Laboratory experiments are highly controlled, with virtually every variable able to be set by the experimenter and participants able to be randomly sorted. The control comes at a loss of ecological validity--in other words, people may not act in a laboratory as they do normally.

Field experiments are much like laboratory experiments, but they take place in a natural environment. For example, there is a famous experiment in which a psychologist asks for directions and swaps with an assistant when the person being questioned is not looking to see whether or not that person will realize he or she is now talking to someone else. Because this takes place in a normal, everyday context, it can be more certain that people's reactions are natural but the study is less precise because the researchers cannot control exactly how many distractions there are to finely test how many it takes for a person to lose attention.

Natural experiments are when a condition in the real world approximates an experiment without the researchers actually setting one up. For instance, they might compare countries that have enacted gun control and ones that have not. In this way, the countries without gun control act as a quasi-control group. This is useful for conducting research that would be unethical or impractical to manipulate through an experiment but the tradeoff is that it loses a lot of ability to control for confounding variables.