- Joined
- Nov 9, 2008
- Messages
- 90
- Reaction score
- 0
Anyone know of some resources so I can teach it to myself? Thanks!
Construct validity
- Thread starter INTPThinker
- Start date
What do you want to know about it? Are you designing a test, evaluating research, trying to answer a trivia question?
Quite simply, construct validity is an overall measure of whether or not a measurement instrument measures what it is supposed to measure. It consists of two "sub-validities" (that's not a real word, BTW): convergent validity (think, "coming together") and discriminant validity (think, "where's Waldo?").
Convergent validity measures the extent to which a measure meshes with similar scales. For example, if I'm creating a new depression scale, I would compare it established scales (e.g. BDI-II) to see if it measures depression in a way that is similar to other depression instruments.
Discriminant validity is the extent to which a measure diverges or differs from other tests. For example, I want to be sure my new depression scale measures depressive symptoms while excluding anxiety symptoms (because of comorbidity issues).
There is also nomological validity, but it can involve some high-level math and stats and it probably isn't necessary for most applications unless you're trying to publish a new scale or something (in which case I doubt you would be asking about construct validity
.
For MOST purposes, if you establish both convergent and discriminant validity, you have construct validity. I can be more specific if you like, but you have to ask a more specific question. Good luck!
Quite simply, construct validity is an overall measure of whether or not a measurement instrument measures what it is supposed to measure. It consists of two "sub-validities" (that's not a real word, BTW): convergent validity (think, "coming together") and discriminant validity (think, "where's Waldo?").
Convergent validity measures the extent to which a measure meshes with similar scales. For example, if I'm creating a new depression scale, I would compare it established scales (e.g. BDI-II) to see if it measures depression in a way that is similar to other depression instruments.
Discriminant validity is the extent to which a measure diverges or differs from other tests. For example, I want to be sure my new depression scale measures depressive symptoms while excluding anxiety symptoms (because of comorbidity issues).
There is also nomological validity, but it can involve some high-level math and stats and it probably isn't necessary for most applications unless you're trying to publish a new scale or something (in which case I doubt you would be asking about construct validity
.For MOST purposes, if you establish both convergent and discriminant validity, you have construct validity. I can be more specific if you like, but you have to ask a more specific question. Good luck!
