2012 ).

We propose that an adequate sample size to reach saturation in focus group research depends on a range of parameters and is likely to differ from one study to the next. Therefore, providing universal sample size recommendations for focus groups studies is not useful. Instead we present a range of parameters based on our study findings that influence saturation in focus groups, which can be used to estimate saturation across different studies ( Table 3 ). These parameters include the study purpose, code characteristics, group composition, and desired type and degree of saturation. Each parameter needs to be considered individually, but the estimated sample size is determined by the combination of all parameters rather than by any single parameter alone. For example, one parameter may suggest a smaller sample size, but collectively they may indicate a larger sample is needed. Therefore, researchers need to assess each study by its specific characteristics and how these may influence saturation to determine an appropriate sample size. Although saturation is ultimately determined during data collection, these parameters provide guidance on identifying and justifying a sample size a priori, such as for a research proposal, but they can equally be used to justify the basis on which saturation was assessed or achieved in a completed study. Often the justifications for sample sizes or reaching saturation are absent in published qualitative research, perhaps because there is little empirical guidance on how to do this. Sample size estimates also need to remain flexible to allow the inductive process to be used during data collection; often this is achieved by identifying a range rather a fixed number when a sample size is proposed in advance (e.g., 4–6 focus groups).

Parameters influencing saturation and sample size for focus group discussions

It is also important to remain pragmatic on the degree of saturation sought (e.g., 80% or 90%). While it is near impossible to reach total saturation, since there is always the potential to discover new things in data, this is also not the objective of saturation ( Corbin and Straus 2008 ; Saunders et al 2017 ). It is not reaching a particular benchmark that is critical but reaching a point where it is determined that new discoveries do not add further insights, thus reaching a point of ‘diminishing returns’ in terms of developing a sufficiently robust understanding of the phenomenon ( Mason 2010 ). While this assessment can only be made during data review, our study provides useful guidelines on when this may occur in focus group data.

Study Limitations

To assess the effect of focus group order on saturation, we used a hypothetical randomization of focus group order, rather than repeating the process of code development using the randomized order of focus group discussions. While the benefits of this approach outweighed the risk of bias had the same researchers repeated the code development process, we could have recruited another group of researchers to conduct this task. Additionally, there is a chance that our findings are influenced by coding preferences and practices of the researchers involved. For example, researcher’s coding style (e.g., lumper vs. splitter) could affect the number and scope of codes developed, and other researchers might have had slightly different findings regarding timing of saturation had they taken a very different approach.

With this study, we contribute empirical research to identify influences on saturation in focus group research. We examined two approaches to assessing saturation and use our results to develop parameters of saturation that may be used to determine effective sample sizes for focus group studies in advance of data collection. Our results show that reaching code saturation captures the breadth of issues and requires few focus groups, while achieving meaning saturation requires more focus groups for greater depth and understanding of these issues. We also identify the strong influence of demographic strata of focus groups on saturation and sample size. If saturation continues to be hailed as the criterion for rigor in determining an adequate sample size in qualitative research, still further research is needed to examine the nature of saturation in different types of data, data collection methods, and research approaches.

The authors declare that there is no conflict of interest.

i This is sometimes called ‘information redundancy’.

Contributor Information

Monique M. Hennink, Hubert Department of Global Health, Rollins School of Public Health, Emory University, 1518 Clifton Road, Atlanta, GA 30322, USA.

Bonnie N. Kaiser, Department of Anthropology, University of California, San Diego (previous) Duke University (current)

Mary Beth Weber, Hubert Department of Global Health, Rollins School of Public Health, Emory University.

• Barbour R (2007). Doing Focus Groups. The SAGE Qualitative Research Kit . Sage Publications: London. [ Google Scholar ]
• Birks J, and Mills J (2011) Grounded Theory: A Practical Guide . Sage Publications:London. [ Google Scholar ]
• Bowen G (2008). Naturalistic inquiry and the saturation concept: A research note . Qualitative Research , 8 , 137–152. doi: 10.1177/1468794107085301. [ CrossRef ] [ Google Scholar ]
• Bryman A (2012). Social Research Methods . Fourth Edition, Oxford University Press, Oxford UK. [ Google Scholar ]
• Carlsen B, Glenton C (2011). What about N? A methodological study of sample-size reporting in focus group studies . BMC Medical Research Methodology , 11 , Article 26. doi: 10.1186/1471-2288-11-26. [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Charmaz K (2014). Conducting Grounded Theory. A Practical Guide through Qualitative Analysis . Second Edition. London; SAGE. [ Google Scholar ]
• Coenen M, Coenen T, Stamm A, Stucki G, and Cieza A (2012). Individual interviews and focus groups with patients with rheumatoid arthritis: A comparison of two qualitative methods . Quality of Life Research , 21 :359–70. doi: 10.1007/s11136-011-9943-2. [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Corbin J and Strauss A (2008). Basics of Qualitative Research: Techniques and Procedures for developing grounded theory . Sage: Thousand Oaks, CA. [ Google Scholar ]
• Corbin J and Strauss A (2015). Basics of Qualitative Research: Techniques and Procedures for developing grounded theory . Fourth Edition. Sage: Thousand Oaks, CA. [ Google Scholar ]
• Crandall JP, Knowler WC, Kahn SE, et al. (2008). The prevention of type 2 diabetes . Nat Clin PractEndocrinolMetab . 2008: 4 ( 7 ):382–393. doi: 10.1038/ncpendmet0843. [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Fern E (2001). Advancing Focus Group Research . Sage: Thousand Oaks: CA [ Google Scholar ]
• Francis J, Johnson M, Robertson C, Glidewell L, Entwistle V, Eccles M, Grimshaw J (2010). What is an adequate sample size? Operationalising data saturation for theory-based interview studies . Psychology & Health , 25 , 1229–1245. doi: 10.1080/08870440903194015. [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Glaser B, Strauss A (1967). The discovery of grounded theory: Strategies for qualitative research . Chicago: Aldine. [ Google Scholar ]
• Greenbaum T (2000). Moderating Focus Groups A Practical Guide for Group Facilitation . Sage: Thousand Oaks: CA [ Google Scholar ]
• Guest G, Bunce A, Johnson L (2006). How many interviews are enough? An experiment with data saturation and variability . Field Methods , 18 , 59–82. doi: 10.1177/1525822X05279903. [ CrossRef ] [ Google Scholar ]
• Guest G, Namey E, McKenna K, (2016). How many focus groups are enough? Building an Evidence Base for Non-Probability sample sizes . Field Methods , Vol 29 ( 1 ), 3–22. doi: 10.1177/1525822X16639015. [ CrossRef ] [ Google Scholar ]
• Gujral UP, Pradeepa R, Weber MB, Narayan KM, Mohan V. (2013). Type 2 diabetes in South Asians: similarities and differences with white Caucasian and other populations . Ann N Y Acad Sci . 2013: 1281 :51–63. doi: 10.1111/j.1749-6632.2012.06838.x. [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Hennink M, Kaiser B, and Marconi V (2016). Code saturation versus meaning saturation: How many interviews are enough? Qualitative Health Research , Vol 27 ( 4 ). doi: 10.1177/1049732316665344. [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Kerr C, Nixon A, Wild D (2010). Assessing and demonstrating data saturation in qualitative inquiry supporting patient-reported outcomes research . Expert Review of Pharmacoeconomics & Outcomes Research , 10 , 269–281. doi: 10.1586/erp.10.30. [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Krueger R and Casey M (2015) Focus Groups: A Practical Guide for Applied Research , Fifth edition. Thousand Oaks, CA: Sage Publications. [ Google Scholar ]
• Malterud K, Siersma V, Guassora A (2016). Sample size in qualitative interview studies: Guided by information power . Qualitative Health Research . Vol 26 ( 13 ), 1753–1760. doi: 10.1177/1049732315617444. [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Mason M (2010). Sample size and saturation in PhD studies using qualitative interviews . Forum for Qualitative Health Research , 11 ( 3 ), art 8. doi: 10.17169/fqs-11.3.1428. [ CrossRef ] [ Google Scholar ]
• Morgan D (1997) Focus Groups as Qualitative Research, Second edition, Qualitative Research Methods Series , vol. 16 Thousand Oaks, CA: Sage Publications. [ Google Scholar ]
• Morse J (1995). The significance of saturation [Editorial] . Qualitative Health Research , 5 , 147–149. doi: 10.1177/104973239500500201. [ CrossRef ] [ Google Scholar ]
• Morse J (2000). Determining sample size [Editorial] . Qualitative Health Research , 10 , 3–5. doi: 10.1177/104973200129118183. [ CrossRef ] [ Google Scholar ]
• Morse J (2015). Data were saturated … [Editorial] . Qualitative Health Research , 25 , 587–588. doi: 10.1177/1049732315576699. [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Nelson J (2016). Using conceptual depth criteria: addressing the challenge of reaching saturation in qualitative research . Qualitative Research . 17 ( 5 ). doi: 10.1177/1468794116679873. [ CrossRef ] [ Google Scholar ]
• Patton M (2015) Qualitative Research and Evaluation Methods , 4rd edition. Thousand Oaks, CA: Sage Publications. [ Google Scholar ]
• Saunders B, Sim J, Kingstone T, Baker S, Waterfield J, Bartlam B, Burroughs H and Jinks C (2017). Saturation in qualitative research: exploring its conceptualization and operationalization . Qual Quant , doi: 10.1007/s11135-017-0574-8. [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]