Advanced statistical methods 1600-SZD-WM-ZMS
This is a course on regression analysis for applied social sciences. It starts with ordinary least squares (OLS) regression and a profound analysis of its various assumptions to then familiarize students with regression models for some non-continuous dependent variables (e.g. count, multiple discrete choice). It combines computer labs with seminars to teach students how to use statistics and avoid common errors. It uses STATA and various additional software that can be installed within It (the so-called ados) to demonstrate how to use statistics in a transparent and correct manner. The seminar component shall make students sensitive to the common pitfalls seen in statistical analyses, such as focusing on R-squares, the use of standardized rather than unstandardized regression coefficients or the wrong interpretation of interaction effects in linear and non-linear models. It shall also discuss the importance and relevance of the issue of replication of statistical
results.
Type of course
Course coordinators
Learning outcomes
Has an in-depth understanding of ordinary least squares (OLS) regression and its assumptions Knows how to deal with the cases where OLS assumptions are violated
Knows how to avoid the common pitfalls in using OLS
Has an in-depth understanding of interaction effects in OLS regression
Has an in-depth understanding of logistic regression as a modelling approach in situations where the dependent variable is a binary one Knows how to interpret interaction effects in logistic regression and the difference between interaction effects in linear vs. non-linear models
Understands the problem of ceiling/floor effects in logistic regression and knows how to deal with it
Knows how to estimate the main types of discrete choice models, including multinomial logistic regression and conditional (fixed-effects) logistic regression
Is aware of the meaning and importance of independence of irrelevant alternatives (IIA) assumption Knows how to test the IIA assumption using Martin and Stevenson’s IIATEST software
Has an in-depth understanding of Poisson regression as a modelling approach in situations where the dependent variable is a count Understands the problem of over-dispersion in Poisson models
Has an in-depth understanding of fixed-effects linear regression
Knows how to use Tomz, Wittenberg and King’s CLARIFY software for computing and presenting statistical results
Knows how to use Iacus, King and Porro’s coarsened exact matching (CEM) software for creating matched (quasi-experimental) data sets Understands the importance of the issue of replication in empirical social sciences
Assessment criteria
Written research paper, active participation in classes
Bibliography
Rabe-Hesketh, Sophia, and Anders Skrondal. Multilevel and longitudinal modeling using Stata. STATA Press, 2008.
King, Gary. "How not to lie with statistics: Avoiding common mistakes in quantitative political science." American Journal of Political Science (1986): 666-687.
Brambor, Thomas, William Roberts Clark, and Matt Golder. "Understanding interaction models: Improving empirical analyses." Political Analysis 14, no. 1 (2006): 63-82.
Ai, Chunrong, and Edward C. Norton. "Interaction terms in logit and probit models." Economics Letters 80, no. 1 (2003): 123-129. King, Gary, Michael Tomz, and Jason Wittenberg. "Making the most of statistical analyses: Improving interpretation and presentation." American Journal of Political Science (2000): 347-361.
Górecki, Maciej A. "Electoral context, habit-formation and voter turnout: A new analysis." Electoral Studies 32, no. 1 (2013): 140-152. King, Gary. "Replication, replication." PS: Political Science & Politics 28, no. 03 (1995): 444-452.
Martin, Lanny W., and Randolph T. Stevenson. "Government formation in parliamentary democracies." American Journal of Political Science (2001): 33-50.
Additional information
Additional information (registration calendar, class conductors, localization and schedules of classes), might be available in the USOSweb system: