This 3-day master class with Prof. Thomas Cook deals with ways of making better causal
inferences in social and behavioural research. The masterclass participants will discuss the
implementation of randomized experiments in non -laboratory settings and then the design
and analysis of various kinds of quasi-experiments. Special emphasis is placed on
regression-discontinuity designs, interrupted time-series work and various methods for validly
(and invalidly) matching non-equivalent groups. A final section of the course deals with what
to do when matching is not possible but different groups must nonetheless be compared.