External validity is a property which enables research studies to be generalized to a larger population. External validity is an issue when constructing experimental and non-experimental research designs. External validity means how precisely the data as well as your conclusions drawn from the data (e.g., Change in A leads to change in B) represent what goes on in the larger population.
What are the Threats to External Validity in Research ?
Interactive effects of experimental arrangements: If the performance of individuals in an experimental group was influenced (favorably or adversely) by specific features of the experiment, or by the fact that it was seen by them as an experiment, results from the experimental group will not apply to samples from the general population who will receive the intervention in a nonexperimental setting. For example, lecturers in an experimental training course may change more than later trainees who basically receive a course on the same material. The more you regulate a situation in order to get valid data, the less the situation is like real life, and as a result, the outcomes you got in the contrived setting will not happen outside of it.
Selection biases: Selection biases are one of the most significant threats to external validity. When the sample which is being researched doesn’t represent the population that the investigator wishes to make generalisations to, there is a selection bias. Where selection bias happens, it is not easy to argue that the results which come from a biased sample can be generalised to the broader population. Example: The outcomes of an experiment in which teaching technique is the experimental treatment, used with a class of low achievers, don’t generalize to heterogeneous ability students. One method to deal with selection bias is to use random sampling to ensure that study samples are equivalent to the general population.
Reactive or interactive effects of testing: “Reactive” effects of testing implies that a pre-test by itself influences post-test performance. Example: An experiment in remedial reading instruction has an impact which doesn’t happen when the remedial reading program, which is the experimental treatment, is applied in the regular program. “Interactive” effects of testing implies that a pre-test influences how individuals are impacted by an intervention. If the performance of an experimental group after an intervention has been impacted by the pre-test, the findings will not apply to the general public that is not going to receive a pre-test. As a result, it may be crucial that you measure the impact of pre-testing itself.
History effects: History effects mean events which take place in the environment that alter the conditions of a research, impacting its end result. Such a history event can occur prior to the start of an experiment, or between the pre-test and post-test. History events are usually framed as risks to internal validity instead of external validity. They impact the results of an experiment in a fashion which threatens its internal validity when the history effect (a) modifies the scores on the independent and dependent variables, and (b) alters the scores of one group more than other.
External validity is a key criterion in research. Keep in mind that all of the threats to internal validity are also threats to external validity in research.