The paper highlights the importance of both local effects and market structure in determining the switching decision in retail electricity markets, and also introduces a new definition of household as the bundle of features of its members. We apply a Bayesian mixed Logit model to analyze the switching behavior in the Italian electricity retail market. The data set includes social, demographic, geographic and economic information regarding 18,448 households and 37,217 individuals together with their opinions, expectations and choices regarding national services. Results show that heterogeneity in the geographic and social context of choice and in the household composition has a major impact on the significance and on the identification of switching determinants implying that there is no one-size-fits-all policy to encourage switching and that further efforts should be dedicated to the definition of an optimal scale of analysis.