the validity of a measurement is the correlation between this measurement and the true measurement of the construct.
this is impossible because we can never have a "true" measurement of a abstact construct that is why we call it abstract
construct validation
since we cant have a direct measurement of an abstract construct, we take an indirect approach. we take an "end-run" around the problem by measuring a construct in multiple ways
each measurement provides a different perspective on the construct
correlate variable that we can measure
content validity
looks at the applicability and breadth of the measurements of the construct
criterion-related validity
two types: concurrent criterion validity
predictive criterion validity
Concurrent criterion validity
the criterion is measured at the same time as the test
Predictive criterion validity
the criterion is measured later, after the test
ex. pregnancy test, fighter jet pilot test, firefighter test
convergent validity
multiple tests of the same thing
usually involves correlating a new test of a construct(ex. intelligence) with a well established test of the same construct
discriminant validity
measurements should not correlate with different and irrelevant constructs
in other words, your measurement should discriminate between what you want to measure and things you dont want to measure
experimental manipulation
use our measurement as a dependent variable in an experiment
create an independent variable that should have an effec on our measurement (DV)
ex. we created a measurement for mood. we then manipulate mood in an obvious way and see whether it influences our measurement
we do an experiment to test whether an independent variable that is known to have a effect on an dependent variable that we are not sure about