Let's say I have one variable called "perceived image", which is measured with 6 items (questions) that I adapted from previous literature. If I have a Cronbach's alpha > 0.7 (for reliability), do I need anyway to run a validity test? In case I must run one, which one? (I am using SPSS). Thanks!
1
-
$\begingroup$ I see there were some votes to close this as unclear - but i don't see why. If any of the people who voted to close it could explain what isn't clear, that would be good (and might convince me to vote the same way). $\endgroup$– Peter FlomCommented Sep 10, 2019 at 10:11
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
1
First, Cronbach's alpha is about reliability, not validity.
Second, whether you have to test the validity of a variable is up to you and your "boss" (or dissertation committee, journal editor, actual boss or whatever).
But, really, you should want to know if your variables are valid. If they are not, nothing you do will mean what you think it means. A full discussion of validity testing would be too long for an answer here. I will note the following simplified definitions:
Reliability is about whether your variable measures anything accurately.
Validity is about whether your variable measures what you think it measures.
answered Sep 5, 2019 at 11:41
-
$\begingroup$ thanks for your clarification. it does help me a lot! $\endgroup$– TomCommented Sep 10, 2019 at 7:54