NCRM videos

What is an experiment? by Peter John

27-08-2014

An experiment is 'a study in which an intervention is deliberately introduced to observe its effect' (Shadish, Cook and Campbell 2002: 12). The researcher observes whether any change that occurs during the experiment is attributable to the intervention or treatment. Although natural experiments do occur - where by accident people or areas get a different intervention - the more common method is where a researcher or agency randomly allocates people or units to control and treatment groups, in what are often called randomised controlled trials (RCTs). Randomisation ensures that there are no differences in observed outcomes between those in the treatment or control groups other than from the treatment so the researcher may observe an unbiased estimate of the intervention (see figure below). Other experimental research designs include quasi-experiments (which lack random assignment) and design experiments (qualitative research in which an intervention in adapted over time in response to observed outcomes).