Combining Choice and Response Time Data: A Drift-Diffusion Model of Mobile Advertisements.

This article focuses on applying a two-stage leaky drift-diffusion model (DDM), a type of sequential sampling model, to analyze smartphone users' binary choices and response times when interacting with mobile video advertisements. Using a large dataset of over 230,000 ad impressions from a mobile ad network, the study finds that incorporating endogenous response time data alongside choice data improves the estimation of users' preferences for advertised apps, as validated by positive correlations with out-of-sample engagement metrics such as advertiser bids, app installs, and in-app purchases. The model distinguishes between decision processes during the ad-play stage and the postad stage, revealing that users' attention and utility signals differ across these stages. Counterfactual simulations suggest that making ads partially nonskippable modestly increases click-through rates and revenue, primarily by affecting users less persuaded by the ad, who generate lower advertiser value. Overall, the research demonstrates the methodological advantages of integrating response times into economic choice models in real-world mobile advertising contexts.