The types of social outcomes that are studied in policy research and public and non-profit administration are usually determined by a complex set of factors. It is consequently difficult to understand the effect of a given policy or change in a specific factor on outcomes, and the conditions under which estimated effects can be considered causal. Experimental scientists solve this problem with careful laboratory controls and randomization. Most social scientists and policy analysts do not have the luxury of randomization and a controlled laboratory setting and must analyze pre-existing observational subject to confounding with multiple other variables. When common statistical tools such as multiple regression are misapplied or research designs are flawed, results can be biased and/or inefficient. This course trains participants in recognizing the proper use and abuse of basic methods and design in quantitative research. It also briefly introduces participants to new developments and methods in the social sciences, as well as their promise and drawbacks.
Main learning objectives:
In this course, participants will not be enabled to develop and run statistical models themselves, but become qualified to evaluate reports and develop the skills required in management positions to identify flaws in research design, violations of statistical assumptions, and other pitfalls that are often hidden in quantitative reports. Upon completing this course, participants will know better when, and when not, to trust the numbers.
Topics: The (sometimes flawed) practice of science; research design; descriptive statistics; statistical inference; common threats to internal and external validity; causality; communicating with data.
Teaching style: Lecture, discussion, group work, worksheets
Instructor
- Mark Kayser , Professor of Applied Methods and Comparative Politics