Experimental Designs

This entry introduces the principles of experimental designs, the role played by randomisation in establishing causal inference, and thus the importance of these designs in establishing what works in policy and practice. It explains how randomisation helps to eliminate alternate explanations for any relationship established between an intervention (independent variable) and an outcome or set of outcomes (dependent variables), in other words how randomisation deals with threats to internal validity. It discusses some of the issues that decision-makers need to consider before assuming that the findings of one experimental study will be applicable to other populations in other contexts, and what researchers can do to maximise a study’s generalisability or external validity.
This entry discusses ethics of randomisation and the role of theory, and introduces a range of practical and technical considerations, such as how and when to randomise, the choice of control conditions, units of randomisation, the implications of the choice of study design on sample size estimates and analyses, and the importance of transparency and explicitness in reporting studies using experimental designs. The entry concludes with an overview of some criticisms of randomised controlled trials and their deployment in policy and practice.