Program Schedule
May 11th, 2023
9:00-9:30 – Morning Gathering
9:30-9:40 – Opening Remarks
9:40-10:15 - Dan Braha, Identification of Social Influence Effects: A Complex Systems Perspective
Social influence has a significant impact on human behavior and decision-making. Influence sources can be independent of social context such as attitudes and personality, or products of social context such as family and friends. It is essential to distinguish intrapersonal from interpersonal social influence. This presentation will examine a novel method for identifying and analyzing the extent of interpersonal social influence based on large-scale observational data, as well as how it might be applied to two key political and economic issues: elections and financial market crises. The new approach for detecting and measuring social influence can be applied to a wide range of other social and managerial settings.
10:15-10:40 – Coffee Break
10:40-11:15 - Ana-Andreea Stoica, Diversity and inequality in information diffusion on social networks
Online social networks often mirror community formation in real-world networks (based on common demographics, interests, or affinities). Such patterns are often picked up and used by algorithms that leverage social data for the purpose of providing recommendations, diffusing information, or forming groups. In this talk, we'll discuss the influence maximization problem where multiple communities exist, showing that common centrality metrics may exclude minority communities from adopting the information diffused. Using the preferential attachment model with unequal communities, we'll characterize the relationship between homophily, network centrality, and bias through the power-law degree distributions of the nodes, and study the conditions in which diversity interventions can actually yield more efficient and equitable outcomes. We find a theoretical condition on the seedset size that explains the potential trade-off between outreach and diversity in information diffusion. To wrap up, we’ll discuss a novel set of algorithms that leverage the network structure to maximize the diffusion of a message while not creating disparate impact among participants based on community affiliation.
11:15-11:50 Sagit Bar-Gill Did you Know you are a Micro-Influencer? The Impact of Influence Awareness on Online Content Consumption
In recent years, micro-influencers have become an integral part of social media marketing strategies harnessing their local influence to drive brand outcomes. In this study, we posit that small-scale influencers’ awareness of their influencer status may be enhanced by targeted messaging, such that perceived influence affects online activity. We study the impact of perceived influence and its interaction with actual social media influence on news exploration and consumption in an online experiment setting. We find that enhancing individuals’ awareness of their social media micro-influence increases exploration intensity and content consumption. This effect is largely driven by individuals with relatively high influence levels. The results suggest that the micro-influencer status is, at least partly, subjective and manipulable, and will inform the design of digital content platforms, as well as influencer-based marketing strategies.
12:00-13:00 - Lunch
13:00-13:35 - Dana Turjeman, Detecting and Fixing Selection Bias in "Likelihood to Recommend" Surveys
Data fusion combines multiple datasets to make inferences that are more accurate, generalizable, and useful than those made with any single dataset alone. However, data fusion poses a privacy hazard due to the risk of revealing user identities. We propose a privacy preserving data fusion (PPDF) methodology intended to preserve user-level anonymity while allowing for a robust and expressive data fusion process. PPDF is based on variational autoencoders and normalizing flows, together enabling a highly expressive, nonparametric, Bayesian, generative modeling framework, estimated in adherence to differential privacy – the state-of-the-art theory for privacy preservation. PPDF does not require the same users to appear across datasets when learning the joint data generating process and explicitly accounts for missingness in each dataset to correct for sample selection. Moreover, PPDF is model-agnostic: it allows for downstream inferences to be made on the fused data without the analyst needing to specify a discriminative model or likelihood a priori. We undertake a series of simulations to showcase the quality of our proposed methodology. Then, we fuse a large-scale customer satisfaction survey to the customer relationship management (CRM) database from a leading U.S. telecom carrier. The resulting fusion yields the joint distribution between survey satisfaction outcomes and CRM engagement metrics at the customer level, including the likelihood of leaving the company’s services. Highlighting the importance of correcting selection bias, we illustrate the divergence between the observed survey responses vs. the imputed distribution on the customer base. Managerially, we find a negative, nonlinear relationship between satisfaction and future account termination across the telecom carrier's customers, which can aid in segmentation, targeting, and proactive churn management. Overall, PPDF will substantially reduce the risk of compromising privacy and anonymity when fusing different datasets
13:35-13:55 – Coffee Break
13:55-14:30 - Daniel Romero, Networks and Identity Drive Geographical Properties of the Diffusion of Linguistic Innovation
Adoption of cultural innovation (e.g., music, beliefs, language) is often geographically correlated, with adopters largely residing within the boundaries of relatively few well-studied, socially significant areas. These cultural regions are often hypothesized to result from either (i) identity performance driving the adoption of cultural innovation or (ii) homophily in the networks underlying diffusion. In this study, we show that demographic identity and network topology are both required to model the diffusion of innovation, as they play complementary, interacting roles in producing its spatial properties. We develop an agent-based model of cultural adoption and validate geographic patterns of transmission in our model against a novel dataset of innovative words that we identify from a 10% sample of Twitter. Using our model, we are able to directly compare a combined network + identity model of diffusion to simulated network-only and identity-only counterfactuals -- allowing us to test the separate and combined roles of network and identity. While social scientists often treat either network or identity as the core social structure in modeling language change, we show that key geographic properties of diffusion actually depend on both factors. Although network and identity each give rise to similar pathways of transmission between USA's counties, each one also influences different mechanisms of diffusion. Specifically, we find that the network principally drives spread to and from urban counties via weak-tie diffusion, while identity plays a disproportionate role in transmission to and from rural counties via strong-tie diffusion. Our work suggests that models must integrate network and identity in order to understand and reproduce the adoption of innovation.
14:30-15:00 – Coffee Break
15:00-15:35 - Elad Yom Tov, The in-situ effect of offensive ads on search engine users
Unscrupulous advertisers may try to increase attention to search ads by using offensive ads, known in other ad channels to increase attention and recall at the detriment to both individuals and society. In my talkI will describe our investigations into whether such ads, when shown to search engine users, may have a similar effect. We developed search scenarios, each with 4 versions of the search results page (SERP), where some of the ads were changed to be irrelevant and/or offensive. Crowdsourced judges assessed the number of annoying, offensive and irrelevant ads in each condition and found a strong correlation between the reported number of annoying ads and the actual number of offensive and irrelevant ads, suggesting people conflate these attributes. The effect of assessing these search scenarios was such that those who assessed the SERPs for themselves (1st person) reported lower positive affect and higher negative affect than judges asked to imagine the results were provided to someone else. In the latter case offensive ads also lead to slightly lower positive and higher negative affect. Finally, in a recall test, only 6% of judges reported seeing an offensive ad when using search engines. Therefore, our work should further detract advertisers from using offensive ads since, in addition to previously documented adverse effects, we show that such ads have a small but statistically significant negative effect on people’s emotional experience.
15:35-16:00 – Coffee Break
16:00-16:35 - Moses Miller, Sharing Atoms, not Bits: Exploring Social Influence in the Food-Sharing Economy
We explore the social influence mechanisms in the social sharing economy using unique data from a food-sharing platform (OLIO). OLIO is a peer-to-peer food-sharing platform, active in over 60 countries, that connects neighbors and local businesses to share surplus food and reduce waste. The platform has over six million users and has shared 25 million food portions. We analyze the activity in the network, both online messages and offline item pickups, over a five-year period. The study examines the network characteristics affecting sharing performance and the conditions under which interactions within the network may be translated to the physical efforts required to sustain the network. The research illustrates how the change in dyad tie temperature over time may affect the overall sustainable influence of sharing hubs in the network. Specifically, results show that a higher influence in the network results from a higher temperature (volume of messaging activity on network ties) rather than a higher degree centrality. More specifically, we investigate how changes in temperature can result in an increase in the entropy of the system, leading to more frequent and motivating food-sharing behaviors. Similarly, we present the relationship between activity and the distance associated with physical efforts in food sharing, such as the transportation of food items between network members. Our results may contribute to a deeper understanding of the factors influencing food-sharing behaviors in online platforms and inform the development of more sustainable sharing economy initiatives.