ensuring all choices are randomly selected by chance (using random name generator) rather than being determined by the researcher - most common example, when using separate groups of participants in different conditions, ensuring random allocation of participants to conditions to avoid researcher bias