Methods of estimating causal effects 1600-SZD-WM-MEEP
The purpose of the course is to describe assumptions and way of using basic statistical methods allowing to estimate the size of effects in causal analysis. They will cover the following topics:
Linear regression - assumptions and their diagnostics, criteria of selecting control variables to achieve "unconfoundedness", use of interactions to model heterogeneity of individual causal effects. Parametric version of regression discontinuity design (RDD).
Logistic and probit regression - assumptions and their diagnostics. Marginal effects of various types (AME, MER, MEM) and its applications for estimating size of causal effects.
Use of panel and hierarchical data - controlling factors from different levels including ways to control for unobservables. Correlation of errors and its importance. Mixed effect regression models with a random effect for constant term. Variants of the difference-in-differences (DiD) method using panel data.
Methods based on the estimation of the probability of intervention: propensity score matching (PSM) and inverse probability of treatment weighing (IPTW) - variants, possibilities of combining with regression analysis.
Modeling error in the predictor: instrumental (IV) variables by the two-stage least squares (2SLS) method and (very short) introduction to structural equation models (SEM).
Classes will consist of a theoretical part, in which the assumptions and properties of the presented methods will be discussed, and a workshop part, in which participants will practice using these methods with selected packages of the R statistical environment.
Type of course
Course coordinators
Learning outcomes
WK3
UW1
Assessment criteria
Description of requirements related to participation in classes, including the permitted number of explained absences:
Participants should be able to perform and interpret results of the basic OLS regression and ANOVA analysis, as well as being able to perform basic data manipulation operations in the R statistical environment.
There is 1 absences allowed.
Assessment Tasks:
class attendance, active participation in workshop, preparing short report in which participant will demonstrate application of one of the methods described during the course to a research question chosen by himself
Learning Outcomes Assessment:
attendance list, workshop participants' activity, assessment of report
Assessment criteria:
active participation in classes 50%, assessment of report 50%
Bibliography
Aiken, L.S., West, S.G. (1991). Multiple Regression: Testing and Interpreting Interactions. Newbury Park, London, New Delhi: Sage Publications.
Angrist, J.D., Pischke, J.S. (2009). Mostly Harmless Econometrics: An Empiricist's Companion. Princeton University Press.
Biecek, P. (2011). Analiza danych z programem R: Modele liniowe z efektami stałymi, losowymi i mieszanymi. Warszawa: PWN.
Fox. J. (2003). Effect Displays in R for Generalised Linear Models. Journal of Statistical Software 8(15).
Fox, J., Weisberg, S. (2018). Visualizing Fit and Lack of Fit in Complex Regression Models with Predictor Effect Plots and Partial Residuals. Journal of Statistical Software 87(9). https://www.jstatsoft.org/v087/i09
Lumley, T. (2010). Complex Surveys: A Guide to Analysis Using R. Hoboken: Wiley.
Additional information
Additional information (registration calendar, class conductors, localization and schedules of classes), might be available in the USOSweb system: