Abstract
Qualitative researchers face an enduring question: How many interviews do I need? While a variety of guidelines exist, there is limited consensus over which specific factors should determine the number of interviews required. We examined the determination of interview sample sizes in 562 qualitative studies across six high-impact management and organizational journals over a decade. Our findings reveal considerable variance in interview numbers, yet limited information is often provided on the criteria used to determine them. To promote clearer alignment between sample sizes and methodology, we examined studies with detailed descriptions of their interview sampling. We identified specific “sampling moves” used to determine the number of interviews, categorized into three types—opening, focusing, and closing sampling moves—that researchers use to establish confidence in the sample and support theoretical insights. By implication, our study refutes the notion of a “magic” interview number. Instead, sampling moves are heuristic tools that qualitative researchers can thoughtfully adapt to their analytical aims when determining appropriate sample sizes.
Introduction
From planning research projects within doctoral training programs to journal review processes, the question of “How many interviews do I need?” confronts researchers, reviewers, and editors. The answer is usually ambiguous and situationally varying. From our own experiences as qualitative researchers, we have observed divergent guidance regarding the number of interviews needed in a study. In a decision letter from a top journal, an editor emphasized that manuscripts rarely had fewer than 50 interviews. We have also encountered PhD students being advised to conduct more than 30 interviews for their research to be publishable. Similarly, textbooks and articles on qualitative research are full of recommendations that allude to an ideal number of interviews and tend to cite criteria drawn from quantitative methods (Emmel, 2013; Muellmann et al., 2021; Wutich et al., 2024). Recommendations for a minimum or maximum number of interviews in the research methods literature range from scholars who suggest that 12 interviews in a homogeneous group of individuals will be sufficient (Guest et al., 2006) up to a maximum of 25 interviews (Kvale & Brinkmann, 2009), with a range of between 60 and 120 proposed by others (Adler & Adler, 2012).
Conversely, a common refrain is that no ideal number of interviews exists. The answer to how many interviews are needed is, usually, “it depends” (Baker & Edwards, 2012, p. 25). While such flexibility may be used to communicate that an interview sample size cannot be decided in advance, extant guidelines indicate a variety of different criteria that can potentially inform interview sample sizes, such as the chosen methodology or research design (Creswell & Poth, 2016; Marshall et al., 2013; Patton, 2014; Ryan & Bernard, 2003). For instance, a criterion derived from grounded theory is that the required number of interviews is established at the point of “theoretical saturation,” where the recruitment of additional study participants offers no new insights, such that no predetermined or ideal number can be said to exist (Glaser & Strauss, 1967). Across many methodologies, however, precisely which factors drive optimum interview sample sizes remain a conundrum.
Given the conflicting and often ambiguous responses to the question of what should determine the sample size in an interview study, it is unsurprising that researchers welcome the clarity of an “ideal” predeterminable interview number. Yet, we view the alluring lore of ideal numbers as problematic. Ideal interview numbers can be viewed as a form of isomorphic template, revealing a homogenization of qualitative research studies (Köhler et al., 2022). As Köhler et al. (2019, p. 3) argue, we are witnessing an inclination toward “quantifying qualitative data over delving deeply into the rich qualitative data.” Such a creep toward quantitative standards is problematic, not only because the desired richness of qualitative data may not be realized (Emmel, 2013; Glaser & Strauss, 1967; Morse, 2007), but also because stressing quantity over quality of empirical material may provide a “misleading impression of robustness” (Alvesson, 2003, p. 28). This, in turn, may indicate that researchers are striving for causal probability rather than analytic plausibility (Flyvbjerg, 2006). Further, authors problematize pursuing a predetermined number as a “precondition for quality” (Small & Calarco, 2022, p. 19) or objectivity (Ranganathan & Benson, 2020). Some commentators even question researchers’ abilities to handle large interview volumes with sufficient analytical depth and sensitivity (Kvale & Brinkmann, 2009; Pettigrew, 1990).
Despite the wide range of recommendations regarding what sample sizes should depend on—such as research design and epistemological assumptions—we lack systematic knowledge of how these recommendations compare with the actual number of interviews reported and what researchers rely on to determine their sample size. There has been a dearth of investigation into how reported interview samples relate to foundational methodological approaches in management and organizational journal publications; furthermore, how sampling practices have potentially evolved beyond these approaches (Pratt et al., 2022). Hence, there is a need to qualify what interview samples depend upon. One of the few studies considering this topic is Saunders and Townsend’s (2016) review of organization and workplace studies, which provides insights into interview numbers, reporting practices, and justification strategies. However, Saunders and Townsend (2016) did not explicitly examine the relationship between sample sizes and methodological approaches, specifically the actual ways through which interview samples are accounted for. Hence, establishing insights between reported interview numbers and methodological guidelines remains an understudied, yet ongoing concern for researchers, reviewers, and editors. This study is, therefore, guided by two research questions: “How are methodological choices and interview sample sizes related?” and “How do authors account for their number of interviews?”
To engage with these research questions, we conducted a literature review (Kunisch et al., 2023) of 562 qualitative, interview-based studies in six high-impact management and organizational journals across North America and Europe over a 10 year period (2010–2019), including the Academy of Management Journal (AMJ), Administrative Science Quarterly (ASQ), Organization Science (Org. Sci.), Strategic Management Journal (SMJ), Journal of Management Studies (JMS), and Organization Studies (OS). Our analysis is two-fold. First, we analyzed how interview numbers compare across methodological approaches. Our analysis shows a considerable variation in interview samples across studies and designs, with individual studies ranging from 2 to 682 interviews. Most studies do not provide explicit justifications for their interview numbers, and we find no or weak relationships between the number of interviews and various methodological approaches. Second, we explored a subset of exemplar articles that were particularly detailed in their descriptions of interview sampling “moves” (Hansen et al., 2025; Pratt et al., 2022). We identified a series of opening, focusing, and closing moves that serve as heuristic tools for authors to account for their choice of interview sample size for the phenomenon in focus.
Drawing on these findings, our study contributes to the current understanding of what could determine the selection of interview samples in qualitative studies in the field of organizational and management research. We problematize the notion of a “magic” interview number for qualitative research and show that reported interview sample sizes are often unmoored from their methodological foundations in researchers’ sampling accounts. Moreover, by tracing the sampling moves used by exemplary studies, we offer alternative methodological guidelines to determine appropriate interview samples in future studies. While these studies offer important insights into interview sampling, we recognize that published articles capture only a partial view of the whole sampling process. Hence, we caution against treating this as a checklist or “template” to follow (Köhler et al., 2022); rather, we offer a framework that can be built upon, with guiding questions that researchers, reviewers, and editors can use to reflect on the quality and rigor of interview-based studies.
Background: Interview Samples in Qualitative Research
The interview is the most popular method employed across the social sciences (Hughes, 1971; Kvale & Brinkmann, 2009; Whittle & Reissner, 2025). It has many incarnations, including the semistructured interview (Kvale & Brinkmann, 2009); the active interview (Holstein & Gubrium, 1995); the narrative interview (Jovchelovitch & Bauer, 2000); and the unstructured interview (Kvale & Brinkmann, 2009). These interviews are conducted via various media, including in-person, automated, and online formats (Salmons, 2017). The interview is still widely deployed in research and is considered a powerful method for exploring the organizational worlds of various subjects (Hallin et al., 2024) and examining the social construction of meaning (Whittle & Reissner, 2025). Nonetheless, how to determine and account for interview sample sizes—that is, who and how many to interview—remains a core question (Emmel, 2013; Patton, 2014; Romney et al., 1986; Wutich et al., 2024). Qualitative researchers face an array of recommendations for interview numbers—some suggest predetermined ideals, typically expressed as ranges or thresholds to be met, while others advocate for numbers that are situationally dependent.
Predetermined Ideal Numbers: Estimating a Range
The literature on research methods offers a range of recommendations on the minimum and maximum numbers of interviews. Some scholars recommend that graduate students “sample between 12 and 60” (Adler and Adler, 2012, p. 10), while others suggest that a suitable range is “20 for an M.A. thesis and 50 for a Ph.D. dissertation” (Ragin in Adler & Adler, 2012, p. 34). Similar norms exist for more experienced researchers. Francis et al. (2010) suggest a minimum sample size of 13 interviews, while Marshall et al. (2013) suggest that “grounded theory qualitative studies should generally include between 20 and 30 interviews” and “single case studies should generally contain 15 to 30 interviews” (p. 20). Others have alluded to a figure of 35 as an apt number for many ethnographies and grounded theory studies (Bernard, 2000). These ranges or thresholds seem to echo the idea that a sample is often “either too small or too large,” making it difficult to generalize or to perform an in-depth analysis (Kvale and Brinkmann, 2009, p. 113).
Along those lines, Guest et al. (2006, p. 73) question how to determine nonprobabilistic sample sizes and analyze 60 interviews, from which they infer 36 codes that appeared “with a high frequency in the transcripts.” Of these, 34 (94%) had already been identified in the first six interviews, while 35 (97%) were identified only after 12 interviews. Accordingly, they explain that to represent and “to understand common perceptions and experiences among a group of relatively homogeneous individuals, twelve interviews should suffice,” and already after six interviews, most of the codes (73%) are uncovered (Guest et al., 2006, p. 79). In a similar vein, it is suggested that 10 interviews can provide “stable” insights into a community's readiness to tackle health problems, albeit leaving some issues unidentified (Muellmann et al., 2021). While this finding is contested (Emmel, 2013; Sim et al., 2018), Guest and co-authors urge researchers to be cautious, arguing that this benchmark does not apply to unstructured, exploratory interviews, where new questions continually arise, whereby saturation becomes harder to establish and more challenging to achieve. Other scholars have taken different perspectives on the question of an ideal number. Some have pointed out the need to convince an audience of the legitimacy of a sample (Bryman, 2012; Patton, 2014), instead pointing to the fact that most studies may need a certain threshold amount to generate “respect” (Charmaz, 2012, p. 21). Bryman (2012) cites Warren's (2002) recommendation that “for a qualitative interview study to be published, the minimum number of interviews required seems to be between twenty and thirty” (p. 425). Others offer the advice that “fewer than 60 interviews cannot support convincing conclusions and more than 150 produce too much material to analyze effectively and expeditiously” (Gerson & Horowitz, 2002, p. 223).
Similarly, scholars have pointed to differences in epistemological and ontological orientations that can add nuance, while still providing a specific number. For instance, suggestions have been made for how many participants there should be in a phenomenological study, ranging from 2 to 10 (Boyd, 2001), 6 to 12 (Thomas & Pollio, 2002), and 5 to 25 (Beitin, 2012). Charmaz (2006, p. 114) has pointed out how grounded theory studies based on a constructivist ontology may require as few as 25 interviews for a small project, although to make arguments that challenge “established research,” a researcher typically needs more. While these perspectives offer diverse suggestions, a common thread is the attempt to outline a predeterminable range or threshold, in contrast to perspectives that stress situational variation that is not predeterminable.
“It Depends”: Situational Variations in Interview Numbers
In contrast to scholars seeking to estimate an ideal or a predefined range of interview numbers, other scholars stress that no such number can truly be predetermined. Instead, what determines the number of interviews required “depends” (Baker & Edwards, 2012, p. 25) on the needs of the study and the process of the research (Collins et al., 2007; Miles et al., 2013; Patton, 2014; Ryan & Bernard, 2003). As Patton (2014, pp. 242–243) notes, “There are no rules for sample size in qualitative inquiry. Sample size depends on what you want to know, the purpose of the inquiry, what's at stake, what will be useful, what will have credibility, and what can be done with available time and resources.”
One dimension that literature highlights as important when determining sample sizes is the focus of a study. For instance, if scholars are studying an evolving phenomenon then the duration of a study is important and the interview sample is likely to be extended (Creswell & Poth, 2016; Patton, 2014; Small & Calarco, 2022). Similarly, when considering the comparative dimensions of a study, for instance in studies of multiple sites, comparing cases or multiple levels, such as micro and macro (Ragin & Becker, 1992; Flick, 2012, 2008; Ragin & Becker, 1992), larger numbers of interviews are required to gathering insights across different actors (Baker & Edwards, 2012; Becker, 2012; Kuzel, 1992; Wutich et al., 2024). The complexity of multilevel phenomena means that researchers often need to gather interviews from more individuals, necessitating a larger sample size (Bryman, 2012). As Bryman (2012, p. 425) notes, “if several comparisons are likely to be wanted—between males and females, different age groups, different types of research participants in terms of locally relevant factors—a larger sample is likely to be necessary.” Relatedly, the homogeneity of a group may also matter. For example, a homogenous group may require fewer interviews than a heterogeneous one (Eisenhardt & Graebner, 2007).
The mix of data sources also shapes the sample. The interview may be used mainly to verify or “member check” observations (Charmaz & Belgrave, 2012), or it may serve as the primary method, which would require conducting more interviews (Creswell & Poth, 2016). Hence, the use of data sources other than interviews also shapes how many interviews are needed. In cases of extreme observational immersion, even a single key informant has yielded substantial insight (Whyte, 1943).
Another dimension that literature considers when determining samples is research designs. Grounded theory and case study methodologies, which draw on sampling principles of an inductive nature (Eisenhardt, 1989), often require data collection to cease at the point of theoretical saturation. Saturation is commonly defined as the point at which researchers determine that no new properties, dimensions, or relationships emerge from additional interviews (Strauss & Corbin, 1998). At this point, researchers stop collecting data, including interview data (Glaser & Strauss, 1967; Patton, 2014). While seeking saturation is not to be conflated with the accurate representation of a given phenomenon, but rather with the ambition to advance the ability to generate theory (Glaser & Strauss, 2004), it may take many months of research before all pertinent “categories are saturated” (Strauss, 1987, p. 24). This may require larger interview sample sizes than narrative analysis and discursive inquiries (Charmaz, 2012). Indeed, it has been argued that any “failure to reach data saturation has an impact on the quality of the research conducted and hampers content validity” (Fusch & Ness, 2015, p. 1408). Emphasizing sample sizes, as determined by saturation, thus shifts attention to the point at which no new information is expected to emerge.
Scholars have further argued that epistemological and ontological positioning should be the critical starting point when considering research designs and interview samples (Alvesson, 2003; Hallin et al., 2024; Langley & Meziani, 2020; Riessman, 2008). For example, Hallin et al. (2024) highlighted how ontological and epistemological assumptions, such as realist versus constructivist ontologies and positivist versus relativist epistemologies, shape the selection and use of interviews. Even within the same research design, such as grounded theory, the ontological and epistemological assumptions of the researcher can shape how interviews are used and thus sampled. As Cunliffe (2011, p. 659) and Locke (2001, pp. 12–13) note, although grounded theory has roots in objectivism, it is also used by subjectivist researchers, including those following social constructivist (Charmaz, 2006), discursive (Gagnon, 2008), and/or institutional (Grandy & Wicks, 2008) traditions, highlighting how sampling decisions and saturation thresholds may diverge across epistemic positions. A key question here is, given researchers’ diverse ontological perspectives, what does the data represent to them? (DeCelles et al., 2021).
Discursive and narrative inquiry often raise similar arguments, emphasizing how smaller sample sizes can be appropriate when the analytic focus is on meaning coherence and discursive patterning (Fairhurst & Uhl-Bien, 2012). Some researchers argue that the analytical intensity of such approaches can limit feasible interview counts (Charmaz, 2012, p. 21). Within discursive and narrative perspectives, the interview is not merely a site of data extraction but rather a situated interaction through which meaning and the phenomenon under study are actively co-constructed (Fairhurst & Putnam, 2004; Grint, 2005; Hallin et al., 2024). What the interviewee is doing in the interview may not be merely describing a problem—they are setting it, framing it, and justifying action through how they talk (Schön, 1983). Accordingly, in constructivist ontologies, smaller samples may be warranted when the aim is to build a compelling interpretation by showing how parts of the material fit together into a coherent account. Claims can be further buttressed through structural corroboration—assembling converging evidence across interview talk, and where relevant, documents and observations (Eisner, 2017; Jacobs, 1990). This understanding suggests that even a limited number of interviews can yield powerful insights when they illuminate broader structures or mechanisms (c.f. Popper, 1957). However, this stance contrasts with many inductive and grounded theory approaches that emphasize theoretical saturation (Charmaz, 2012).
Despite tensions arising across the methodological literature above, there is little investigation of how the differing recommendations we have highlighted compare with the actual number of interviews conducted across published management and organizational research. Given the diversity of methodological recommendations, it is important to examine how researchers navigate these options, and how sampling practices have potentially evolved beyond or find themselves in tension with these guidelines (c.f. Pratt et al., 2022). Based on this, we ask two research questions. First, “How are methodological choices and interview sample sizes related?” and second, “How do authors account for their number of interviews?”
Methodology
To examine our research questions, we conducted a literature review (Kunisch et al., 2023) to classify and explain interview sample sizes across qualitative research studies published in AMJ, ASQ, Org. Sci., SMJ, JMS, and OS over a period of 10 years (2010‒2019). Inspired by prior studies (Gibbert & Ruigrok, 2010), we consulted journal rankings (i.e., the ABS and ABDC lists) and chose to focus on the highest-ranked management and organization journals with a reputation for publishing highly cited qualitative work. While high-quality scholarship and high journal impact factors should not be conflated (Köhler et al., 2020), we made this decision given the influence and impact of these outlets in the management and organizational research field.
Data Collection
We used the Business Source Premier website to identify relevant studies and searched for articles including the search terms “interviews” and “qualitative research.” We also added “ethnography,” “grounded theory,” “narrative,” and “process studies” to ensure that we included relevant articles. We searched “all text”—including title, abstract, keywords, and the article text—to ensure that all published articles were considered. This broad search yielded 1,780 candidate studies across the six journals selected.
Many of the initial 1,780 articles proved irrelevant for our purposes, and we excluded conceptual papers; those qualitative studies not using interviews; studies mentioning the keywords “qualitative” or “interviews” outside the context of a specified method; and mixed-methods studies combining quantitative and qualitative approaches. While mixed-method studies were of interest, we decided to exclude them, as these studies mostly relied on the quantitative part and used interviews only as a supplement (e.g., Tasselli, 2015). Ultimately, we included 562 articles in the dataset. See Table A1 in the supplemental material for a complete list of articles.
Data Analysis
Having identified 562 qualitative interview studies, we organized our data in an Excel spreadsheet to ensure a transparent coding process. All the authors were involved in coding the articles with the help of a research assistant. Consequently, at least two people read and coded each article. When disagreements occurred regarding the coding of an article, the authors collectively reread the article to reach a consensus. The analysis took place through three stages.
First Stage: Creating an Overview of the Interview Studies
We began reading and analyzing the entire sample set, providing immediate notes on how the articles used and accounted for interviews. To generate an overview of the interview studies, we relied on descriptive statistical analysis. We established the means, medians, and standard deviations of interview sample sizes across each journal, the distributions within and across the six journals, and examined the development in interview numbers over time. 1 This provided important insights into whether there was an “ideal” range or threshold of interviews in leading journals or whether there was simply a high variability in sample sizes.
Second Stage: Comparing Interview Numbers and Methodological Approaches
To further examine the relationship arising between interview sample sizes and methodological approaches, we coded the 562 articles on a range of topics derived from the methodological literature. Drawing inspiration from the literature, while stressing that the complexity of multilevel phenomena requires a larger interview sample size (Bryman, 2012), we coded for a range of parameters. The first aspect was the level of analysis, indicating whether the interviews were conducted at multiple levels or only at a single level (e.g., employees or top managers). Secondly, we coded for site, to search for any variance arising when studying single or multiple organizations or units (Marshall et al., 2013). For example, multisited studies could involve multiple organizations or units, potentially resulting in more interviews than single-sited studies involving a solitary organization or unit (Patton, 2014). Third, we coded for time to understand whether studies and data collection of longer duration tended to increase the number of interviews (Bryman, 2012). Fourth, we coded for a combination with other methods, as the interview method may serve different purposes and is often used in combination with other qualitative data sources, such as observations or documents (e.g., Elmholdt et al., 2025; Gill & Gill, 2024). If interview data sources are the primary source, it could potentially explain a higher number of interviews (Creswell and Poth, 2016)
As our fifth parameter, we coded for research design. Based on a pilot reading of 50 articles, we divided our findings into seven research design categories, namely case studies, grounded theory, process studies, ethnography, discourse, narrative, and other. 2 Case studies were the largest category and included studies that explicitly use the term case study, which would usually refer to Eisenhardt (1989) and Yin (1998) for methodological inspiration. Grounded theory was the second largest, and these studies often include references to Strauss and colleagues (e.g., Strauss & Corbin, 1998) and Gioia et al. (2012). Ethnography was the third largest category, with Van Maanen (1979) often cited. We used similar criteria for process studies, often referring to Langley (1999), narrative, and discourse studies. The latter two would use less consistent and more varying references in their methodology, such as Taylor and Van Every (2000) in discursive studies (e.g., Cornelissen, 2012) or Barry and Elmes (1997) within narrative studies (e.g., Sonenschein, 2010). While many studies combine research designs, we searched in our coding for the dominant research design articulated in each article. For example, if a case study used grounded theory, we would code for the dominant reference and categorization. Reflecting on prior work, we expected discourse, narrative, and ethnographic studies to include fewer interviews than grounded theory, case studies, and process studies (Charmaz, 2012; Marshall et al., 2013).
A sixth parameter in our coding addressed ontology/epistemology, 3 which is often emphasized as key across designs in methodological textbooks (e.g., Charmaz, 2006; Van Maanen, 1979). We coded the studies for mentions of ontology/epistemology given repeated and sustained recommendations for authors to state their assumptions, across articles and editorials (e.g., Cunliffe, 2011; Hallin et al., 2024; Jarzabkowski et al., 2021; Morgan & Smircich, 1980). We drew inspiration from established work within qualitative studies (Guba & Lincoln, 1994; Whittle & Reissner, 2025), which highlight the critical importance of “the nature of reality” (ontology) and “what can be known” (epistemology) in determining samples and the data drawn from interviews. Yet, in our study, epistemological and ontological commitments were rarely explicit and proved challenging to infer. For instance, while some versions of grounded theory may be said to reflect a realist ontology and epistemology, it has also been employed from alternative positions (Charmaz, 2005; Cunliffe, 2011). Similarly, while the social construction of knowledge is widely recognized in qualitative studies, few studies explicitly stated whether they aligned with constructivist or constructionist ontologies. Yet, there are a variety of such ontologies (Cunliffe, 2008; Czarniawska, 2003), which place distinct emphases on the roles of language (e.g., salient in sensemaking and discourse accounts) and everyday interaction (e.g., salient in practice-theoretical and symbolic interactionist accounts) in constituting social reality and research phenomena. Accordingly, mapping clear epistemological and ontological stances proved challenging. Therefore, we decided to use only explicit statements of ontology and epistemology as proxies. We coded the articles based on what they explicated in terms of philosophical commitment (e.g., using phrases like practice ontology, process ontology, interpretivism, or social constructivism), yet, as most studies did not state any specific commitment, we categorized them as unspecified.
For our seventh parameter, given our interest in understanding how interview sample sizes are determined in qualitative studies, we also coded for whether and how the authors justified their number of interviews. This coding relates to the methodological criteria of saturation, although we considered other arguments, such as meeting a stated research purpose (Saunders & Townsend, 2016). We coded all those that did not explicitly account for this as unspecified.
Third Stage: Analyzing Interview Sampling Moves
During the coding process, we read through the methods sections of each paper and, while observing that sampling explanations were often silenced (Hansen et al., 2025), we started to notice those papers that articulated interview sampling more clearly. We wanted to maximize the richness of our sample and inspired by other similar methodological studies (Lé & Schmid, 2022; Locke et al., 2022) we curated a subsample of 60 articles that articulated more detailed descriptions of sampling methodologies. Hence, we focused on those rarer, more detailed accounts of how authors determined interview sample sizes, with the caveat that this was still often subtle. We began to identify the “moves” (Hansen et al., 2025; Pratt et al., 2022) through which interview numbers were accounted for. As we read the articles, we noted how the interviews were selected; how trustworthy the final sample was; and the authors’ justification for concluding the sampling.
We started to notice a certain sequence of sampling moves, which could be abstracted into three overall categories, specifically: (1) Opening moves, in which researchers emphasized their openness to discovery and the need to use interviews to explore the research phenomena of interest; (2) focusing moves, which delineate how the research process transitions from exploratory breadth to analytical depth, narrowing in on aspects of a phenomenon through targeted data selection to deepen theoretical insights; and (3) closing moves, which describe how researchers articulate the sufficiency of their data, often signaling that further collection is either unnecessary or constrained. Closing moves serve to establish the trustworthiness of findings and the sample size, thus concluding the study. The three categories of sampling moves encapsulated subactions, which were motivated through different accounts of why such a move was needed, as well as further explanations of how this was done. We concluded this part of the analysis by confirming our findings of moves within the larger sample. Thus, we tested our moves and actions on 20 randomly selected articles from the sample. These additional articles revealed no further additions.
Findings: Interview Numbers and Sampling Moves in Qualitative Research
In the following sections, we first provide an overview of interview numbers in the 562 studies in our sample; second, we reveal how the methodological approaches relate to interview sample sizes; and third, we analyze interview sampling moves.
Interview Samples Overview
Summary of Interview Numbers.
Distribution of number of interviews 2010-2019 (Y-axis = number of studies, X-axis = number of interviews). Interviews counted in intervals of 10, from 0 to 10).
Our mapping of interview numbers also indicates disparities in methodological expectations across the journals and within geographic areas (e.g., North America vs. Europe). While, at an aggregated level, there is a notable difference between North American journals (mean of 74 and median 59) and European journals (mean of 48 and median 40), the journals we examined also varied remarkably from a mean of 46 and a median of 38 in Org. Stud. to a mean of 96 and a median of 77 in SMJ (Table 1).
Beyond cross-journal and regional differences, our mapping also revealed temporal patterns. While the median number of interviews in articles was 40 in 2010, it was 61 in 2019 (Table 1). This indicates a marginal linear effect across the 10-year sample (Figure 2), which could signal a push toward larger numbers of interviews. Yet, when examining linear trends a bit closer, North American journals appear to be trending upward, while European journals remain relatively stable, with a slight decrease in numbers (Figure 2). However, one should be cautious about overstating this point, as we continued to observe great variation in the numbers of interviews, ranging from 9 to 89 in 2010 to between 15 and 315 in 2019 across articles published in North American journals, relative to 10 to 146 in 2010 and 3 to 181 in 2019, respectively, across articles published in European journals. Such great variance demonstrates that no ideal range or threshold is expected.
Development in interview numbers (2010-2019).
Interview Samples and Methodological Approaches
Interview Numbers and Methodological Choices (N = 562).
* ≤ 15 / > 200 interview.
Overall, our findings indicate that no clear (or only weak) relationships exist between the number of interviews and the choices of methodology across levels of analysis, number of sites, duration of the study, inclusion of other methods, research design, and ontology/epistemology. Regarding levels, we did not find any evidence to support the contention that multilevel studies require more interviews than single-level studies. As seen from Table 2, the mean number of interviews is quite similar across these two approaches, 64 and 54, respectively. Regarding other methods, our findings do not support the notion that combining interviews with other methods (e.g., observations or documents) can explain smaller samples. Studies relying on interviews (only), have on average approximately the same number of interviews as studies that combine interviews with other sources of data (Table 2).
Regarding research design, we see only minor variations in terms of the mean and median number of interviews (Table 2). The median number reveals relative similarity across the research designs. Case studies and grounded theory comprise the largest groups of studies. Discourse studies generally have fewer interviews and range from a fine-grained analysis of 26 interviews on the role of metaphorical terms and expressions in sensemaking (Cornelissen, 2012), toward a larger analysis of 113 interviews and additional data sources used to explore the role of concepts in strategic sensemaking (Jalonen et al., 2018). Yet, we only found a limited number of discourse studies, especially in North American journals (N = 5, Europe N = 10), which constrains our conclusions. In narrative studies, the number of interviews is higher than in other research designs, although a relatively small number of studies and outliers again influence the findings. When we excluded outliers, we found that the number of interviews accounted for less explanatory power.
For ontological/epistemological assumptions, our analysis revealed that these positions are rarely made explicit in relation to sampling interviews. While those studies addressing ontology and epistemology did so in a large part to analyze and interpret their research findings (e.g., Schakel et al., 2016; Sheep et al., 2017; Tukiainen & Granqvist, 2016), such considerations were largely inferred “between the lines” rather than explicitly stated as epistemological or ontological “modes” of operation (Whittle & Reissner, 2025). A few studies noted that their ontology had influenced the sample size. For example, one study noted in their methodology how their practice ontology required them to study and understand phenomena as continuously evolving. Thus, the authors had to observe and conduct interviews over time (Jarzabkowski et al., 2019, p. 857). Yet, because most studies’ justifications and sample-size determinations were indirect, it was difficult to draw firm conclusions about the relationship between ontology/epistemology and interview sample size (Table 2).
While levels, other methods, research design, and ontology/epistemology revealed limited variance, the number of sites (in those studies relying on interviews conducted at multiple sites) and duration of the study provided only a weak explanation for increasing interview numbers. Similarly, the study duration appears to matter when dividing studies into intervals (Table 2), with shorter studies having fewer interviews than longer ones. Yet the different periods across studies make direct comparisons difficult.
The overarching observation across the studies is, however, the large variation in interview samples. Our findings thus resonate with the variety of methodological recommendations outlined above, which suggests that a wide range of interviews can be acceptable, depending on the specific research design employed (e.g., Charmaz, 2012, 2014; Marshall et al., 2013). Yet, our findings also highlight how explanations of the factors influencing interview sample sizes are often silenced. As Hansen et al. (2025) observe, silencing practices that objectify qualitative research, such as marginalizing contextual nuance and researcher decision making, are pervasive. These practices contribute to a lack of clarity around choices—a lack of transparency, which we contend includes sample size. Notably, 84% of the articles in our sample do not explicitly justify the number of interviews (Table 2). When studies justify their number, they often do so through the notion of theoretical or data saturation. However, it is mostly unclear whether such saturation refers directly to the number of interviews, more generally to all data sources, or specifically to a part of the analysis or iteration arrived at through the process of thematic coding. Given this general ambiguity, we decided to focus on studies that most clearly articulated their interview sampling approach.
Interview Sample Size “Moves”
We found that there is no “ideal” number of interviews for publishing articles and that average numbers are misleading because of the considerable variance they mask. We therefore contend that asking, “How many interviews do I need?”, is not the right question for qualitative researchers to pose. Instead, it is important to understand how and why certain differences within studies can lead to different interview sample sizes.
Looking across exemplary studies from our sample, we identified how researchers explained their perspective on using interviews in terms of their “sampling moves.” These fall into three broad categories—opening, focusing, and closing moves—each reflecting a distinct approach to account for the interview sample. Rather than relying on simple counts, these moves reveal how researchers progress and contextualize their sampling of interviews over the course of study. Table 3 summarizes each of the moves and their respective motives (the “why”) and details their association with a series of related actions (the “how”). Table 4 provides an illustrative example demonstrating how these different moves are employed in a single study.
Sampling Moves.
Detailing Sampling Moves Through Illustrative Examples.
Sampling moves represent fundamental explanations relevant across a range of studies and ontological-epistemological positions. For instance, sampling moves could be employed in both representational and constitutive views on interview talk (e.g., Fairhurst & Uhl-Bien, 2012), without requiring the researcher to shift to discourse or conversational analysis fully. Appreciating sampling moves encourages researchers to recognize that in interview research, what counts as “data” is not just the content of what is said, but also how, when, and why things are said. Again, interviews can serve different purposes, which require different analytic approaches and thus appropriate sample sizes. Hence, sampling moves help clarify these purposes and approaches and bring the inquiry forward.
Opening Moves
Opening moves refer to the ways researchers justify sampling, so the study remains open to discovery. Such opening moves are particularly salient in inductive and ethnographic research, where identifying and refining the research phenomenon is recognized as a crucial part of the research process (Pink et al., 2015). Researchers use opening moves as a means for advancing the study in terms of: (1) understanding context, (2) fine-tuning research questions, and (3) refining the interview protocol.
Understanding Context
This move expands the understanding of the research context and lays the groundwork for further data collection. This is exemplified by the use of exploratory or pilot interviews, which can facilitate access to and help identify relevant participants for the study (Canales, 2016; Crosina & Pratt, 2019; Hengst et al., 2020; Smets et al., 2012). Crosina and Pratt (2019) explain: “following some exploratory interviews (N = 4), we embarked on a broader data collection effort” (p. 72), indicating that these preliminary interactions provide essential contextual insights that justified and shaped later interview sampling decisions. Similarly, Smets et al. (2012) conducted 12 pilot interviews to ensure appropriate case selection, while Canales (2016) demonstrates that opening interviews are not merely preliminary steps but rather pivotal moments that reveal latent opportunities for inquiry. In his study on the evolution of Mexican small-to-medium enterprise (SME) financing, early exploratory interviews were essential for mapping the institutional terrain by identifying key stakeholders, detecting subtle variations among banks and states, and constructing a comprehensive chronology of events. Hence, such an opening move for understanding context not only orients the research but also ensures that the subsequent inquiry is firmly rooted in the realities of the field (Battilana & Dorado, 2010).
Fine-Tuning the Research Question
Opening moves can play an important role in refining the research question by revealing emergent themes that are hard to anticipate. Initial interviews can reveal new dimensions of the phenomenon, prompting researchers to adapt or refine their research question (Koppman et al., 2016; Ladge et al., 2012; Porter et al., 2018; Soderstrom & Weber, 2019). Porter et al. (2018, p. 877) noted that early interviews with key actors in the climate change debate unexpectedly surfaced the centrality of controversies surrounding factual errors by the Intergovernmental Panel on Climate Change. This insight led them to recalibrate their research question to focus on how authoritative strategies manage dialectical tensions. Similarly, Ungureanu et al. (2019, p. 1332) found that preliminary sampling in their study of hybrid partnerships uncovered recurring patterns of “vicious circles of decision” that were not initially anticipated. These early interactions prompted the researchers to refine their research questions to better capture the complexity of decision dysfunctions in hybrid contexts.
Refining the Interview Protocol
Exploratory interviews can also be an opening move that are instrumental in refining the interview protocol, such as revealing which questions are unclear and need further modification (Crosina and Pratt, 2019). Cardador and Pratt (2018, p. 2057) studied employee identification in a police officer context and explained that, during the first round of interviews, informants frequently expressed feeling “part of the cops,” prompting a deliberate revision of the protocol for subsequent rounds to ascertain whether this emergent theme was critical. Petriglieri and Obodaru (2019) describe how their interview protocol stabilized over time in their study of dual-career couples and identity formation. They note that: “By the time we interviewed the 26th couple, the protocol was stable and remained so for the rest of data collection” (p. 702). Similarly, Klingebiel and Meyer (2013, p. 138) emphasized that “the responses of initial interviewees allowed us to sharpen our emphases in subsequent interviews.” By analyzing early interview insights, they were able to adjust their questioning to better capture managers’ evolving awareness of uncertainties and adaptive strategies. As such, early interviews can be justified as an important opening move and means of refining the protocol for subsequent interviews.
As these examples show, opening moves can be critical in advancing a study while supporting flexibility and exploration. Such moves were mostly salient in the early phases of the studies reviewed, but may not be limited to this part of the research process as researchers can return to opening moves later in the research process.
Focusing Moves
Focusing moves explain why and how interviews should narrow in or “focus” on a particular aspect of a phenomenon. While opening moves are important in the early stages of the study, focusing moves become salient as the research question(s) and phenomena are clarified, and the sampling either becomes theoretical or seeks confirmation of initial findings through additional interviews. Hence, focusing moves represent a shift from initial exploratory sampling to more targeted sampling as the phenomenon of interest becomes clearer (Beane, 2019). We identified four focusing moves: (4) stratifying the sample, (5) bounding the sample, (6) curating the sample, and (7) decomposing the sample.
Stratifying the Sample
This move depicts how researchers construct analytical leverage, as they iterate between theory and data, by selecting interviewees across relevant categories (Glaser, 1978). Authors may allude to theoretical qualification as they select interviewees through iterative sampling (Gehman et al., 2013), progressive sampling (Bresman, 2013), or purposeful sampling (Vough, 2012). Purposeful sampling is often used to stratify interviewees along dimensions such as gender, status, or age (Ben-Menahem et al., 2016; Detert & Treviño, 2010; Trost, 1986). Others used a combination of purposeful sampling via snowballing techniques and randomized solicitation (Detert & Treviño, 2010). Detert and Treviño (2010) qualified their interview sample by selecting specific groups to gain multiple perspectives on leadership. In their study of employee voice in a Fortune 500 corporation, they systematically selected leaders at the top of focal units. They then randomly sampled subsets of direct reports down to the lowest hierarchical levels: “Within each unit, we purposefully selected leaders at the top of the focal units and then randomly selected subsets of direct reports down to the bottom of each hierarchy, attending to gender and tenure only to get a diverse mix of respondents at each lower level” (Detert & Treviño, 2010, p. 252). This sampling ensured that the study captured a diverse range of perspectives on how leadership influences employees’ willingness to speak up.
Another way of stratifying the sample can be found in interview studies, where sampling is shaped by broader ontological and epistemological orientations, notably in discursive and narrative studies. In such studies, researchers do not aim for statistical representativeness; instead, they structure the corpus for meaningful comparison by selecting interviewees across settings and roles that are expected—given the study's assumptions about language and meaning—to generate analytically contrastive accounts. Cornelissen (2012) illustrates this approach. He conducted 13 interviews with corporate communication professionals across six organizations, stratifying the sample by organizational context and role. This choice reflects that these professionals routinely navigate tensions between personal commitments and social accountability, making them likely to produce rich accounts of the phenomenon. The sample is therefore stratified to enable comparison across comparable cases. The discursive patterns that drive theorization—metaphors, narrative absences, and rhetorical moves—are then inferred analytically from these accounts (see also Alvesson & Robertson, 2016). Through this analysis, Cornelissen identifies coherent discursive strategies (e.g., strategic shifting, framing, and narration) that go together structurally and render complex social processes intelligible. In such a study, a small, purposefully structured sample is thus not a limitation but a condition of possibility for tracing how meaning coheres—and varies—across situated accounts.
While examples like Cornelissen are rare, they nevertheless demonstrate how epistemological and ontological commitments implicitly inform sampling rationales and the kinds of contrasts researchers build into the sample. More broadly, we note that interview-based studies differ in whether they treat language and interaction as being constitutive of reality—as in Cornelissen (2012)—or as representations of an external social world, underscoring how interview samples are selected and made to count, and thereby justified, within different paradigms (cf. Fairhurst & Uhl-Bien, 2012). Echoing Cunliffe (2011, p. 648), these perspectives point to the need to “situate our work in careful and informed ways,” as researchers make choices that affect a study's focus and sampling, even when those choices are not explicitly stated.
Bounding the Sample
Another focusing move regarding sample sizes revolves around empirical opportunities and challenges (Creed et al., 2010; Schakel et al., 2016). One common justification for a constrained sample size is the difficulty of accessing suitable participants. These sampling challenges may include limitations in access to participants, the nature of the research phenomenon, or the depth of engagement required with each interviewee. Scholars use various arguments to qualify datasets, particularly smaller ones, to ensure theoretical robustness despite practical constraints. Difficulties in gaining access to interviewees also reveal important contextual insights. For instance, such barriers may speak to the scarcity or secrecy of the phenomenon (Anteby, 2024). Peticca-Harris et al. (2016) describe access limitations as a key factor in shaping their final sample, while Dhalla and Oliver (2013) faced difficulties securing interviews with senior bank officials due to the sensitivity of their study on noncompliance within the banking sector. The authors reported that one bank refused to participate in their study. Similarly, Hirst and Humphreys (2015) encountered temporal constraints when interviewing professional service employees, as participation was viewed as “nonproductive time,” thereby making employees hesitant to participate in the study. These examples highlight how practical constraints can shape the sample size.
At times, scholars compensate for constrained samples by collecting richer, more detailed data in each interview. Creed et al. (2010, p. 1358) illustrate this in their study of gay, lesbian, bisexual, and transgender (GLBT) ministers as institutional change agents, which relied on only 10 interviews, yet provided in-depth analysis. Creed et al. (2010) acknowledged that their study relied on fewer interviews than often expected in qualitative studies but justified this by emphasizing their depth of engagement with each participant. They argue that fewer interviews enabled them to explore the experience of institutional contradiction in greater detail. Ultimately, bounding the sample empirically seeks to demonstrate that, while a large interview sample can be desirable, it is not always necessary, and a focused sample can be sufficient.
Curating the Sample
Curating data describes another focusing move, one which appears particularly salient to scholars managing large interview numbers while navigating the challenges of ensuring sufficient analytical depth. When spanning many interviews, some researchers employ selective strategies to highlight the most relevant, thereby curating only those interviews they wish to foreground (Battilana & Dorado, 2010; Courpasson et al., 2012; Helfen & Sydow, 2013; Mazmanian & Beckman, 2018). While it is established that not all interviews or interview quotes can be represented in an article (Pratt, 2009), the sheer volume of data can make it challenging to analyze the material in sufficient depth. Thus, the process of curation becomes relevant. Battilana and Dorado (2010, p. 1421) illustrate this move as they conduct 78 interviews across two organizations but focus on only four interviews from one organization and seven from the other to avoid “data asphyxiation.” Similarly, Courpasson et al. (2012) refined their study on productive resistance by selecting seven long-form interviews with key participants rather than relying on their entire dataset of 170 interviews. This approach to curating the interview sample enabled them to showcase the nuances of productive resistance through everyday engagements while ensuring that the empirics remained manageable for theorizing. Although curating data can be considered an indirect sampling move, articles are often slices of larger research programs. Thus, curating data may serve as another way to explain why more or fewer interviews are needed.
Decomposing the Sample
A final type of focusing move involves demonstrating rigor through the reporting of sample sizes and data collection methods. This detailing often involves decomposing the interview sample across different stages, levels, and thematic units (Cardador & Pratt, 2018; Dalpiaz et al., 2016; Helfen & Sydow, 2013). Where the stratifying move explains the selection of interviewees across categories, decomposition is a focusing move concerned with reporting, which disaggregates the achieved interview sample to show the evidence base. Cardador and Pratt (2018, p. 2057) exemplify this approach by providing an explicit breakdown of their interview sample, specifying that “Nine interviews were conducted in round 1, 10 were conducted in round 2, and 22 were conducted in round 3” (p. 2057). This systematic detailing allows the reader to trace the study's evolving focus and understand how emerging insights influenced subsequent rounds of interviews and, consequently, the final sample size. Similarly, Rerup and Feldman (2011) reported conducting “a total of 109 semistructured interviews with a cross-section of 44 different informants,” further specifying that these included “91 one-on-one interviews with 33 individuals, 14 group interviews with 25 individuals, and 4 telephone interviews with 3 individuals” (p. 829). Often, the inclusion of a table serves to break down interviews across different organizational roles and stages, clarifying the composition and size of the sample. By decomposing the interview sample across time, organizational levels, and thematic categories or units of observation, researchers demonstrate their evolving focus and signal how their findings are grounded in the interview data.
Closing Moves
The final category of moves relates to explanations of why and how more interviews are either unnecessary or not possible. Hence, such moves work to “close” the sampling process. We identified three types of closing moves, which included: (8) validating findings and theorizations, (9) establishing an endpoint, and (10) clarifying limits.
Validating Theorizations
One move in determining the final sample size is validating insights through member checks or triangulation with other data. This move seeks to demonstrate the credibility of the findings and theorizations by cross-referencing interview data with additional sources, such as returning to informants for validation, integrating historical or documentary evidence, or using interviews as a secondary data source for augmentation. Some scholars refer to these as “validity checks” (Ozcan & Gurses, 2018) or “member checks” (Charmaz, 2005), a process which involves re-engaging interview participants to assess whether the emerging interpretations accurately reflect their experiences.
Crosina and Pratt (2019) demonstrate how member checks can be conducted with interviewees. They describe how they conducted a second round of interviews with 27 informants to address key questions that emerged during the first analysis phase and note how: “These latter interviews also served as ‘member checks’” (Crosina & Pratt, 2019, p. 74). The return process ensured that the researchers’ interpretations remained aligned with the participants’ experiences and that they had conducted sufficient interviews to answer the research question. Member checks can also be done with complementary actors. In their study of how employees identify with customers, Cardador and Pratt (2018) employed a member check as a validity measure after analyzing their initial dataset. In developing their theoretical model, the authors conducted interviews with five additional customers “to verify employee interpretations of their relationships with customers” (Cardador & Pratt, 2018, p. 2060). Similarly, Smets et al. (2012) studied practice-driven institutional change among banking lawyers and used interviews with five elite respondents to evaluate their emergent theorizing. Validating theoretical constructs with additional participant perspectives, member checks, and similar approaches can be important for testing the sufficiency of the sample size. While validation can take different forms, depending on the researcher's approach, it can refine and eventually confirm theoretical constructs while demonstrating the robustness of the findings (Crosina & Pratt, 2019; Hatch & Schultz, 2017; Rerup & Feldman, 2011).
Establishing an Endpoint
Another closing move is determining a point at which conducting interviews should stop, driven by the researchers’ analysis or interpretation rather than participants’ confirmation and verification. The most common type is “theoretical saturation,” the point at which additional data no longer contribute novel insights to the study's emerging theoretical framework (Glaser & Strauss, 1967). Some studies exemplify this move by explicitly articulating when and why data collection ceased, demonstrating that their findings were theoretically comprehensive (Cardador & Pratt, 2018). For instance, Petriglieri and Obodaru (2019) noted: “By the 40th couple, no new topics and themes emerged, suggesting that we were approaching theoretical saturation (Glaser & Strauss, 1967). After interviewing ten more couples confirmed it, we stopped collecting data” (Petriglieri & Obodaru, 2019, p. 702). This iterative approach demonstrates that theoretical saturation is not a fixed number of interviews, but rather an emergent assessment based on whether new insights continue to shape the study's conceptual categories. This highlights how qualitative researchers often iteratively assess their findings, using each phase of data collection to determine whether further interviews will meaningfully refine their theoretical framework.
Several studies indirectly accounted for the adequacy of their sample based on the emergence and repetition of themes, not necessarily by articulating intercoder reliability and claiming theoretical saturation (Sheep et al., 2017) or member checks (Cardador and Pratt, 2018), but also by validating patterns of key concepts across multiple interviews (e.g., Brown & Toyoki, 2013; Hoedemaekers & Keegan, 2010). For example, a study on identity work and legitimacy among prisoners evaluated their data by noting that “all types of identity/legitimacy talk were well represented” in their sample (p. 881). They provided counts showing the number and percentage of inmates (out of 44) who referenced different forms of legitimating and delegitimating identity work, stating that “most made reference to several of them” (p. 881). This demonstrates an assessment of the data's coverage of the phenomenon based on the prevalence and representation of different thematic categories across the interview sample. Similarly, Hoedemaekers and Keegan (2010) used a cutoff point of 10 interviews to substantiate a signifier's centrality. Signifiers found across the interviews in sufficient frequency and distribution were incorporated into their list of key signifiers, such as “to work together,” which was highly central to various data sources, including interviewees’ talk. This explicitly uses the presence and spread of themes across the sample of interviews as a criterion for determining the adequacy of the sample size.
Other more discursive-oriented scholars in our sample focus on the internal coherence of discursive constructions, both within and across accounts, to theorize from the sample (e.g., Cornelissen, 2012; Dameron & Torset, 2014). Their validation rests on demonstrating a coherent discursive structure, one that organizes their phenomena across multiple accounts. These accounts allude to a constitutive view on talk (Fairhurst & Uhl-Bien, 2012) and a structural corroboration (Jacobs, 1990), whereby researchers establish convergent lines of analysis drawn from a spectrum of methods or data (see also Fairhurst, 1993). Such approaches point out that their interpretive logic justifies their sample size in supporting a validity claim—not through frequency or saturation—but rather via the coherence of the accounts and their internal logic (c.f. Grint, 2005; Jacobs, 1990; Schön, 1983). In this way, an endpoint can take different forms depending on the researcher's assumptions.
Clarifying Limits
A final closing move concerns how some scholars described how their research had been circumscribed or curtailed by various factors, such as running out of access, financial constraints, or the time required to continue the study. We previously noted access challenges as part of the focusing move (see bounding the sample), but here we draw attention to how scholars close their sampling efforts in response to external limitations (Dick, 2015; Porter et al., 2018). An illustrative example comes from Dick (2015, p. 905), who explained: “deviation from the intended sample happened because the participating forces experienced considerable difficulties in getting people to participate in the research and the project timing was bounded by the requirements of the funding council.” This quote illustrates how pragmatic concerns, such as participant recruitment difficulties and externally imposed deadlines, can delimit the interview sample. While likely a common experience in practice, such a closing move was seldom referred to explicitly in published accounts.
Overall, the three categories of sampling moves reflect the strategies identified in a focused selection of 60 qualitative interview studies from our sample of high-impact management and organizational journals. Inspired by Pratt et al. (2022), we suggest that applying these moves uncritically or as a rigid template could hinder rather than enhance research. Instead, we consider these moves as illustrative resources, or heuristic tools, that qualitative researchers can thoughtfully adapt and expand to their analytical needs. Most researchers tend to employ both opening and closing moves, with focusing moves positioned in the middle, oscillating between the two. Our goal is not to create another template of required steps for authors. Rather, we offer a framework with guiding questions that researchers, reviewers, and editors can engage with and advance when evaluating the quality of qualitative interview research. Although the studies we examine provide rich examples of interview sampling, published articles capture only a partial view of how sampling unfolds in practice—and additional “sampling moves” beyond those we detected are likely relevant. Hence, our aim is not to prescribe standardized methodological steps but to provoke reflections by providing guiding questions that encourage researchers, reviewers, and editors to engage critically and creatively with the quality and trustworthiness of qualitative interview-based research.
Discussion: Determining Interview Samples
In this study, we set out to examine how interview sample sizes and methodological choices are related, and how scholars account for their interview numbers. Our findings reveal a high variance in interview numbers, and most articles in our sample provided only limited details about the factors or considerations that determined the final number of interviews. We observed no (or only weak) relationships between the number of interviews and various methodological approaches. To advance our analysis further, we then focused on revealing the sampling moves that are often only subtly deployed to account for the use and number of interviews employed.
Our findings have two important implications, which encourage both caution and inspiration. First, qualitative research studies appear to have become partially unmoored from their methodological foundations in how interview sampling is accounted for. To explain this unmooring further, we turn our attention to what we term politics and pragmatism. Second, despite ongoing concerns about the homogenization of qualitative research and the creep of quantitative measures, our analysis reveals considerable variety in interview sample sizes across studies. This suggests a more complex picture, with evidence of both homogeneity and heterogeneity. We propose that attending sampling moves can be a source of inspiration and a way forward, helping researchers reflect on their research and reconnect methodological foundations with interview numbers, while encouraging novelty in research methods.
Unmoored From Methodology: Politics and Pragmatism
One explanation for the weak relationships between the number of interviews and methodological approaches is politics, understood as, “actions undertaken to acquire, enhance, and use power to obtain preferred outcomes in situations having dissensus on choices” (Pfeffer, 1981). Researchers are aware of their audiences (e.g., journals, editors, reviewers, and readers in general) and may choose to sample in ways that are deemed credible for their audience (Emmel, 2013; Patton, 2014). The pervasive requirement of publish or perish encourages qualitative researchers to perform actions that comply with editors’ and reviewers’ expectations of high-quality research to obtain publications (Emmel, 2013; Köhler et al., 2022). High quality may be confused with quantification because numbers “tend to acquire an aura of objectivity” (Ranganathan & Benson, 2020, p. 576). The trend toward more interviews may support such an aura. Quantification facilitates easy comparison and evaluation, and “imposes a universal definition of ‘high quality’” (Espeland & Stevens, 1998; Ranganathan & Benson, 2020, p. 578). Qualitative researchers have long been charged with basing their theorizing on “thin evidence,” an imputation that necessarily precludes them from producing high-quality research (Gioia et al., 2012, p. 18). This implicit association between quality and quantification may help to explain why researchers could feel a pressure to emphasize total interview numbers rather than to detail their methodological processes and ontological and epistemological stance. As Pratt et al. (2020, p. 3) warned, “incorporating logics from quantitative research into qualitative research has political as well as methodological implications for the field.” Indeed, quantification of interviews risks overlooking the logic that interviews are not only a means for finding standardized patterns, but also spaces where diverse and contextually adaptable meanings are constructed and understood (Jacobs, 1990; Schön, 1983).
Another explanation for the weak relationship between the number of interviews and a range of methodological approaches is pragmatism, suggesting that determining interview samples is often an emergent and opportunity-driven process. Given the varied and conflicting guidance available, researchers must often make situational judgments that do not neatly fit into existing guidelines. Hence, sampling decisions are guided not only by principles of, for example, theoretical saturation, but also by emerging opportunities and practical constraints such as time, access, resources, and chance (Strauss & Corbin, 1998). From this perspective, it can be challenging to remain faithful to rigid methodological rules that may not consider such constraints, and qualitative researchers often decide on interview sample sizes based on opportunities encountered during the research process. This is the experience of the authors of this article, as well as in several studies we examined. For instance, Wijaya and Heugens’ (2018) ethnographic study of institutional reproduction in eleven Pentecostal churches utilized opportunities that emerged naturally in the field, which could not have been predetermined. In this way, the number of interviews reported in research articles may reflect the tension between researchers’ practical experiences of performing research and adherence to methodological requirements. In this regard, our proposed use of sampling moves rejects a predetermined perspective and complements authors like Guest et al. (2006), who offer evidence of saturation thresholds in relatively homogeneous samples. Yet, because samples in organization and management studies are often heterogeneous, we highlight the repertoire of sampling moves researchers employ across such varied contexts.
Reconnecting Research Methodologies to Interview Numbers Through Sampling Moves
As discussed above, there is considerable evidence of pressure on qualitative researchers to adhere to quantitative and homogenizing standards. For instance, the drive to be replicable (Pratt et al., 2022) or to adopt specific templates for publication (Köhler et al., 2022; Pratt et al., 2022). Indeed, Köhler et al. (2022, p.2) refer to the survey results of the IFSAM (International Federation of Scholarly Associations of Management, 2021), which stated that “management knowledge had become homogenized and formulaic, with researchers compelled to craft their work to conform to a particular style and format.” However, our findings suggest that not all facets of qualitative research face the same conformance pressures. Qualitative research enjoys considerable variance in terms of interview sample sizes. Our findings thus provide a more nuanced account and cautious stance on the sometimes-assumed creep of quantitative standards into qualitative research. This nuance is important because it encourages scholars to avoid potentially premature and polarized conclusions regarding the state of qualitative management research.
Drawing on our analysis, we propose that a focus on sampling moves—opening, focusing, and closing—can help foster a closer connection between research methodologies and the use of interviews in publications. As Patton (2014, p. 163) noted, one should strive to account for the appropriateness of the methodology in that “The point is to do what makes sense, report fully on what was done, why it was done, and what the implications are for the findings.” We claim that focusing on sampling moves will help qualitative researchers do so in tailored ways. Some studies may remain in the opening phase for longer to allow for a more refined focus on the phenomenon under investigation (e.g., exploratory ethnographies), while others may engage in closing moves earlier (e.g., studies with predefined theoretical frameworks). Most researchers navigate all three moves at different stages, adjusting their justifications as their study evolves.
Specifically, we claim that by being cognizant of various sampling moves and clearly communicating their decisions in relation to methodological approaches, researchers can better determine and justify their interview samples. This reflects our belief that methods sections in articles should provide more consideration of the process of using interviews and acknowledge that multiple methodological factors inform interview sample sizes. Such reflections also involve bringing to the fore the ontological and epistemological commitments that may, directly or indirectly, shape the sampling process (Cunliffe, 2011). For example, interview studies may not aim for generalization; instead, interviews often seek to illuminate differences in personal accounts, interpretations, or lived experiences (e.g., Gill, 2014). In this sense, epistemological positioning—not just pragmatic or methodological considerations—guides the determination of an appropriate interview sample. While we do not observe systematic variance across epistemic or epistemological communities with their traditions, our study does indicate important sensitivities in this regard. For example, open-ended, exploratory interviews were more commonly used in inductive studies aligned with interpretivist traditions. At the same time, focused or curated sampling moves often reflected commitments to internal coherence and the possibility of structural corroboration (Jacobs, 1990). At the same time, we recognize that studies with fewer interviews, while epistemologically justified, may face challenges in the publication process if the expectations of journal editors or reviewers are not fully aligned with such approaches.
As summarized in Table 5 and inspired by Mees-Buss et al. (2022), we further propose a list of indicative questions to ask within each move, which we introduce below.
Sampling Moves and Probing Questions.
The questions raised in Table 5 highlight a core message derived from the sampling moves: interview samples often emerge over time as researchers grapple with the practical realities of data collection, analysis, and theoretical discovery (Abbott, 2004). A focus on interview sampling moves seeks to unmask the complexity of the research process and provide insight into the depth of knowledge the researcher has developed. This is critical, as the quality of interviews is more important than the number (Corbin & Strauss, 1996).
In attending to opening, focusing, and closing moves, it is important to recognize that interviews often serve multiple purposes within the research process. Several studies used interviews as pilot studies as an opening move, while other interviews may have been used to build contextual insights or to stratify and explore multiple, yet distinct groups (e.g., Smets et al., 2012). Accounting for such moves can also demonstrate the degree of exposure to the phenomenon under study. As Small and Calarco (2022, pp. 18–19) point out, researchers can be exposed to a phenomenon under study in many ways, and accounting for that is key to building trust in a researcher's interpretations. Detailing the differing reasons for interviews and how they were used elucidates how researchers arrived at their sample size. Further, it can avoid the sometimes misleading implication that all interviews serve equally to substantiate the claims presented in a study.
Focusing on sampling moves recognizes that researchers often navigate real-world complexities that require adaptation, compromise, or improvisation in response to unforeseen challenges. Accounting for such challenges serves as the very foundation of focusing moves. Certainly, highlighting such challenges can illuminate not only methodological choices, thereby underpinning a study, but also reveal critical insights into the context of the phenomena under investigation (e.g., the secrecy of the context; Anteby, 2024).
Limitations
Our analysis is based on six high-impact management and organizational journals over a decade. It is important to note that there are many other journals across the social sciences that we have not analyzed. Articles in these other journals may have employed interviews in ways we have not appreciated in our study, including the adoption of distinct epistemological and ontological stances. For instance, semistructured interviews, which dominate management and organizational research, may be less prevalent in other fields. Future research could further unpack our work on interview samples, for instance, by assessing the use of interviews in other disciplines. We believe there is a considerable opportunity to deepen our understanding of this topic by interviewing scholars from different disciplines about their “moves” and use of interviews and research purposes. Given our discussion of how power and politics influence sample sizes, an associated line of inquiry could involve asking researchers how their methodological decisions emerge and are sustained within academic culture through peer pressure, institutional norms, or the editorial preferences of journals. For instance, our experience is that many methodology sections are omitted from published versions, potentially leaving critical details out for a broader audience. Finally, our reviewed articles were all published prior to the COVID-19 pandemic, when nearly all interviews were conducted face-to-face (Sah et al., 2020). The rise of online interviewing and the use of large language models to process interview data may alter both the dynamics of data collection and, potentially, the number of interviews considered sufficient—a point future research could productively explore.
Conclusion
Interviews remain a core method in organizational research, and yet a key question persists: How should interview sample sizes be determined? In this study, we have explored how interview sample sizes relate to methodological choices and how researchers account for their interview samples. Our analysis of North American and European organization and management journals reveals substantial variation in interview sampling and, in many cases, a lack of explanation or clear connection to methodological approaches. To address this, we examined how authors account for the number of interviews in their published work. Based on this, we identified what we term sampling moves—the practical ways in which interview studies progress and determine their sample. These moves shift the focus away from numbers and fixed guidelines, toward assessing qualitative interview studies on their own methodological terms, in configurations tailored to the studied phenomena.
While classical methodological guidelines offer specifications or guardrails, new criteria are emerging (Small & Calarco, 2022), and researchers are also developing their own strategies. It is our contention, therefore, that qualitative scholars should continuously interrogate, question, and challenge the norms of interview numbers. We hope our results will help stimulate critical reflection on the use of interviews. New ways of conducting qualitative research and interviews will continue to emerge, for example through new ways of doing interviews and large language models. We believe such developments require researchers to be even clearer about their sampling moves so others can understand and assess different approaches. Our study offers guidance for researchers, reviewers, and editors when gauging whether an interview sample size is appropriate.
Supplemental Material
sj-pdf-1-orm-10.1177_10944281261424516 - Supplemental material for “How Many Interviews Do I Need?” An Examination of Interview Numbers and Sampling Moves in Qualitative Research
Supplemental material, sj-pdf-1-orm-10.1177_10944281261424516 for “How Many Interviews Do I Need?” An Examination of Interview Numbers and Sampling Moves in Qualitative Research by Kasper Trolle Elmholdt, Michael Gill and Jeppe Agger Nielsen in Organizational Research Methods
Footnotes
Acknowledgements
The authors are grateful to Tine Köhler and Anne Smith for their thoughtful editorial guidance, and to the three anonymous reviewers for their constructive engagement and insightful feedback throughout the development of the paper. The authors also thank participants in the Academy of Management Annual Meeting 2021 session, ‘Collecting Data with Interviews: Concerns and Guidance for Researchers.’
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Supplemental Material
Supplemental material for this article is available online.
Notes
Author Biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
