Advanced Methods (General)
At least one of the three technical skills courses must be an advanced general analysis course chosen from the following (or an approved comparable course under the rubric COMPSCI, STATISTC, or BIOSTATS)
DACSS 713 Advanced Statistical Methods
This course will build on students' previous foundations in probability, statistical inference, and linear regression. An introduction to generalized linear models (GLMs) and multilevel (mixed effects/hierarchical) models will be followed by additional advanced topics at the discretion of the instructor. These will include special cases of GLMs and multilevel models and may also consider measurement of latent variables (e.g. factor analysis, IRT).
DACSS 756 Machine Learning for Social Sciences
This course will provide an overview of machine learning (ML) with special attention to social and behavioral analytics applications. Machine learning combines insights from artificial intelligence, probability theory, statistical inference, and information theory to help automate tasks involving pattern recognition, prediction, and classification. "Learning" is analogous to "inference" in statistics, and the modern statistical toolkit includes various machine learning methods developed to handle large (and messy) datasets. The course focuses on statistical learning and is a good second or third course in statistical methods for graduate students in the social and behavioral sciences. We will examine key supervised and unsupervised learning techniques and reflect upon appropriate and inappropriate applications of such approaches for those seeking to understand the social world. We shall also discuss the ethical issues involved in automated analysis and computer-assisted decision-making, including how they may sometimes help overcome human biases and, in others, only reinforce these tendencies. Sample Syllabus
Note: If the student is already taking one of the above courses in place of DACSS 603, they are required to take a second from among the following advanced methods electives list.
Advanced Methods (Electives)
At least one from this group:
DACSS 690AB Agent Based Modeling for Social Complexity Research
This course introduces students to computational approaches for studying complex social systems, combining foundations from complexity science with agent-based modeling (ABM) as a tool for theory construction, exploration, and policy-relevant reasoning. The course begins by developing core concepts of complexity - nonlinearity, feedback, emergence, attractors, and sensitivity to initial conditions - and uses these ideas to motivate why many social and policy phenomena cannot be adequately understood through linear, equilibrium-based, or purely variable-centered models. Topics may include social diffusion, threshold behavior, contagion, collective action, enforcement, network influence, delay, and accumulation. Through hands-on labs using NetLogo and/or BehaviorSpace, students explore how design choices, scale, and interaction structure shape system behavior. The course is designed for students in the social sciences, public policy, and related fields. No prior programming experience is required.
DACSS 695N Social Network Analysis
This is a course on network analysis. The study of networks across the sciences has exploded recently. In this course, we will cover network scientific theory as it applies to the social sciences, network data collection and management, network visualization and description; and methods for the statistical analysis of networks. The course will make extensive use of real-world applications and students will gain a thorough background in the use of network analytic software. Most of the applications discussed will be drawn from political science, but this course will be relevant to anyone interested in network analytic research.
DACSS 790N Network Inference
This course covers various approaches for network inference and delves into the following questions: How do networks we observe emerge? Under what conditions do they change (or not)? What are the network outcomes individuals get based on their structural positions and roles? How do information and resources move from one spot to another within and beyond networks? Where does the flow stop? What are the mechanisms that lead to changes to networks? Do people form (or dissolve) social relations because of shared similarities (or outright differences) with others, or because they tend to be influenced by others already in their networks? Extending the discussion on descriptive and structural characteristics of network data in Social Network Analysis (DACSS 695N), this course introduces statistical frameworks with which network dynamics can be investigated. Recapitulating issues that make it hard to marry conventional frameworks for statistical inference with network data, such as autocorrelation and relational dependency, this course starts with mathematical models that utilize some mechanisms of network formation and continues on to statistical models. In particular, this course introduces some well-known statistical network models such as exponential random graph models (ERGMs), network regression models, latent space models, and stochastic actor-oriented model (SAOMs), as well as models that consider temporality in statistical frameworks such as temporal exponential random graph models (TERGMs) and diffusion in networks.
DACSS 758 Text as Data
With the recent explosion in digitized text availability, social scientists increasingly turn to computational tools for analyzing text as data. In this course, students will first learn how to convert text to formats suitable for analysis. The course will introduce and proceed through tutorials on various natural language processing approaches to treating text as data. This will include relatively simple dictionary approaches for measurement, supervised learning approaches for document classification, vector representations, contextualized embeddings, and more.Sample Syllabus
DACSS 790T Advanced Text as Data (to be renamed Large Language Models)
Computational social scientists are increasingly leveraging the wealth of digital text along with powerful computing resources for the analysis of "text-as-data." In this 3-credit graduate course, we tackle advanced approaches for the systematic analysis of text, starting with general principles and progressing through a variety of approaches that better account for the complexity of text than simple bag-of-words models.. We'll explore more sophisticated representations like word embeddings, both static and contextual, and their applications in social science research. Throughout, we will build from an understanding of the underlying neural networks and deep learning techniques by creating our own models, before introducing transfer learning with pre-trained language models. The course concludes by focusing on how text is unlocking new possibilities in causal inference and careful research design.
DACSS 690D Spatial Data Analysis
This course introduces students to spatial analytics using both GIS and R, with primary focus on social scientific research, social justice concerns as well as social and public policy related inquiries. Exposure would also be given to the theoretical underpinnings of robust analysis as they pertain to spatial data analytics. Descriptive statistics in the context of spatial data would be the focus of the first third of the term, followed by an introduction to OLS and relevant estimation aims such as minimization of residuals. The last third of the term would be dedicated to various techniques of introducing robustness in the specification of spatial OLS. Instruction of topics will be conducted in a self-contained manner, however, some basic knowledge of R along with probability and statistics would be needed to prosper in the course. Evaluation will primarily be based on two class projects that engage the concepts introduced throughout the term.
DACSS 790D Temporal Dynamics: Time Series, Panel Data, and Event History
This course provides an introduction to statistical methods for analyzing temporal data, including time-series analysis, panel data modeling, and event history analysis. It equips students with the skills to model and interpret data that evolves over time and includes techniques for causal inference in temporal settings. Students will begin with foundational methods in time-series analysis, including descriptive tools, autoregressive integrated moving average (ARIMA) models, and autoregressive distributed lag (ADL) models. Key topics include stationarity, autocorrelation, unit root tests, intervention analysis, and spectral analysis. Advanced methods such as volatility modeling (ARCH/GARCH) and causal inference techniques will also be introduced. The course then transitions to panel data analysis, focusing on mixed-effects models and regression techniques for handling repeated measures. The final part of the course covers event history analysis, where students will learn statistical techniques for modeling time-to-event data, including discrete and continuous time models, duration dependence, time-varying covariates, competing risks, and repeated events. Throughout the semester, students will apply these methods to real-world datasets using R, emphasizing practical implementation and interpretation. By the end of the course, students will have a comprehensive toolkit for analyzing temporal data and conducting causal inference across a variety of social science disciplines, such as political science, public policy, and economics.
DACSS 790C Causal Inference
As social scientists, we not only want to identify correlations and patterns in data but also want to explain why those patterns exist. In this course, students will first learn the fundamentals of causal inference, including key concepts and the directed acyclic graph (DAG) as a broadly applicable modeling framework. From there, the course will introduce and proceed through tutorials on a variety of causal inference approaches. This will include methods such as natural and field experiments, mediation analysis, instrumental variables, difference-in-differences (DID), propensity score matching, regression discontinuity, and synthetic control.
DACSS 690E Experiments for the Social Sciences
From policy nudges to marketing to social mobilization, experimental research helps shape many aspects of political and social life. In the course, you will get a chance to participate in, discuss, and learn to design and analyze a wide range of experimental treatments and outcomes. In addition, the course will introduce students to computational tools and software used to design and analyze experiments. Topics include measurement and estimation of semantic priming, moderation and mediation models, conjoint design and analysis, cloud-hosting of response-time studies, mixed effects, and API access to crowd-sourcing platforms such as mTurk. Sample Syllabus
Other Technical Skills Electives
DACSS 585: Intro to GIS
This class serves as an introduction to Geographic Information Science (GIS). GIS is the science of spatial relationships, linking data to locations to explore relations between objects. Based in geographic thought and emerging from initial applications in natural resource management, GIS has evolved to be a universally applicable way of thinking and set of knowledge, skills, and practices. The goals of this course are to teach you basic GIS concepts through practice and theory, to enable you to make useful and meaningful contributions to various disciplines through spatial analysis. Throughout this course, you will be challenged to not only think spatially, but apply spatial analysis techniques within GIS. Sample Syllabus
DACSS 690C: Computational Social Science Methods
This course reviews different computational social science methods that are applied under different academic and professional situations. This includes different but complementary methods to format and explore data as tables, maps, graphs, and text. The course also includes basic methods for decision-making support such as optimization and social simulation. The course includes the use of version control applications, as well as promoting the practice of reproducibility and transparency. The course makes use of Python and R, including NetLogo for social simulation, Gephi for network exploration, and Git for version control. Those languages are not a pre-requisite for the course, as they will be used as templates.
DACSS 690R: Data Preprocessing
This course gives students the tools to collect, organize, cleanse, format, integrate and transform their data so that it is ready for analytical work in the social sciences. The course covers R and Python in parallel, so that students become familiar with these popular and powerful languages, while comparing that each one offers for the pre processing stage.
DACSS 611: Introduction to Python for Data Science
Python has gained immense popularity as a programming language due to its ability to handle diverse types of data, powerful libraries for data analysis, robust support for tasks such as web scraping and data extraction from online sources, and its widespread use in machine learning and deep learning communities. Python is known for its readability and ease of use, making it a favorite among beginners and seasoned programmers alike. This introductory course on Python for data science will focus on the essential tools that are particularly beneficial for social data scientists and data professionals. This course will provide you with a solid foundation in Python and equip you with the necessary skills to effectively work with data using Python.
DACSS 690V: Data Visualization for Policy, Management, and Social Research
This course gives students the tools to show insights to political or scientific communities, while presenting different strategies to avoid biased interpretations. Given the overwhelming computational toolbox for displaying information, the course follows a ’keep it simple’ approach from the beginning, starting from foundational topics relating color, nature of data, and the brain; and takes students to build their own visualization tools. Emphasis is placed on complex data such as networks, geography and multivariate models. While the course uses R, it makes no emphasis on programming and more on the building of templates to produce information.
DACSS 695SR: Survey Research Methods
This course will focus on advanced survey design and analysis topics. Topics covered include different approaches to sampling, how to construct and use survey weights, and tools for analyzing and enriching survey data, including approaches to conducting matching and multiple imputations, as well as the construction and analysis of panel data. The course will also focus on designing and analyzing survey experiments. Sample Syllabus