In the academic and scientific arena, students and researchers often find themselves involved in quantifying or validating concepts as part of their clinical and laboratory research work. The objective is to ensure that the methods and measurements being used lead to accurate and reliable results that can be validated and replicated. This article examines the concept of validity, specifically discriminant validity, discusses how it differs from convergent validity, and shares tips on how to measure discriminant validity effectively.
What is discriminant validity?
Discriminant validity assesses whether two tests that measure different constructs are unrelated and whether they have any correlation. In other words, this type of validity determines whether measurements or concepts that are supposed to be different from one another (unrelated) are, in fact, not related. It assesses whether a measurement tool distinguishes between different constructs, ensuring that each construct remains unique and independent. By confirming that constructs are distinct, researchers can ensure that their findings are both precise and meaningful, thereby strengthening the overall validity of their studies.
Discriminant validity emerged as an essential concept in 1959. It revealed the extent to which an idea or measure is distinct from others. For example, when implementing and assessing an employee satisfaction survey, the scale measuring job satisfaction should not inadvertently measure workplace stress as well. In this case, two different constructs are being measured: job satisfaction and motivation.
We can measure job satisfaction based on how satisfied employees are with their jobs. Here, factors like pay, facilities available at work, and professional relationships are assessed. Under workplace motivation, the aspects that are evaluated include perks in terms of salary bonuses, prospects for professional growth, etc. Once the survey is administered under the two different constructs, the responses will be analyzed to understand the relationship between the two.
There is a very low correlation between the scores of the two constructs of job satisfaction and workplace motivation, as they are two distinct constructs. Therefore, it is clear that there is high discriminant validity in the test or measure employed. In other words, discriminant validity helps researchers to establish variance in different concepts, leading to accurate and reliable tests or survey results.
Discriminant vs. Convergent Validity
Discriminant validity and convergent validity are two essential components of construct validity in social sciences. Construct validity refers to the extent to which a test accurately measures the concept or theory it was designed to assess. Researchers evaluate both discriminant and convergent validity together to determine whether a test truly measures the intended construct or trait.
Convergent validity examines whether a test correlates with other tests that measure the same or similar constructs. On the other hand, as we have seen, discriminant validity evaluates whether tests measuring different constructs display little or no correlation, as expected. Students and researchers need to note that for a test or measure to demonstrate high construct validity, it must exhibit strong discriminant validity and convergent validity in equal measure.
Keep in mind that it is important to evaluate convergent validity first before assessing discriminant validity. Research teams should develop methods to incorporate discriminant and convergent validity during the test or measure construction phase.
Measuring Discriminant Validity
Having discussed what discriminant validity is and how it differs from convergent validity, let us now try to understand how discriminant validity is measured. Similar to other validity methods, discriminant validity can be measured using various statistical methods. One can use correlation analysis (Pearson’s (r)) to calculate the correlation coefficient between two measures.[1] This correlation analysis can demonstrate high discriminant validity with a low correlation coefficient that is below 0.85.
Another method used to measure discriminant validity is the Fornell-Larcker Criterion, which is a method used in structural equation modelling (SEM). The Fornell-Larcker Criterion method compares the square root of the average variance extracted (AVE) for each construct with the correlation with other constructs.[2] Researchers can also employ cross-loading analysis where, for example, survey questions are primarily associated with the intended constructs rather than other constructs. Here, the indicators or test questions will be significantly loaded on the intended construct and much less on the others.
Discriminant validity is also referred to as divergent validity in specific disciplines. As with other types of validity, discriminant validity is also critical in ensuring meaningful and accurate measurement in research. If researchers fail to use discriminant validity, there is a possibility of having too many concepts or constructs that appear similar and can also combine together, leading to problematic and inconclusive results.
With discriminant validity, researchers and students can ensure that they work with robust theoretical frameworks while enhancing their research and study findings. This significantly contributes to the overall research process while also leading to the conceit, testing, and development of new theories and ideas. When discriminant validity is established in a research process, the research team can be confident in the credibility of their study findings.
They will also be assured that the tests and measures they have introduced are accurately focused on the intended constructs. It should also be emphasized that by integrating different types of construct validity in the tests and measures used in research, the findings and conclusions reached will also be widely accepted, thereby increasing the impact of the research endeavour.
References:
- https://www.sciencedirect.com/topics/computer-science/pearson-correlation
- https://www.analysisinn.com/post/discriminant-validity-through-fronell-larcker-criterion/#:~:text=The%20Fronell%2DLarcker%20criterion%20is,construct%20and%20any%20other%20construct.
R Discovery is a literature search and research reading platform that accelerates your research discovery journey by keeping you updated on the latest, most relevant scholarly content. With 250M+ research articles sourced from trusted aggregators like CrossRef, Unpaywall, PubMed, PubMed Central, Open Alex and top publishing houses like Springer Nature, JAMA, IOP, Taylor & Francis, NEJM, BMJ, Karger, SAGE, Emerald Publishing and more, R Discovery puts a world of research at your fingertips.
Try R Discovery Prime FREE for 1 week or upgrade at just US$72 a year to access premium features that let you listen to research on the go, read in your language, collaborate with peers, auto sync with reference managers, and much more. Choose a simpler, smarter way to find and read research – Download the app and start your free 7-day trial today!
