We analyze the selection of high efficiency windows by builders of new housing units in the United States from 2000 to 2010. Windows are among the five most important technologies impacting energy use in structures. Focusing on windows provides insights into the decisions that result in energy efficient houses and the factors affecting those decisions, which can be muted or completely missed when looking at building ratings or other aggregated estimates. The study analyzes a large data set for the continental United States, applying the Least Absolute Shrinkage and Selection Operator (LASSO) model selection and cross validation of the training set model with a randomly selected validation data set. Our findings strongly support the importance of climate and energy costs in decisions on energy efficient housing, with important but smaller effects for public policies and incentives. We also find that taxing and insurance policies that increase the overall costs of construction can have negative impacts on the diffusion of energy efficient products. (C) 2015 Elsevier B.V. All rights reserved.