We present an experimental method for optimizing compliance promotions, which are promotions in which customers must voluntarily participate (e.g. physician detailing, loyalty reward offers, retention campaigns, coupons). This entails an experiment to exogenously vary promotion features, a means to identify which promotion features can be causally extrapolated, an approach to extrapolate those causal effects, and an optimization over the promotion features, conditioned on the extrapolation.

We explore the impact of digitization on community participation across various interest communities. Overall, digitized events have lower prospective participation compared to in-person ones. However, participation varies significantly based on event characteristics and interest topics, suggesting a need for tailored digital community-building strategies.

Direct buy advertisers procure blocks of ad impressions at fixed rates from publishers. To find sites where their ads are effective, they must try many publishers, leading to wasted resources. By pooling advertiser information and recognizing the similarity in performance of ads with similar visual elements, we show substantial gains in advertiser welfare and publisher revenue can be achieved.

We develop an experimental paradigm for studying exposure to fake news. The paradigm works without experimenter deception and does not harm participants. We are still in the process of validating the paradigm.

We develop a dynamic learning model and fit it to browsing and link data from celebrity news sites. We simulate how banning links affects consumer browsing and find that linking increases celebrity news consumption, especially among consumers who browse the least.

Experimental samples that are too small to detect true effects have plagued the behavioral sciences for years. This study shows how much costs matter for sample size selection by analyzing data describing experiments conducted at a behavioral lab. Click on this card for a link to the manuscript, and an explanation of why I am no longer trying to publish this paper.

We present a predictive model of local demand for movies with two unique features. First, we allow consumers’ heterogeneity to depend on their voting tendencies. Second, instead of relying on genre classifications to characterize movies, we estimate latent movie attributes.

We examine the history of keywords used by articles published in “Marketing Science” to develop insights on the evolution of marketing science as a field.

Pay-for-performance programs score hospitals using process measures organized by therapeutic area. We propose an alternative method for scoring hospitals. Our method reflects how hospitals organize care, is better aligned with patient outcomes, and leaves room for hospitals to improve.