B-M Model, Knowledge, Trait, Pool


  • BK Mahalder
    bidyuth.mahalder@gmail.com (##journal.primary_contact##)
    Food and Agriculture Organization of the United Nations (FAO)
  • MB Ahmed Khulna University, Khulna, Bangladesh
  • H Bhandari International Rice Research Institute (IRRI) Bangladesh Office, Dhaka, Bangladesh
  • MU Salam 115/3 East Bhurulia, Joydebpur, Gazipur, Bangladesh
  • S Chakraborty Faculty of Engineering and Infromation Technology, University of Technology Sydney, Australia


Quantifying knowledge on agriculture can have many benefits to stakeholders. While many knowledge-based systems exist in modern days for farmers’ decision support, specific models are lacking on how knowledge traits can impact on agricultural production systems. This study employed modelling technique, supported by field data, to provide a clear understanding and quantifying how knowledge management in production practices can contribute to rice productivity in the environmentally stressed southwest Bangladesh. This research accounted for ‘Boro’ rice as the target crop and ‘BRRI dhan28’ as the test variety. The ‘B-M Model’ was developed following the principle and procedure from published literature, ‘brainstorming’ and data from field surveys. Three knowledge management traits (KMT) were defined and quantified as the inputs of the model. Those are: self-experience and observation (SEO), extension advisory services (EAS) and accessed information sources (AIS). The yield influencing process (YIP), the intermediate state variable of the model, was deduced by accounting for the two dominant agronomic practices, seedling age for transplanting and triple superphosphate (TSP) application. ‘Knowledge drives farmers’ practice change which in turn influences yield’ was composed as the theoretical framework of the ‘B-M Model’. The model performed strongly against an independently collected field data set. Across the 180 farmers’ data, the average relative rice yield (RRY) predicted by the model (0.705) and observed in the field (0.716) was close (root mean squared deviation (RMSD) = 0.018). The difference between predicted and observed RRY was not statistically different (LSD = 0.03), indicating the model fully captured the field data. A regression of predicted and observed RRY explained 96% variance in observation, further proving the model’s strength in estimating RRY in a wider range of farmers’ rice yield. In a normative analysis, the practicality and usefulness of the model to stakeholders were simulated for the understanding of how much achievable yield could be expected by changing farmers’ knowledge pool (the sum of three KMT) on rice production practices, and at what combination(s) of KMT to be considered at strategic hierarchy to materialize a targeted achievable yield. To the best of the knowledge, a model quantifying rice yield in relation to knowledge management trait does not exist in literature. Upon successful testing under diverse yield scenarios using multiple and sophisticated statistical tools that enhanced the credibility of the model, it is concluded that the model has the potential to be used for identifying quantitative pathways of farmers’ knowledge acquisition for practice change leading to improved productivity of rice in the southwest region of Bangladesh.