Experimentation Without Randomised Controls

Abstract

Background

In an experimental study, researchers often have the ability to assign different treatments. This is often done with randomisation. There are many settings; however, where it is not desirable to use randomisation. It is unclear how to best design an experimental study without randomisation while still providing persuasive evidence.

Objectives

The aim of this study was to outline several approaches, broadly classified as quasiexperimental studies, where researchers can use methodologically sound alternatives to randomisation.

Results

The interrupted time series, phased inventions, withdrawal design, waiting list control group, stepped wedge design and regression discontinuity all represent approaches where careful nonrandom allocation to treatment groups can produce high-quality research findings.

Conclusion

Quasiexperimental studies can produce rigorous research findings. The allocation to treatment groups and the times of evaluation need to be carefully designed. Proper use of these quasiexperimental approaches can enhance research options in settings where the research team has control of allocation but finds randomisation to be problematic.