The course discusses problem of statistical inference in the social sciences and techniques for improving the validity of these inference. Problems of statistical inference mainly occur because theories remain partly inconclusive in respect to model specification and thus model uncertainty emerges. Discussed techniques include (but are not limited to): randomized trials, regression discontinuity, matching, instrumental equation and structural equation models, and robustness tests.
The course also invites to participants to present their own work and discusses research design alternatives, analytical options, and communication strategies for empirical analyses.
The course will help Ph.D. students to develop research designs that allow deriving statistical and causal inferences from empirical analysis. Students will learn about the different methodologies, techniques and research designs for statistical and causal inferences.
Participation is compulsory and provides the basis for assessment. According to university rules, the attendance requirement is met if a student is present at least 80 percent of the time.
The course relies on a combination of seminar style teaching and participants' presentations of research designs or research papers. The students' research is a constitutive part of the content of the seminar. Students thus ought to be willing to present their own research.
Active participation in classroom discussions (20 percent), the presentation of a research design or of a research paper (50 percent) and homework (30 percent).
Appointments upon request. Please use email: firstname.lastname@example.org to arrange an appointment.