Conference Showcases Statistical Social Science

By Tory Brykalski - On November 6, 2015, the Institute for Social Sciences hosted its inaugural "Statistical Methodology in the Social Sciences" conference. Led by Professor of Economics A. Colin Cameron, the day-long event featured presentations of quantitative research by scholars from eight different UC Davis departments. Faculty and graduate students from eighteen centers, departments and graduate groups took the opportunity to meet fellow statistical methodologists and establish interdisciplinary connections, as well as to hear about research topics ranging from Obamacare to gender dynamics in virtual worlds.

Michael Levy, a PhD candidate in Environmental Science and Policy, gave a talk entitled “Estimation of Network Tie Probability with Exponential Random Graph Models.” Levy’s work, which primarily explores how people think about and solve sustainability challenges, uses statistical data to make sense of how social networks make connections. How do wine industry experts, for example, disseminate information about sustainability? How do different nodes in the social network—growers, producers, etc.—pick up and use this information?

Two models

Levy draws on two different network models to explore how information travels. The first assumes that information is free-flowing, dependent on functional connections. In this model, a lack of information can be solved by building more connections to different nodes within the network.

The second model assumes that information is not free-flowing, and that blocks in the network—mistrust, misrepresentation of data, etc.—prevent different nodes from receiving it. The solution in this kind of network is to find ways to forge links around the blocks.

Important questions

Levy’s research raises important questions about information and networks, and how the former travels through the latter. For social scientists interested in disseminating what we learn about the world and how it works, addressing those questions—as well as those asked by the conference’s other speakers—is essential.

How should we share our findings? By sending them to a node in a network to which we are already peripherally related, in order to build trust? Or by sending them to an entirely new network? Levy suggested that it may be the first, though his ongoing project may yet prove otherwise.

The conference featured the following presentations:

Siwei Liu, Human Ecology 
Dynamical Correlation: A New Method for Measuring Similarity with Multivariate Intensive Longitudinal Data

Shu Shen, Economics 
Distributional Tests for Regression Discontinuity: Theory and Empirical Examples 

Xiaoling Shu, Sociology 
Historical and Cohort Changes in Gender Attitude in the U.S.: An Age-Period-Cohort Analysis of Macro- and Micro-level Dynamics during 1972-2012

Cindy Shen, Communications
Gender dynamics in virtual worlds : Analyzing behavioral data from US and China

Dalia Ghanem, Agricultural and Resource Economics
Testing Identifying Assumptions in Nonseparable Models

Chris Hare, Political Science 
Comparing Parametric and Nonparametric Methods for Estimating Voter Locations in Ideological Space

Michael Levy, Environmental Science and Policy 
Estimation of network tie probability with exponential random graph models            

Emilio Ferrer, Psychology
Statistical and Exploratory Models for Studying Dynamics in Social Interactions

Adrienne Hosek, Political Science
Insuring Obamacare on the Health Insurance Marketplace