Analysing household energy-saving behaviours is crucial to improve energy consumption predictions and energy policy making. How should we quantitatively measure them? What are their determinants? This study explores the main factors influencing residential energy-saving behaviours based on a bottom-up multivariate statistical approach using data from the recent French PHEBUS survey. Firstly, we assess energy-saving behaviours on a one-dimension scale using IRT. Secondly, we use linear regression with an innovative variable selection method via adaptive lasso to tease out the effects of both macro and micro factors on the behavioural score. The results highlight the impact of five main attributes incentivizing energy-saving behaviours based on cross-variable analyses: energy price, household income, education level, age of head of household and dwelling energy performance. In addition, our results suggest that the analysis of the inverted U-shape impact of age enables the expansion of the energy consumption life cycle theory to energy-saving behaviours. (C) 2016 Elsevier B.V. All rights reserved.