Quantitative/ Study design
Randomised experiment
Use random assignment to test effects.
Randomly assign eligible units to intervention conditions, then compare outcomes. Randomisation balances causes of outcomes in expectation, supporting a causal comparison when the design and analysis remain valid.
WHEN IT FITS
You need an intervention effect and assignment is ethical and feasible. Specify the population, intervention, comparator, and outcome before starting.
Strengths
- Provides a strong basis for causal identification.
- Reduces confounding through random assignment.
- Can estimate uncertainty around a prespecified effect.
Limitations
- Attrition, nonadherence, and spillovers can threaten inference.
- Assignment may be costly, infeasible, or unethical.
- Results may not transport to other people or settings.
Know the boundary
Random assignment is different from random sampling. A randomised experiment can support causal inference without representing the whole population.
USED ACROSS
Banking & financeBusiness & MBAPsychologyComputer scienceEducation