My research combines economic and behavioral insights with advanced computational methods in causal inference and machine learning to tackle important managerial problems or answer questions of general interest in the social sciences. To this end, I use various forms of online/field/natural experimental designs for valid causal identification.

Working Papers

• Native ads and Credibility of Online Publishers – Reject and resubmit from Management Science

(Joint work with Manon Revel, Dean Eckles, Adam Berinsky, and Ali Jadbabaie)

This project explores how native ads impact publishers’ credibility perceived by their audience. Increasingly, news organizations have turned toward online advertising to increase their revenue and provide financial stability. Native advertising is a format of advertisement where the ads blend in with news and take the format and appearance of the non-advertising content that surrounds it. These ads are designed to have a clickbait style to maximize click-through rates, while the content could be often of low quality. As a result, many publishers using clickbait native ads may trade short-term revenues for long-term audience trust.

We tested this hypothesis through two large randomized survey experiments. Analyzing 1.4 million collected ad headlines, we found that over 80% of these ad headlines use a clickbait style, and that politics is among the most common topics in ads. Following the prevalence study, we used a sophisticated survey experiment in which we randomized real publishers’ brands, articles, and ad content to measure the causal impact of clickbait and political ads in implicitly deteriorating readers’ trust in publishers.

Findings from these experiments provide evidence that even a one-time exposure to clickbait and, to a lesser extent, political ads, significantly reduces readers’ perception of the publishers’ credibility. Furthermore, exploring heterogeneity of effects, we find that the least well-known publishers are the ones most negatively affected by these ads. Thus, online publishers, the public, and policy-makers perhaps ought to pay closer attention to the ads that appear alongside online news.

• Habits in Consumer Behavior: Evidence From Store Closures – Submitted to Management Science

(Joint work with Dean Eckles and Ali Jadbabaie)

This project studies the role of habits in the widespread experience of shopping in grocery stores. The idea that retail purchases are often habitual is broadly consistent with the finding in economics and marketing that consumers’ choices exhibit inertia. However, much prior work has simply distinguished all inertia from other sources of auto-correlated choices such as preference heterogeneity. We borrow insights from social psychology literature on habits to shed some light on the habitual aspect of this revealed inertia in choices in a setting of broad interest in marketing, economics, and public policy.

To this end, we use store closures as a shock that partially disrupts households’ shopping behavior to identify the role of formed shopping habits in repeated brand purchases. Our identification strategy is based on a combination of different households being exposed to store closures at different times, as well as within-household variation in how much that household is exposed to a closure for a particular product category. We apply a two-way fixed effects model in a difference-in-differences with a staggered timing framework to understand the resulting habit discontinuity. The results show that temporarily following the closure, households are more likely to purchase something other than their favorite brand in frequently purchase product categories, while they form new shopping habits in newly visited stores after a while. This effect remains significant even after accounting for reduced availability of their favorite brands.

The statistical evidence indicates the significant role of shopping habits in households’ decisions, and is inline with theoretical psychological predictions. Moreover, we find that the temporary disruption in habits results in a lasting impact on households’ brand choices, demonstrating the downside of habits leading to persistence of sub-optimal behavior. This would have immediate implications for firms who could benefit from understanding (or discovery) of these shopping habits by incentivising stores to keep the placement of their brands consistent inside the store, and also help policymakers design and implement more effective policies to promote healthy shopping habits.

Collaboration with the World Bank

I enjoy working on eclectic problems that fascinate me. I also have several projects in the pipeline that I am keen to advance. For example, since the Summer of 2021, I have been working as a short-term consultant at the World Bank, collaborating on two projects with the development data team under the supervision of Sam Fraiberger. 

• The first project is an RCT in the field, studying the impact of empathy training for Pakistani civil servants on their online pro-social behavior. We compute various indicators of pro-social language in their tweets using the BERT model and estimate the causal effect of the effective altruism training on their tweets’ content. For this project, I had to deal with many causal inference challenges. First, we had a small number of clusters, so using asymptotic inference methods was not feasible and I tried clustered bootstrap and randomization inference to have valid inference. Second, we had to account for spillover effects due to the interference among participants in the Twitter network.

• We have also been working on an online experimental study investigating persuasion gaps across stances on polarizing topics. So far, we have launched two surveys measuring how people anticipate others with alternative stances will update their beliefs after reading a snippet on economic inequality or climate change. Next, we are going to design an experiment to measure the actual belief update compared to the expected change. The results will shed light on how well people can predict the persuasiveness of new information for people who agree/disagree with them, which would impact their sharing behavior on social media platforms and hence the resulting polarization. 

Future directions

In the future, I am interested in exploring the links between social media habits and mental health problems. Digital technologies in general, and social media in particular, constitute an over-increasing and undeniable part of our daily lives. Yet, the scale of their impact on various behavioral and cognitive abilities are understudied. I believe leveraging my experience in behavioral sciences and experimental design, I can contribute to this area which is of paramount importance to society and policy-makers. I’m looking forward to collaborating with my future colleagues from diverse backgrounds.