Mixed Methods Single Case Research: State of the Art and Future Directions

Historical Roots of MMSCR

Prototypically, research at the single case level is conducted using either a predominantly qualitative approach or a predominantly quantitative approach. We will refer to these two approaches as the qualitative case study and the quantitative single case experiment. ¹

A qualitative case study can be used to conduct an in-depth and rigorous empirical investigation of a particular phenomenon by studying one identified “case” within its context. The “case” is supposed to contain a manifestation of the phenomenon and can refer to any unit: a person, a part of a person (e.g., a brain region), a group of persons (e.g., a family or a religious community), an organization (e.g., a company or a school), a set of procedures (e.g., a public program or a community intervention), an event (e.g., a political meeting), or an artifact (e.g., a building). Well-known examples of qualitative case studies are the studies conducted by Freud, Pavlov, Watson, and Piaget in the field of psychology, by Frederic Le Play and the “Chicago School” in the field of sociology, and by Bronislaw Malinowski and Margaret Mead in the field of anthropology (Merriam, 2009; Stake, 1995; Swanborn, 2010; Yin, 2017).

Quantitative single case experimental designs can be used to study the causal relationship between the independent and the outcome variable for a single case, when an independent variable is experimentally manipulated and an outcome variable is repeatedly measured under the various levels of the independent variable for this case. Well-known pioneers of the single case experimental approach in the behavioral sciences include Fechner, Wundt, Ebbinghaus, Stratton, Skinner, and Sidman (Barlow, Nock, & Hersen, 2009; Gast & Ledford, 2014; Kazdin, 2011). ²

Although case study research has a long tradition of including both quantitative and qualitative approaches, the history of science shows that most often one of the approaches is just given a subsidiary role and that the quantitative data and methods are commonly used to locate the case within a larger sample or a general population (for an overview, see Kitchenham, 2010; Maxwell, 2016; Yin, 2017). Case studies in which quantitative and qualitative approaches are given an equal status, and in which experimental manipulations and statistical analyses are directed exclusively at the single case level are rare. However, this equal status mixed methods approach to research at the single case level dates back to, at least, the end of the 18th century, when narrative materials and descriptive interpretation were incorporated into single case experiments. Influential examples are the single case experiments of Edward Jenner at the end of the 18th century that contributed important knowledge on immunization and the ultimate eradication of smallpox, starting with the case of James Phipps (Riedel, 2005).

The more explicit historical roots of MMSCR date back to the late 1970s and 1980s, in the field of psychotherapy. Whereas in the 1960s and early 1970s empirical psychotherapeutic research was often focused exclusively on either process or outcome, in qualitative case studies and quantitative single case experiments respectively, in the late 1970s and 1980s researchers started to jointly apply qualitative and quantitative approaches to focus on the integration of both process and outcome in the study of single cases in complex therapeutic situations (Gaston & Marmar, 1989; Kiesler, 1973). Such an MMSCR approach avant la lettre allowed practitioners and researchers to study change processes as well as outcomes in detail. In the same vein, methodologists such as Campbell (1979) and Howard (1983) stressed the importance of methodological pluralism in idiographic psychotherapy research and evaluation research. In his discussion of the case study approach, Campbell (1979) concluded “Qualitative knowing cannot be replaced by quantitative knowing. Rather, quantitative knowing has to trust and build on the qualitative, including ordinary perception. We methodologists must achieve an applied epistemology which integrates both” (p. 66).

The Distinct Nature of MMSCR

The exploration of MMSCR, and its distinct nature, is important from the perspective of qualitative case study research, from the perspective of single case experimental research, and from the perspective of mixed methods research. We will take each of these perspectives in turn.

For qualitative case study research, it is important to realize that quantitative and experimental research is not antithetical to idiographic research or to research with a focus on one case. In fact, single case experimental research has this focus on the case, and in terms of “unit of research” is located at exactly the same level as qualitative case study research. In this sense, they might be naturally compatible partners. Furthermore, the addition of baseline measurements, time-series data, controlled intervention, and formal data analysis might strengthen or enrich the understanding of “the case,” certainly if questions regarding efficacy, explanation, and prediction are involved (Barlow et al., 2009; Kazdin, 1981, 2011).

From the perspective of single case experimental research, the addition of an integrated qualitative case study might be crucial to give context, to put narrative flesh on the bones of the experimental and statistical analysis, and ultimately to provide a basis for replication and generalization. Integration of observations, interview material, and document information is especially relevant if the research question is not merely “Does it work?” but also involves issues of process, feasibility, appropriateness, and meaningfulness (Pearson, Wiechula, Court, & Lockwood, 2005). In handbooks on single case experimental research (e.g., Barlow et al., 2009; Gast & Ledford, 2014; Kazdin, 2011), cursory reference to the potential of additional qualitative data is given. For example, Kazdin (2011) argued as follows:

However, the presentation of qualitative methods and techniques in these handbooks is never elaborate, nor is it made clear how quantitative and qualitative research strands can be integrated. It is at this point that the mixed methods literature comes to bear.

From the perspective of mixed methods research it is important to thematize MMSCR as a distinct approach in order to disentangle the level dimension (case vs. group level) and the methodological approach (qualitative vs. quantitative). Too frequently, research at the case level is exclusively conceptualized as qualitative research and research at the group level is conceptualized as quantitative research, with the prototypical example of the mixed methods study as “cases within surveys” or “surveys within cases” (Yin, 2006, 2017). Also for mixed methods research, there is a growing awareness that the intensive study of a single case has its own merit, and yields evidence that an extensive study of a sample cannot provide (Van Ness, Murphy, & Ali, 2017). Mixed methods research can move beyond the discursive and isolated remarks about the necessity, the benefits or the expediency of combining single case experiments and qualitative case studies, and offer an organized and integrated conceptual framework and methodological toolbox (Creamer, 2018; Creswell, 2016; Mertens et al., 2016; Plano Clark & Ivankova, 2016).

What Kinds of Knowledge Can MMSCR Produce?

An important epistemological question related to MMSCR is what kinds of knowledge MMSCR can produce. We distinguish three distinct kinds of knowledge that MMSCR can produce. First, MMSCR can produce answers to specific research questions concerning the case(s) under study. For instance, MMSCR can be used to quantitatively and qualitatively assess the effectiveness, feasibility, appropriateness, and/or meaningfulness of a treatment, program, or intervention for a single case. From a practical point of view, MMSCR can be preferred over group research when rare or unique conditions are involved, or when it is impossible to obtain a large homogeneous sample of cases under similar conditions. For example, case studies are recommended in disability research if atypical disorders or uncommon comorbidities are under investigation (Zhan & Ottenbacher, 2001). This use of MMSCR is analogous to what Stake (1995) called an “intrinsic case study.” In this first type of MMSCR, we have intrinsic interest in the peculiarities and particularities of the case under investigation, and the question of broader applicability or generality is irrelevant or only of secondary importance.

Second, MMSCR can be used for the development of new theory. According to Eisenhardt (1989), essential features of case studies aimed at theory development are: theoretical sampling, using multiple data collection methods, involving multiple investigators, conducting iterative within-case and cross-case analyses using divergent data analysis techniques, and applying replication logic across cases to develop, extend, and sharpen theory. These features particularly hold for MMSCR: Several methodologists argued that a mixed methods study can be strengthened by integrating multiple data collection and data analysis methods, and by involving multiple researchers in a single study (e.g., Fielding, 2012; Flick, Garms-Homolova, Herrmann, Kuck, & Röhnsch, 2012). Furthermore, the ability to build strong theories improves when a multiple-case design is used and when literal or theoretical replications are involved (Yin, 2017). Stake called this the “instrumental case study” because the case is not studied for its own sake, but because it is instrumental for something else, namely theory development.

Third, MMSCR can be used for verifying existing theories and empirically testing specific hypotheses. Yin (2017) emphasized that case studies are not only useful for exploration but also for verification and testing. This goal can be accomplished by starting from theoretical propositions derived from or related to the research question and by linking the data to each of these propositions. In this sense, an MMSCR can falsify theoretical propositions and stimulate theory building by identifying and studying a “black swan” or a “critical case” for hypotheses that may be hard to “prove.” An example is the case of Genie, who allowed testing the “critical period hypothesis” formulated by Eric Lenneberg relating to children’s language acquisition. Genie was taken into custody at the age of 13 years by social workers. Until then, she had spent much of her life chained to a chair in her bedroom, without much human interaction, and she could not speak, walk, or respond to other people. Her case attracted psychologists and linguists, who were interested in finding out whether she could still learn to speak (Kulkarni, 2012). Lenneberg’s “critical period hypothesis” stated that the critical period for language acquisition lasts until the age of 12 years: After the onset of puberty, one is no longer able to learn and utilize language in a fully functional manner (Cherry, 2016). Genie’s case corroborated Lenneberg’s hypothesis: Despite being given an enriched learning environment from the age of 13 onward, she was not able to learn to speak, because she had missed the “critical period” (Cherry, 2016).

What Are Possible Philosophical Underpinnings for MMSCR?

Integration of the quantitative and qualitative research strands in mixed methods research involves not only integration of methods and techniques but also integration at the level of methodology and underlying philosophy (or paradigm; Fetters & Molina-Azorin, 2017). By defining MMSCR as research in which single case experimental and qualitative case study methodologies and their accompanying sets of methods and techniques are integrated, we already referred to the first two integration dimensions. In this section, we want to explore the third dimension and reflect on the way in which four philosophical stances (constructivism and interpretivism, critical-transformative-feminist approaches, pragmatism, and postpositivism) are compatible with MMSCR. Our position is that users of MMSCR should at least be aware of their dominant view on what reality is (ontology), what knowledge is (epistemology), what the underlying values are (axiology), and which research logic is preferred (methodology). We agree with Hesse-Biber (2010b) that each of the philosophical stances will lead to a different approach of the research endeavor, but we do not rule out the possibility that two or more stances are taken by different members of a research team and that different assumptions prevail in different phases of the study (Creswell & Plano Clark, 2018). Furthermore, our exploration of the four philosophical stances is not meant to be limitative. Other stances, such as critical realism (see, e.g., Maxwell & Mittapalli, 2010), or other combinations of stances can be connected to MMSCR.

What Kinds of MMSCR Have Been Applied in the Empirical Literature?

So far, two kinds of MMSCR have been applied in the empirical research literature: qualitative case studies in which quantitative research components are brought in and single case experiments that are complemented with qualitative data. With regard to the first type of MMSCR, in 1981, Kazdin explicitly promoted to bring in quantitative research components in qualitative case studies. Moreover, Kazdin (1981) argued that the collection of qualitative as well as quantitative data is what enables researchers to draw valid inferences from case studies. He for instance stated that a case study without quantitative data “usually precludes drawing firm conclusions about whether change occurred” (Kazdin, 1981, p. 188). In addition to Kazdin’s work, other influential examples of quantitatively informed case studies are Strupp’s (1980a, 1980b, 1980c, 1980d, 1990) studies on time-limited psychotherapy.

Second, single case experimental researchers have argued that in addition to collecting quantitative data to document changes in the outcome variable(s) caused by the manipulation of the independent variable, qualitative data can be collected to document the feasibility, appropriateness, and/or meaningfulness of the treatment or intervention under study (e.g., Kellett & Hardy, 2014). Furthermore, in addition to assessing the statistical significance of an intervention effect by means of collecting quantitative data and conducting statistical analyses, qualitative data can be collected to additionally study the practical significance of the intervention (Kirk, 1996; Pearson et al., 2005; Thompson, 2002). Practical significance refers to the size of the improvement in the outcome variable(s) as well as to the practical importance of this improvement (Kazdin, 1999; Perdices & Tate, 2009). In 1977 already, Kazdin advised single case researchers to assess the practical importance through social validation: He recommended supplementing quantitative single case experiments with qualitative data on whether clinical, social, or applied importance had been achieved. A recent example of a single case experiment complemented with qualitative data is the study of Kellett and Hardy (2014): The authors conducted a single case experiment on the treatment of paranoid personality disorder with cognitive analytic therapy, but supplemented the experiment with the collection of qualitative data on the participant’s experiences with cognitive analytic therapy. In addition, qualitative research components were used to study whether clinical change had taken place, and to further study the identified change mechanisms.

MMSCR in Current Practice and Applications

Whereas the single case experimental and qualitative case study methodologies typically appear to exist on “separate islands” in the methodological literature, in practice they are often combined. For instance, in the domain of challenging behavior among persons with intellectual disabilities, a practitioner can quantitatively measure the amount of challenging behavior shown by a client before, during, and after implementing an intervention (e.g., teaching alternative skills), plot the collected data on the challenging behavior level over time, and evaluate whether the intervention has an effect on the challenging behavior of the client, thereby applying visual and/or statistical analysis techniques (see, e.g., Heyvaert, Wendt, Van den Noortgate, & Onghena, 2015; Zhan & Ottenbacher, 2001). However, when defining the challenging behavior and selecting an appropriate intervention, “quantitative”and“qualitative” information is taken into account that relates to the history, present behavior, and context of the client (e.g., previous interventions with this client and their effects, contextual factors that appear to influence the challenging behavior, contextual factors that might impede or facilitate certain interventions; e.g., Heyvaert, Saenen, Maes, & Onghena, 2014, 2015). In addition, perceptions about the “effects” of the intervention might be qualitatively expressed: Does the client and the caretakers and family feel that the intervention is helpful, effective, desirable, feasible, and useful in day-to-day practice? How do they believe the intervention can be optimized? Furthermore, practitioners may qualitatively link data points of noteworthy clinical change (whether in the direction of improvement or deterioration) to their notes on the course of the intervention, which may provide a basis for evaluating the impact of specific components of the intervention, as well as the impact of external events (Dattilio, Edwards, & Fishman, 2010). In addition to this MMSCR example in the field of special education, practitioners in many other applied fields combine qualitative and quantitative data and methods at the case level, but most commonly without looking at the mixed methods literature for a methodological justification or optimization (see, e.g., Gannotti, Gorton, Nahorniak, & Masso, 2013).

How to Conduct MMSCR?

A large diversity of mixed methods designs have been applied in actual empirical research and even a large diversity of typologies have been proposed for organizing these designs (Creswell & Plano Clark, 2018; Leech & Onwuegbuzie, 2009; Tashakkori & Teddlie, 2010; Teddlie & Tashakkori, 2009). All these mixed methods designs can be used for the combination of qualitative case studies and single case experiments or for the integration of qualitative case study components and single case experimental components. It is beyond the scope of this article to explore all possibilities, critiques, and alternatives (see, e.g., Maxwell, Chmiel, & Rogers, 2015; Maxwell & Loomis, 2003), but we will focus on one recently published mixed methods single case study that demonstrated the versatility of integrating qualitative and quantitative research components at the case level.

Van Ness et al. (2017) gave an example of a mixed methods sequential single case study about the effectiveness and acceptability of a humming intervention for an older person with moderate dementia. The aim of the intervention was making caregiver feeding more effective and making patient eating more enjoyable. Van Ness et al. (2017) opted for a mixed methods approach for this single case study because they wanted to test the effectiveness of the intervention while attending to the preferences and concerns of the person involved. Their research design is depicted in Figure 1.

OPEN IN VIEWER

Figure 1.

Schematic representation of the mixed methods sequential design combining qualitative case study and single case experimental components in the example of Van Ness et al. (2017).

Figure 1 shows that the research design was constructed in three stages, with qualitative case study components in the first and third stage, and a single case experiment in the second stage. In the first stage, they interviewed the person in an open-ended qualitative fashion to determine his musical preferences. In a second stage, a caregiver feeding the person was videotaped for 30 noon mealtimes over the course of a calendar month: 12 consecutive days without intervention, followed by 18 consecutive days with the humming mealtime intervention. The video observations were subsequently converted into measurements using the Edinburgh Feeding Evaluation scale, with smaller scores indicating fewer feeding/eating problems. The starting day of the intervention was randomly determined, with a restriction that there should have been at least 5 days in each phase. The humming intervention took into account the musical preferences determined in the first stage of the study.

Data analysis consisted of presenting the data in tabular format, calculation of descriptive statistics, and the performance of a randomization test. Results showed that the average score during the preintervention phase was 14.3 and the average score during the humming phase was 9; which represented a statistically significant reduction in problems according to a 5% level randomization test.

In a third stage, a qualitative interview was conducted to inform the degree of belief that the researcher should give to assessing the weight of evidence against the null hypothesis that there was no difference in means between the preintervention and intervention phases. Questions were asked regarding key experimental assumptions and regarding the quality of the experience of eating, about how well the musical preferences were satisfied, and related qualitative questions.

Van Ness et al. (2017) explicitly framed their approach in the pragmatic tradition, with abundant references to William James, John Dewey, and Charles Sanders Peirce, and with the latter as the main guide:

Whereas Van Ness et al. (2017) framed their approach as “especially appropriate for early-stage research interventions intended to generate explanatory hypotheses” (p. 342), we want to join Stake (1995), Eisenhardt (1989), and Yin (2014) in their advocacy for the valid use of a case study to answer specific research questions concerning the particular case under study, to develop theory, or to test theory (see section “What Kinds of Knowledge Can MMSCR Produce?”).

Variations on the basic design as in Figure 1 can easily be constructed. Variations are possible within each stage or between stages. For example, within the single case stage it would have been easy to gain design power by randomly alternating days with and without humming if it is assumed that the intervention has an immediate, concurrent, and temporary effect (Manolov & Onghena, 2017; Onghena & Edgington, 1994). If a design with one large intervention phase is preferred, it would have been appropriate to include a posttreatment phase (Barlow et al., 2009; Gast & Ledford, 2014; Kazdin, 2011). Other choices can be made between stages: Stages can be in different orders (e.g., QUAN → QUAL → QUAN) or stages can be added (e.g., QUAN → QUAL → QUAN → QUAL).

Because qualitative case studies and quantitative single case experiments are both located at the single case level, we reckon that there is substantial potential to design fully integrated mixed methods studies (Creamer, 2018; Fetters & Molina-Azorin, 2017). For example, the single case experimental design components could be completely nested within a qualitative case study, in the sense that the experiment is just part of a larger case study, and that qualitative observations and interviews with peers continue while the experiment is running. To emphasize this potential for integration, we prefer the research stages as proposed by Nastasi, Hitchcock, and Brown (2010). Figure 2 provides a six-stage summary of these stages. In the first stage, the context is given: the theories, research, practice, policy, and worldviews that guided the research. In the second stage, the purpose of the study is delineated and the specific research questions are formulated. The third stage pertains to the data collection: sources, instruments, duration, and number of measurement occasions. The collected data are analyzed, evaluated, and interpreted in the fourth and fifth stage, respectively. Finally, in the sixth phase the study is reported and the conclusions loop back to theory, research, practice, and/or policy (and makes part of the first stage of follow-up studies). The most important elements of this design are the two-pointed arrows at each stage. These arrows represent the potential for mixing the qualitative and quantitative components at each stage of the research process.

OPEN IN VIEWER

Figure 2.

Six-stage summary of mixed methods design, proposed by Nastasi, Hitchcock, and Brown (2010).

Trustworthiness and Validity of MMSCR

Because MMSCR is not (yet) part of a well-established methodological canon, researchers who consider this approach may have to anticipate skeptical readers and fellow scientists. Therefore, we recommend paying special attention to trustworthiness and validity issues while planning, conducting, and reporting the study. Before embarking on an MMSCR study, it is wise to reflect on the methodological quality of the single case experimental component, the qualitative case study component, and on the added value of the mixed methods approach.

First and foremost, it is crucial to check whether the proposed approach is aligned with the research questions and the envisioned results and implications. Although the arguments for trustworthiness and validity will differ from study to study, it is recommended practice to match the research design and analysis to the research questions and ambitions (Creswell, 2013, Creswell & Plano Clark, 2018).

Second, the single case experimental component should conform to current standards and guidelines for designing and analyzing single case experiments. An example of standards and guidelines for the domain of special education can be found in Kratochwill et al. (2010, 2013). These standards and guidelines assign the categories of “Meets Standards,”“Meets Standards With Reservations,” and “Does Not Meet Standards” to each single case experimental design. Furthermore, outcome measures that meet the design standards (with or without reservations) can be categorized as demonstrating “Strong Evidence,”“Moderate Evidence,” or “No Evidence” (see also Smith, 2012).

Third, the qualitative case study component can bolster the trustworthiness and validity by incorporating the participant’s sense making and perspective on the effect of the intervention, by the addition of an in-depth description of the characteristics of the participant and the context (and thereby facilitating transferability), by guiding the choice and interpretation of the intervention and outcome measures (as demonstrated in the study by Van Ness et al., 2017), and by critically examining the researcher’s role using qualitative techniques such as member checks, negative case analyses, audit trails, and prolonged engagement (Hitchcock et al., 2010; Nastasi & Schensul, 2005).

Fourth, the added value of the mixed methods approach can be demonstrated by concentrating on the use of multiple sources of evidence as a tactic for addressing concerns regarding trustworthiness and validity (Heyvaert, Hannes, Maes, & Onghena, 2013). As Kitchenham (2010) concluded “The combination of qualitative and quantitative techniques enhances legitimation as the qualitative analyses involve descriptive precision and the quantitative analyses ensure numerical precision” (p. 562). If the single case experimental component is properly conducted, then the addition of a qualitative case study component can only increase trustworthiness and validity. MMSCR borrows strength from both quantitative and qualitative research traditions to build a convincing case.

Future Directions for MMSCR

Throughout this article, we consecutively discussed the historical roots and the distinct nature of MMSCR, the kinds of knowledge MMSCR produces, its philosophical underpinnings, the kinds of MMSCR that can be found in the literature and in current practice, and illustrate how MMSCR can be conducted and evaluated in terms of trustworthiness and validity. To address complex and multifaceted research questions that involve both quantitative and qualitative empirical research components related to a single case, researchers can combine elements from the single case experimental and qualitative case study methodologies in a mixed methods single case study. Mixed methods single case studies can be used to answer research questions that relate to particular cases, but also to develop and to test theories from individual cases.

In this section, we want to point to three remaining methodological challenges for MMSCR. A first challenge pertains to the development of a critical appraisal tool for MMSCR. The section on “Trustworthiness and Validity of MMSCR” already presented some general considerations and principles that are crucial for the evaluation of quality and evidence in MMSCR, and the next step would be to organize these considerations and principles in a critical appraisal instrument that allows a rating of specific studies in terms of quality and validity. Solid building blocks for such an instrument can already be found in the methodological literature: (a) in the guidelines for conducting single case experiments and n-of-1 trials provided by Tate et al. (2008, 2013, 2016a, 2016b), Shamseer et al. (2015), and Vohra et al. (2015); (b) in the discussion of the quality of qualitative research in Burns (1989), Mays and Pope (2000), Morse, Barrett, Mayan, Olson, and Spiers (2002), and Hannes, Lockwood, and Pearson (2010); and (c) in frameworks for evaluating the integrative quality of the mixed methods approach (Heyvaert, Hannes, et al., 2013; O’Cathain, 2010; Pluye, Gagnon, Griffiths, & Johnson-Lafleur, 2009). In our opinion, the availability of a critical appraisal tool geared toward MMSCR would be an asset for researchers who consider this approach. The development of such an appraisal tool would be an interesting direction for further methodological research.

A second challenge relates to the team work that is involved in conducting mixed methods single case studies, analogous to the team work that is involved in many mixed methods studies in general (see, e.g., O’Cathain, Murphy, & Nicholl, 2008). Various people might be involved when designing, conducting, validating, and reporting on a mixed methods single case study. First, methodologists as well as researchers with substantive expertise on the phenomenon of interest might be involved in order for the mixed methods single case study to be methodologically sound and trustworthy as well as topically rich and well-grounded in the existing knowledge base on the phenomenon. Second, methodologists specialized in qualitative case studies as well as methodologists specialized in single case experiments might be part of the research team. Third, stakeholders and practitioners might enrich the research team with their real-life experience with the phenomenon of interest and with the case(s) included in the study that represent the phenomenon. They might show themselves indispensable members of the research team in for instance the formulation of relevant research questions and hypotheses, and critically interpreting the study’s findings (Torrance, 2012).

In the context of health sciences research teams, Curry et al. (2012) suggested several guiding principles to deal with differences within the team, to trust the other members of the team, to create a meaningful group and develop a common language, to handle conflict and tension, and to treat leadership as a role rather than an individual characteristic, balancing issues of relationship and task. In addition, Bowers et al. (2013) pointed out that funding agencies and teams should take into account the additional cost of coordination, schedule dedicated time for planning and collaborating across researchers, fund support for in-person meetings, and make optimal use of resources. Furthermore, as Maxwell (2016) observed, sometimes the close involvement of the complete team with the collection and analysis of both the qualitative and quantitative data, rather than seeing these as separate “strands” of the research, might be beneficial for the integration of the two approaches. We would speculate that the integration of the single case experimental approach and the qualitative case study approach is function of the quality of collaboration, communication, and integration of the team, but in the end, this is an empirical question. From our own experience, the coordination and integration of divergent assumptions regarding research, different worldviews and philosophical orientations of the different team members is an important consideration. It would be interesting if future research addressed this challenge of working in teams when conducting MMSCR.

A third challenge for further research is the application of mixed methods research synthesis (Heyvaert, Hannes, & Onghena, 2017; Heyvaert, Maes, & Onghena, 2013) for multiple case studies and single case experiments (see Stake, 2006). Cases can be recruited and integrated in one encompassing MMSCR study or cases can be (theoretically or literally) replicated, possibly reported by different research teams and synthesized based on the reports (Yin, 2017). A decision on the number of cases to include in such a multiple MMSCR study depends on a number of factors that may be given different weights in different research domains and for different topics: the size and stability of the phenomenon under investigation, the availability of cases, the research resources, the feasibility, standards, and guidelines within research domains, and the purpose of the replication (replication until convergence and saturation is reached, falsification or corroboration, aggregation and/or generalization). Examples of sequential analysis and statistical power analysis guiding the decision on the number of cases to include, can be found in Kuppens, Heyvaert, Van den Noortgate, and Onghena (2011) and Heyvaert et al. (2017). The use of meta-analytical tools for MMSCR, and the use of mixed methods research synthesis in particular, might be particularly promising to reach the goal, set by Van Ness et al. (2017): the development of “research programs leading to replicable and generalizable knowledge” (p. 344).