Validity Flashcards
What is validity?
Does it measure what it intends to measure?
Face/ Content validity
Whether the measure APPEARS to measure what it should.
Judged by experts or experienced clinicians
Problems with face/ content validity
:( susceptible to experimenter bias
:( doesn’t always reflect accurate knowledge
:( may be a new area of research
:( lacks empirical support
Criterion validity
The extent to which a variable is related to a relevant outcome- is this measure related to something it should be?
1) CONCURRENT VALIDITY- measure compared against another standard criterion for that construction (e.g. a self-report measure of anxiety compared with clinician ratings)
2) PREDICTIVE VALIDITY- score on the measure compared with score obtained on a future criterion (e.g. score on a new reading test use to predict how many books a child is reading 6 months later)
For both, correlations are used to assess the degree of criterion validity
Construct validity
1) CONVERGENT- degree to which new scale correlates with measures of a similar construct (e.g. new depression scale with a standard, widely used depression scale)
2) DISCRIMINANT- doesn’t correlate with measures of a dissimilar construct (e.g. depression uncorrelated, or only weakly correlated, with IQ)
Multitrait-multimethod matrix
(Campbell & Fiske, 1959)
Model for testing construct validity (convergent and discriminant)
Used to measure multiple traits with multiple methods
Monotrait-heteromethod shows convergent validity (correlates the same trait measured with different methods- should be a higher correlation across the diagonal of the box than all others in the box,
Monotrait-monomethod (highest correlation should be the top diagonal as this is correlated with itself on the same method)
Heterotrait-monomethod shows divergent validity (different traits with the same method- different traits should not correlate highly with each other)