Convergent and Discriminant Validity
If you're struggling to understand how convergent and discriminant validity work together to help establish construct validity, you are not alone! The vast majority of people who try to learn about construct vality initially have trouble getting a handle on the way convergent and discriminant validity work.
In an effort to help you pin down the meanings of convergent and discriminant validity, I'd like to share a paragraph that comes out of a book authored by Anne Anastasi, a highly respected individual who made enormous contributions to the field of testing many years ago. Her discussion of convergent and discriminant validity was brief and clear, and it remains instructive to this day.
Here's Anastasi's paragraph:
"In a thoughtful analysis of construct validation, D. T. Campbell (1960) pointed out that, in order to demonstrate construct validity, we must show not only that a test correlates highly with other variables with which it should theoretically correlate, but also that it does not correlate with variables from which it should differ. In an earlier article, Campbell and Fisk (1959) described the former process as convergent validation and the latter as discriminant validation. Correlation of a quantitative reasoning test with subsequent grades in a math course would be an example of convergent validation. For the same test, discriminant validity would be illustrated by a low correlation with scores on a reading comprehension test, since reading ability is an irrelevant variable in a test designed to measure quantitative reasoning."
OK. That's the paragraph I wanted to share. I hope you found it as clear and as instructive as I did. I also hope it drove home the point that discriminant validity is demonstrated by low (near zero) correlations, not high negative correlations.
P.S. If you haven't done so already, check out the 2nd of Chapter 4's "on-line resources." Via a link, you'll be taken to a page of another Website contains a more lengthly (and more colorful) discussion of convergent and discriminant validity.
Copyright © 2012
Schuyler W. Huck