Synthetic survey data in renewable energy acceptance research: A validity audit using PLS-SEM

Engineering notes

Engineering notes will be added by the Power for Solar editorial team.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为光伏效率、钙钛矿太阳能电池、储能技术、太阳能热利用、BIPV、并网技术等高价值论文补充中文说明。

Original abstract

Social acceptance is widely recognised as a critical determinant of household renewable energy adoption, particularly for residential solar photovoltaic uptake, yet empirical models often struggle to explain behavioural intention under conditions of uncertainty. Structural equation modelling (SEM) studies frequently report weak explanatory power for adoption intention, raising questions about whether such limitations reflect model inadequacy or the contingent nature of real-world decision-making. At the same time, synthetic survey data generated by large language models are increasingly proposed to augment behavioural analysis, despite limited evidence regarding how such data behave under established modelling frameworks. This study compares decision structures derived from human survey data collected in Australia with those obtained from a synthetic large language model using two structurally identical partial least squares SEMs. Both models are evaluated using structural estimation, bootstrapping, and out-of-sample predictive assessment. The interview-informed conceptual model captures relationships among awareness, trust, digitalisation, contextual uncertainty, household dynamics, and purchase intention. Results show that the human dataset exhibits strong measurement validity but weak explained variance for awareness and purchase intention, reflecting the contingent and situational nature of household decision-making. The synthetic dataset produces a decision structure with slightly higher explanatory power yet remains limited in capturing contextual fragility. These findings indicate that synthetic data can support structural alignment testing but do not replicate practical uncertainty. In this study the synthetic data are evaluated as a candidate augmentation tool for acceptance research rather than as a behavioural model of households, and the comparison is treated as a structural stress test rather than as a test of measurement equivalence. The study highlights the importance of interpreting low explanatory power as a substantive feature of household energy decisions and cautions against treating synthetic data as behavioural substitutes in social acceptance research.

Links and sources

Comments