Lecture series generously sponsored by Yahoo!
Videos of some CSSI seminars are available here
Online reputation management: Estimating the impact of management responses on hotel reviewsJanuary 30, 2015 • 12:00pm-2:00PM
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30
Abstract: Failure to meet a consumer’s expectations can result in a negative review, which can have a lasting, damaging impact on a firm’s reputation, and its ability to attract customers. To mitigate the reputational harm of negative reviews many firms have adopted the strategy of responding to them. How effective is this reputation management strategy? We empirically answer this question by exploiting a difference in managerial practice across two hotel review platforms, TripAdvisor and Expedia: while hotels regularly respond to their TripAdvisor reviews, they almost never do so on Expedia. We exploit this distinction to identify the causal impact of management responses on reputation using difference-in-differences. We find that responding hotels see an average increase of 0.12 stars in the ratings they receive after they start responding. Moreover, we show that this increase is not due to hotel quality investments. Instead, we find that the increase is consistent with a shift in reviewer selection: consumers with a poor experience become less likely to leave a review when hotels begin responding. Our findings suggest that while management responses are an effective way to manage reputation, they can also obscure the measurement of a firm’s true quality by causing unfavorable reviews to be underreported.
Joint work with Giorgos Zervas (Boston University School of Management)
Bio: Davide Proserpio is a fourth year Ph.D. candidate in the Department of Computer Science at Boston University where he is advised by Professor Sharon Goldberg and John Byers and he frequently collaborates with Professor Giorgos Zervas. His current research involves leveraging concepts from computer science, statistics and economics to study complex social systems. Davide received his bachelor in telecommunication engineering from Politecnico di Milano (Milan, Italy) and his master in engineering from Carlos III University (Madrid, Spain).
CSSI talks hosted by the Department of Resource Economics:
Monday, January 25, 2015 • 9:45 a.m.–11:00 a.m. • Stockbridge Hall Room 303
Thursday, January 29, 2015 • 10:00 a.m.–11:15 a.m. • Stockbridge Hall Room 303
Wednesday, February 4, 2015 • 10:30 a.m.–11:45 a.m. • Stockbridge Hall Room 303
Friday, February 6, 2015 • 10:30 a.m.–11:45 a.m. • Stockbridge Hall Room 303
Monday, February 9, 2015 • 10:30 a.m.–11:45 a.m. • Stockbridge Hall Room 303
Hadley Room, Campus Center 10th Floor
Description: Research Computing encompasses many of the most important skills, tools and resources drawn upon in contemporary social science. From PhD students just beginning their training to senior faculty looking to expand upon the toolkit used in their established research agendas, understanding the myriad research computing resources available to social scientists on campus can be critical to successful scholarship. In this event, organized by the Institute for Social Science Research and co-sponsored with the Office of Research Development and Computational Social Science Institute, representatives from organizations on campus will present the many resources and opportunities available to graduate students and faculty through their respective organizations. The event will include a 30min lunch, then a brief presentation from each of four panelists, followed by a 30-40min Q&A. The objective is to provide social scientists with a comprehensive overview of research computing resources and opportunities available on campus. Panelists represent the Institute for Social Science Research (ISSR), the Office of Information Technology (OIT), and the Massachusetts Green High Performance Computing Cluster (MGHPCC). Resources and opportunities discussed will include software and methods instruction and assistance (ISSR); hardware, data storage and management (OIT); and high performance computing resources (MGHPCC).
Seminar cancelled due to snow preventing travel. To be rescheduled.
Heninger Distinguished Professor in the Department of Public Policy at the University of North Carolina
Intellectual Merits and Broader Impacts in Social, Behavioral, and Economic SciencesFriday, March 6, 2015 • 12:00 p.m., Lunch will be provided
Massachusetts Room at Mullins Center
Abstract: In an event co-sponsored with the Institute for Social Science Research, Department of Landscape Architecture and Regional Planning, and Office of Research Development, Maryann Feldman, NSF Program Officer for the Science of Science and Innovation Policy (SciSIP) program, will present information about NSF funding opportunities in the Social, Behavioral, and Economic (SBE) sciences.
Bio: Maryann Feldman is the Heninger Distinguished Professor in the Department of Public Policy at the University of North Carolina and winner of the 2013 Global Award for Entrepreneurship Research, presented by the Swedish Entrepreneurship Forum and the Research Institute of Industrial Economics. Feldman's research and teaching interests focus on the areas of innovation, the commercialization of academic research and the factors that promote technological change and economic growth.
Babur De los Santos
Assistant Professor of Business Economics and Public Policy, Kelley School of Business, Indiana University
E-Book Pricing and Vertical RestraintsMonday, March 9, 2015 • 10:00-11:15 a.m.
Stockbridge Hall Room 303
Abstract: This paper empirically analyzes how the use of vertical price restraints has impacted retail prices in the market for e-books. In 2010 five of the six largest publishers simultaneously adopted the agency model of book sales, allowing them to directly set retail prices. This led the Department of Justice to file suit against the publishers in 2012, the settlement of which prevents the publishers from interfering with retailers’ ability to set e-book prices. Using a unique dataset of daily e-book prices for a large sample of books across major online retailers, we exploit cross-publisher variation in the timing of the return to the wholesale model to estimate its effect on retail prices. We find that e-book prices for titles that were previously sold using the agency model decreased by 18 percent at Amazon and 8 percent at Barnes & Noble. Our results are robust to different specifications, placebo tests, and synthetic control groups. Our findings illustrate a case where upstream firms prefer to set higher retail prices than retailers and help to clarify conflicting theoretical predictions on agency versus wholesale models.
Computer Science graduate student recruitment event
Hidden State Inference and Prediction under a Mixed Effects Hidden Markov Model for Multivariate Longitudinal DataWednesday, April 8, 2015 • 10:00 a.m.
Arnold House, Room 113
Abstract: Mixed effects hidden Markov models (MHMMs) have been applied to both univariate and multivariate longitudinal data, with particular applicability in studies of addiction. In such studies, the longitudinal data are collected to capture some feature of the underlying disease process (usually abstinence from substance use), which is often not directly measurable. MHMMs are able to model the heterogeneity in the longitudinal trajectories which may be due to dynamic changes in the underlying disease states via the hidden Markov process, and also any between-subject differences through random effects. We apply this approach to bivariate longitudinal data generated from a smoking cessation clinical trial, where smoking status was monitored longitudinally during the study through patient self-report and physiological monitoring (carbon monoxide). The focus of this talk will be on inference drawn from the hidden states, with particular attention paid to their possible interpretation, when compared to the study’s outcomes, and differences between the results of univariate and multivariate models. A carefully motivated simulation study shows that, even when the observed data is contaminated by an unmodeled state-specific process, our multivariate approach performs better in predicting the underlying true hidden states when compared to the individual univariate models. These results suggest that using a multivariate response MHMM may have particular relevance in studies of addiction where observed data may be contaminated.
Estimating Population Susceptibility in Dynamic Models of Infectious DiseaseFriday, April 17, 2015 • 12:00pm-2:00PM
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30
Abstract: Epidemics of communicable diseases place a huge burden on public health infrastructures across the world. Advanced warnings of increases in disease incidence can, in many cases, help public health authorities allocate resources more effectively and mitigate the impact of epidemics. A central challenge in modeling infectious diseases is accurately estimating the (largely unobservable) population of individuals who are susceptible to infection. In this talk, I will review methods used to infer the susceptible fraction from aggregate time-series data and incorporate these estimates into models of disease transmission. Additionally, I will propose a new approach to accounting for the susceptible population over time based on observed case data. This method provides a simple way to include complex dynamics in otherwise standard statistical time-series models. Using over four decades of surveillance data on dengue fever infections from the Ministry of Public Health in Thailand, I will illustrate the ability of these methods to draw inference about mechanistic disease transmission models and predict the future spread of disease.
Bio: Dr. Nicholas Reich is an Assistant Professor of Biostatistics at UMass-Amherst. His research focuses on developing statistical models for analyzing infectious disease time-series data. His recent research, in collaboration with the Ministry of Public Health in Thailand, has yielded important insights into the complex global dynamics of dengue fever and was featured in the New York Times.
Life Sciences Laboratories, 6th Floor, University of Massachusetts Amherst
This event will bring together leaders in academia, industry, and government. Space is limted and registration is required. For additional information or to discuss any special needs, please contact the Office of External Relations and University Events at 413.577.1101 or firstname.lastname@example.org.
For additional details and to register please visit http://ds.cs.umass.edu/launch.