Income, housing unit size, number of appliances owned and other components of socio-economic status (SES), are variables generally considered to influence residential energy use. Using structural equation modeling of SES indicators from the U.S. residential energy consumption survey (RECS), we provide evidence that SES, usually modeled as a latent construct with effect indicators, is better conceptualized as including at least some causal indicators. We studied the mediating effect of housing unit size and number of owned appliances on the relationship between SES, household size and residential energy consumption (REC). We found that household size was positively associated with REC, as a direct effect. SES had a strong impact on REC, while being mediated by housing unit size and number of appliances owned. In conclusion, research must take the latent nature of SES into account to uncover its total influence on REC. (C) 2019 Elsevier B.V. All rights reserved.