Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Additional File 1: Editorial positions on qualitative research and sample considerations (where available). (DOCX 12 kb)
GUID: EB541695-F99A-46B8-BFC7-B574CE51A628 Additional File 2: List of eligible articles included in the review (N = 214). (DOCX 38 kb) GUID: 28B341FB-4F28-4AA2-89CD-77D9199F3B2E Additional File 3: Data Extraction Form. (DOCX 15 kb) GUID: CAB83FF8-FECC-4005-9828-58BFB5D0647B Additional File 4: Citations used by articles to support their position on saturation. (DOCX 14 kb) GUID: EBF2927F-78E5-4DCA-A5F3-A9CC1A9059CFSupporting data can be accessed in the original publications. Additional File 2 lists all eligible studies that were included in the present analysis.
Choosing a suitable sample size in qualitative research is an area of conceptual debate and practical uncertainty. That sample size principles, guidelines and tools have been developed to enable researchers to set, and justify the acceptability of, their sample size is an indication that the issue constitutes an important marker of the quality of qualitative research. Nevertheless, research shows that sample size sufficiency reporting is often poor, if not absent, across a range of disciplinary fields.
A systematic analysis of single-interview-per-participant designs within three health-related journals from the disciplines of psychology, sociology and medicine, over a 15-year period, was conducted to examine whether and how sample sizes were justified and how sample size was characterised and discussed by authors. Data pertinent to sample size were extracted and analysed using qualitative and quantitative analytic techniques.
Our findings demonstrate that provision of sample size justifications in qualitative health research is limited; is not contingent on the number of interviews; and relates to the journal of publication. Defence of sample size was most frequently supported across all three journals with reference to the principle of saturation and to pragmatic considerations. Qualitative sample sizes were predominantly – and often without justification – characterised as insufficient (i.e., ‘small’) and discussed in the context of study limitations. Sample size insufficiency was seen to threaten the validity and generalizability of studies’ results, with the latter being frequently conceived in nomothetic terms.
We recommend, firstly, that qualitative health researchers be more transparent about evaluations of their sample size sufficiency, situating these within broader and more encompassing assessments of data adequacy. Secondly, we invite researchers critically to consider how saturation parameters found in prior methodological studies and sample size community norms might best inform, and apply to, their own project and encourage that data adequacy is best appraised with reference to features that are intrinsic to the study at hand. Finally, those reviewing papers have a vital role in supporting and encouraging transparent study-specific reporting.
The online version of this article (10.1186/s12874-018-0594-7) contains supplementary material, which is available to authorized users.
Keywords: Sample size, Sample size justification, Sample size characterisation, Data adequacy, Qualitative health research, Qualitative interviews, Review, Systematic analysis
Sample adequacy in qualitative inquiry pertains to the appropriateness of the sample composition and size. It is an important consideration in evaluations of the quality and trustworthiness of much qualitative research [1] and is implicated – particularly for research that is situated within a post-positivist tradition and retains a degree of commitment to realist ontological premises – in appraisals of validity and generalizability [2–5].
Samples in qualitative research tend to be small in order to support the depth of case-oriented analysis that is fundamental to this mode of inquiry [5]. Additionally, qualitative samples are purposive, that is, selected by virtue of their capacity to provide richly-textured information, relevant to the phenomenon under investigation. As a result, purposive sampling [6, 7] – as opposed to probability sampling employed in quantitative research – selects ‘information-rich’ cases [8]. Indeed, recent research demonstrates the greater efficiency of purposive sampling compared to random sampling in qualitative studies [9], supporting related assertions long put forward by qualitative methodologists.
Sample size in qualitative research has been the subject of enduring discussions [4, 10, 11]. Whilst the quantitative research community has established relatively straightforward statistics-based rules to set sample sizes precisely, the intricacies of qualitative sample size determination and assessment arise from the methodological, theoretical, epistemological, and ideological pluralism that characterises qualitative inquiry (for a discussion focused on the discipline of psychology see [12]). This mitigates against clear-cut guidelines, invariably applied. Despite these challenges, various conceptual developments have sought to address this issue, with guidance and principles [4, 10, 11, 13–20], and more recently, an evidence-based approach to sample size determination seeks to ground the discussion empirically [21–35].
Focusing on single-interview-per-participant qualitative designs, the present study aims to further contribute to the dialogue of sample size in qualitative research by offering empirical evidence around justification practices associated with sample size. We next review the existing conceptual and empirical literature on sample size determination.
Qualitative research experts argue that there is no straightforward answer to the question of ‘how many’ and that sample size is contingent on a number of factors relating to epistemological, methodological and practical issues [36]. Sandelowski [4] recommends that qualitative sample sizes are large enough to allow the unfolding of a ‘new and richly textured understanding’ of the phenomenon under study, but small enough so that the ‘deep, case-oriented analysis’ (p. 183) of qualitative data is not precluded. Morse [11] posits that the more useable data are collected from each person, the fewer participants are needed. She invites researchers to take into account parameters, such as the scope of study, the nature of topic (i.e. complexity, accessibility), the quality of data, and the study design. Indeed, the level of structure of questions in qualitative interviewing has been found to influence the richness of data generated [37], and so, requires attention; empirical research shows that open questions, which are asked later on in the interview, tend to produce richer data [37].
Beyond such guidance, specific numerical recommendations have also been proffered, often based on experts’ experience of qualitative research. For example, Green and Thorogood [38] maintain that the experience of most qualitative researchers conducting an interview-based study with a fairly specific research question is that little new information is generated after interviewing 20 people or so belonging to one analytically relevant participant ‘category’ (pp. 102–104). Ritchie et al. [39] suggest that studies employing individual interviews conduct no more than 50 interviews so that researchers are able to manage the complexity of the analytic task. Similarly, Britten [40] notes that large interview studies will often comprise of 50 to 60 people. Experts have also offered numerical guidelines tailored to different theoretical and methodological traditions and specific research approaches, e.g. grounded theory, phenomenology [11, 41]. More recently, a quantitative tool was proposed [42] to support a priori sample size determination based on estimates of the prevalence of themes in the population. Nevertheless, this more formulaic approach raised criticisms relating to assumptions about the conceptual [43] and ontological status of ‘themes’ [44] and the linearity ascribed to the processes of sampling, data collection and data analysis [45].
In terms of principles, Lincoln and Guba [17] proposed that sample size determination be guided by the criterion of informational redundancy, that is, sampling can be terminated when no new information is elicited by sampling more units. Following the logic of informational comprehensiveness Malterud et al. [18] introduced the concept of information power as a pragmatic guiding principle, suggesting that the more information power the sample provides, the smaller the sample size needs to be, and vice versa.
Undoubtedly, the most widely used principle for determining sample size and evaluating its sufficiency is that of saturation. The notion of saturation originates in grounded theory [15] – a qualitative methodological approach explicitly concerned with empirically-derived theory development – and is inextricably linked to theoretical sampling. Theoretical sampling describes an iterative process of data collection, data analysis and theory development whereby data collection is governed by emerging theory rather than predefined characteristics of the population. Grounded theory saturation (often called theoretical saturation) concerns the theoretical categories – as opposed to data – that are being developed and becomes evident when ‘gathering fresh data no longer sparks new theoretical insights, nor reveals new properties of your core theoretical categories’ [46 p. 113]. Saturation in grounded theory, therefore, does not equate to the more common focus on data repetition and moves beyond a singular focus on sample size as the justification of sampling adequacy [46, 47]. Sample size in grounded theory cannot be determined a priori as it is contingent on the evolving theoretical categories.
Saturation – often under the terms of ‘data’ or ‘thematic’ saturation – has diffused into several qualitative communities beyond its origins in grounded theory. Alongside the expansion of its meaning, being variously equated with ‘no new data’, ‘no new themes’, and ‘no new codes’, saturation has emerged as the ‘gold standard’ in qualitative inquiry [2, 26]. Nevertheless, and as Morse [48] asserts, whilst saturation is the most frequently invoked ‘guarantee of qualitative rigor’, ‘it is the one we know least about’ (p. 587). Certainly researchers caution that saturation is less applicable to, or appropriate for, particular types of qualitative research (e.g. conversation analysis, [49]; phenomenological research, [50]) whilst others reject the concept altogether [19, 51].
Methodological studies in this area aim to provide guidance about saturation and develop a practical application of processes that ‘operationalise’ and evidence saturation. Guest, Bunce, and Johnson [26] analysed 60 interviews and found that saturation of themes was reached by the twelfth interview. They noted that their sample was relatively homogeneous, their research aims focused, so studies of more heterogeneous samples and with a broader scope would be likely to need a larger size to achieve saturation. Extending the enquiry to multi-site, cross-cultural research, Hagaman and Wutich [28] showed that sample sizes of 20 to 40 interviews were required to achieve data saturation of meta-themes that cut across research sites. In a theory-driven content analysis, Francis et al. [25] reached data saturation at the 17th interview for all their pre-determined theoretical constructs. The authors further proposed two main principles upon which specification of saturation be based: (a) researchers should a priori specify an initial analysis sample (e.g. 10 interviews) which will be used for the first round of analysis and (b) a stopping criterion, that is, a number of interviews (e.g. 3) that needs to be further conducted, the analysis of which will not yield any new themes or ideas. For greater transparency, Francis et al. [25] recommend that researchers present cumulative frequency graphs supporting their judgment that saturation was achieved. A comparative method for themes saturation (CoMeTS) has also been suggested [23] whereby the findings of each new interview are compared with those that have already emerged and if it does not yield any new theme, the ‘saturated terrain’ is assumed to have been established. Because the order in which interviews are analysed can influence saturation thresholds depending on the richness of the data, Constantinou et al. [23] recommend reordering and re-analysing interviews to confirm saturation. Hennink, Kaiser and Marconi’s [29] methodological study sheds further light on the problem of specifying and demonstrating saturation. Their analysis of interview data showed that code saturation (i.e. the point at which no additional issues are identified) was achieved at 9 interviews, but meaning saturation (i.e. the point at which no further dimensions, nuances, or insights of issues are identified) required 16–24 interviews. Although breadth can be achieved relatively soon, especially for high-prevalence and concrete codes, depth requires additional data, especially for codes of a more conceptual nature.
Critiquing the concept of saturation, Nelson [19] proposes five conceptual depth criteria in grounded theory projects to assess the robustness of the developing theory: (a) theoretical concepts should be supported by a wide range of evidence drawn from the data; (b) be demonstrably part of a network of inter-connected concepts; (c) demonstrate subtlety; (d) resonate with existing literature; and (e) can be successfully submitted to tests of external validity.
Other work has sought to examine practices of sample size reporting and sufficiency assessment across a range of disciplinary fields and research domains, from nutrition [34] and health education [32], to education and the health sciences [22, 27], information systems [30], organisation and workplace studies [33], human computer interaction [21], and accounting studies [24]. Others investigated PhD qualitative studies [31] and grounded theory studies [35]. Incomplete and imprecise sample size reporting is commonly pinpointed by these investigations whilst assessment and justifications of sample size sufficiency are even more sporadic.
Sobal [34] examined the sample size of qualitative studies published in the Journal of Nutrition Education over a period of 30 years. Studies that employed individual interviews (n = 30) had an average sample size of 45 individuals and none of these explicitly reported whether their sample size sought and/or attained saturation. A minority of articles discussed how sample-related limitations (with the latter most often concerning the type of sample, rather than the size) limited generalizability. A further systematic analysis [32] of health education research over 20 years demonstrated that interview-based studies averaged 104 participants (range 2 to 720 interviewees). However, 40% did not report the number of participants. An examination of 83 qualitative interview studies in leading information systems journals [30] indicated little defence of sample sizes on the basis of recommendations by qualitative methodologists, prior relevant work, or the criterion of saturation. Rather, sample size seemed to correlate with factors such as the journal of publication or the region of study (US vs Europe vs Asia). These results led the authors to call for more rigor in determining and reporting sample size in qualitative information systems research and to recommend optimal sample size ranges for grounded theory (i.e. 20–30 interviews) and single case (i.e. 15–30 interviews) projects.
Similarly, fewer than 10% of articles in organisation and workplace studies provided a sample size justification relating to existing recommendations by methodologists, prior relevant work, or saturation [33], whilst only 17% of focus groups studies in health-related journals provided an explanation of sample size (i.e. number of focus groups), with saturation being the most frequently invoked argument, followed by published sample size recommendations and practical reasons [22]. The notion of saturation was also invoked by 11 out of the 51 most highly cited studies that Guetterman [27] reviewed in the fields of education and health sciences, of which six were grounded theory studies, four phenomenological and one a narrative inquiry. Finally, analysing 641 interview-based articles in accounting, Dai et al. [24] called for more rigor since a significant minority of studies did not report precise sample size.
Despite increasing attention to rigor in qualitative research (e.g. [52]) and more extensive methodological and analytical disclosures that seek to validate qualitative work [24], sample size reporting and sufficiency assessment remain inconsistent and partial, if not absent, across a range of research domains.
The present study sought to enrich existing systematic analyses of the customs and practices of sample size reporting and justification by focusing on qualitative research relating to health. Additionally, this study attempted to expand previous empirical investigations by examining how qualitative sample sizes are characterised and discussed in academic narratives. Qualitative health research is an inter-disciplinary field that due to its affiliation with medical sciences, often faces views and positions reflective of a quantitative ethos. Thus qualitative health research constitutes an emblematic case that may help to unfold underlying philosophical and methodological differences across the scientific community that are crystallised in considerations of sample size. The present research, therefore, incorporates a comparative element on the basis of three different disciplines engaging with qualitative health research: medicine, psychology, and sociology. We chose to focus our analysis on single-per-participant-interview designs as this not only presents a popular and widespread methodological choice in qualitative health research, but also as the method where consideration of sample size – defined as the number of interviewees – is particularly salient.
A structured search for articles reporting cross-sectional, interview-based qualitative studies was carried out and eligible reports were systematically reviewed and analysed employing both quantitative and qualitative analytic techniques.
We selected journals which (a) follow a peer review process, (b) are considered high quality and influential in their field as reflected in journal metrics, and (c) are receptive to, and publish, qualitative research (Additional File 1 presents the journals’ editorial positions in relation to qualitative research and sample considerations where available). Three health-related journals were chosen, each representing a different disciplinary field; the British Medical Journal (BMJ) representing medicine, the British Journal of Health Psychology (BJHP) representing psychology, and the Sociology of Health & Illness (SHI) representing sociology.
Employing the search function of each individual journal, we used the terms ‘interview*’ AND ‘qualitative’ and limited the results to articles published between 1 January 2003 and 22 September 2017 (i.e. a 15-year review period).
To be eligible for inclusion in the review, the article had to report a cross-sectional study design. Longitudinal studies were thus excluded whilst studies conducted within a broader research programme (e.g. interview studies nested in a trial, as part of a broader ethnography, as part of a longitudinal research) were included if they reported only single-time qualitative interviews. The method of data collection had to be individual, synchronous qualitative interviews (i.e. group interviews, structured interviews and e-mail interviews over a period of time were excluded), and the data had to be analysed qualitatively (i.e. studies that quantified their qualitative data were excluded). Mixed method studies and articles reporting more than one qualitative method of data collection (e.g. individual interviews and focus groups) were excluded. Figure 1 , a PRISMA flow diagram [53], shows the number of: articles obtained from the searches and screened; papers assessed for eligibility; and articles included in the review (Additional File 2 provides the full list of articles included in the review and their unique identifying code – e.g. BMJ01, BJHP02, SHI03). One review author (KV) assessed the eligibility of all papers identified from the searches. When in doubt, discussions about retaining or excluding articles were held between KV and JB in regular meetings, and decisions were jointly made.