1 Introduction

Causal reasoning, as we define it here, refers to the cognitive activities we engage in when figuring out the effects of specified causes, and how these effects are brought about, but also when identifying causes that may produce specified effects. These activities usually involve making sense of the broader causal setting, i.e., the context, in which the causes and effects of interest operate. It also involves choosing a research design and methods, collecting or generating data, interpreting results and, finally, justifying causal claims.

Causal reasoning thus does not only occur when we interpret data to test whether there is a causal relationship; it takes place throughout the entire research process and it can be very diverse. The potential outcomes framework introduced in Chap. 4 (Rubin, 2004, 2005) which is currently receiving lots of attention (e.g., Kimmel et al., 2021), is only one way to reason about causal relations. It is rooted in a particular idea of causation that is associated with the experimental method, and puts much emphasis on the design of a study, because the quality of the design determines the ability to make a causal claim. There are also other ideas about causation, which have been formalised to various degrees. The potential outcomes framework is formalised mathematically, other ideas about causation less so. Different approaches also put different emphasis on the concept of a causal relation, which methods are suitable, and what is considered appropriate evidence for a causal claim (Illari and Russo, 2014).

In addition, different approaches may differ in how researchers build their causal models, and whether they search for causes in a single case or for general causes across a population of cases, cf. Sect. 2.3.

This chapter aims to introduce this broader picture and invites the reader to explore the variety of ways sustainability researchers reason about causation. An understanding of this diversity and ways to navigate it is important for inter- and transdisciplinarycollaboration and for assessing causal claims and their consequences for action. Both collaboration and reflexivity are critical for enhancing understanding of social-ecological systems and for finding appropriate solutions for sustainability problems. To this end, we discuss how causal reasoning proceeds in a study, illustrate the diversity of causal reasoning with some examples and conclude with pointing to some tools and further readings that help to clarify causal reasoning.

Making sense of and analysing causation in complex social-ecological systems (SES) is an emerging research frontier. This chapter provides some initial ideas, but a more thorough treatment is beyond its scope. For a deeper exposure to particular aspects of this broad frontier, we refer the reader to the literature (cited throughout the text and in the suggested readings below).

2 Causal Reasoning About Social-Ecological Systems

Researchers study social-ecological systems with the aim to enhance understanding of pressing environmental problems and potential solutions to address them (Kates, 2011). We want to understand what causes a problem, such as biodiversity loss, or the deterioration of a freshwater lake and what can be done about it. The field is inter- and transdisciplinary. It involves several disciplines from the social and natural sciences and the humanities and it co-produces knowledge with practitioners and stakeholders in participatory and change-making processes (Lang et al., 2012; Norström et al., 2020; Chambers et al., 2021). Both understanding and action are thus key goals in SES research and are intricately linked. Pursuing both goals requires causal reasoning. This causal reasoning can vary significantly among diverse actors that bring their different backgrounds, experiences and values to the study of SES and their search for solutions. How to deal with the plurality of causal understandings and the co-production of causal knowledge in inter- and transdisciplinary processes is an important challenge and research frontier (Schlüter et al., 2023b; Caniglia and Schlüter, 2023).

Researchers with different backgrounds bring different world-views and epistemologies to the study of causation in complex SES. Causal questions such as whether a cause produces the intended effect, which causal processes have generated an outcome of interest, or, more generally, questions about how SES work, will be answered differently across diverse disciplines. This is so because they build on different beliefs about what should be considered a cause-effect relationship, they differ in what kind of relations are considered interesting and important or they differ fundamentally in their views of the world.

Different research traditions also have different normative standards about what counts as acceptable evidence for causal claims, how this evidence should be collected and what can be generalised from particular studies. These norms are associated with certain epistemologies and preferences for approaches and methods, such as viewing experiments as the golden standard for causal inference, versus viewing in-depth historical studies that trace causal pathways as the best way to understand causation.

Bridging these different views and approaches into dialogue in ways that respects their differences and bridges them where possible is important because no single approach to causation can deal with all problems or aspects of SES. The complex, multi-scalar and social-ecologically intertwined nature of SES pushes the limits of the reasoning and methods used. In addition, the field requires approaches that move beyond a conception of linear causality towards conceptions that acknowledge the complex nature of SES (Preiser et al., 2021; Geels, 2022).

One way sustainability researchers have dealt with the complexity of sustainability problems and the challenges of interdisciplinary collaboration is through the construction and use of frameworks, which has led to their proliferation (Biggs et al., 2022, ch.3). Frameworks are collections of concepts that are considered to be most relevant for studying a particular phenomenon. Frameworks often also include ideas about causal relations, e.g., how a change in one element affects another element. A prominent example of a framework to study collective action is the SES framework proposed by Ostrom (2007). That framework has been developed to provide a comprehensive set of variables that have empirically proven to be relevant for explaining cases of successful collective action for managing a common pool resource. Ostrom’s framework refrains from specifying causal relationships, this is the function of theories. However, it does assume relationships between variables at the highest level, e.g. that the resource system, resource units, governance system and users have a direct causal influence on interactions and outcomes. An overview of the most common frameworks in SES research can be found in Biggs et al. (2021).

3 Causal Reasoning in a Study

When we conduct a study, the causal reasoning we use is shaped by processes that take place outside the framework of the study, because the research context and the scientific and practical backgrounds of those involved (Fig. 9.1; Box 1, 2) set the stage for the study. The causal inferences made in the study are influenced by the participants’ causal understanding of the social-ecological system in which the phenomenon is embedded (Fig. 9.1; Box 2). This, in turn, is influenced by disciplinary backgrounds, experiences, literature, selected theories, frameworks, and scientific norms of what is acceptable and desirable in scientific practice in a given community. Furthermore, participants’ world-views, positionalities (i.e., gender, cultural background, class, country of origin) and everyday experience (Fig. 9.1; Box 1) informs and motivates research goals.

Fig. 9.1
A block flow chart of different instances of causal reasoning during a research process. The steps are as follows. Specifying focus of causal inquiry, making sense of the causal configuration, designing study and collecting data, interpreting data, making causal claims, worldview, causal understanding, research goals.

Different instances of causal reasoning during a research process. Causal reasoning occurs during different stages of a research process. It is influenced by the worldview, positionality, and experiences of those involved in the study. Literature, scientific norms, theories and frameworks also influence causal reasoning and the prior causal understanding of the phenomenon of interest and the social-ecological system in which it is embedded (Box 1, 2). The research goals (Box 3) influence the causal questions a study asks, e.g. whether the aim of the study is to measure the effects of specific causes or identify the causes of specific effects (Box A). A study can address several of these questions. Once the goal of the causal inquiry has been set, the next step involves making sense of the causal configuration (Box B). This informs the design of the study and data collection or generation (Box C). The final steps are taken when researchers interpret the results (Box D) and justify their causal claims (Box E). The new understanding of the system and the causal configuration gained may feed back into the broader context of the study (Box 1, 2)

Full size image

The goal of a study, e.g. whether the aim is to predict (what may happen in the future?), intervene (what is the best way to bring about a desired effect?), explain (why and how did something happen?), or attribute responsibility (what cause was decisive in bringing about an effect?) shapes the subsequent causal reasoning.

The research goals also influence the focus of the causal inquiry, i.e. what kind of causal questions will be prioritised, e.g. whether a study focuses on the effects of specific causes, how effects are brought about, or the causes that bring about specific effects (Fig. 9.1; Box A. i–iii). The goals and questions, together with the background understanding and position of the researcher (Fig. 9.1; Box 1–2) influence how researchers make sense of the causal configuration of the phenomenon of interest (Fig. 9.1; Box B).

4 Causal Configuration

Social-ecological phenomena are complex, they are composed of a variety of elements and interactions that are organised in a specific way in both time and space. This is the causal configuration of the phenomenon of interest.

Researchers normally begin a study with a mental model of the causal configuration of interest (Fig. 9.1; Box B). This model is informed by the researcher’s prior understanding of the system at hand.(Fig. 9.1; Box 2). Then new knowledge is generated, resulting in an updated model of the causal configuration.

For example, if we are interested in the governance of an eel fishery, prior knowledge of local institutions, eel biology, fishing styles, and changes in landings informs our mental model of the causal configuration. We then learn new details about the causal configuration through the study, such as the diversity of fishers’ livelihoods and adaptation strategies to financial and climatic shocks, competition, and incentives. This new knowledge results in a more elaborated model of the causal configuration.

The representation of the causal configuration made by the research or co-production team, their methodological and theoretical background and data accessibility, inform the selection of methods, possible intervention, and data collection (Box C). The design of the study and the methods used strongly influence the causal interpretation of the results.

5 Interpreting Results

After data are obtained and processed (Fig. 9.1; Box C), causal reasoning focuses on interpreting data as evidence (Fig. 9.1; Box D). This depends on background information about the causal configuration (Fig. 9.1; Box 2) and is a critical step to figure out whether the data is evidence of a causal relation or not. Scientists might give different reasons, provide different interpretations, or favour one instead of another. However, it is possible to identify some commonly used schemes of reasoning supporting causal conclusions, such as (i) that the cause precedes the effect (cf. Sect. 3.4), (ii) that a correlation is an indicator of a causal relation (cf. Sects. 7.1–7.3), (iii) that the cause and the effect are linked through a mechanism (cf. Sect. 8.4), (iv) that manipulating the cause will change the effect in otherwise invariant conditions (cf. Sects. 6.1, 7.1 and 7.2), or (v) that the most likely causal explanation is the one that best makes sense of all the available evidence (cf. Sect. 8.4).

6 Making and Justifying Causal Claims

The final stage of a causal study is to write a report where the conclusions are drawn and the arguments for these conclusions are given (Fig. 9.1; Box E). The reasoning at this stage builds upon the prior stages of the study, but often reshapes and refines it. The crucial point of this stage is to provide justificatory support for claims.

The strength assigned to a causal claim should match the support provided by the evidence. For instance, when claiming a causal relation between two (quantitative or qualitative), variables, it is not enough to refer to an observed correlation between them. At most one may claim that there might be a causal relation. Causal claims might vary regarding their strength, specificity, and scope. Compare these four claims:

  • Baseline claim. High trust among male small-scale fishers in Kino bay, Mexico, seems to lead to an increase in their income.

  • Stronger claim. High trust among male small-scale fishers in Kino bay, Mexico leads to an increase in their income.

  • More specific claim. High trust among male small-scale fishers in Kino bay, Mexico, leads to an increase in their income of 20%.

  • Wider scope claim. High trust among small-scale fishers around the world seems to contribute to their income.

Each of these claims require different kinds of evidence.

When communicating research or interpreting other people’s research one needs to be aware that, claims are differently justified and differently interpreted. In inter- and transdisciplinaryspaces there can be tensions between the evidence provided for causal claims and scientific standards for justification of causal claims.

7 Diversity of Causal Reasoning

Depending on the goals of a study, the problem to be investigated and the chosen approach, each case of causal reasoning outlined in Fig. 9.1 will be unique. To illustrate this diversity, we explore causal reasoning in five exemplary studies from SES research. Example 1 is a case of statistical causal inference that examines whether a community monitoring program can reduce groundwater extraction from aquifers, improve water quality, and increase user satisfaction in Costa Rica (Carpio et al., 2021). Example 2 aims to explain the synchronicity of recent global crises, such as the 2008 food-energy crisis and the financial-energy crisis (Homer-Dixon et al., 2015). Example 3 examines the case of the Baltic cod collapse (Lade et al., 2015). Example 4 studies the mechanisms that may explain the emergence of self-governance arrangements in fisheries in Mexico (Lindkvist et al., 2017) and example 5 examines how a practice-based approach to sustainability interventions can support workable solutions in ever-changing contexts (West et al., 2019).

7.1 Study 1: Quantifying the Effect of Community-Based Monitoring on Groundwater Management: A Statistical Causal Inference Approach

The goal of Carpio et al. (2021) was to investigate whether there is a causal relation between an externally driven community monitoring program and improved groundwater management in rural Costa Rica, and if so, to quantify the effect (Fig. 9.1; Box 3). It thus asks the causal questions ‘what are the effects of a specified cause, i.e., the community-based monitoring’, and ‘what are the magnitudes of these effects and how are they brought about?’ (Fig. 9.1; Box A).

The authors develop their understanding of the causal configuration that underlies the effect of community-based monitoring on groundwater management (Fig. 9.1; Box B) using literature from three empirical and theoretical fields: common pool resources, community-based environmental monitoring and citizen monitoring of public services (Fig. 9.1; Box 1, 2). This knowledge was used to specify a hypothesised mechanism through which monitoring (i.e., interventions) influences the quality of water management. The causal configuration informed the development of three hypotheses about the effects of monitoring.

These hypotheses were tested using a randomised experimental design where the causal variable, i.e. community-based monitoring, was externally manipulated through applying an intervention to some communities but not others (Fig. 9.1; Box C). This approach assumes that through manipulating the community-based monitoring (the assumed cause) we can obtain knowledge about its connection to the assumed effect, and that randomisation eliminates the influence of contextual variables and makes the communities comparable. The monitoring intervention was applied to communities that were randomly selected, but not to those in the control group and data on the primary outcomes and intermediate variables were collected (Fig. 9.1; Box C). The data was then interpreted using counterfactual reasoning, i.e., the changes in outcome variables between treated and control units were compared (Fig. 9.1; Box D). Final and intermediate outcomes that are part of the mechanisms were also measured and analysed statistically (cf. Sect. 6.7: Causation, Manipulation and Intervention).

The experimental results provided some evidence for the causal claim that community monitoring improves groundwater management (Fig. 9.1; Box E) because the impacts of the intervention point in the right direction (communities with monitors pumped less, had better water quality and higher customer satisfaction), but impacts after 1 year of the program were modest. The authors also found evidence consistent with their theory of change, but the effects of the program on the intermediate outcome variables were small and imprecisely estimated. The conclusions were justified by experiments and the specification, and partial verification, of a plausible mechanism (cf. Sect. 8.4: Causal Explanations and Mechanisms).

However, no alternative mechanism was discussed. In their discussion, the authors reflected on factors that could make the intervention more successful, e.g. why a particular causal pathway was not very strong and how it could be strengthened, and on the implications of the results for action. There is no discussion on how this study may have changed the causal understanding of the system or phenomenon of interest.

7.2 Study 2: Synchronous Failure: The Emerging Causal Architecture of Global Crisis

The researchers who conducted this study (Homer-Dixon et al., 2015) were interested in explaining the synchronicity of recently emerging world crises (Fig. 9.1; Box 3). Since we, the co-authors of this book, did not conduct the study, we can only speculate about how the authors’ personal trajectory shaped their understanding of the system prior to conducting the study (Fig. 9.1; Box 1 and 2).

This study has two parts, in the first part authors constructed a plausible causal model of world-scale crisis synchronicity, and in the second they validated the causal model with empirical evidence from case studies.

The focus of this causal inquiry was on how synchronicity of global crises emerge (Fig. 9.1; Box A.ii). To build the model of the causal configurations responsible for this outcome, authors looked at processes that have shaped human-nature interaction during the last decades. They argued that, as the scale of human activity, resource use, and world connectivity has increased, the flows of information, matter and energy between subsystems have become more intense, as well as their proneness to crises. Then, the authors represented these features in three stylised and interconnected models inspired by complexity theories.

As an example, the long fuse big bangcaptures the non-linear behaviour and configurational change of world subsystems—like the energy, food or economic subsystems—when their coping capacity is exceeded. Simultaneous stresses on subsystems erodes their capacity to endure stress, which eventually leads to a big bang and ramifying cascades, which captures the way in which crisis propagation happens across interconnected subsystems.

Overall, their causal model proposes that the synchronicity of world crises is a consequence of three factors: (i) the simultaneity of stresses across world sub-systems, (ii) their homogeneous proneness to crisis, and (iii) the tight connectivity that allows for crisis propagation (Fig. 9.1; Box B). This hypothetical model is an example of reasoning in terms of INUS conditions (cf. Sect. 5.6).

In the second part of the argument, the authors looked at two case studies of simultaneous global crises, the 2008–2009 food-energy crisis and the financial-energy crisis in the same years (Fig. 9.1; Box C). The authors interpretation of these case studies consisted in mapping them out onto their proposed model (Fig. 9.1; Box D). The model advanced by the authors at the beginning of the argument was not changed, but it was updated in regards to its empirical support; the authors claimed that ‘recent global crises reveal an emerging pattern or architecture of causation that will increasingly characterise the birth and progress of crises in the future’. This claim gets justificatory support from the plausibility of the model of synchronous crises, it’s consistency with theories, and the empirical illustration. However, it makes two assumptions: that there are no alternative explanations and that the current global trends will continue (Fig. 9.1; Box E). The case studies are examples of qualitative research, which provides rich information about the mechanisms responsible for the outcome (cf. Sect. 5.3: Causation in Qualitative Studies).

7.3 Study 3: Exploring the Importance of Social Processes for the Collapse of the Baltic Cod Stocks: A Modelling Approach

The goal of Lade et al. (2015) was to assess the role of social processes, such as fishers’ decision making and actions, government decisions and market dynamics, for the collapse of the Eastern Baltic codpopulations in the 1980s (Fig. 9.1; Box 3). The focus of the causal inquiry was thus to identify the causes of specified effects (Fig. 9.1; Box A.iii). The authors made sense of the causal configuration that may underlie the cod collapse through a collaborative process where the authors brought different ecological, economic and social-scientific expertise about Baltic cod fisheries to the discussions (Fig. 9.1; Boxes 1 and 2).

Together, they built a causal loop diagram specifying key feedbacks (cf. Sect. 5.4) that were hypothesised to have influenced the cod collapse (Fig. 9.1; Box B). Based on this diagram a generalised dynamical systems model was developed and parameterised for a situation before and during the beginning of the collapse, using fishery data, literature and expert knowledge of the research team (Fig. 9.11; Box C). A stability analysis of the modelled system separated in the social part, the ecological part and the coupled system before and during the collapse was conducted. This was done to assess the impact of the social system on the collapse and identify which feedbacks had the largest effect (Fig. 9.1; Box C).

The authors compared model versions, where the social and the ecological systems were decoupled, with a version of a coupled system, both before and after the collapse, to evaluate the causal influence of social processes on the collapse of the cod stocks. This is an example of the use of counterfactual reasoning (cf. Chap. 4) within a model, or rather different model versions, that represent the counterfactual situation. Based on a comparison, the authors developed causal knowledge about which social processes contributed to the shift in the Baltic Sea ecosystem (Fig. 9.1; Box D). Using the an analysis of the feedback mechanisms, (cf. Sect. 8.5.2: Feedback mechanisms), the authors explained the model outcomes. They made the causal claim that the adaptivity of external fishers (i.e. fishers that came to the Baltic Sea from Sweden’s West Coast) initially stabilised the ecosystem despite changing environmental conditions for a certain period of time.

7.4 Study 4: Explaining Emergent Patterns of Self- Governance Arrangements in Small-Scale Fisheries: A Modelling Approach

Lindkvist et al. (2017) made a modelling study aimed to investigate the conditions under which either cooperative or non-cooperative forms of self-governance emerge in a typical fishing community in northwestern Mexico (Fig. 9.1; Box 3). It asks two causal questions; what are the causes of specified effects and how are they brought about (Fig. 9.1; Box A.i–ii)?

The study drew on frameworks and theories such as the SES framework (Ostrom, 2007), institutional analysis, collective action theory, common pool resource theory, and complex adaptive systems theory, all of which the researchers in the team previously had used (Fig. 9.1; Box 1, 2). Thus, the researchers’ backgrounds influenced the study through their previous engagement with these theories and frameworks, but also through prior knowledge of the case and their experiences of working with fishers and in fishing communities (over 20 years for one co-author). This informed how they defined the social-ecological system and phenomena of interest, the research goals, and the assumptions of which variables matter for cooperatives such as the form of self-governance to persist over time (Fig. 9.1; Box B).

Against this background and based on data collected in previous studies the authors built a model that can be used as a virtual laboratory to answer the following research questions: (i) How do micro-level factors related to trust—such as the reliability of fishers, and loyalty between fish buyers and fishers and between members in cooperatives—affect the emergence and persistence of different self-governance arrangements? (ii) How does environmental variability affect whether cooperatives or patron-client relationships emerge as the dominant form of self-governance? (iii) How stable are these two self-governance arrangements and what causes them to fail?

Using the case knowledge and an agent-based model, the authors were able to discover and reason about specific mechanisms that explain how effects were produced. In the model one can change different variables and observe their effects on the emergence and persistence of different self-governance arrangements. This indicates which variables one can manipulate through different policies or interventions in relation to the desired outcome in reality. (cf. Sect. 6.7: Causation, Manipulation and Intervention). Additionally, the model setup includes several feedback mechanisms at the level of individual agents, such as the reinforcing feedbackloop where increased loyalty results in less cheating, which in turn increases loyalty (cf. Sect. 8.5.2: Feedback mechanisms). These feedbacks became important parts of the explanation why under some conditions cooperatives could survive while not in others.

The model showed that high diversity in fishers’ reliability and low initial trust between cooperative members make the establishment of cooperatives difficult. In contrast, patron-client relationships are more flexible in choosing whom to work with and can better cope with this kind of diversity. However, once established, cooperatives are better equipped to handle seasonal variability in fish abundance and provide long-term security for the fishers. Through these types of causal findings gained from analysing and testing the model, combined with case based knowledge, the researchers could uncover and reason about specific causal mechanisms that help explain how certain effects are produced (cf. Sect. 8.4: Causal Explanations and Mechanisms).

The primary aim of the model was to investigate under which conditions different causes would, or would not, lead to certain effects and provide causal explanations for why and how (Fig. 9.1; Box A.ii). The causes and effects of interest to explore in the first place were, however, derived from prior case knowledge, theories and frameworks (Fig. 9.1; Box 2). The study design and the choice of agent-based modelling as a method (Fig. 9.1; Box C), were also a result of previous experience of the researchers involved (Fig. 9.1; Box 1–2). The interpretation of the results and the causal claims made (Fig. 9.1; Boxes D, E) were based on the key factors and processes in the model, but situated against background knowledge of the author team. The knowledge about the causal configuration of the social-ecological fishery system contributed to deeper knowledge about how policies could support specific governance structures in theory and practice (Fig. 9.1; Box E, 1, 2).

7.5 Study 5: Addressing the Challenges of Climate Adaptation: A Practice-Based Approach to TransdisciplinarySustainability Interventions

The goal of the transdisciplinary ‘Future proofing Conservation project’ (van Kerkhoff et al., 2019) was to develop new ways of addressing the challenges posed by climate adaptation for protected area policy-makers and managers in Colombia. A practice-based approach to sustainability interventions is compatible with the assumption that for many sustainability problems ‘optimal’ solutions hardly exist and that problem formulations often are unclear and contested. It thus challenges linear assumptions about knowledge and action and suggests that ‘the primary task of participants in sustainability interventions is to arrive at workable solutions to situations of dynamic complexity that are fundamentally open-ended and unpredictable’ (West et al., 2019).

Accordingly, this example emphasises that causal reasoning in the context of complex SES requires collaborative and participatory processes involving the stakeholders affected in a particular place. Strictly speaking, collaboration is not only required when defining the causal configuration (Fig. 9.1; Box B), but already when characterising the prior causal understanding of the system (Fig. 9.1; Box1, 2). In the process, climate science acquires a new role, ‘not as a solution-provider (“let’s wait until the scientists tell us what to do”), but as a knowledge base that conservation governance practitioners need to act upon (“we are knowledgeable actors”)’ (op.cit., 547–8). Accordingly, the study encourages stakeholders to conceive of adaptation not as a simple ‘once and for all’ application of knowledge but as a continuously evolving practice. It thus highlights the importance of the feedback from the study itself to a continuously evolving causal understanding (Fig. 9.1; Box1, 2). There is a strong emphasis of practice-based approaches on this last point: Acting and knowing are merged in practice and as such ‘....the final methodology can be regarded as encouraging practitioners to think about climate adaptation as a practice, rather than a task. As a practice it is ongoing, deliberative and potentially transformative, framed by learning and dialogue rather than the application of technical solutions’ (op.cit., 548).

8 Summary of Examples

These examples show that causal reasoning can be done in many different ways and it is strongly influenced by the goal of the study, by who is involved and what theories and frameworks, literature, scientific norms, and experiences, they bring to the table. For example, the first study builds on the potential outcomes framework (cf. Chap. 4) and research in economics in order to quantify the effect of an intervention using an experimental design that compares treatment with control units. The last study builds on practice theory and research in the humanities in order to build causal understanding through collaborative processes where scientists and non-scientists make sense of causality while engaging with the complexity of the problem and potential solutions. Here causal understanding is dynamic and continuously co-produced through the practice of problem solving. The examples illustrate the use of the causal concepts introduced earlier in the book, but also show that causal reasoning in SES research makes use of a broader set of concepts than what we could discuss in this introductory text.

Differences of causal reasoning and resulting causal claims between studies may arise because of different foci, e.g. on singular versus general causation (cf. Sect. 5.2), different data, e.g. quantitative versus qualitative, or different goals, e.g. evaluating the magnitude of a causal effect versus developing causal explanations. Study 1 for example makes use of the causal ideas of intervention and potential outcomes using an experimental design to collect quantitative data on a population of cases, study 2 applies INUS conditions and Hill’s criteria using qualitative data. Studies 3 and 4 use counterfactual reasoning and manipulation in the context of modelling with the aim to identify mechanisms that bring about the phenomena of interest in the modelled system. The last study illustrates a focus of causal inquiry that lies specifically on how the causal configuration constitutes and re-constitutes in processes of transdisciplinary collaboration over the practice of climate change adaptation. So doing, this approach goes beyond the distinction made between causal and constitutive explanations made in Chap. 8 and explores how these interrelate and condition each other.

It is important to realise that the goals of a causal inquiry and choices made early on in a study direct causal reasoning and create path dependencies. For example, taking a systemic view and choosing a modelling approach will shape the process of making sense of the causal configuration differently than if a researcher takes a practice-theoretical view choosing a participatory approach. The five examples show that all elements of the causal reasoning processes are considered, but each study has a different emphasis. For example, in study 1, on the effectiveness of community monitoring, the authors put most emphasis on justifying their causal claims through scrutinising the design of their experiment and finding evidence, for the proposed mechanism that links the intervention to the outcomes. In study 4, on the emergence of self-governance, the focus is on understanding the conditions and the mechanism that explain why cooperatives rarely dominate. In study 5, on climate change adaptation, the authors emphasise the collaborative, practice-based and continuously evolving nature of causal reasoning. Study 3 puts much emphasis on building a comprehensive representation of the causal configuration through integrating interdisciplinary expert knowledge. Study 2 puts much emphasis on constructing an archetypical representation of the problem that is then tested in two case studies.

The five examples show that all elements of the causal reasoning processes are considered, but each study has a different emphasis. The five examples employ different methods for their causal inquiry, from experiments (study 1), dynamical systems and agent-based modelling (studies 3 and 4), participatory processes (study 5) and a combination of theoretical deliberations and case studies (study 2). These methods not only allow them to do different things, e.g., only the first method allows quantifying a causal effect, or only the modelling methods allow investigating how the system changes over time as a consequence of interactions between agents and their environment or feedbacks between system elements. They are also grounded in different assumptions of what is considered appropriate evidence for a causal claim. Finally, approaches and associated methods differ in their assessment of the causal configuration, from a focus on a single cause-effect relationship embedded in a larger causal configuration to a systemic view that incorporates more aspects of the larger causal configuration. The degree of formalisation of the causal configuration also varies, which has effects on which methods can be applied.

Our description of the studies also illustrate the difficulties of eliciting information about the worldviews, positionalities, and experiences that underlie causal reasoning because they are rarely made explicit. This lack of transparency is problematic because it limits our ability to assess the scope, quality and compatibility of a causal claim, or an approach to studying causation for a particular problem at hand, or for the integration of approaches or knowledge.

Tools that support eliciting underlying ontological and epistemological assumptions e.g., (Eigenbrode et al., 2007; Hazard et al., 2019)), and, more specifically, tools that facilitate dissecting the causal reasoning of different approaches help to increase transparency (we will briefly introduce this tool below). This is important in order to enable inter—and transdisciplinary collaborations across different traditions of causal reasoning.

9 Navigating the Diversity of Causal Reasoning

In this chapter, we have discussed and illustrated that causal reasoning takes place during all phases of a research process and that it is diverse and depends on the backgrounds, positionalities and prior knowledge of those involved in the study (Fig. 9.1). Awareness of the many steps in which causal reasoning manifests itself enables researchers to articulate, understand and reflect on the causal reasoning that underlies a particular study. This is important for inter- and transdisciplinary collaboration and for assessing the scope and validity of a causal claim, and its consequences for intervening in a SES to bring about a desired effect.

In order to make explicit how, specifically, the background, theories, research goals, etc. (everything in Boxes 1–3) influence causal reasoning activities during the different research phases, we have developed a guide called CoMap (Hertz et al., 2024). CoMap specifies five elements that together constitute causal reasoning: the conceptualisation of causation a study builds on, its analytical focus, the theories and frameworks used, the selected methods and causal notions. These elements are interdependent and influence each other, and their interplay is shaped by the purpose of an analysis. In addition, typically, one of these elements—which may be called ‘entry point’—is particularly important in that it designates an element that orients or exerts influence on the other elements in that these need to ‘align’ with it.

Through making these choices and path dependencies explicit, this guide reveals how causal reasoning looks like, that is, it becomes apparent which choices can and need to be made by researchers in assembling a study. The examples presented above show how, accordingly, causal reasoning might vary considerably. This means that through the process of eliciting causal reasoning we become aware of each other’s ‘blind spots’. That provides the basis for (1) a reflection on the assumptions underlying the causal reasoning of research approaches, for (2) engaging in inter- and transdisciplinary collaboration, either by developing a common research approach, or (3) by relating different research approaches to each other.

10 Summary

This chapter characterise causal reasoning as the cognitive activities we engage in, implicitly or explicitly, when studying relations between causes and effects. We engage in causal reasoning during the entire research process of a study, from developing the causal questions, making sense of the causal configuration of the phenomenon of interest, designing the study and choosing appropriate methods, interpreting results to make and justify causal claims. In our causal reasoning we draw on our theoretical and methodological background, including ideas about causation, on previous experiences, the literature, the norms of our community and our previous understanding of the SES in which the phenomenon of interest is embedded. Differences in the goals of a causal inquiry, such as prediction, explanation, or intervention, and in the background conditions that shape causal reasoning, produce a large diversity of causal reasoning strategies. These different strategies produce different causal understanding, which has consequences for what can be done with the knowledge, e.g., when designing interventions.

Further Reading

Study Questions

  • Have you encountered different ways of causal reasoning? If so, which ones?

  • How does your worldview, positionality, and experiences influence your causal reasoning?

  • How has the literature, theories, and frameworks you have engaged with influenced your causal reasoning?

  • How have the scientific norms at your home institution hindered or enabled your ability to engage in different types of causal questions?

  • Can you think of examples from your own experience where your understanding of the causal configuration changed after you conducted a study and observed a particular phenomenon? Did this experience change your worldview and/or what literature, theories, and frameworks you engage with?

  • Do you see any other interesting differences or similarities between the examples presented in Sect. 9.4?