Breaking biases and building momentum for transforming agricultural research for development practices: recommendations and research opportunities

According to the UN Global Stocktake Report 2023, we are not on track to achieve the Sustainable Development Goals (SDGs) by 2030¹. Current food systems perpetuate inequalities and fail to address rising food insecurity. Our agricultural systems, which are trying to feed over 8 billion people worldwide, are one of the main factors that breach our planetary boundaries². Agricultural Research for Development (AR4D) is essential in providing evidence-based solutions to address current social and environmental risks and provide possible positive outcomes of agrifood systems³. To be effective, these solutions need to be driven by end-users’ needs and aspirations. While AR4D organizations are increasingly adopting a holistic approach that integrates diverse actors making up Agricultural Innovation Systems (AIS), there has been less attention to the reflexivity and positionality of AR4D actors who implement these systems within their organizations.

This perspective paper examines the urgent need for a paradigm shift among AR4D actors to effectively contribute to the achievement of the SDGs by 2030. Since the establishment of the Millennium Development Goals and their successor, the SDGs, social objectives such as women’s empowerment and reducing inequality have been central to AR4D missions. Donors and governments are increasingly investing in agricultural innovation as a strategy to achieve these social outcomes⁴. As a result, there have been ongoing AR4D discussions on how to more effectively link agricultural research discovery and innovation to social and environmental outcomes. AIS is a widely utilized and continually evolving concept that positions agricultural research within a dynamic system of innovation actors^5,6. These dynamics, which include interconnected actors and embedded power structures, are reflected in AR4D practices and conceptual frameworks to envision research that will achieve stated SDG goals towards a sustainable agricultural transformation. However, AR4D actors have struggled to incorporate inclusive and diverse evidence-based practices resulting in outcomes that do not neglect or under-serve marginalized groups and individuals. This paper posits that internalized AR4D biases and misaligned incentives among funders and researchers have gone unacknowledged and unaddressed, affecting the AR4D outcomes they aim to achieve.

Upstream AR4D actors significantly impact the directionality of innovation and scaling processes, but are susceptible to historical and path dependent biases sometimes ingrained in colonial, patriarchal, and other non-evidence-based ideologies and frameworks. They affect how AR4D actors perceive challenges, the strategies they adopt to tackle them, the innovations that emerge, and the people who benefit from these innovations⁶. Factors such as funding sources, team composition, and selection of participants for inclusion in decision-making processes affect innovation results either positively or negatively^7,8. Prioritizing socially marginalized populations is essential for progress on the largely social nature of many SDGs, which necessitates the involvement of multiple types of expertise and stakeholders throughout the AIS⁹.

Current approaches to innovation systems describe ‘development challenges’ and resulting innovation as exogenous to the AIS they are part of. In this way, they obscure the potentially pivotal roles and feedback provided by diverse local actors and power structures that affect decision-making throughout the innovation and research processes^10,11. Agrifood actors should be seen as research partners along with scientists, recognizing that end-users are innovators in their own right^12,13,14,15. However, end-users, such as farmers, are not a homogenous group and social differences should be appropriately considered and acknowledged. In this way, there is a need to “democratize” knowledge systems, allowing for more collaborative research and fostering genuine co-production among different and diverse agri-food stakeholders¹⁶. This is true not just for the process but the definition of a positive outcome, that is in how and who defines terms such as “empowerment” and “transformation.” Further, despite decades of calls to “mainstream gender”, the contributions and aspirations of women across AIS, in all their diversity, often remain at the margins of mainstream institutional practices or outsourced to “the gender expert” rather than being integrated into AR4D initiatives¹⁷.

To progress towards SDGs, we propose modifications to the system’s default operations to acknowledge and consider social differences amongst upstream actors (e.g. donors and high-level researchers). During their attempts to achieve positive social AR4D outcomes, current AR4D actors encounter barriers throughout the innovation process due to biases in upstream governance, a lack of interdisciplinary approaches in science, a focus on technological advancements and widget counts rather than outcomes, inadequate evaluation frameworks, and well-documented inclusivity issues. These barriers result in a dearth of investment in inclusive approaches, interdisciplinary and transdisciplinary research, and in understanding – perhaps long term – systems approaches to foster complex and long-term impacts. While current conceptualizations of AIS may acknowledge the pivotal roles of local actors, actual approaches and funding to support local leadership are lacking. Further, there is a failure to address normative and power structures that affect decision-making both ‘in the field’ and within upstream AR4D practices due to a lack of reflexivity.

This work provides five recommendations for transitioning AR4D towards more impactful innovation and scaling for desired positive social outcomes, that stem from the literature review, as well as the experiences and expert knowledge of the authors, while also pointing to future research frontiers within each topic. The authors build on a review of recent articles exposing the shortfalls of research in development efforts, notably Abera et al. ¹⁸, Brock et al. ¹⁶, Schutter et al. ¹⁹, and McGuire et al. ²⁰. Recommendations to further examine how upstream AR4D actors in conjunction with downstream actors, can use agricultural innovation to more effectively drive inclusive positive social change include:

1) addressing upstream biases for greater downstream effectiveness, 2) strategically integrating social and natural sciences through interdisciplinary and transdisciplinary teams, 3) prioritizing outcome-based over innovation-driven approaches to scaling, 4) measuring complex change by applying dynamic and empowerment evaluation approaches, and 5) acknowledging and acting on social differentiation throughout the innovation and scaling process. By better understanding and addressing these factors, AR4D researchers and practitioners can foster transformative change in AIS so as to more effectively target the SDGs.

Addressing upstream biases for greater downstream effectiveness

Upstream AR4D decision-makers, such as government bodies or donors, significantly influence the directionality of innovations, which includes the perception of challenges they tackle, their strategies to tackle them, and who the ensuing innovations benefit. It is imperative to formally consider the contextual power dynamics within AIS and their related impacts on effectively achieving desired positive social outcomes. Although there have been frequent demands for innovation and scaling processes that are more user-led and human-centered, such as grassroots and farmer-inclusive innovation and local leadership building, there have been limited long-term approaches to implement these recommendations at the highest institutional levels. Identifying the correct challenge-solution sets requires processes that go beyond current AIS frameworks. Decision-makers often prioritize specific demands, which can result in ‘elite capture’, where powerful groups dominate decisions for their own benefit, or ‘maladaptation’, where policies neglect marginalized communities and cause further exclusion or environmental harm.

Such unacknowledged biases, combined with a propensity to oversimplify intricate, situation-dependent interactions, has led to numerous unintentional outcomes that have worsened inequality within communities and had lasting environmental repercussions²¹. Some of these negative outcomes of AR4D research and implementation may be influenced by factors beyond AR4D actors’ control. However, influential recurring patterns, such as women’s unequal access to information, capital, and market networks²² have been consistently highlighted by extensive gender research and feminist theory and can immediately be accounted for and addressed. Furthermore, there is an extensive evidence base that demonstrates restrictive cultural norms that result in marginalized groups’ limited ability to benefit from innovations compared to more powerful actors²³. Yet, AR4D upstream actors often show resistance to employing gender-focused research and feminist methodologies to tackle research and societal issues aimed at improving the lives of marginalized populations. This resistance frequently manifests as a lack of attempts to reverse – or understand – practices that create and sustain social inequalities. As AR4D actors, we can improve our upstream accountability and regain significant agency in preventing unintended consequences. Through strengthening AR4D institutions capabilities to promote research processes that genuinely consider social differences, co-production, and knowledge democratization^16,24. Simultaneously, AR4D actors can take increasing responsibility for the subsequent effects in the field—conducting relevant ex-ante assessments and building in socio-ecological safeguards²⁵.

Some tools, including the Quality of Research for Development in the CGIAR Context (Qo4RD)²⁶, the Scaling Scan²⁷ and GenderUp²², have been developed for this purpose. For instance, the Qo4RD framework²⁷, was designed to evaluate research for scientific legitimacy and credibility among other criteria; while GenderUp and Scaling Scan focus on inclusivity and relevance. Nevertheless, it is important to acknowledge that these tools facilitate current AR4D approaches – which are based in a ‘technology transfer’ approach rather than inclusive AIS processes.

To address biased approaches in AR4D and foster reflexivity, it is essential to consider the influence of power dynamics and unintended consequences on scaling outcomes. Calls for AIS transformation must be accompanied by a commitment to shifting perceptions, challenging existing standards, and questioning development agendas that perpetuate colonial, top-down methods. AR4D actors should cultivate self-awareness, acknowledging how their biases, privileges, and chosen methodologies shape outcomes and influence the broader system.

Future Research: Assess the impact of upstream and midstream AR4D dynamics on downstream outcomes, identify necessary socio-ecological safeguards to prevent adverse effects, and explore potential incentives for more inclusive and effective AR4D practices. Consider strategies to create agricultural innovation systems that prioritize the participation and leadership of local actors, especially marginalized groups, and investigate how to incorporate local knowledge, cultural practices, and community priorities into the innovation process. Support capacity strengthening programs to equip AR4D researchers and practitioners with skills in interdisciplinary collaboration, cultural competency, and participatory research methods. Encourage both North-South and South-South knowledge exchange networks to share best practices and lessons learned.

Strategically integrating social and natural sciences through interdisciplinary and transdisciplinary teams

Over the years, agricultural research has centered around technology transfer-driven methods that aimed to boost yield and profit by improving varieties and management practices focused on commodities²⁸. Research methods were designed primarily to meet biophysical objectives, often overlooking the needs and priorities of specific end-users and system implications^29,30. While there have been positive shifts, few social scientists are involved in most research and projects focused on technocratic solutions, and even fewer gender and inclusion experts³¹, with claims that interdisciplinary teams ‘delay outputs’³². Social and natural sciences often address distinct and different types of problems. However, it is through the merging of the two, along with the incorporation of transdisciplinary perspectives, that scientific research can effectively address complex global challenges and achieve impact at scale.

Scientists often view attaining the SDGs through their own disciplinary lens, but rarely can a single discipline lead to success for the highly complex human-planet challenges we face. This realization is perhaps driving increased calls for adopting “systems” approaches that integrate different perspectives, and encompass different scales (such as local and global), disciplines (such as social and natural sciences), and expertise type (such as user, private sector, and public sector)^33,34. Adopting a systems perspective requires researchers to acknowledge that challenges and solutions are interrelated and are often perceived differently depending on the stakeholder’s viewpoint³⁵. This necessitates a critical perspective that involves identifying boundary decisions, exposing power and motivation sources, and utilizing second-order thinking to evaluate biases and worldviews that influence choices, including our own as system participants³⁶. There are frameworks available to bridge disciplines and sectors that include social and environmental perspectives. One such model is “doughnut economics”, which prioritizes social guarantees within planetary boundaries, aiming for regenerative and distributive actions to enhance global well-being³⁷. Comprehending the social and environmental trade-offs and alternative paths toward sustainability can result in well-informed choices and reduced risks in research and projects, promoting transparency with stakeholders regarding preferred goals and operationalizing mechanisms.

Scientific advancements in processes and methodologies can enable more comprehensive responses to intersecting societal challenges and socio-ecological systems³⁸. This was demonstrated by the EAT-Lancet Commission on sustainable diets, which presented an interdisciplinary framework that combines health, environmental sustainability, and socio-economic factors to suggest diets that promote human and planetary health while supporting global sustainability goals². By embracing trans- and interdisciplinary perspectives, AR4D actors can intentionally facilitate transformative sustainability changes that minimize negative impacts and maximize positive social outcomes and generate SDG co-benefits. This prioritizes place-based and impact-focused approaches alongside, or even over, refining scientific methodologies and technological advancements at the expense of system innovation.

Incorporating a broader range of actors, both local and external, fosters more diverse innovation and research, enabling the generation of a wider array of solutions and systems to achieve shared goals. The AR4D system is often entrenched in established processes, and transforming it requires integrating diverse perspectives to challenge and enrich these approaches.

Future Research: Understand and devise techniques to proficiently incorporate varied viewpoints; examine the advantages and disadvantages of pursuing a single discipline versus inter- and transdisciplinary approaches. Research methodologies and frameworks that effectively integrate social sciences, natural sciences, and humanities into agricultural research. Additionally, investigate how interdisciplinary and transdisciplinary teams can work together to address complex agricultural challenges while considering social differences and power dynamics.

Prioritizing outcome-based over innovation-driven approaches to scaling

Outcome-based innovation and scaling prioritize the result rather than the use of the innovation or intervention itself. This thereby provides a holistic perspective to problems, considering alternative outcomes, and prioritizing societal impacts, using a ‘mission-oriented innovation system’ framework³⁹. In the realm of systems thinking, the complexity of situations often presents itself with numerous, interacting elements, unclear boundaries, and a myriad of perspectives defining both problems and solutions, all of which interact to produce outcomes that stakeholders wish to improve upon⁴⁰.

Identifying intentional and unintentional positive and negative societal outcomes is an essential but often sidelined aspect of AIS, raising concerns regarding shared goals, selection procedures, and monitoring systems, which can be addressed through interdisciplinary and collective scoping of challenges⁴¹. Additionally, scaling core and supporting innovations together as ‘socio-technical innovation bundles’ recognizes the importance of a systems approach to achieve certain social goals^13,42 Further, it emphasizes AR4D actors thinking about innovation and curating innovation for the purpose of an outcome and innovating for that, rather than inserting innovation where they think it might best fit or in a ‘gender-blind’ manner.

These outcomes can vary depending on the context and stakeholders involved. Engaging this approach requires strong collaborative frameworks that consider various viewpoints, possible trade-offs, and intricate connections between various outcomes. Additionally, comprehensively monitoring, evaluating, and implementing adaptive management strategies is necessary to understand the ways in which innovations contribute to these outcomes, as the authors explain further in the next recommendation. Successfully implementing and navigating outcome-based scaling requires advanced methodologies to understand and evaluate systems approaches, and a thorough comprehension of contextual factors and representation of voices to ensure meaningful and sustainable impact across diverse dimensions of development.

Future Research: Investigate the benefits and challenges of outcome-based scaling, as well as techniques for integrating holistic perspectives into innovation and ways to empower beneficiaries by giving them greater choice and agency. Increase systems literacy among AR4D actors to improve capacity to intervene in situations of complexity for improved outcomes.

Measuring complex change by applying dynamic and empowerment evaluation approaches

Ensuring food and nutrition security, environmental health and biodiversity conservation, gender equality, and make progress across 17 SDGs simultaneously is complex. The pressure to show contributions to impact can encourage strategies focused on “low-hanging fruit” and easily quantifiable outputs, which can discourage instead of encouraging novel approaches. To adequately address this complexity, innovation and scaling endeavors must adopt systems approaches and utilize metrics and frameworks to quantify systems change. Understanding and influencing directionalities in agricultural systems becomes challenging without indicators that analyze systemic impact dynamics⁴¹. Overcoming data challenges necessitates adapting indicators to diverse contexts and prioritizing real-world inquiries over data availability.

Key performance indicators for AR4D interventions or research often prioritize quantity over quality, resulting in inconsistent and, at times, unfavorable development outcomes. The standard project duration of 3–5 years promotes the production and measurement of faster deliverables, such as farmers trained, instead of measuring social change, which can take longer and be impacted by more confounding variables. The need for tax-funded projects to “succeed” also creates a perverse incentive to avoid risky behaviors like exploring new research niches or partnerships. This incentive can hinder the desire to “fail forward” or implement lessons learned from failures, resulting in overcommitment to less viable routes. There should be a shift towards adaptive project management guided by outcome-oriented monitoring instead of the current output-focused approaches (see 2.3 above). This suggests that the theories of change or impact pathways devised initially may develop in different directions and, therefore, achievement indicators need to be adjusted accordingly.

Inclusive monitoring involves adding specific data points that capture the transformative goals we aim to achieve. For instance, promoting social inclusion requires incorporating dimensions of gender and intersectionality. As a baseline, collecting sex-disaggregated data is crucial to measuring meaningful societal changes. Currently, systematic methods such as reflexive, dynamic, transformative, and design-driven outcome mapping and evaluation approaches exist to assess complex, long-term transformations^43,44. However, such rigorous methods to evaluate the effects of interventions on marginalized populations remain underutilized in AR4D efforts. This gap poses a challenge in understanding interventions’ long-term effectiveness and short-term aftermath.

Following an empowerment evaluation model⁴⁵, for example, during project design, AR4D actors can establish goals for what they want to achieve (e.g., increased nutrition) and anti-goals for things they wish to avoid (e.g., increased marginalization of populations) that align with a recognized desired directionality²⁰. This involves creating suitable and adaptable indicators to better understand the innovation’s impact on end-users and anticipate any negative outcomes that may arise from scaling. Establishing goals and indicators also involves creating a monitoring, evaluation, and learning plan that can adapt to iterative research plan changes and scaling strategies while accounting for intended and unintended system changes over time and space – and includes intended ‘beneficiaries.’

Future Research: Incorporate metrics and indicators to assess systemic impacts while exploring the impact of inclusive and collaborative approaches in achieving long-term systemic change. Encourage donor and researcher investment in alternative monitoring and evaluation methods that consider complexity and transformative change. Create comprehensive frameworks that avoid ‘win-washing’ to evaluate agricultural innovations’ enduring and dynamic effects on social outcomes, incorporating methods to assess equity, resilience, and the empowerment of marginalized groups, along with other social dimensions impacted by AR4D interventions.

Acknowledging and acting on social differentiation throughout the innovation and scaling process

Although AR4D actors pursue goals to improve livelihoods of specific marginalized populations, they often fail to prioritize the “for whom” when designing innovations and scaling strategies, at the outset that can effectively achieve these goals. Restrictive social norms⁴⁶ and gender-blind programmatic design unintentionally marginalize women and other social groups in the agricultural development context⁴⁷. Current scaling initiatives often neglect socially differentiated impacts and do not foresee long-term consequences or trade-offs for socially disadvantaged groups²¹. Social differentiation must be considered within innovation and scaling processes to ensure they are genuinely inclusive. This means employing deliberate processes and interventions that address the needs of the most marginalized groups and setting inclusive goals that aim to transform societal norms.

Intersectionality, a feminist research methodology developed by Crenshaw in 1991⁴⁸, emphasizes the significance of comprehending how various aspects of social differentiation influence individuals’ experiences and outcomes. It emphasizes ethical practices by adopting the precautionary principle of “do no harm,” which dictates that research should not pose any additional risks to the individuals involved – or not involved⁴⁹. Acknowledging intersectionality requires tailored scaling approaches that consider evidence that incorporates users diversity into innovation and scaling strategies that respond to their unique needs and cultural landscape. Notably, AR4D aims beyond ‘do no harm’, underscoring the necessity of incorporating more socially differentiated approaches.

An innovative approach to addressing gender and social inclusion requires a paradigm shift towards considering diverse intersectional profiles and their unique needs and capacities. For instance, farmers’ identities can be better understood by using typology approaches, enabling tailored interventions that respond to farmers’ needs and capacities^50,51,52. For example, “women” is a deeply heterogeneous group, and depending on other social variables such as income and ethnicity will determine a sub-group’s challenges and opportunities. Tailored approaches can enhance social impact by recognizing and tackling the diversity within communities.

There is thus a need to introduce intentional interventions and inclusive goals to change social norms, consider social differentiation factors, such as gender, ethnicity, and socio-economic status, at all innovation and scaling process stages, and employ participatory methods that empower underrepresented groups in decision-making and guarantee a fair distribution of benefits.

Future Research: Further develop and test strategies for incorporating social differentiation in innovation and scaling; assess the role of typology approaches and application of social theories in creating tailored interventions for diverse groups of farmers. Embolden gender-transformative approaches that tackle the root causes rather than the symptoms of inequality.

As a multidisciplinary and diverse group of researchers, the authors identify critical challenges that hinder progress within AR4D and opportunities to accelerate momentum towards achieving the SDGs by 2030. To foster more impactful innovation, scale for sustainable development efforts, and transform AR4D practices, we propose a baseline set of recommendations. These recommendations emphasize interdisciplinary collaboration and inclusive, systems-based approaches that are ready for immediate implementation. By adopting these strategies AR4D practitioners can effectively navigate complexities, prioritize social inclusion, drive systemic change, and promote the development of equitable agricultural systems. Further, these recommendations are valuable not only to achieve the SDGs but also as essential goals in their own right.