Research

WORKING PAPERS

Content Generation on Social Media Platforms: The Role of Negative Peer Feedback

with Jessica Fong and S. Sriram (Job Market Paper)

In recent years, user-generated content (UGC) on social media platforms has received increased scrutiny for creating a polarizing environment wherein users post extreme opinions and/or receive negative backlash from dissenters. In this paper, we ask how receiving negative peer feedback, in the form of downvotes, affects echo chamber formation and polarization by influencing UGC. We focus on two aspects of UGC: (a) propensity to post (i.e., incidence) and (b) the strength of the opinion expressed in subsequent posts, which we measure with the texts’ extremity (i.e., intensity). Using data from Reddit, we find that receiving negative (and positive) peer feedback increases users’ subsequent posting activity, relative to no feedback. Additionally, we do not find evidence that negative feedback exacerbates echo chamber formation by driving away users with unpopular opinions. However, receiving negative peer feedback moderates extreme sentiments such that when the initial opinion is extreme, users reduce the intensity of their subsequent opinions upon receiving negative feedback. This suggests that negative peer feedback can serve as the whip that regulates the polarization of opinions by encouraging users to moderate their tone. These effects of negative feedback are consistent with users attempting to maintain their reputation.

More Progress, Less Attrition: The Role of User-Generated Content in Online Education

with Ali Goli, S. Sriram, and Pradeep K. Chintagunta

Content platforms routinely face the important challenge of managing user attrition. Usually, managing attrition involves investing in content improvement, altering the monetization approach, or employing promotions. In this paper, we investigate whether platforms can leverage user engagement with discussion forums as a vehicle for retaining customers. We use a unique dataset from Coursera, a leading online education platform, for this study. Coursera's core offering consists of course material that includes lecture videos and supplementary readings. In addition, the platform provides space for user-generated content (UGC) through discussion forums related to course materials. We develop a measure to capture the relevance of the content on the forum homepage to a visitor and leverage the exogenous variation in that content as a potential shifter of engagement with UGC. This, in turn, allows us to identify the causal effect of a user's engagement with UGC on their likelihood of attrition from the course. Our results suggest that engagement with UGC facilitates a shift in users' learning pathways, indicating that they are able to move on to other topics after overcoming prior learning obstacles. Our analysis suggests that a policy that ranks UGC in the discussion forum in terms of relevance to each user can result in a 23.06% improvement in user retention relative to the existing UGC ranking policy of the firm. Thus, in the specific context of online education platforms, increasing user engagement with UGC can be a powerful lever that these platforms can use to improve customer retention.

WORK IN PROGRESS

The Role of Personalized Engagement-Optimizing Recommendation Systems in the Polarization of Beliefs

with Jessica Fong and S. Sriram

Social media platforms can choose a variety of different ways to measure engagement, such as dwell time, #likes, score = #likes – #dislikes, #shares, #comments, etc. Different ways of measuring and then optimizing engagement can lead to different sets of content items being selected for recommendation and also different content item sequences being selected for recommendation. This variability is crucial as both the content feed that is selected to be recommended and the order in which content items within the feed are recommended can significantly shape users’ beliefs on contentious topics. Thus, in this paper, we study how different choices available to platforms for measuring and then optimizing engagement affect the polarization of beliefs. Consistent with prior literature on polarization, our preliminary finding suggests that while users are more likely to like pro-attitudinal content items, users tend to spend more time on counter-attitudinal content items. We then conduct a simulation to evaluate the risk of amplification of belief polarization associated with different types of personalized engagement-optimizing recommendation systems. Interestingly, the preliminary results suggest that engagement-optimization is indeed accompanied by a risk of amplifying the polarization of beliefs, however, dwell time maximization is accompanied by a lower risk of belief polarization followed by like maximization and then score maximization.

Do Digital Payments Increase the Consumption of Vice Goods? Evidence from the Introduction of UPI Payments in India

with S. Sriram

The Government of India launched the Unified Payments Interface (UPI), an instant real-time cashless digital payment system in August 2016. The proliferation of UPI throughout India gave more than 260 million consumers the experience of making purchases digitally in a cashless manner for the first time in their lives. In addition, by enabling money transfer between the bank account of a consumer and that of the merchant using a single unified interface, UPI increased both the transparency of transactions and budget salience at the point of transaction. Thus, on the one hand, increased transaction transparency and increased budget salience suggest that consumers would curb their expenditure while using such a cashless digital mode of payment. On the other hand, prior research on consumer purchase behavior has shown that plastic-based cashless modes of payment result in consumers experiencing weaker pain of payment while partaking in a purchase, which in turn, increases their shopping expenditure. In this paper, we first collect data from two sources: (i) Cashless transactions made in different geographic districts around the nation on the largest UPI-based payment app, and (ii) survey-based reported household expenditures in different product categories. Next, we describe the association between the amount of household shopping expenditure and the volume of UPI-based cashless digital transactions. Our findings suggest that on average, there is a positive association between the amount of household shopping expenses and the increased proliferation of the UPI-based cashless digital mode of payment. Interestingly, we also find a positive association between the share of expenses on vice goods and the increased proliferation of the UPI-based cashless digital mode of payment. This finding provides suggestive evidence in favor of the pain-of-payment account. Our research aims to highlight the importance of the need to integrate insights from marketing research into policy design so that financial development initiatives don’t end up hurting the people that they are meant to help. 

PUBLICATION

“Unpacking overuse of androgen deprivation therapy for prostate cancer to inform deimplementation strategies”  Ted A. Skolarus, Sarah T. Hawley, Jane Forman, Anne E. Sales, Jordan B. Sparks, Tabitha Metreger, Jennifer Burns, Megan V. Caram, Archana Radhakrishnan, Lesley A. Dossett, Danil V. Makarov, John T. Leppert, Jeremy B. Shelton, Kristian D. Stensland, Jennifer Dunsmore, Steven Maclennan, Sameer Saini, Brent K. Hollenbeck, Vahakn Shahinian, Daniela A. Wittmann, Varad Deolankar, S. Sriram

Implementation Science Communications, 5(1), 1-13 (2024).

© June 2024 by Varad Deolankar