Abstract
This paper presents a methodology for predicting a product’s adoption and social impact using agent-based modeling (ABM) and neural networks to aid in decision-making related to the design and implementation of the product in a sociotechnical system. The collection of primary data on the social impact of a product is also outlined. Although this paper illustrates the method for improved cookstoves in Uganda, the general method can be applied to a wide range of contexts. A field study was carried out in Uganda, consisting of two phases of data collection. The data from the field work were used to train a neural network to predict if an individual would adopt an improved cookstove. Data collected from surveys and the trained adoption model were used to create an ABM to estimate adoption rates and social impacts experienced by households that had adopted technology and to assess social impact indicators. The contributions of this article are a method for collecting primary social impact data on a product and how to integrate those data into a predictive agent-based social impact model. This methodology also enables the examination of leverage points in the sociotechnical system to improve the social impact of a product as it is implemented in society.