A social experiment is a research project conducted with human subjects in the real world. It typically investigates the effects of a policy intervention by randomly assigning individuals, families, businesses, classrooms, or other units to different treatments or to a controlled condition that represents the status quo.
The qualifier “social” distinguishes a policy experiment from a “clinical” experiment, typically a medical intervention within the subject’s body, and also from a laboratory experiment, such as a university psychology faculty might conduct under completely controlled conditions. In a social experiment, randomization to assigned treatment is the only element in the subject’s environment that the researchers control. All other elements remain exactly what they were
Social experiments are often referred to as “the gold standard” for program evaluation and reform processes. In measuring the impact of a social program, the researcher has to assess what the outcomes of the relevant population would have been in the absence of the program.Almost every naturally occurring comparison group, however, will differ from the composition of a non-random treatment group, usually because of selection bias (outside of an experiment, people choose to receive the treatment or choose not to). Randomization creates a control group that is statistically identical in a large sample with the group that is assigned to receive the treatment, and in principle there is no selection bias