47th GESIS Spring Seminar (2018)

February 19 - March 9, 2018, Cologne, Germany

 

The GESIS Spring Seminar (formerly ZA Spring Seminar) offers three consecutive one-week courses in advanced methods of quantitative data analysis for Social Scientists. It has been taking place in Cologne annually every spring for more than 45 years.

February 19-23, 2018

Lecturers:

Prof. Ian Brunton-Smith, PhD, United Kingdom

Dr. Nigel de Noronha, United Kingdom

 

Short Course Description:

This course introduces methods for modelling multilevel data structures, for example pupils nested within schools, or individuals nested within neighbourhoods. Starting with basic concepts in multilevel modelling and the fundamentals of random intercept and random coefficient models, the course will then cover more advanced topics including: nonlinear models for binary responses, repeated measures, multivariate models, cross-classified models, and spatial data structures.The course will use MLwiN.

February 26 - March 2, 2018

Lecturers:

Asst. Prof. Scott J. Cook, PhD, United States

Prof. Jude C. Hays, PhD, United States

 

Short Course Description:

This course focuses on detecting, estimating, and analyzing models of spatially dependent data. Spatial interdependence  that the actions, outcomes, behaviors of some units are affected by those of other units  is ubiquitous throughout the social sciences. It includes not simply geographic space, but any means by which we can conceive of units being linked (e.g., cultural ties, political affiliations, economic relationships). As such, many of the most interesting phenomena in political science have a theoretically meaningfully spatial component: contextual or network effects on individual voting behaviors and opinions; strategic decision making amongst two or more actors (e.g., countries in a conflict, parties in an election, votes in a legislature); the diffusion of demonstrations, riots, coups, and wars, etc. This course demonstrates how to effectively model such dependence using spatial and spatiotemporal econometric models.

March 5-9, 2018

Lecturers:

Dr. András Vörös, Switzerland

Prof. Tom A.B. Snijders, PhD, Netherlands

 

Short Course Description:

Stochastic actor-based models for network dynamics are models for statistical inference for network panel data, i.e., repeated measures of a network, or of network and behaviour, or of multiple networks, on a given group of actors (where some turnover of the group is allowed). This methodology combines network analysis and statistical inference by representing network dynamics by simulation models, akin to agent-based simulation, but with a flexibility that allows the connection with empirical data, expressing the operation of several 'mechanisms' jointly, and testing of hypotheses while controlling for other mechanisms that may also be operating. Statistical procedures for applying these models are implemented in the R package RSiena. Siena stands for Simulation Investigation for Empirical Network Analysis. The course will give an explanation of the model, and how it is applied to longitudinal panel data of networks, which may be combined with actor attributes, and/or with other networks as co-evolving dependent variables. There will be practical exercises with RSiena. Participants are encouraged to bring their own data for analysis.