Sampling in design research: Eight key considerations

How a research team deﬁnes their study sample can be decisive in shaping impact on both practice and theory. However, sampling in design research faces several major challenges, including diverse terminology, limited prior literature, and lack of common framework for discussing sampling decisions. We address these challenges by bringing together guidance from across related research ﬁelds as well as cross-referring to examples from published design research. We oﬀer a structured process for sample development and present eight key sampling considerations. The paper contributes to research method selection, development, and use, as well as extending discussions surrounding knowledge construction, standards of reporting, and design research impact.

H ow a research sample is defined forms a key element in establishing the scope of a study and in shaping its potential impact on both theory and practice.Sampling affects both scientific rigour and, in many cases, the perceived value and practical impact of research (Douglas, Noble, & Newman, 1999;Wacker, 2008); and thus forms a key link between considerations related to knowledge construction, research methods, and research impact (Cash, Daalhuizen, & Hay, 2022).However, these key considerations are often left implicit or under-acknowledged in both setting and reporting the work.
Design research (the study of design) brings together a diverse set of philosophies and approaches, including constructivism, participatory research through design, and qualitative and quantitative theory and methodology development.While this diversity gives rise to much of the richness within the field it also means that any methodological discussion must be carefully contextualised.Further, overarching methodologies such as the Design Research Methodology (DRM) by Blessing and Chakrabarti (2009) or Design Science (Hevner, 2007), require detailed support for specific aspects of the research process in order to provide a robust body of methodological knowledge.Thus, we focus on qualitative and quantitative studies engaged with the scientific theory development cycle, i.e. contributing to theory building and/or theory testing (Cash, 2020;Wallace, 1971, p. 18).This provides a foundation for interpretation and adaptation of methodological discussions found in a similar context in related fields, as well as implications for other research approaches sharing similar concerns.
The methods relevant to this context range from qualitative interview and case studies (where sampling can relate to, for example, the selection of specific cases, as in the work of Crilly and Moros¸anu Firth (2019)), to smaller scale quantitative or mixed method studies (where sampling can relate to the specific definition of study participants, as in the study by Nelius, Doellken, Zimmerer, and Matthiesen (2020)), or to large scale quantitative studies (where sampling can be used to reflect whole segments of a population, as in the work of Graff, Meslec, and Clark (2020)).While each method emphasises different aspects of sampling, there are three key challenges to their adaptation and application in design, which we aim to address in this paper.
First, there is a high degree of terminological diversity in the reporting of studies and across the literature.This can contribute to difficulties in study interpretation as well as a perception of siloing between studies (Le Dain, Blanco, & Summers, 2013), such as in experimental versus real world settings (Ball & Christensen, 2018;Crilly, 2019a).Thus, there is a need to clarify relevant sampling terminology.
Second, there has been little specific discussion of sampling terminology or considerations in the design research literature.Hence, the reporting of sampling decisions is often implicit and can appear to be something of a methodological 'black box'.This hampers efforts to develop methodological rigour, as well as the examination of potential synergies across design research approaches.
Finally, most relevant sampling guidance is found in the social science and psychology literatures.This typically emphasises scientific concerns and neglects interactions with practitioners and the design framing.This can lead to a perception of conflicting directives, hindering application to design research.Thus, there is a need to set out a framework to facilitate more structured and detailed discussion and development of sampling considerations in this context.

Design Studies Vol 78 No. C Month 2022 1 Approach and definitions
Given the above need, we take a first step towards unpacking key sampling considerations.Importantly, while some of these considerations are relevant to a range of research approaches, we do not provide a universal guide to sampling in design research.Further, due to the need to synthesise and abstract sampling knowledge it is necessary to adopt a theory-driven approach to identifying sampling considerations.Specifically, we develop an initial set of considerations adapted from related fields.This establishes a foundation for discussion and a potential point of departure for future review, metaanalysis, and refinement, as well as examination of intersections between design and other fields (McComb & Jablokow, 2022).This follows similar work in related fields where synthesis of methodological knowledge provides an essential basis for subsequent review and analysis (Onwuegbuzie & Leech, 2007), as well as the approach taken by Cash (2018Cash ( , 2020) ) in his synthesis and meta-analytical review of theory-development in design research.Thus, our approach has two components: i) to suggest an initial terminology, literature, and framework for discussing sampling in this context, and ii) to contribute to a broader discussion of methodological and research synergies across design research.
To this end, we survey, distil, and adapt sampling guidelines from over twenty research fields, all with a common foundation in theory development.These distilled guidelines form the basis for each of our considerations.To concretise these considerations and bring them into the design research context, we provide illustrative examples drawn from the Design Studies journal.Importantly, due to the formative nature of this area of design research methodology we do not carry out a systematic review, rather we contextualise and exemplify our considerations.However, before we start, it is first necessary to establish some basic definitions relevant to sampling discussions in this context (Robson & McCartan, 2011, p. 276): population: the total set of relevant cases (e.g.all designers working with digital technologies); and sample: a subset of the population (e.g.designers working with digital technologies in one company).
Sampling in this context, thus reflects an interplay between population and sample, forming a bridge between theoretical and methodological discussions rooted in the theory development cycle.

Sampling: eight key considerations
Design research studies are often motivated by challenges in theory or practice and conclude with contributions that close knowledge gaps, and subsequently enable the next generation of research and insight.This corresponds to a Sampling in design research typical progression from qualitative theory building to more quantitative theory testing in an iterative cycle (Cash, 2020;Wallace, 1971, p. 18).Our purpose and the nature of this contribution provides a common foundation for sampling decisions (Wacker, 2008).Thus, we illustrate a double-loop sampling process, Figure 1.
The inner loops in Figure 1 comprise: a definition loop used to describe and scope a sample, linked to a refinement loop pointing to potential combinatory strategies.However, before entering these loops a researcher must consider both the scientific and practice concerns that define the design framing of the research.Based on the balance between scientific and practice concerns, a researcher might enter the definition loop at any point.For example, a researcher focused on testing general hypotheses, might start at 'theoretical framing' thereby needing the most statistically representative sample possible, prioritising scientific concerns; while a researcher focused on theory building, might also start at 'theoretical framing', but with the aim to identify a particularly interesting sample, balancing scientific and practice concerns.Finally, a researcher focused on issues of application in collaboration with a specific company, might start at 'sample size', prioritising practice concerns, with a secondary reflection on scientific concerns.Whatever the starting point, researchers should examine all considerations to fully delineate their sample.

Framing design research: scientific and practice concerns
Essential to design research is its potential for both scientific and societal impact.Therefore, before entering the double-loop process it is vital to establish the design research framing in terms of both good scientific conduct and link to practice.These form foundational considerations that inform how all other considerations are interpreted and applied.

Good scientific conduct and ethical appropriateness
Key to any research endeavour is a firm grounding in good scientific conduct and research integrity.This positions scientific and ethical appropriateness as the first foundational consideration; which concerns both the integrity of the research and the individuals involved, thereby ensuring sample members are fairly treated and protected (Kitchenham et al., 2002;Onwuegbuzie & Collins, 2007).Focusing on research integrity, this includes ensuring the sample represents the intended population, management of possible biases, and transparency in research decision making, reporting, and communication of limitations (Onwuegbuzie & Collins, 2007).Any mismatch between the proposed sample and population can introduce bias (e.g.overrepresentation of familiar and research-aware participants or excluding portions of the sample or population).A useful and relevant overview can be found in the American Educational Research Association (2011) code of ethics.Particularly critical for design researchers, is the involvement of, and engagement with, practitioners, where there might be ongoing collaborations or limited access to possible sample members.These concerns span the whole research process and can appear to be in tension with practice concerns, thus necessitating careful review and reporting.For example, Ventrella, Zhang, and MacCarty (2020) provide some discussion of ethical concerns impacting their sample.However, they also highlight the need for greater discussion of ethical considerations in general.Given the limited scope here, we recommend the above noted code of ethics, as well as the works of Rosenthal (1994), Robinson (2014), andCreswell (2012).
Consideration 1: Scientific good conduct: what ethical concerns are relevant?

Ensuring impactful research: linking to practice
Key to any design research endeavour is balancing scientific and practice concerns.This positions linking to practice as the second foundational consideration; which puts an emphasis on connecting sampling considerations to a study's framing in design, as well as understanding how sampling can support communication, involvement, and impact in practice (Sjøberg et al., 2002).For example, if a researcher aims to impact practice by helping improve organisational processes in a collaborating company, then their research must foster longeterm relationships, company buy-in and (hopefully enthusiastic) engagement, as well as other aspects of organisational change management Sampling in design research (Chakrabarti & Lindemann, 2016) not necessarily required by the contribution to knowledge (Section 2.2.1).Further, design researchers examine a wide range of e often messy e design situations using a variety of approaches and perspectives, which influence where to enter the sample definition loop (Figure 1), as well as how the specific considerations should be approached.
Here, practitioners can have diverse roles in, and expectations for, research impact, which must be carefully managed in the context of theory development studies.However, impact can be as simple as providing structured feedback and research will often receive greater attention if the impact on practice is explicit, as discussed by Zielhuis, Sleeswijk Visser, Andriessen, and Stappers (2022).As such, design framing and engagement with practitioners is essential (Cooper, 2019;Easterbrook, Singer, Storey, & Damian, 2008).
Consideration 2: Design framing: what type of impact on practice do you hope to achieve?

Definition loop
There are five main considerations in this loop: theoretical framing, scope, generalisation approach, sample schema, and sample size.

Theoretical framing
In the theory development context, appropriateness of a population/sample e both qualitative and quantitative e is typically defined with respect to its intended contribution to knowledge (Wacker, 2008).This influences choices for population/sample, as well as the degree of granularity expected in their definition (Lynch, 1999;Wacker, 2008).
In general, the number of potential factors in any situation far exceeds reporting or reasonable conceptualisation.Therefore, definition of population/sample is typically limited to only those factors that affect the concepts under study (Lynch, 1999;Wacker, 2008) and is a key part of the coupling between the research question and method (Goldschmidt & Matthews, 2022).This gives rise to a 'previous literature convention' (Wacker, 2008), i.e. if a factor has not been related to the theory of interest in a study then it is not required for defining the population/sample.For example, there are multiple factors linked to designer expertise, including diverse skills and domain knowledge.As such, while Kavakli and Gero (2001) are able to robustly differentiate novice (second year students) and expert (more than 25 years of experience) designers' sketching processes, the lack of granularity in the sample definition means that more detailed interpretation of what specific factors determine these differences is impossible.Thus, the more mature the body of theory, the more specifically relevant factors can be identified and defined (Cash, 2018;Melnyk & Handfield, 1998).

Design Studies Vol 78 No. C Month 2022
Cash (2018) describes levels of theory development in a cycle from 'discovery and description' to 'extension and refinement'.In the early phases of this cycle, where specific factors are typically ill-defined, research focuses on rich, reflective description of phenomena and context.This places an emphasis on being able to explain the significance of a sample in context.For example, Carlson, Lewis, Maliakal, Gerber, and Easterday (2020) link their sampling decisions to their theory building aim, delineating the scope of their work and limiting the granularity of their sampling criteria.Subsequently, they use theoretical sampling to refine, develop, or refute insights from their initial rich data.Together, these support their qualitative description of metacognitive processes underpinning novice design work.
In the later phases of this cycle, where factors are typically well defined, research focuses on specific definition and control.This places an emphasis on being able to define and isolate the key factors linking a population and sample.Research in this context typically employs quantitative studies that can only function when the number of variables can be limited (Easterbrook et al., 2008;Melnyk & Handfield, 1998).For example, Vandevenne, Pieters, and Duflou (2016) detail a range of general and specific aspects of domain knowledge in their sample, which might impact the outcomes of their experiment.As such, they can limit the number of variables introduced by the prior knowledge of their sample participants, and thus more clearly elucidate their Sampling in design research hypotheses.Table 1 summarises the maturity of theory in terms of this cycle (Cash, 2018).
Consideration 3: Theoretical framing: where in the theory-building/theorytesting research cycle is current knowledge?

Generalisability
One of the major challenges in discussing 'generalisability' is that the term has various meanings (Lee & Baskerville, 2003;Onwuegbuzie & Leech, 2007), and, as typically used, conflates two dimensions that more broadly define the domain of a contribution (Wacker, 2008): The extent to which a contribution applies to existing populations ("who" factors such as nationality, organisation, or experience) Abstraction: The extent to which a contribution is bounded by time and space ("when" and "where" factors such as location, timing, or scale) Theory that is both general and abstract, applies to all people at all times in all places, and is often referred to as general or universalistic (Bello, Leung, Radebaugh, Tung, & Van Witteloostuijn, 2009;Gainsbury & Blaszczynski, 2011;Stevens, 2011).However, theory that applies to particular individuals in an explicit time and place is referred to as specific or particularistic (Stevens, 2011).When an intended contribution is universalistic, sampling becomes more about verifying the supposed universality when considering well defined factors.For example, Blizzard et al. (2015) use a stratified random sample across college students in the U.S., in order to evaluate universalistic design thinking traits.In contrast, the more specific a contribution, the greater the priority on being sensitive to emerging contextual factors that might impact the findings.For example, Crilly and Moros¸anu Firth (2019) focus on a small set of specific cases in order to develop rich contextualised insights.Thus, before considering the sample, it is necessary to scope the population.Hence, an initial consideration is definition of a population based on desired generalisability and abstraction (Lee & Baskerville, 2003;Onwuegbuzie & Leech, 2007;Robinson, 2014).
Consideration 4: Scope: how general and abstract is the intended contribution?
Given a population, the sample depends on the variables and relationships in focus as well as the planned generalisability and abstraction of the contribution (Bello et al., 2009;Lynch, 1999;Wacker, 2008).Numerous metaanalyses have demonstrated that this link must be evaluated on a variableby-variable, relationship-by-relationship basis (Dasgupta & Hunsinger, 2008;Peterson, 2001).For example, in the design fixation literature, there are a wide range of variables and relationships that can vary substantially across studies (Crilly, 2019b;Vasconcelos & Crilly, 2016).A sample is thus defined based on the intended applicability of a contribution across the wider population (Robinson, 2014;Tuckett, 2004).This leads to a number of approaches, both qualitative and quantitative, each with implications for sampling (Onwuegbuzie & Leech, 2007).Thus, the second consideration here is the identification of an approach based on the desired generalisability and abstraction within the population, as outlined in Table 2 (Polit & Beck, 2010;Stevens, 2011;Wacker, 2008).
Consideration 5: Generalisation approach: what type of generalisability and abstraction is desired in the population?
With respect to Considerations 4 and 5 it is pertinent to contrast the generalisation approaches of qualitative and quantitative studies.For the qualitative, we highlight the work of Crilly and Moros¸anu Firth (2019) who use thematic analysis to distil insights from select, in-depth cases.Although not explicit in their text, they employ analytical generalisation (Table 2), developing their insights by moving between their results, prior literature, and theory.In contrast, Graff et al. (2020) quantitatively examine a scale for evaluating perceived analogical communication.They test the robustness of this scale across three large samples to develop external statistical generalisation (Table 2), using a carefully defined population composed of graduate students.

Sampling schema and size
How the sample will be collected (sampling schema) and size link the prior considerations to research method (Onwuegbuzie & Collins, 2007).Schema are split into probability and non-probability (Daniel, 2012;Onwuegbuzie & Sampling in design research Collins, 2007), corresponding to statistical and analytical/cases-to-case generalisation (Table 2) (Onwuegbuzie & Leech, 2007;Polit & Beck, 2010).For example, Crilly and Moros¸anu Firth (2019), use a number of criteria to develop a purposive sample of cases with specific characteristics of interest (e.g. the development of radically new, physical, commercial products, with potential for in-depth data access).This constrains the scope of their investigation and provides a basis for developing rich, in-depth insights.In contrast, Blizzard et al. (2015) use a probability sample in order to develop a general understanding of design thinking traits across universities, and student groups in the U.S.This serves to delineate the scope of their claims and provide a concrete basis for identifying sample participants.Probability samples use mathematical rules to ensure that everyone in a population has the same chance of being included in the sample, while non-probability samples include individuals based on a range of criteria.Non-probability can thus be further decomposed as (Creswell, 2012;Daniel, 2012): purposive (purposeful, judgemental, selective or subjective): based on the characteristics of the population and research purpose (considerations 3-5), quota: based on a stratified quota, and convenience: based on availability.
An overview of possible schemas is given in Figure 2 (Onwuegbuzie & Collins, 2007;Teddlie & Yu, 2007).Critically, each schema has implications for both the interpretation of the study and results as well as the practical identification of participants e as highlighted by Hay, Duffy, Gilbert, and Grealy (2022) in the design cognition context e yet few design research studies explicitly report the specific schema adopted.
Consideration 6: Sample schema: what schema fits your theoretical framing, generalisation approach, and research method?
Sample size is based on research purpose, generalisation approach, and research method.As such, both qualitative and quantitative research methods can draw on small or large samples depending on the other considerations in Figure 1, as well as the type of data collected and analysis approach used (Onwuegbuzie & Collins, 2007;Sandelowski, 1995).A number of authors provide minimum size guidelines for specific research methods, for example interview studies with >12 participants, one-tailed experiments with >21, and case studies with >4, summarised in Figure 3 (Onwuegbuzie & Collins, 2007).However, it is typical that qualitative samples should be large enough to support saturation (further data collection would only confirm the results already identified) and small enough to deliver rich insight (Onwuegbuzie & Leech, 2007;Sandelowski, 1995;Teddlie & Yu, 2007).This trade-off between breadth and depth is clearly illustrated by Crilly and Moros¸anu Firth's (2019) deliberate constraint of their sample to only three cases, which they are able to Design Studies Vol 78 No. C Month 2022 analyse in great depth.Similarly, quantitative samples should typically meet the statistical requirements of the generalisation approach, such as significance and statistical power.This is key when comparing experiments and their effects, as illustrated in the meta-analyses of design fixation by Sio, Kotovsky, and Cagan (2015) and Vasconcelos and Crilly (2016).While the numbers given for different approaches in Figure 3 are guidelines only, and sample size should always be justified with respect to the specific study, they provide an important point of reference and help normalise sample size discussions across studies within a field.For example, Nelius et al. (2020) use an experimental study in a small-scale theory building mode, which can lead to confusion if Sampling in design research experimental and quantitative are conflated.Hence they explicitly contextualise their sample size discussion with respect to the guideline of Onwuegbuzie and Collins (2007).As such, this is key to the transparent positioning of the study in the wider research context, as well as in understanding the scope of the work.
Consideration 7: Sample size: what size fits with your generalisation approach and research method?

Refinement loop: sampling strategy
Key to understanding the value of studies in context e particularly with respect to their contribution to the wider qualitative, quantitative theory development cycle e is that no sample can fulfil all possible research demands.Therefore, when looking beyond a single study, it is necessary to consider how combinations of samples can be used to mitigate individual weaknesses and maximise collective strengths (Onwuegbuzie & Leech, 2007;Teddlie & Yu, 2007).This mirrors discussions that illustrate the strengths of combinatory mixed methods research designs (Cash & Snider, 2014;Hanson, Creswell, Plano Clark, Petska, & Creswell, 2005;Onwuegbuzie & Leech, 2006;Tashakkori & Teddlie, 2008).For example, when dealing with more constrained or quantitative studies it is often necessary to prioritise either internal or external validity (Gainsbury & Blaszczynski, 2011) (i.e.internal validity: the extent to which evidence supports conclusions (usually causal) within the context and integrity of a specific study; external validity: the extent to which conclusions from a specific study apply to other contexts with implications in the wider world).This leads to sampling strategies such as that by Cash, Hicks, and Culley (2013), who combine three studies: i) a qualitative study with a practitioner sample fully embedded in context, ii) a semi constrained study with a practitioner sample emphasising external validity, and iii) a semi constrained study with a student sample emphasising internal validity.While each study is limited in isolation, together they provide a rich picture linking real world practitioner and laboratory-based student practice.Another example of effective sample combination can be found in the work on Ventrella et al. (2020), who employ multiple samples in order to develop and evaluate a novel sensor system.Here, the combination of samples provides several insights and allows for a progression in research objectives from understanding the specific challenge to evaluating the final usability of the sensor.There are many recommendations for combinatory sampling, ranging from specific schema analogous to those outlined in Figure 2, to large-scale strategies that shape the whole research design, summarised in Table 3.
Consideration 8: Sampling strategy: (when using multiple studies) what combination of sample schema provide the best balance with respect to all prior considerations?

Limitations and further work
Before discussing implications, it is necessary to highlight two main limitations of this work.First, sampling is a broad topic, with common considerations but also many field-specific adaptations.While our work has focused on key considerations for one aspect of design research, further work is needed to i) examine the specific instantiation of these considerations across the design

Implications and reporting
The considerations detailed in Section 2, draw a link between intended contribution to knowledge and specific research sample.In Table 4 we connect each consideration to recommended actions, relevant resources in this article, and suggested reporting, as well as providing examples of current good practice in design.Here, we reiterate the foundational nature of considerations 1 and 2 in establishing the design framing, and subsequently informing the interpretation and application of all other considerations.Reporting the considerations explicitly as exemplified in Tale 4 is key to supporting meta-analysis or literature review (Chai & Xiao, 2012;Vasconcelos & Crilly, 2016), and is also key to discussions of method appropriateness (Goldschmidt & Matthews, 2022), limitations, and research quality (Prochner & Godin, 2022).For example, Graff et al. (2020) provide a careful discussion of the strengths and weaknesses of their student sample, and reflect on the potential transferability of the insights to other populations, based on the theoretical variables at play and how they are expressed in different groups.
In addition to Table 4, the considerations also point to two implications related to typical sampling discussions in the theory development context.First, the strengths and weakness of samples employing, for example, students versus practitioners, can be potentially leveraged via variable-by-variable generalisation (Section 2.2.2), and sample combination with respect to the wider theory development cycle (Section 2.3).This aligns with results of multiple meta-analyses from across domains, including social science (Peterson, 2001), information and management (King & He, 2006;Schepers & Wetzels, 2007), and software engineering (Hannay, Dyb a, Arisholm, & Sjøberg, 2009).This leads to a summary of student sample pros and cons (Table 5).

Sampling in design research
Second, scientific and practical concerns need to be combined to effectively balance the needs of impact on theory and practice (Section 2.1).This leads to three main insights: i) in most cases sampling decisions are driven by research purpose and theoretical framing; ii) mixed methods may be considered in any context; iii) practice may be engaged by design researchers to allow for feedback and to increase research credibility, transparency, and Optionally form a sampling strategy and complete the definition loop for each sample Table 3 In method introduction and sample definition "we used a design science approach that integrated rapid ethnographic and sensor-based methods in a multi-site case study and roughly followed the stages of the design process.
Table 2 shows the overall progression of research phases, research goals, and methods used."(Ventrella et al., 2020, p. 88) Design Studies Vol 78 No. C Month 2022 comprehensibility, particularly if evidence is drawn from student samples (Sjøberg et al., 2002).

Conclusion
Effective sampling forms an essential element in developing design research that impacts both theory and practice.Here, we have brought together guidance from diverse research fields, to take a first step towards developing a terminology, literature, and framework for discussing key sampling considerations in design research.The double-loop sampling process and associated considerations (Figure 1) are a first of its kind in design research, and constitute a call to action, highlighting the need for further examination and theorising around the impact of sampling on design research claims.In doing so, we aim to contribute to a broader discussion of methodological and research synergies across design research approaches.
Ultimately, any study will inevitably have limited insight, and thus transparency in the reporting of methodological and sampling considerations is increasingly important, particularly in the diverse context of design research.Thus, we encourage researchers to report these considerations in their papers, to make justification, assumptions, and limitations explicit.

Figure 1
Figure1The double-loop sampling process, with consideration numbers

Figure 2 A
Figure2A typology of sampling schema split into probability and non-probability, with branching sub-types and summary definitions; numbering denotes alternative names: 1 Random, 2 Deviant, Outlier, 3 Heterogeneous, 4 Complete collection, 5 Theory-based, Concept, 6 Chain, Network, Reputational, 7 Volunteer; )denotes that the schema is implemented after data collection has begun.

Figure 3
Figure 3 Sample size and research method

Table 1
The theory-building/theory-testing research cycle and its implications for sample definition

Table 2 A
typology of generalisation approaches in terms of generalisability and abstraction Case-to-case transfer (see also transferability) Making generalisations from one case to another based on in-depth descriptions of each specific case that allow readers to make inferences about applicability to other contexts Low/Low focused on within sample understanding Internal statistical generalisation Making generalisations from selected participants to the overall sample.This requires mature theoretical definition of variables relevant to within sample variation

Table 3
An overview of combinatory sampling strategies ) elaborate and detail their application and reporting, and iii) explore commonalities and differences across design research philosophies and approaches.Second, sampling discussions in related fields have built on a critical interplay between conceptual development used to synthesise knowledge, and critical literature review used to interrogate this.Due the lack of prior discussion or guidance in the design literature, our work necessarily focused on conceptual development i.e. synthesizing the considerations.This points to the subsequent need for i) systematic review of sampling practices across design research, ii) meta-analysis of interactions between sampling, method use, and knowledge outcomes, and iii) evaluation of potential areas requiring further adaptation of or new considerations.Thus, despite these limitations, our work provides an important point of departure and comparison for further work, and mirrors developments in related fields where guidelines are iteratively proposed, interrogated, and refined.

Table 4
Summary of actions, resources, reporting guidance, and design research examples for the eight considerations.References are omitted from the quoted examples for clarity