At ZSI, we offer expertise in both experimental and quasi-experimental research designs, providing robust frameworks for investigating causal relationships. Experimental designs are characterised by the fact that the research subjects are randomly assigned to a test or control group and all other conditions are controlled as far as possible. In this way, confounding variables can be minimised and the effect of an intervention can be estimated as precisely as possible. In quasi-experimental designs, random assignment is not possible. Instead, techniques such as propensity matching, regression discontinuity and instrumental variables are used to come as close as possible to the advantages of experimental designs. Quasi-experimental designs are particularly suitable for study designs in which randomisation is not possible, for example when comparing two predetermined groups, such as school classes.
