Access to Agricultural Extension Services and Adoption of Improved Agricultural Practices among Smallholder Sunflower Farmers: Evidence from Dodoma and Singida Regions, Tanzania

Abstract

The role of agricultural extension services (AES) in promoting the adoption of improved agricultural practices (IAPs) is widely recognised. However, existing studies often conceptualise extension access as a binary condition—whether farmers have contact with extension services or not-overlooking variation in how access occurs in practice. This study examined how differences in the frequency of interpersonal contact and extension communication modes jointly shape adoption behaviour among smallholder farmers. Drawing on the Diffusion of Innovations Theory, the study conceptualised extension access along two dimensions: frequency of interpersonal contact and modes of communication. A cross-sectional research design employing quantitative and qualitative approaches was used. Quantitative data were collected from 385 smallholder farmers through structured questionnaires and analysed using descriptive statistics and binary logistic regression in IBM SPSS version 23. Qualitative data were obtained through focus group discussions and were analysed thematically to complement quantitative findings. Findings indicate that adoption of IAPs remains relatively low, with only 41.8% of farmers classified as adopters. Binary logistic regression results revealed that the frequency of extension contact was significantly associated with IAPs adoption (OR = 1.902, p < 0.05). In contrast, digital and mass-media communication channels did not show statistically significant associations after controlling for interpersonal contact. Qualitative findings further revealed that farmers valued repeated interpersonal engagement because it enhances trust and continuous technical guidance. The findings suggest extension systems to strengthen interpersonal engagement while using digital and mass-media channels as complementary support mechanisms. However, the study is limited by its reliance on cross-sectional data, which constrains causal inference.