In research that uses statistics as the primary form of measurement, validity is an important aspect of the research design. Validity refers to the accuracy of a measure. A measure is valid, in general, when it measures what it is designed to measure. Internal and external validity are the two primary types of validity. However, they reflect different aspects of a study's design and results.
In order to understand internal and external validity, it is important to understand what the term validity itself means with regard to research design. Validity has to do with the strength of conclusions made about research. Often, validity is referred to as how much truth exists within a given research design. A researcher who asks, "Does my study measure what it says it measures?" is looking to make her study valid.
Overall, there are several different types of validity, although, internal and external validity are the two primary types. Other types of validity include conclusion validity (which looks at the relationship between a specific program and research outcomes) and construct validity (which looks at how concepts are defined in the study versus what is actually measured).
Internal validity can also be thought of as causal validity. That is, high internal validity can show strong evidence of causality. For instance, if there are two variables, and one appears to cause the other, the extent to which this relationship is true, depends on the internal validity of the design of the research. This is true for both qualitative and quantitative research designs.
There are different types of "threats" to internal validity; there are certain events or circumstances that can lower the internal validity of a study. For example, a single group threat occurs when only one group is being studied. Internal validity will be low for a single group, since it is difficult to know if the research affected those participants, or if results were due to their being members of that group. The use of control or comparison groups can reduce single-group threat.
External validity has to do with whether or not the study as a whole can be generalised. If study results can be generalised across other populations of people, settings, outcomes, times and treatment variations, then the study is thought to have good external validity. This is true for both quantitative and qualitative research designs.
There are certain factors that can threaten external validity, such as how the subjects are selected. If subjects are not selected randomly, then the results cannot be generalised, since results will only refer to people from a select group.
- Comstock/Comstock/Getty Images