An integrative modelling approach to understanding human responses to hydrogen energy technologies
Project lead
Dr Mitchell Scovell, mitchell.scovell@csiro.au
Lead researchers
Mr Daniel Herr
Dr Slava Vaisman
Challenge
Past research has shown that many theories of human behaviour help to explain people’s acceptance of energy technologies. There is, however, currently a lack of research drawing on this research to develop models for predicting future human responses to technological developments.
This project aims to fill this gap by using knowledge from the fields of social and computer sciences to explain and predict likely human responses to future hydrogen technology developments.
What we are doing
The primary objective of the project is to create and test comprehensive models of human behaviour, enhancing our ability to understand and predict human responses to the emerging hydrogen industry. With such knowledge, decision-makers should be able to better anticipate and address potential concerns before problems arise. Beyond its practical problem-solving focus, the project also seeks to advance fundamental knowledge in the realms of social and computer sciences and to promote more conceptual and methodological rigor in the quantitative social science and the machine learning literatures.
The main aim of the project is to develop and test integrative models of human behaviour to better explain and predict human responses to the emerging hydrogen industry. While this project does have focus on solving practical problems, there will also be a focus on sharing knowledge between social science and computer science disciplines to develop new methodological approaches.
Dr Mitch Scovell (L) and PhD candidate Daniel Herr (R) looking at statistical computer models.
One specific focus of the project is using theoretically informed Bayesian networks to understand the causal mechanisms of human responses to hydrogen. An advantage of this approach is that the model can be queried to answer counterfactual questions. In other words, we can ask the question: what might happen if things were different? This has many practical implications as we can explore how changes in policy, messaging or technological development might change how people think and behave.
The project will support the responsible development of the hydrogen industry in several ways. The outcomes will help to ensure that investment in this new industry is being appropriately spent in that people are willing to buy, use and live near these technologies. The findings will also inform relevant stakeholders (e.g., industry, government, and policymakers) about the important factors that influence hydrogen acceptance and demand, and how to best address factors that may impede the successful development of the industry. Finally, the findings will help to ensure that hydrogen energy technologies are implemented in a way that aligns with community values and expectations.
Beyond the practical implications, the project aims to promote more conceptual and methodological rigor in the quantitative social science and the machine learning literatures. Whilst social scientists have the necessary theoretical understanding of how people think and behave, many social scientists lack the formal training in mathematics, statistical theory and data science required to build and test formal representations of their theories. Similarly, statisticians/machine learning researchers often lack the relevant background knowledge in the theoretical and empirical concerns of the social sciences to make scientifically valid statistical inferences. By drawing on knowledge in both fields this project aims to bring insights from both disciplines to help develop new research approaches.
Outcomes to date
The team has prepared and submitted the findings from two subprojects. The first subproject used a theoretically informed Bayesian network to explore causal associations between various psychological factors and local hydrogen hub acceptance. The resulting model can be used to explore the implications of what might happen to local acceptance if we were to intervene (e.g., using change to policy or messaging) on specific psychological factors. We found that intervening on specific psychological factors (e.g., risk perception and the perceived economic benefits) has a relatively strong influence on local acceptance.
The second project focused on addressing problems with measurement error in statistical models. Social scientists are faced with the common challenge of trying to measure unobservable factors (e.g., attitudes or beliefs) using observable data (e.g., responses to a survey). It is, therefore, important to have statistical approaches that can account for measurement error to get the most accurate inferences from data. In this project, the team developed a novel way to model degradation processes (e.g., degrading attitudes over time) using cross entropy methods to address the measurement error problem.
Project finish date
July 2026
Relevant project publications
- Herr, Daniel; Vaisman, Radislav; Scovell, Mitchell and Nikolai Kinaev (under review). On alternative Monte Carlo methods for parameter estimation in gamma process models with intractable likelihood. IEEE Transactions on Reliability
- Herr, Daniel; Scovell, Mitchell; Vaisman, Radislav (under review). Hydrogen technology acceptance with Bayesian networks. Technology in Society