Suzanna De Boef
I am an associate professor in Political
Science at Penn State University, former Director of the Social Science
Statistics Partnership—now QuaSSI, and associate editor of Political Analysis. I
also recently edited The Political Methodologist. My research interests
span statistical methods and the dynamics of opinion and vote choice. See links to my current research
projects, preprints and reprints, courses I teach and my Curriculum Vitae.
Contact
information:
320 Pond Laboratory
University Park, PA 16802
Office: 814-863-9402
Email: sdeboef@psu.edu
My research interests span the fields of methodology and political behavior. The questions that have motivated my research in both fields share a concern with the analysis of dynamic processes. Methodologically, I have devoted much of my research to identifying and explicating methods appropriate for the analysis of dynamic processes so that analysts can reliably test hypotheses about change across subfields and increasingly across disciplines. Substantively, my contributions include findings on how both opinions and votes are affected by changing economic perceptions, economic conditions, and the political landscape. The cornerstones of my current research agenda are a large methods project that focuses on event history models for repeated events and projects dealing with the interaction of economic conditions and attitudes with political opinions and behavior. Other projects include work that examines the effects of issue framing on public opinion both on individual opinions and on mass opinion over time. The focus of this project is effect of the innocence frame on attitudes toward the death penalty.
Current Research Projects:
Group Economic Performance, Economic Accountability and Economic Voting: This project, with Jonathan Nagler, argues that self-interested and rational voters look for economic indicators that provide them with information about growth and about how growth will be distributed. People care not just about how big the economic pie is, but as in all politics, they care about what part of it they get. This should affect vote choice and, more generally, evaluations of the incumbent. We examine presidential approval over time across different demographic groups of voters, and over different measures of economic performance.
Multilevel, Stratified Frailty Models and the Onset of Civil War (with Janet Box-Steffensmeier, Kevin Sweeney, and Kyle Joyce). The past decade has seen an increase in the number of studies of civil conflict in the comparative politics and international relations literature. We caution against the robustness of current empirical results for two reasons. Civil conflict is an extremely heterogeneous event with several causes that are measured inconsistently or unmeasured in the extant literature (may well be immeasurable). These covariates may exist both at the country and regional levels. Additionally, the occurrence of civil conflict in a state may affect the likelihood of additional, future conflict, creating event (or occurrence) dependence. Current analyses, we argue, do not adequately capture these theoretically and empirically important features of the process. We recommend a statistical technique to cope with possible heterogeneity and event dependence in the context of multiple levels of analysis and apply it to the study of civil conflict, offering a strategy for the analysis of repeated events processes that promises to produce more robust results in this important literature.
The
Discovery of Innocence: The Americans and the Death Penalty, 1960-2005
(with Frank R.
Baumgartner and Amber E.
Boydstun). This project that traces changing issue definitions and changing
politics associated with the death penalty. We seek to demonstrate the degree to
which the new “innocence” frame is displacing the traditional
“morality” frame and the consequences of this change for public
opinion and public policy relating to the death penalty. The link includes the book manuscript
(forthcoming Cambridge University Press early 2008), prospectus, and
replication files.
Taking Time Seriously (with Luke Keele). Applied time series analysis tends to neglect basic aspects of dynamic specification—estimating overly restrictive models, fundamentally misunderstanding equilibrium, and under-interpreting dynamic effects. Efforts to test theories about the dynamics of politics suffer. This paper presents a strategy to ensure rich, reliable dynamic specification. (Forthcoming in the American Journal of Political Science.)
Strategic Framing and Cognitive Response to the Death Penalty (with Frank R. Baumgartner, Amber E. Boydstun, Frank Dardis, and Fuyuan Shen), forthcoming Mass Communication and Society. This paper presents a theory of framing based on the nature of challenging arguments. Specifically our interest is in exploring how individuals think about frames that challenge their predispositions conditional on the dimension of the argument. We conduct an experiment and present evidence on the effectiveness of three different frames relating to the death penalty "innocence" frame, which cuts across the traditional pro/anti-morality dimension of debate.
Event Dependence and Heterogeneity in Duration Models: The Conditional Frailty Model (with Janet Box-Steffensmeier and Kyle Joyce), Political Analysis 15(3): 237-256. We introduce the conditional frailty model, an event history model that separates and accounts for both event dependence and heterogeneity in repeated events processes. Event dependence and heterogeneity create within subject correlation in event times thereby violating the assumptions of standard event history models. Simulations show the advantage of the conditional frailty model. Specifically they demonstrate the model's ability to disentangle the sources of within subject correlation as well as the gains in both efficiency and bias of the model when compared to the widely used alternatives, which often produce conflicting conclusions. Two substantive political science problems illustrate the usefulness and interpretation of the model: state policy adoption and terrorist attacks. Replication materials and a web appendix are also available.
Repeated Events Survival Models: A Conditional
Frailty Model (with Janet
Box-Steffensmeier). Repeated events processes are ubiquitous across a great
range of important health, medical, and public policy applications, as well as
political science, but models for these processes have serious limitations.
Published in Statistics
in Medicine Vol, 25, No. 20
(December), pp. 3518-3533, the paper focuses on models developed and
applied in the biostatistics literature. The methods and models are highly
relevant for the study of repeated events across disciplines. In the paper we
recommend a robust strategy for the estimation of effects in medical
treatments, social conditions, individual behaviors, and public policy programs
in repeated events survival models under three common conditions: heterogeneity
across individuals, dependence across the number of events, and both
heterogeneity and event dependence. We develop a new model for repeated events
processes that accurately accounts for the various conditions of heterogeneity
and event dependence by using a frailty term, stratification, and gap time
formulation of the risk set. We
examine the performance of these models and others that are commonly used in
applied work using
The Dynamics of the Gender Gap (with Tse-min Lin and Janet Box-Steffensmeier), American Political Science Review Vol 98, No.4 (August), pp. 515-528. We illustrate and explain gender differences in partisanship over time, finding that the gender gap in partisanship increases as objective economic conditions worsen, as the political environment grows more conservative, as the percentage of women with a college degree grows, and as the gap in the economic circumstances of men and women grows.
The Political (and Economic) Origins of Consumer Confidence (with Paul Kellstedt), American Journal of Political Science. Vol 48, No. 4 (October 2004), pp. 633-649. We argue that politics is important for understanding consumer sentiment beyond what we know from economic conditions. Specifically, we demonstrate a direct effect of political evaluations of the president’s management of the economy, the party of the president, extraordinary political events, and monetary policy, as well as an indirect effect of media coverage of the economy, on consumer sentiment, after controlling for economic conditions.
Modeling Equilibrium Relationships: Error Correction Models with Strongly Autoregressive Data (Figures 1 and 2), Political Analysis, Vol 9, No. 1, (Winter 2001), pp. 78-94.
Persistence and Aggregations of Survey Data over Time: From Microfoundations to Macropersistence. Electoral Studies, Vol 19, No. 1, (March 2000), pp. 9-29.
Testing for Cointegrating Relationships with Near-Integrated Data, with Jim Granato, Political Analysis, Vol 8, No. 1, (Winter 2000), pp. 99-117.
Near-Integrated Data and the Analysis of Political Relationships, with Jim Granato, American Journal of Political Science, Vol. 41, No. 2. (Apr., 1997), pp. 619-640.
Tracking Opinion Over Time: A Method for Reducing Sampling Error, with Don Green and Alan Gerber, Public Opinion Quarterly, Vol 53, No. 2 (Summer 1999), pp. 178-192. See http://research.yale.edu/vote/samplemiser.html to use the Kalman Filter algorithm presented in the paper.
The Dynamic Structure of Congressional Elections, with Jim Stimson, The Journal of Politics, Vol. 57, No. 3. (Aug., 1995), pp. 630-648.
Recent and Current Courses:
PLSC 427: Political Opinion: Fall 2007. PLSC 427: Spring 2000. This course begins with a focus on the nature and development of individual opinions, practical issues associated with gauging opinions through survey data, and the analysis of individual opinions. We then turn to mass opinion asking how it relates to individual opinions, how it moves over time, and how it drives politics. Considerable time is spent working with survey data on American political attitudes.
PLSC 502: Statistical Methods for Political Research: Fall 2005. Part of the required doctoral methods sequence in political science, the course begins with a review of essential topics in probability theory, statistical inference and hypothesis testing and covers regression analysis and simple extensions. The course focuses on how to use statistics to evaluate hypothesized causal relationships—how to estimate parameters and test hypotheses about them using regression, understand the statistical properties of the OLS estimator under the set of Gauss-Markov assumptions, and understand the statistical and intuitive meaning and consequences of violations of these assumptions, as well as learn a variety of ways to diagnose and correct for violations of the assumptions so that regression remains an appropriate tool for inference in practice.
PLSC 597C: Writing and Professional Development in Political Science: Fall 2007.Fall 2006. Fall 2005. Fall 2004 The course will focus on the challenges that lie in the last few years of graduate school and beyond: comprehensive exams, the dissertation, publishing and the job market. Class sessions will cover a variety of topics including writing for publication, criticizing as a professional, writing reviews, writing the dissertation, finding grant opportunities, writing grant proposals, preparing for comprehensive exams, responding to anonymous reviews and editor letters, creating posters, presenting your research, networking/presenting yourself to the profession, and preparing dossiers.
PLSC 597B: Monte Carlo Analysis: Spring 2005.
PLSC 597C: Teaching, Writing, and Professional Development: Spring 2005. The first of two courses in a required series on professional development in the graduate program in political science, this course focuses on issues related primarily to success during graduate school. Topics covered include: planning your graduate school years, putting together a curriculum vita, preparing for candidacy exams, selecting MA topics, and writing the MA; teaching topics including planning a course and recitation sections, preparing lectures or discussion classes, promoting active learning, dealing with problems in the classroom, as well as grading and evaluating learning; and writing research papers topics such as defining problems, crafting arguments, and outlining and revising manuscripts. Link to resources.
PLSC 597B: Advanced Quantitative Analysis:
Fall 2004. This course is an advanced graduate seminar on Statistical
Methods. Currently numbered 597, this course will become 504 and is designed to
be the third course in our basic graduate methods sequence. The course focuses
on extensions of the classic linear model and will cover time series and
limited dependent variable models.
PLSC 542: American Political Behavior: Spring 2004. This course is an advanced graduate seminar on American political behavior. The course focuses on classic readings and current controversies in theory and research on electoral behavior.
PLSC 597C: Pooled Time-Series and Cross-Sectional Analysis, Spring 2004. This course is designed to teach students to model relationships involving data that has both a time serial and cross sectional dimension. Examples include modeling voting outcomes across states and over time, attitudes across individuals and over time, budgets across agencies and over time, etc. These types of data allow us to conduct dynamic comparison, but they also present unique difficulties for the analyst. Link to course page:
PLSC 597C: Teaching and Professional Norms in Political Science, Fall 2003, Fall 2001 and Spring 2002. Welcome to graduate school and to the profession of political science! This course introduces incoming graduate students to the many roles they are likely to assume and tasks ahead in the first two years of academic life.
PLSC 597E: Time Series Analysis, Fall 2002, Fall 2000. This course considers statistical techniques to evaluate social processes occurring through time, covering time series econometrics, Box-Jenkins methods, and topics including: Granger causality and vector autoregression, pooled time series analyses, and cointegration techniques.
PLSC 503: Multivariate Analysis, Spring 2002. The main body of this course focuses on intermediate to advanced level single-equation regression techniques. More advanced topics such as maximum likelihood estimation and its uses (e.g., logit and probit, models for count data), and time series techniques will comprise the last half of the course.
PLSC 542: American Political Behavior: Fall 2001. This course is an advanced graduate seminar on American political behavior. The course focuses on a set of recent themes in theory and research on electoral behavior.
PLSC 497C: American Political Behavior, Fall 2002.
PLSC 405: The American Presidency: Spring 2000.
