Hilton San Francisco & Renaissance Parc 55 Hotel
San Francisco, CA | August 14-17, 2004

Event History Analysis

Monday, August 16, 2:30-4:10 p.m.

Leaders: Lawrence L. Wu and Jui-Chung Li, New York University

This seminar will provide a brief introduction to event history methods, with an emphasis on on continuous-time models and estimation techniques. Topics include the exploratory analysis of event history data, nonparametric methods, right censoring, maximum likelihood estimation, alternative specifications for a time dependent baseline hazard rate, time-varying covariates, multiple transition and competing risk models, and left truncation and left censoring.

Return to top

HIPAA Guidelines and Research in Medical Sociology

Sunday, August 15, 10:30 a.m. – 12:10 p.m.

Leaders: Richard Wagner, University of California, San Francisco; and Emily Kolker, Brandeis University

This seminar will discuss the practical implications for sociological research of the Health Insurance Portability and Accountability Act (HIPAA), also known as “The Privacy Rule.” HIPAA regulates disclosures of patients’ health information. The provisions and terminology of HIPAA relating to use of health information in research will be explained. Variations in institutions’ implementation of HIPAA will be discussed, along with what sociologists must do to gain access to and protect information in different settings. We will consider how researchers may identify and contact potential subjects in compliance with HIPAA. The interface between HIPAA and human subject protection regulations and Institutional Review Board (IRB) approvals will be explored.

Return to top

Integrating Quantitative and Qualitative Approaches

Monday, August 16, 8:30 – 10:10 a.m.

Leader: Lisa Pearce, University of North Carolina

This seminar is for sociologists interested in designing research that incorporates a mix of survey and ethnographic methods. There will be a brief introduction about the theoretical motivations and assumptions of an integrative approach, but the session will quickly move forward on the pragmatic assumption that mixing methods is possible and beneficial. The session will largely focus on practical applications of mixed method research for studies ranging from individual dissertations to large-scale team research projects. Specific topics will include sample design strategies, analytical approaches, how to present and publish findings, and how to obtain funding for mixed method research.

Return to top

Methodologies in Consumer Behavior Research

Monday, August 16, 2:30 – 4:10 p.m.

Leader: Leora Lawton, Population Research Systems

Sociologists are well-suited to study consumer behavior. A read of a consumer behavior textbook reveals that the field is essentially applied social demography. Nevertheless, few sociologists work in this area, perhaps due to underlying biases about working in the business sector, and perhaps due to lack of training. In this workshop, we will discuss the applications of qualitative and quantitative methods as applied to consumer behavior research, and go through two examples: customer satisfaction, and new product development. Participants will receive handouts.

Return to top

Multi-level Models

Sunday, August 15, 2:30 – 4:10 p.m.

Leader: William M. Mason, University of California, Los Angeles

Return to top

Qualitative Reasoning and the Internet: Problems, Progress, and Promise

Saturday, August 14, 2:30 – 5:30 p.m.

Leaders: Ed Brent, University of Missouri and Idea Works, Inc.; and Pawel Slusarz, Idea Works, Inc.

This seminar is for sociologists interested in using computers to reason about qualitative and unstructured data in teaching, research, and practice, with particular emphasis on the Internet. It will show how recent advances improve the efficiency of qualitative coding, provide tools to help generalize and refine concepts, and can be adapted for diverse practical applications. Examples focus on the Internet, describing ongoing projects using Qualrus™ that analyze open-ended responses to questions in web surveys in both French and English; code multimedia (including HTML web pages, rich text, graphics, audio, and video); grade essay questions submitted by students via the Internet; perform literature reviews of abstracts, full-text documents, and references from e-journals and digital libraries; and interpret unstructured questions about census data submitted to a web site. Several of these projects also illustrate strategies for collaborating in team research via the Internet from geographically dispersed sites.

Return to top

Simulations

Sunday, August 15, 8:30 – 10:10 a.m.

Leader: Michael Macy, Cornell University

What do flocks of birds, traffic jams, fads, drinking games, forest fires, riots, and residential segregation have in common? Traditionally, sociologists have tried to understand social life as a structured system of institutions and norms that shape individual behavior from the top down. In contrast, agent modelers suspect that much of social life emerges from the bottom up, more like improvisational jazz than a symphony. This seminar is intended for sociologists interested in learning how to model social life based on agent-level interaction, where an agent might be an individual, a firm, or a node in a network. The seminar will progress through a series of questions, from epistemological to methodological:

1. What are the theoretical and practical differences between modeling the interaction of factors (e.g., multivariate models) and the interaction of actors?

2. What are the strengths and weaknesses of agent-based models, compared to other model-building techniques (data-based, mathematical, and natural language)?

3. Is the agent-based approach relevant for macro sociologists? How can “bottom-up” models be used to study structural influences on macrosocial dynamics?

4. How should we evaluate agent-based models? Should we strive for simplicity or for realism? If the outcomes depend on the assumptions, how can one have surprising results? How should agent-based models be tested empirically? Should the assumptions be realistic? Should models be calibrated for predictive accuracy? Or should models be used to search for clues about puzzling outcomes?

We will address these questions as we carefully inspect some very simple agent-based models, including Schelling's "neighborhood segregation," Axelrod's "evolution of cooperation," and a Bush-Mosteller stochastic learning model of Prisoner's Dilemma.

Return to top

Social Network Analysis

Tuesday, August 17, 8:30 – 10:10 a.m.

Leader: James Moody, Ohio State University

Return to top

An Overview of Structural Equation Models

Monday, August 16, 10:30 a.m. – 12:10 p.m.

Leader: Kenneth Bollen, University of North Carolina, Chapel Hill

The workshop provides an overview of Structural Equation Models(SEMs). Sometimes called "LISREL," this general statistical model includes ANOVA, multiple regression, path analysis, factor analysis, and many other procedures as special cases. A distinguishing feature of the model is that it allows multiple indicators of latent variables (concepts) and it can estimate relationships that take account of measurement error. The seminar will examine the major subtypes of the model and the primary steps in applying it. We will discuss examples of programs and outputs from one or more of the major SEM packages. Participants should have a good background in multiple regression and should be familiar with matrix notation.

Return to top