Specifically I'm thinking of a simplified division whereby validity is divided into:

  1. Construct validity
    1a. Convergent validity
    1b. Discriminant validity

  2. Criterion related validity
    2a. Predictive validity
    2b. Concurrent validity

This division leaves out some common concepts (e.g. face validity, other types of criterion validity), but it's for undergraduates taking their first course in statistics. I'm required to teach using this division.

I'm looking for examples, mnemonics, diagrams, and anything else that might help me explain the division in a memorable and intuitive way.

One thing I'm particularly struggling with is a clear way to explain the difference between concurrent validity and convergent validity, which in my experience are concepts that students often mix up.

  • 1
    $\begingroup$ One thing that people often misappreciate, in my own view, is that they think construct validity has no criterion. But any validity must have a criterion. Either external or internal. A construct is an internal criterion, and an item is being checked to correlate with that criterion, the latter must be therefore modeled. If one doesn't formulate the internal criterion as such self-contained entity the checking of correlations within the set of items will be an assessment of interitem homogeneity/interchangeability which is one of facets of reliability, not validity. $\endgroup$
    – ttnphns
    Jun 2, 2015 at 7:05
  • 1
    $\begingroup$ There's not going to be one correct answer that will be memorable and intuitive to you, I'm afraid. But there are innumerable book chapters, articles, and websites on this topic. Good luck. $\endgroup$
    – rolando2
    Jun 2, 2015 at 20:40

1 Answer 1


Both convergent and concurrent validity evaluate the association, or correlation, between test scores and another variable which represents your target construct. Here is the difference:

Concurrent validity tests the ability of your test to predict a given behavior. For instance, verifying whether a physical activity questionnaire predicts the actual frequency with which someone goes to the gym. You could administer the test to people who exercise every day, some days a week, and never, and check if the scores on the questionnaire differ between groups.

Convergent validity examines the correlation between your test and another validated instrument which is known to assess the construct of interest. In this case, you could verify whether scores on a new physical activity questionnaire correlate to scores on an existing physical activity questionnaire.

Here is an article which looked at both types of validity for a questionnaire, and can be used as an example: https://www.hindawi.com/journals/isrn/2013/529645/ [Godwin, M., Pike, A., Bethune, C., Kirby, A., & Pike, A. (2013). Concurrent and Convergent Validity of the Simple Lifestyle Indicator Questionnaire. ISRN Family Medicine, 2013, 1–6. https://doi.org/10.5402/2013/529645]

A book by Sherman et al. (2011) has a chapter which describes the types of validity you mention - which are also part of the 'tripartite model of validity.' You may be able to find a copy here https://www.researchgate.net/publication/251169022_Reliability_and_Validity_in_Neuropsychology

The reference for the chapter is [Sherman, E. M. S., Brooks, B. L., Iverson, G. L., Slick, D. J., & Strauss, E. (2011). Reliability and Validity in Neuropsychology. In The Little Black Book of Neuropsychology (pp. 873–892). Springer US. https://doi.org/10.1007/978-0-387-76978-3_30]

  • $\begingroup$ please add full references for your links in case they die in the future $\endgroup$
    – Antoine
    Apr 15, 2020 at 21:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.