ICPSR Summer Program in Quantitative Methods of Social Research
2021 Summer Workshops by the Hobby School of Public Affairs
The Hobby School of Public Affairs announces the workshop offerings by University of Houston faculty for the 2021 Summer Program in Quantitative Methods of Social Research of the Inter-University Consortium of Political and Social Research (ICPSR).
Due to COVID-19 issues, these workshops will meet online and not physically at the University of Houston.
ICPSR was launched in 1962 as a national initiative for training social scientists in research methods, including mathematical modeling, statistics, data analytics, and experimental design.
ICPSR is hosted at the University of Michigan, and runs its flagship Summer Program in Ann Arbor. The Hobby School joins a selective group of academic institutions in the ICPSR network offering summer training workshops. Other hosts of these workshops include the University of Massachusetts at Amherst, University of California at Berkeley, Indiana University at Bloomington, University of Colorado at Boulder, and the University of North Carolina at Chapel Hill. ICPSR also has three international partners: University of Glasgow, Scotland; University of St. Gallen, Switzerland; and York University, Toronto, Canada.
These workshops aim at training students and professionals in the analytical tools and software used in the analysis of data in the social sciences. The Hobby School will offer five short workshops this summer.
- Linear Regression Analysis in the Social Sciences: May 17-28
- Introduction to R: May 24-28
- Multilevel Analysis: May 24-28
- Introduction to Causal Inference: June 7-18
- Social Science Data and Model Visualization in R: August 16-20
For more information about the ICPSR Summer Program by the Hobby School, please contact Pablo M. Pinto (email@example.com), Director, Center for Public Policy, or Scott Mason (firstname.lastname@example.org), Program Manager 2 at the Hobby School.
Workshops Offered by the UH Hobby School of Public Affairs
Linear Regression Analysis in the Social Sciences
Instructor: Patrick Shea, University of Houston (Department of Political Science) and Sunny Wong, University of Houston (Hobby School of Public Affairs)
Dates: May 17-28, 10:00 a.m. - 1:30 p.m.
Room: n/a (online)
Course Description: This course is designed for participants to develop quantitative research skills with applications to social science topics. Participants will gain an overview of research design, data management, and statistical analysis and interpretations of research findings. The course will be centered around several main topics covering the basic analysis of ordinary least squares (OLS), the technique of estimating bivariate and multivariate regression models, the overall fitness of a regression equation, and the hypothesis and diagnostic testings, and more. This course takes the "learning by doing" approach by discussing the major themes in regression analysis with detailed examples, which show how the subject works in practice using Stata.
Prerequisites: The level of the course will be approximately that of Gujarati and Porter's Basic Econometrics
Instructors: Ryan Kennedy, University of Houston (Department of Political Science)
Dates: May 24-28, 11:00 a.m. - 3:00 p.m.
Room: n/a (online)
Course Description: Most students learn to use R piecemeal, by performing tasks similar to what they have already learned in another program like Stata or SPSS. The result is often frustration on the part of both students and professors. This course takes a different approach. Utilizing the "tidyverse" tools, this course shows how writing code in R can be both easy and intuitive. We will cover graphics, data management and modeling, automated updating of tables, replication using markdown, functional programming, and, time permitting, application design. This course is not about repeating in R what students have already learned elsewhere, but to instead become R programmers capable of approaching any new challenge.
Prerequisites: Basic statistics
Software: R and RStudio (installation instructions will be submitted by instructor prior to start of workshop)Workshop Information:
Instructor: Ling Zhu, University of Houston (Department of Political Science)
Dates: May 24-28, 10:00 a.m. - 2:00 p.m.
Room: n/a (online)
Course Description: This workshop introduces participants to various statistical models suitable for multilevel data. Multilevel analysis is a widely used approach to deal with nested data, which allow researchers to consider both social contexts and micro-level characteristics in their statistical models. Multilevel analysis also provides advantages and flexibility for modeling longitudinal data. This course will focus on teaching participants the full roadmap of multilevel analysis: setting up and managing multilevel data, identifying data structures, choosing the appropriate model specification, evaluating fixed and random effects, interpreting and visualizing statistical results. We will start with applications in the context of hierarchical linear regression models, and then discuss several extensions suitable for time-serious-cross-section data and categorical dependent variables. Participants need to be familiar with the general linear regression approach and linear algebra, but this course does not require prior experience in multilevel data.
Software: R/R-Studio. Applications will be taught primarily using R/R-Studio, but familiarity with R is not required. I will also briefly introduce available STATA modules for commonly used multi-level models.
Introduction to Causal Inference
Instructor: Jessee Stephen, University of Texas at Austin (Department of Government)
Dates: June 7-18, 10:00 a.m. - 1:00 p.m.
Room: n/a (online)
Course Description: Causal inference has become increasingly important in many areas of academic research and applied statistical work. Researchers often care not just about describing relationships between variables, but also determining whether one variable causes another. The most direct method for learning about causality is a randomized experiment. But in many contexts, true experiments can be infeasible, unethical, or impossible.
This course will cover methods for making causal claims with observational, rather than experimental, data. It will cover the potential outcomes framework for thinking about causality and will introduce students to several approaches for estimating causal effects when true randomized experiments are not possible. The topics covered will include regression, matching, instrumental variables, differences in differences, and regression discontinuity designs. Students will also learn how to think about potential problems with causal claims including selection bias, controlling for post treatment variables, and other issues.
Prerequisites: Students taking this course should already be familiar with basic statistics including estimation, inference, and preferably have some knowledge of linear regression and related methods. Applications will be done in R but familiarity with R is not required and I will also mention how many of the methods can be used with Stata.
Software: R and Stata
Social Science Data and Model Visualization in R
Instructor: Boris Shor, University of Houston (Department of Political Science)
Dates: August 16-20, 11:00 a.m. - 3:00 p.m.
Room: n/a (online)
Course Description: This course will introduce social science students to modern methods of exploring and communicating data to a variety of audiences. The primary tools will be the R statistical programming language, RStudio as a development environment, and the tidyverse family of R packages (especially ggplot2) for visualization. The object will be to enable students to feel confident in descriptively exploring their data, communicating that to scholarly and outside audiences, and communicating inferential results for presentation and paper writing purposes. Throughout, good workflow practices for scientific writing will be emphasized.
Prerequisites: A basic knowledge of R is required
Software: R and RStudioWorkshop Information: