Correlation, causation, and dynamics: Methodological innovations in sustainable energy behaviour research

Abstract

The global consensus today states that climate change is due to human activities. The main driver of anthropogenic climate change is rising levels of carbon dioxide in the atmosphere. Encouraging households to engage in sustainable energy behaviours would help mitigate climate change. To understand which factors are related to sustainable energy behaviours researchers have employed different methodological approaches and statistical models. Typically, these methodologies can be classified into correlational research which involves exploring relationships, (field) experiments which are better suited to establish causality, and longitudinal designs which examine how relationships between factors and sustainable energy behaviours change over time. In this thesis, we introduced graphical models and generalised additive models to sustainable energy behaviours research, particularly to answers questions involving exploratory research, causal inference, and capturing differences in dynamic energy use patterns between groups. The first part of the thesis (Chapters 2 and 3) introduces tools for exploratory analyses. Exploratory analyses provide a first understanding of the relationships between items and variables related to sustainable energy behaviours, which enables researchers to better understand the data before opting for more complicated analyses. In Chapter 2, we illustrate how the Gaussian graphical model provides an easy to grasp overview of relationships between items and variables related to sustainable energy behaviours. Further, in new lines of research, so-called causal search algorithms can be used to explore probabilistic causal relationships between variables related to sustainable energy behaviours. To the best of our knowledge, the performance and applicability of causal search methods to sustainable energy behaviours research is yet to be investigated. In Chapter 3, we conduct a simulation study to compare the performance of two constraint-based causal search methods (PC algorithm and LiNGAM algorithm) in sustainable energy behaviours research. The results indicated that researchers must use these methods with care as the two methods tend to be inaccurate and sensitive to errors due to sampling variation. The second part of the thesis proposes novel tools for causal inference. Randomized controlled trials have been strongly advocated to evaluate the effects of intervention programmes on sustainable energy behaviours. While randomized controlled trials are the ideal, in many cases, they are not feasible. Hence, an important question is: which would be an appropriate solution to carefully evaluate effects of intervention programmes on sustainable energy behaviours when randomized controlled trials are not feasible? In chapter 4, we propose a potential solution to this question using graphical causal models and in particular, directed acyclic graphs (DAGs). Lastly, we focus on examining dynamic patterns in sustainable energy behaviours. To mitigate anthropogenic climate change, many households engage in sustainable energy behaviours such as purchasing photo-voltaic panels (PV). Purchasing PV can be a highly effective mitigation strategy, particularly when households utilize their PV in a sustainable way (e.g., use less electricity when PV production is low). However, the literature provides competing explanations on the likelihood that PV owners use their PV in a sustainable way. Extending previous studies which were based on self-reports, in Chapter 5, we use generalized additive models to study to what extent PV owners use their PV in a sustainable way using actual net electricity usage data obtained from smart meters. We did not observe differences between PV owning households and household without PV in net electricity use at times where PV production is low, suggesting that PV owning households do not use their PV is a sustainable way.