Self-selection bias impedes electric vehicle adoption

Currently, electric vehicle (EV) adoption in the United States is transitioning from early experimentation to stabilized adoption. However, barriers to mass adoption remain, particularly related to user perceptions and behavioral differences between non-EV and EV users. This study analyzes survey data from 996 US adults, including non-EV, plug-in hybrid EV (PHEV) and battery EV (BEV) users, using an econometric model that accounts for self-selection bias in evaluating users’ perceived importance of charging time-related features at EV charging stations (EVCSs). Results show predictive performance in both the selection model classifying EV groups (pseudo- \( {{\rm R}^2=0.05\sim 0.06}\) ) and the outcome model estimating perceived importance ( \( {{\rm R}^2=0.36\sim 0.38}\) ). The estimated treatment effects for current PHEV and BEV users are -0.920 and -0.975, respectively, indicating lower perceived importance under their observed EV adoption states relative to their estimated counterfactual non-EV states. In contrast, perceived importance among non-EV users slightly decreases by 0.016 under the PHEV adoption counterfactual and increases by 0.913 under the BEV adoption counterfactual. On average, transitioning from non-EV status to PHEV or BEV status significantly decreases perceived importance by 0.442 and 0.076, respectively, with these shifts larger than or comparable to corresponding observed simple differences. This suggests that charging-time concerns may be less salient under the estimated EV adoption states after accounting for self-selection. Our results indicate that self-selection bias can impede EV adoption by amplifying perceived barriers among non-EV users. Furthermore, by identifying and adjusting for self-selection effects, this research offers insights for designing targeted outreach strategies and infrastructure policies to support mass diffusion of EVs.