External Validity

Posted by: Karen Chapman-Novakofski, PhD, RDN on Monday, October 12, 2020

In JNEB, the Discussion section should compare and contrast the results of the research with other published studies. The reason for this is to provide the readers with an idea of the external validity of the study. That is, can results be generalized to other populations, settings, times? Is the sample representative of a larger whole? External validity is enhanced with randomization, which in turn heightens the representativeness of the sample. Replication also increases external validity.1
 
Comparison to other similar studies is a way of talking about replication, and as such, external validity. This type of interpretation of external validity can be found in Jung et al's 2 article in this issue. The authors compare their results to those of different populations of consumers relative to local food consumption beliefs, noting those where their results with college students are unique. They also point out in their Implications section how external validity may be enhanced by replicating their study with groups of different demographic characteristics and differing cultural backgrounds. The authors don't explicitly state “external validity,” but the meaning is implied.
 
In the article by Ares et al,3 the authors also compare their findings to similar publications, but the reader has to review the references to determine if these are also Uruguayan samples. Unfortunately, the reference titles do not share this information. However, the authors do provide an evaluation of how representative their sample is of the Uruguayan population as a whole. Because it is not, the findings can't be expanded to suggest the findings apply to all Uruguayan adults. However, many of their findings are similar to previously published work, with some unique results.
 
Pulling all these types of replications together is the aim of systematic reviews, for example, Robson et al's systematic review concerning family meal frequency, diet, and family functioning.4 The studies evaluated in this review used diverse terminology and measurement tools, making conclusions more difficult and calling attention to the need for a more uniform study design. However, the review has several findings that are consistent with another systematic review, and by replicating results we can enhance the external validity. The systematic review by O'Shea et al5 highlights the need for consistent reporting in simulation research, but also the need for more robust research in the area of this educational approach.
 
Within a manuscript, the Introduction lays the groundwork, the Methods explain the study and allow for replication, and the Results are always exciting. To me, though, it is the Discussion and Implications sections that offer a clear view of where this research contributes to the science of nutrition education and behavior. We can always read the abstract and get the “bottom line” results, but let's take a fresh look at the discussions. There is so much to learn. And if we want to use our research to develop policy, we must consider external validity.
 
This editorial was originally published in the October 2020 issue (Vol. 52, Issue 10) of the Journal of Nutrition Education and Behavior.