Combining Choice and Response Time Data: A Drift-Diffusion Model of Mobile Advertisements
30 Pages Posted: 19 Dec 2018 Last revised: 22 Dec 2020
Date Written: September 16, 2020
Endogenous response time data is increasingly becoming available to applied researchers of economic choices. However, the usefulness of such data for preference estimation is unclear. Here, we adapt a sequential-sampling model — previously-validated to jointly ex- plain subjects’ choices and response times in laboratory experiments — to model users’ responses to video advertisements on mobile devices in a field setting. Our estimates of utility correlate positively with out-of-sample measures of ad engagement, thus providing external validation of the value of incorporating endogenous response time information into
a choice model. We then use the model estimates to assess the effectiveness of manipu- lating attention towards an advertisement. Counterfactual simulations predict that requiring users to watch some portion of the ad — as is the practice of some online platforms (e.g. YouTube) — generate only modest increases in click-through rates and revenue.
Keywords: Mobile advertising, Attention, Drift-diffusion model, Response times, Video advertisements, Skippable ads
JEL Classification: L81, M37, D03, D83, D87, C15, C22
Suggested Citation: Suggested Citation