The strength of an experiment is its capacity to demonstrate cause-and-effect relationships. In order to confirm cause-and-effect, the researcher must construct a situation where all other possible causes are eliminated. The result may be an artificial situation that is removed from the real world, and hence lacks generalizability (external validity). Experiments are good for internal validity, but often lack external validity.
Quasi-experiments are more limited than true experiments with regard to internal validity because there might be extraneous variables that are associated with a particular treatment level, e.g., in comparing student performance in two classrooms using different training methods, one classroom teacher may be more competent than the other, and be the true cause of the difference. Teacher competence would confound the training method. The experimental groups may differ on variables other than the independent one. Data from quasi-experiments is correlational, not causal.
Single-subject studies are limited in external validity because the findings are based on a single case. Results from field experiments may be more generalizable to other events (higher on external validity) than is the case for lab studies, but there is a trade-off with regard to uncontrolled extraneous variables in the field.
Take self-test #4.
See the Methods Manual: Guidelines for running experiments
On to Summary