Stakeholder involvement in research processes is widely seen as essential to enhance the applicability of research. A common conclusion in the extensive body of literature on participatory and transdisciplinary research is the importance of the institutional context for understanding the dynamics and effectiveness of participatory projects.

Innovation platforms (IPs) form a popular vehicle in agricultural research for development (AR4D) to facilitate stakeholder interaction, agenda setting, and collective action toward sustainable agricultural development. In this article, the authors analyze multilevel stakeholder engagement in fulfilling seven key innovation system functions. Data are gathered from experiences with interlinked community and (sub)national IPs established under a global AR4D program aimed at stimulating sustainable agricultural development in Central Africa.

Agricultural innovation systems has become a popular approach to understand and facilitate agricultural in-novation. However, there is often no explicit reflection on the role of agricultural innovation systems in food systems transformation and how they relate to transformative concepts and visions (e.g. agroecology, digital agriculture, Agriculture 4.0, AgTech and FoodTech, vertical agriculture, protein transitions). To support such reflection we elaborate on the importance of a mission-oriented perspective on agricultural innovation systems.

The latest comprehensive research agenda in the Journal of Agricultural Education and Extension was published in 2012 (Faure, Desjeux, and Gasselin 2012), and since then there have been quite some developments in terms of biophysical, ecological, climatological, social, political and economic trends that impact farming and the transformation of agriculture and food systems at large as well as new potentially disruptive technologies.

Classical innovation adoption models implicitly assume homogenous information flow across farmers, which is often not realistic. As a result, selection bias in adoption parameters may occur. We focus on tissue culture (TC) banana technology that was introduced in Kenya more than 10 years ago. Up till now, adoption rates have remained relatively low.

Classical innovation adoption models implicitly assume homogenous information flow across farmers, which is often not realistic. As a result, selection bias in adoption parameters may occur. We focus on tissue culture (TC) banana technology that was introduced in Kenya more than 10 years ago. Up till now, adoption rates have remained relatively low.

Most micro-level studies on the impact of agricultural technologies build on cross-section data, which can lead to unreliable impact estimates. Here, we use panel data covering two time periods to estimate the impact of tissue culture (TC) banana technology in the Kenyan small farm sector. TC banana is an interesting case, because previous impact studies showed mixed results. We combine propensity score matching with a difference-in-difference estimator to control for selection bias and account for temporal impact variability.