A Theory-Driven Extension Model for Farmer Innovation Adoption and Impact

Thesis Abstract

This study investigates how a theory-driven extension model can enhance farmer innovation adoption and its subsequent impacts on productivity and resilience in agricultural systems. The central problem addressed is the persistent gap between technological innovations developed by researchers and their uptake by smallholder farmers, which undermines potential productivity gains and food security. The aim is to develop and validate an extension model that integrates established behavioral and diffusion theories to predict adoption trajectories and measurable impacts on farm performance. Specific objectives include (i) to synthesize a theory-based framework linking farmer attitudes, social networks, extension activities, and perceived usefulness to adoption willingness; (ii) to quantify the relationships among extension contact frequency, trust in information sources, perceived risk, and adoption timing using a longitudinal design; (iii) to evaluate the impact of adopted innovations on yield, input efficiency, and climate resilience indicators; (iv) to identify contextual moderators (landholding size, access to credit, and market access) that condition the adoption–impact pathway; and (v) to propose a decision-support toolkit for extension agents to tailor interventions. A mixed-methods design is employed, combining a longitudinal survey and qualitative inquiry across five agrarian communities. The population comprises smallholder farmers in the maize–legume belt of a mid-latitude region, with an estimated active population of 1,200 farmers. A stratified random sample of 420 farmers is selected to ensure representation by farm size and gender, with follow-up waves at 12 and 24 months. Data collection instruments include a structured questionnaire measuring constructs drawn from the Diffusion of Innovations and Theory of Planned Behavior, a social network inventory, and farm financial and production records. In-depth interviews and focus group discussions (n=40 individuals; 8 groups) are conducted to illuminate contextual nuances and validate quantitative findings. Instrument validity is established via expert review and pilot testing (Cronbach’s alpha targets of ?0.70 for all scales; test–retest reliability r ? 0.80). Data analysis proceeds in two strands quantitative analysis uses structural equation modelling (SEM) to test the hypothesized pathways among extension inputs, farmer cognitions, and adoption outcomes, and multiple regression to estimate the impact of adoption on yield, input use efficiency, and drought tolerance indicators; qualitative data are subjected to thematic analysis to extract patterns of trust, information channels, and perceived risk, triangulated with survey results. Expected findings include a robust, theory-backed pathway in which extension exposure increases perceived usefulness and trialability, which, coupled with high trust in extension agents and dense farmer networks, accelerates adoption timing and breadth. Adoption is anticipated to significantly improve maize yield and input use efficiency by 12–18% and 8–12% respectively, with gains in climate resilience metrics such as reduced yield variability under drought scenarios. Moderating effects are expected for landholding size and access to credit, with larger farms and credit-secured farmers showing higher adoption rates and greater benefits. The study will contribute to knowledge by integrating diffusion theory and behavioral theory into a single, operational extension model, offering empirical evidence on the causal chain from extension activities to adoption and measurable farm-level impacts. It will also provide a validated SEM framework and a contextual heuristic for extension programming. The practical significance includes a decision-support toolkit for extension agents that translates theoretical constructs into actionable indicators and monitoring protocols, enabling targeted interventions, improved resource allocation, and enhanced farmer learning ecosystems. The main conclusion should be that a theory-driven extension model can more accurately predict adoption and quantify impact when it explicitly incorporates farmer cognition, social contexts, and trust dynamics. Recommendations include scaling the model across diverse agro-ecologies, integrating digital information platforms to strengthen information credibility, and embedding training modules for extension agents focused on social facilitation, risk communication, and network-based outreach.