*Geek Box: Internal vs. External Validity

‘Internal validity’ is the term for the criteria that are considered necessary for a trial to demonstrate causation. Randomisation, strict double-blinding, placebo control, no other variables introduced, a well-defined population, all give a trial high internal validity. However, one limitation of aiming for high internal validity is that we only know that the treatment causes the outcome in that circumstance; we don’t know that the results will automatically apply elsewhere.

The question of whether the results of a trial apply in the real world is known as ‘external validity’. Trials with high external validity are known as ‘pragmatic trials’, which may be randomised trials, but do not require blinding, and are often conducted in the real-world, or a specific clinical setting. Because these trials are conducted in the setting in which the results will be applied, they have high external validity.

There is no right or wrong when it comes to the spectrum of trials from rigorous control to pragmatic. It depends on the research question being asked, and the outcome. Pragmatic trials are underused in nutrition science, but there is more discussion about purposefully setting up more intervention studies on the pragmatic end of the spectrum, as the results will be more applicable in the population [external validity]. However, for other questions – like the effects of a specific diet composition on certain outcomes – high internal validity is still needed, to control factors that may influence the outcome if the trial was conducted in free-living conditions.