A randomized experiment is a type of controlled experiment in which subjects get randomly chosen to be part of either the control or experimental treatments. Typically these two groups are compared twice, before and after the treatment, to see whether or not the means of each group have changed. The objective here is to analyze a single independent variable and keep all other variables constant.
In a randomized experiment, the independent variable being analyzed can be comprised of more than one level. This can be implemented by having multiple experimental groups. For example, pretend that a clinical trial was analyzing a particular drug, in which the control group received a placebo (no drug) and the experimental group received the drug. The different levels which could be applied to the experimental group could be the different doses of the drug each experimental group received. For instance, one group could intake a low dose, another group a moderate dose and the third group a moderately high dose.
Furthermore, in a randomized experiment, there is an assumption held by the researcher that any external factors will influence the treatments equally. Thus, any major differences seen are a result of the independent variable being tested.
The use of randomized experiments is to minimize the bias that can be associated with experimental designs. Therefore, the concept of random assignment is tightly linked to these types of studies because it is essential in randomized experiments that the subjects pertaining to each group are randomly selected. In statistics, randomized experiments are commonly used for research and this makes understanding how to properly design these analyses to be of the utmost importance.