Validity and validation in archetype analysis: practical assessment framework and guidelines

Archetype analysis is a promising approach in sustainability science to identify patterns and explain mechanisms shaping the sustainability of social-ecological systems. Although considerable efforts have been devoted to developing quality standards and methodological advances for archetype analysis, archetype validation remains a major challenge. Drawing on the insights from two international workshops on archetype analysis and on broader literature on validity, we propose a framework that identifies and describes six dimensions of validity: conceptual; construct; internal; external; empirical; and application validity. We first discuss the six dimensions in relation to different methodological approaches and purposes of archetype analysis. We then present an operational use of the framework for researchers to assess the validity of archetype analysis and to support sound archetype identification and policy-relevant applications. Finally, we apply our assessment to 18 published archetype analyses, which we use to describe the challenges and insights in validating the different dimensions and suggest ways to holistically improve the validity of identified archetypes. With this, we contribute to more rigorous archetype analyses, helping to develop the potential of the approach for guiding sustainability solutions.


Introduction
From early mentions in a wide range of disciplines, including philosophy, psychology and arts (Eisenack et al 2021), the concept of archetype has evolved into a novel approach used in sustainability science to generate knowledge across cases and places (Oberlack et al 2019, Eisenack et al 2021). Building on previous assessment frameworks providing comprehensive information to support policy making in the context of social-ecological systems (Ash et al 2010, Fallon et al 2021, archetypes are defined as recurrent patterns in variables and processes that determine the sustainability of social-ecological systems, identified by a diverse portfolio of methods . They can provide guidance on how to address sustainability problems in complex socioecological systems. Archetype analysis aims at striking a balance between two antithetic needs in sustainability sciences: to achieve well-contextualised solutions and to learn across (different) cases (Poteete et al 2010, Beach andPedersen 2016). To that end, archetype analysis provides guidance on how to navigate the middle ground between idiosyncratic approaches, where every case stands on its own, and nomothetic approaches, where universal solutions rely on oversimplified diagnostics (Ragin 2000, Basurto and Ostrom 2009. Given these advantages, archetype analysis has also been suggested as an approach for future research on biodiversity and ecosystem services in the recent IPBES report, especially for scenario development (IPBES 2016).
Due to the complexity of socio-ecological systems (Folke et al 2005, Pahl-Wostl 2007 and the interdisciplinary nature of sustainability sciences (Karlqvist 1999, Lam et al 2014, with its diverging and often conflicting scientific traditions (Nagatsu et al 2020, Boda 2021, no silver bullet exists that balances the benefits of situated knowledge with those of strong generalisation. For that reason, some researchers have gone to great lengths to test the reliability of their findings about archetypes, devising creative solutions that test the consistency of archetypes identified against real world observations (Sietz et al 2012, Vidal Merino et al 2019. Yet, validation in archetype analysis has often been conducted in rather adhoc ways, making it an analytical frontier for advancing the meaningfulness of archetype analysis. More meaningful archetypes, in turn, will make for better integration in decision-making on sustainable development .
The aim of the present paper is to advance archetype analysis by proposing an overarching approach to validation in archetype analysis, focusing on design, analysis and application. Relying on the outcomes of two expert workshops in 2019 and 2021 , Piemontese et al 2021a, we review different types of validity relevant for archetype analysis and integrate them into an overarching framework for archetype validation with six dimensions of validity for the corresponding archetype analysis steps. Subsequently, we apply the framework to 18 published archetype analyses, identify the strengths and weaknesses with respect to the six validity dimensions, and provide guidance for increasing validity.

The six dimensions of validity in archetype analysis
The validation of research results is a common challenge in science. Confounding factors, spurious relations, measurement errors all pose potential threats to research inference, calling for ways of ensuring the validity of research results. Adding to that, sustainability science is characterized by a problemdriven nature (Kates et al 2001, Clark and Dickson 2003, Pauliuk 2020 as well as by pervasive uncertainty (Messerli et al 2019, Clark andHarley 2020). This forces researchers to make compromises between tractability and data availability considerations on one hand, and what is meaningful and actionable for research stakeholders on the other hand. Further challenges are specific to archetype analysis, particularly the challenge of exploring the middle ground between idiosyncratic and nomothetic approaches . This pushes searches to engage with causality rather than simply detecting correlations (Mahoney 2008, Rohlfing andZuber 2021); doing so, however, inevitably forces researchers to engage with mixed methods, raising questions of internal, epistemological consistency for a coherent research design (Morse et al 2006, Seawright 2016, Gibson 2017. With these challenges in mind, archetype validation featured prominently on the agenda of the 2019 and 2021 workshops on archetype analysis , Piemontese et al 2021a. The workshops and follow-up discussions identified six dimensions which characterize the concept of validity in archetype analysis and correspond to six archetype analysis steps, i.e. actual stages of the scientific process involving considerations of validation. The validity dimensions described in detail below are: conceptual; construct; internal; external; empirical; and application validity (figure 1). Among these, conceptual validity corresponds to the step of problem framing, when research questions and framing are formulated. Construct validity relates to the step of attribute selection, which operationalizes the framing into quantifiable attributes. These two analysis steps constitute the foundation for designing the archetype analysis. Following the analysis phase includes the methods of analysis, which uses the attributes to reveal the archetypes, and the generalizability of outcomes. These two steps correspond to internal and external validity, respectively. The last phase concerns the application of the archetypes identified, which embraces the empirical relevance of the archetypes in the context of the issues relevant to society and considered in the framing steps (corresponding to empirical and application validity). The validity dimensions and steps may overlap and interact with each other as explained in later sections, given the circular nature of archetype analysis.

Conceptual validity
Conceptual validity refers to the salience of the research framing, including the formulation of the problem and research questions from the perspective of scientists as well as partners from policy and practice (when relevant). This is a key component of archetype validity since research problems and questions affect all other dimensions of validity at later stages of the knowledge production cycle (Locke 2012).
From a sustainability science point of view, archetype analysis needs to be rooted in research that is relevant to sustainability and guided by appropriate theoretical framing to produce scientificallysound knowledge (Magliocca et al 2018). At the same time, since sustainability science is a problem-driven research field, research questions and framing should build on problems relevant to society, and may also involve non-academic stakeholders in the problem formulation. Conceptual validity in archetype analysis is based on the integration of both scientificallysound research framing and problem-driven research questions.

Construct validity
Construct validity refers to the overall quality, fit, and representativeness of the variables chosen to transform the concepts studied into the attributes used to build the archetypes (Yin 2009, Gerring 2016. Attributes can be characteristics, variables, qualities, factors, or other properties chosen at an intermediate level of abstraction to achieve a balance between case-based validity and generalisation. As pointed out by Eisenack et al (2019), achieving a selection of attributes at an appropriate level of abstraction is a major challenge, making it difficult to provide strict selection procedures.
To ensure a coherent and plausible selection, attributes need to be well justified and properly linked with the conceptual framing and generalizability objectives. In the case of deductive research approaches, where researchers hypothesise based on existing theories or frameworks, the attributes should be properly related to the theoretical and conceptual framing (Maxwell 2010). For inductive research, where researchers look for patterns in the data to develop a theory, the attributes should be empirically grounded with the problem addressed (Blackstone 2012). For example, building the research hypothesis and selecting the attributes may be done through participatory approaches in collaboration with stakeholders (Yin 2009). Other examples of scientificallysound methods for attribute selection are metaanalysis of literature or multi-stakeholder approaches (Oberlack et al 2016).

Internal validity
Internal validity in archetype analysis refers to the appropriateness of the applied method (or combination of methods) for analysing the attributes within the research framing. This is particularly important in archetype analysis, as it can rely on diverse methodological tools from quantitative, qualitative, and mixed methods approaches .
The use of appropriate methods for the goals of the research-methods which represent the state of the art in validation of inferences and are backed by detailed documentation-improves the trustworthiness and replicability of the research approach and results.
For straightforward, single-method analyses, this dimension of validity can be addressed by using conventional method-specific validation procedures (Salciccioli et al 2016). For example, in quantitative archetype analysis methods, such as cluster analysis, there are standard methods to ensure robust cluster results (e.g. based on a consistency test, see Kok et al 2016, Sietz et al 2017, Segnon et al 2021, and to measure the closeness and representativeness of the clusters (Rocha et al 2020). In qualitative archetype analysis methods, quality standards are ensured, for example, via triangulation of information and sources, codebook construction, intercoder reliability checks and assessment of rival explanations (Yin 2009). However, in interdisciplinary or transdisciplinary studies, mixed-method approaches are commonly used to capture and analyse different types of qualitative and quantitative data and provide complementary information to gain insights. Although there are limited specific guidelines to assess internal validity in mixed-method analysis, these typically focus on an integrative framework where methods have a singular internal validation process, as well as the validation of meta-inferences generated through the combination of methods (Tashakkori and Teddlie 2008, Ihantola and Kihn 2011, Neudert et al 2019. Finally, to strengthen internal validity, method-specific techniques for uncertainty analysis should be used to define limitations and increase the reliability and usability of results in those cases where it is possible to do so (Bleijenbergh et al 2011).

External validity
External validity pertains to the generalizability of the claims emerging from archetype analysis, or the confidence in extending the results outside of the study sample (Yin 2009, Maxwell 2010, Magliocca et al 2018. This is a key challenge in archetype analysis, since an important purpose is to find an intermediate level of abstraction able to provide contextspecific generalisations across or even beyond the studied cases. This is particularly important for archetype analyses that aim for the transferability of solutions across contexts. Here, it is important to ensure that a study clearly states the domain of generalizability, both spatially and temporally . However, defining clear boundaries of generalizability is a challenging and debated task (Magliocca et al 2018).
To assess external validity, we suggest that two aspects must be present: (a) the generalizability boundaries need to be stated and discussed in the paper and guide the study design; and (b) at least one method needs to be used to validate whether the claims hold beyond the studied cases, if applicable, and fall within the same archetypes.

Empirical validity
Inspired by validation approaches used in modelling (Bossel 1994), archetypes prove to be empirically valid if they correspond to reported sustainability outcomes (e.g. food or livelihood security), and if the causal mechanisms leading to these outcomes are consistent and plausible (Sietz et al 2012, 2017, 2019, Vidal Merino et al 2019, Sterzel et al 2020, Segnon et al 2021. Empirical validity is important because the term 'archetypes' , rather than 'ideal types' , follows the detailed configurations of factors embodied in archetypes, as opposed to the extreme, hypothetical constructs embodied in ideal types . In both quantitative and qualitative approaches, a real-world check of the archetypes identified can be achieved through stakeholder-based assessments and workshops.

Application validity
Application validity refers to the usefulness of results for application by final knowledge users (Bossel 1994), for example non-governmental organisations or government officials. Application validity is a key dimension in sustainability research with the aim of fostering real-world actions that increase sustainability , for example, the scaling up of sustainable solutions (Coe et al 2014). Applications also include investigations of the relevance of archetypal patterns for improved understanding of sustainability outcomes, e.g. distribution of armed conflicts, complementing traditional statistical methods used to explore the occurrence or lack of armed conflicts (Sterzel et al 2014).
Scaling of sustainability solutions is a prime research frontier in archetype analysis to systematically support the transfer of knowledge and insights from one place to another , Eisenack et al 2021. The scalability of sustainability strategies rests on the assumption that similar social-ecological conditions translate into comparable interventions , Piemontese et al 2020. Archetypes prove to have application validity if the transferability of strategies to foster sustainable land use and other studied phenomena can be shown within a given archetypal pattern and within the stated domain of validity (Václavík et al 2016). This can increase the acceptance of findings derived from archetypes (e.g. potential strategies and solutions) by final users who may use these findings to address real-world problems, guide transformations, or further scientific knowledge and innovation (Bleijenbergh et al 2011).

From concept to application: an operational framework for validity assessment in archetype analysis
The six dimensions of validity are milestones of a highly-interconnected circular process, which are connected to corresponding analytical steps in archetype analysis, designed following the criteria presented by Eisenack et al (2019) (figure 1). As in other circular frameworks within sustainability science (Holling and Gunderson 2002, Pahl-Wostl 2009, Ostrom 2011, the circularity implies that there is neither fixed hierarchy nor a predefined 'first step' among the six dimensions. For example, addressing conceptual validity dimension in the problem framing is not always the first step of the validation process. The problem framing and research questions of the analysis may strongly depend on the needs and purposes of stakeholders (e.g. policy makers, local communities, and practitioners), which depend on the impact that the analysis is intended to produce. In this case, addressing application validity and not conceptual validity may be the first step of the validation process. Each step in this process is broad, in the sense that the analysis is interdisciplinary and often involves iterations between scientists and stakeholders.
Each analytical step can strengthen or weaken the significance of the other steps, hence the relevance of the validity dimensions. The relative significance can generate both trade-offs and synergies among dimensions and depends on the purpose (real-world application or theoretical contribution), approach (inductive or deductive), and the methodological lens (qualitative, quantitative, or mixed-methods; Davies andElder 2005, Koro-Ljungberg 2008). For example, a study designed for real-world application (thus identifying practical knowledge needs first) and consisting of a well-planned participatory approach may increase impact by targeting and framing problems as perceived by stakeholders, thus strengthening the application validity dimension. A stronger application validity demands a clear problem framing, which strengthens the conceptual validity dimension in a way that is immediately understandable and useful for users. This could be the case also for studies where specific methods, either qualitative, quantitative, or mixed, are well-known and suitable for case studies with similar characteristics. In such cases, stronger links between attribute selection and analysis can facilitate the construct and internal validity dimensions, strengthening impact and therefore application validity, particularly when stakeholders are involved. In clustering, for example, establishing internal validity includes the selection of a suitable number of clusters to optimise variance measures or reproducibility. Stakeholder We applied and tested the six validity dimensions on a selection of published archetype analyses across different sustainability research fields. In its application, the six dimensions provide an operational framework to guide the validity assessment in archetype analyses. The framework consists of the definitions of validity standards for the six dimensions, with evaluation criteria ranging from weak ('W'), medium ('M'), to strong ('S'). We draw the evaluation criteria of validity dimensions (table 1) based on the definitions of the six dimensions (see section 2), cross-cutting characteristics of archetype analysis and interdisciplinary requirements of sustainability science. For example, conceptual validity in archetype analysis is based on the level of fit and integration between the two salient conceptual components of (a) scientifically-sound research framing and (b) problem-driven research questions. Thus, we define conceptual validity as weak if both components are absent or poorly addressed, medium if one of the two components is absent or poorly addressed, and strong if both components are properly addressed.

Best-practice guideline to improve validity in archetype analysis
We applied the validity assessment to 18 published archetype analyses (supplementary material, section S2 available online at stacks.iop.org/ERL/17/025010/ mmedia) to demonstrate the usefulness of the framework and provide overarching guidelines and lessons learned to support a wider and informed use of the framework within the archetype analysis research community. Each assessment was performed by one co-author and reviewed by all the co-authors of this paper to minimise biases. Nonetheless, the exemplary assessments are not meant to be completely objective, but rather to show the applicability of the framework and help draw generalised guidelines and insights on its usability. In the following sections we refer to these assessments using a code made of a letter and the name of the first author.
The overall results of the assessments indicate that almost all studies encounter the trade-offs described in section 3; none of the selected archetype analyses scored 'Strong' on all validity dimensions (figure 2). Overall, external and application validity are the weakest dimensions. One reason for this finding may be that deliberate planning of validation is not yet a common good practice in archetype analysis, and many publications do not report on their practice of validation. This points to the need for a comprehensive, multi-dimensional validation plan, designed before the start of the analysis, and for its reporting in publications. The advantage of implementing a validation plan for single studies is that it clarifies, step by step, which phase might be weaker, thus offering an opportunity to increase validity for all Conceptual Both of the two salient conceptual components (scientifically-sound research framing and problem-driven research questions) are absent or poorly addressed.
Either of the two salient conceptual components (scientifically-sound research framing and problem-driven research questions) is absent or poorly addressed.
Both of the two salient conceptual components (scientifically-sound research framing and problem-driven research questions) are properly addressed. Construct No clear connection between the selected attributes and the conceptual framing of the study (e.g. no clear theoretical nor empirical justification of attribute selection).
Attributes and the conceptual framing are connected, but attributes are selected without rigorous methods (e.g. meta-analysis).
Clear connection between the attributes and the conceptual framing and attributes are selected using rigorous methods.

Internal
The method of analysis is inappropriate for the objectives and data used, or the method is not state-of-the-art, or it is unclear how results emerged from primary data.
Methods are appropriate and state-of-the-art, but there is inadequate information on data sources and/or analysis. Method-specific validity checks are not applied or are insufficiently applied. Archetypes correspond to some extent to reported outcomes and are partially consistent with causal mechanisms leading to these outcomes.
Archetypes clearly correspond to reported outcomes and are consistent with causal mechanisms leading to these outcomes. Application The archetype application, e.g. transferability of sustainability solutions within ATs, is not indicated or cannot be proven.
The archetype application, e.g. transferability of sustainability solutions within ATs, is partially indicated, based on independent data, case studies and/or stakeholders' perceptions.
The archetype application, e.g. transferability of sustainability solutions within ATs, is explicitly indicated, based on independent data, case studies and/or stakeholders' perceptions.
dimensions. Using these insights, we developed bestpractice guidelines to improve validation in archetype analysis. These guidelines include considerations for developing a validation plan, as well as strengthening the links and reducing trade-offs between dimensions.

Developing a validation plan
Validation needs to be planned as part of the research design before conducting the actual archetype analysis. The validation plan should outline the desired level of validation with regard to all six dimensions, it identifies in which phase of archetype analysis validation is needed, for which dimensions, and present methods for testing and increasing validity. The validation plan should distinguish between independent and interdependent dimensions. When dimensions need to be validated independently, it could help to use complementary approaches. For example, in quantitative archetype analysis, a viable approach to increase external, empirical, and application validity is to define an independent set of validation data to compare and cross-check the archetypes identified with other sources (e.g. A-Sietz, B-Vidal-Merino, C-Piemontese and M-Segnon). When dimensions are interdependent, the addition of complementary approaches could include details on existing data and/or plan the collection of additional data for validation. For example, the inclusion of stakeholder interviews and workshops can support validation of several dimensions in qualitative archetype analysis (G-Karrash, S-Neudert) and scenario archetypes (E-Pedde, P-Pinnegar). The value of synergies deriving from interdependent dimensions is discussed more in detail in section 4.2.
To design a comprehensive validation plan, we suggest conducting archetype analysis and validation in iterative steps, with sufficient resources and time scheduled to enable several iterations. Decisions in the iterations include: (a) adjusting or adding hypotheses, underlying processes, or data sets used in the archetype analysis to obtain a closer agreement with observed system behaviour; (b) revising archetype results based on insights obtained from validation; and (c) confirming archetype credibility with regard to each validity dimension.

Validation requirements
The first step in designing a validation procedure is defining and weighing the degree to which and how archetypes are to be validated. Desired levels of validation precision include the definition of functional and spatial resolution, user perceptions, and temporal dynamics that the identified archetypes need to reflect. This means that levels of allowable or unavoidable imprecision must be specified. For example in conceptual validity, there appears to be a trade-off between case-specific research questions, which can be validated and refined with stakeholders, and broader-scale, scientifically-driven research questions of broad relevance to the scientific community (e.g. the case-specific framing in S-Neudert vs the broad global scale in T-Oberlack). The problem framing (conceptual validity) in global studies can be informed by literature and/or theories, review of case studies, and stakeholder participation. Iteratively, conceptual validity influences application validity in that a research problem framed with stakeholders enhances the relevance of the research results for stakeholders (e.g. F-Thorn). Transferability (i.e. application validity), typically a key motivation of archetype analysis, depends on the extent to which scientists and practitioners from multiple contexts perceive problem settings to be similar (e.g. see empirical validity in I-Levers).
If well-documented local case studies are used to validate broad spatially-explicit archetypes, the question arises as to whether the small local extent and the high functional resolution of case studies represents the large extent and functional aggregation of broader archetypes. Ideally, there would be a sufficiently-large number of independent case studies available for validation that capture the variations within an archetype. However, these are often not available. Alternatively, available case studies can be used to illustrate important features, processes, and sustainability outcomes (e.g. A-Sietz, C-Piemontese). Moreover, follow-up case studies are of critical importance to revise or validate knowledge about archetypes from previous studies.
Moreover, the level of consistency with independently-reported causal mechanisms and sustainability outcomes (i.e. empirical validity) can be determined by stakeholder evaluation (e.g. I-Levers), inter-coder reliability in model-centred meta-analyses of case studies (Oberlack et al 2016) and statistical approaches (Sietz et al 2012). Working at a local level is advantageous for quantitative validation (Sietz et al 2012) given that limitations in regional or global observational data (e.g. spatial resolution) often constrain such validation on larger scales (e.g. C-Piemontese). Stakeholder evaluation can improve qualitative consistency, e.g. by merging cluster pairs which are qualitatively similar (e.g. I-Levers).
External, empirical, and application validity often require additional research steps, e.g. verifying identified archetypes with additional data or approaches (e.g. external and empirical validity) or consulting stakeholders with research results (i.e. empirical and application validity). If these steps cannot be carried out or reported in an archetype analysis, the limitations and the required additional steps should be mentioned (e.g. empirical and application validity in T-Oberlack).
The validity of archetypes can be improved by investing more time in testing validity and adjusting the analysis. Yet at some point, a further investment results only in a very small improvement in validity (i.e. saturation effect). Therefore, it is important to determine as part of the decision procedure whether the precision level reached is sufficient for the objective of the study.

Directions for future validation 4.3.1. Improving weaker validity dimensions
Application validity is one of the weakest dimensions in the present assessment, with 44% of the studies scoring weak and only 6% scoring strong (figure 2). One likely reason is that assessing the transferability of knowledge across contexts is very challenging within the scope of a single study (Vaclavik et al 2016). Future research should therefore increasingly test whether sustainability solutions can be transferred across cases or regions that share an archetype, i.e. similarities in specific attributes and/or causal effects (Adler et al 2018. To achieve this, researchers would need to test, challenge, or refine earlier identified archetypes, rather than identifying new ones. Such a programmatic approach in archetype analysis will build more cumulative knowledge about archetypes of sustainability over time.
Weak application validity can also arise if the research problem and questions are framed from a purely disciplinary or academic perspective (Beran et al 2021). Future archetype analyses may improve their application validity by embracing transdisciplinary co-creation of knowledge through participatory research approaches (Wuelser et al 2021, Jacobi et al 2022, as discussed in section 4.3.2.
Finally, future studies aiming for high application validity should define application requirements, including clear definitions of intended users, spatial scale, and the temporal horizon of application, because these are prerequisites for determining the degree to which archetypes accurately represent knowledge needs and real-world situations.
The 2nd weakest dimension is external validity (39% scoring weak and only 6% strong, figure 2). While external validity generally increases with the number and diversity of cases considered, rich context-specific knowledge, the idiosyncrasy of each case and high data requirements of each case often limit the total number of cases, thus limiting external validity of a single study (e.g. J-Banson) (Wuelser et al 2021). Therefore, approaches to increase external validity include a careful and unbiased selection of cases from different contexts or the use of different sets or types of data in the delineation and the validation of archetypes respectively.
Another reason for low external validity arises if a study does not specify the conditions under which a knowledge claim is expected to hold true (Magliocca et al 2018). Therefore, a requirement for external validity in archetype analyses is to state the limits of generalizability and establish a clear domain of validity. Furthermore, if future research increasingly adopts a programmatic approach to archetypes, by testing, challenging or refining earlier identified archetypes for new contexts (e.g. different countries), external validity of knowledge claims about archetype will increase cumulatively.
The scale of archetypes needs to be considered in validation best practises. A small set of broad spatially-explicit archetypes may have high internal validity at the specific scale at which they are identified (Sietz 2014, Sietz et al 2017. Yet, archetypes observed at one scale may manifest differently at other scales and findings can only be generalised when underlying relationships are consistent across scales. Nested archetypes identified at various scales are one way to operationalize a multi-scale approach providing refined opportunities for transferring sustainability solutions from one place to another (A-Sietz).

Encouraging participatory approaches to strengthen validity
Many of the weaknesses that we found in our validity assessments could be improved by involving stakeholders in the research process through participatory methods. For example, involving stakeholders from the problem framing and hypothesis formulation phase can increase not only the conceptual validity, but can also enhance the application validity and final utility of archetypes (Vidal Merino et al 2019). Stakeholder involvement can prove useful also for increasing construct validity, because a participatory selection of attributes could be a strategy to identify the most relevant attribute and operationalize the conceptual problem (Piemontese et al 2021b), especially if used after participatory problem framing. For internal validity, participatory methods are only appropriate for specific, mostly qualitative, research methods. Using participatory methods can also benefit the external validity of archetypes, for example, by guiding and assessing the extent of and relevant conditions for generalizability.
A relevant role for participatory approaches is also in improving empirical and application validity. Insights derived from archetype analysis should be translated and diffused in appropriate communication channels in a way that stakeholders can make use of them (Lang et al 2012). Involving final users can increase their acceptance of findings derived from archetypes (e.g. potential strategies and solutions), and facilitate the translation of findings to solve real-world problems or to make scientific progress (Bleijenbergh et al 2011, Jacobi et al 2022. Finally, the acceptance of archetypes is key to generating actionable knowledge. For that, stakeholders should feel ownership over and commitment to the archetypes identified, and consider as legitimate any insights resulting from archetype analysis. Such results are best rendered through transdisciplinary research efforts (Bleijenbergh et al 2011, Lang et al 2012. Nonetheless, using participatory methods is not a panacea to justify choices made in analysis without providing the rationale, motivation, and boundaries for the proper use of participatory methods and careful explanation of how they contribute to archetype validity. We note that participatory methods require sufficient resources, access in social networks and stakeholder interest in research (Tribaldos et al 2020). Finally, archetypes are meant to provide transferable and generalizable knowledge claims to support sustainability. Thus, any generalised claims need to carefully account for a diverse and inclusive group of stakeholders so as to reduce biased information and to produce effective and just recommendations.

Conclusions
Validating archetypes identified in sustainability science is to date an under-utilised opportunity to demonstrate their credibility and hence usefulness for informing decision making. To advance validation in archetype analysis in a structured way, we introduced six dimensions of validity and illustrated how they may be used. We discussed how to analyse validity and plan for archetype validation considering the interdependence of validity dimensions. We provided recommendations to support a more comprehensive validation process, as well as evaluation criteria to identify weak, medium, and strong levels of validity for the six dimensions.
In contrast to the classical notion of validation as a check of results against an objective reality, we provided a holistic approach that seeks to achieve validity throughout the research process. We recommend designing each stage of the validation process in such a way that synergies between the six validity dimensions are considered, thereby increasing the overall validity and societal relevance of resulting archetypes. Future archetype validation can best be advanced based on (a) systematic planning of validation procedure, (b) better availability of functionallyand spatially-resolved data, and (c) encouragement of multi-stakeholder participation. Our framework provides guidance to those seeking to enhance the validity of archetypes by systematically understanding and using the opportunities associated with particular analytical steps.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).