Lecture 12: Validity

validity

The extent to which a test measures what it was designed to measure

Test items (content) → measure (construct)  →

other criteria (measure A, measure B, measure C)

face validity

the appearance of a test seems to promote its accuracy; good for marketing a test

content validity

the extent that a test’s questions/operations/etc…reflect the real range of responses of the entire domain of skills, understandings, behaviors that the test is supposed to measure

the domain of items in content validity

must be carefully specified so responses are representative to the universe of situations sample by the test

early drafts of a test are

piloted (items are analyzed; some stay, some are omitted)

concurrent validity

the degree a test relates to another measure of the same construct (ex: stanford binet intelligence test scores X wechster adult intelligence test scores)

convergent validity

the degree a test relates to other measures that it should, theoretically (ex: does the smith mental ability and resource test (SMART), and intelligence test, correlated with other measures associated with intelligence)

criterion related validity

comparing a test’s scores with other (criteria) representative of measures

types of criteria in criterion-related validity

concurrent (at the same time), predictive (at 2 time points), convergent (theoretically should relate), divergent/discriminant (shouldn’t relate)

construct validity

the degree a test really represents the concept it purports to measure

concepts in construct validity are

indirectly (operationally) measured

construct validity is supoorted by

data analysis (within the test and between other measures) and theoretically plausible

validity (2); research design

do the methods/design of a study lead to truly valid (or statistically significant) results? and eliminate bias from confounding variables (internal validity)?

external validity

can results be replicated

internal validity

eliminate bias from confounding variables

ecological validity

can results be applied outside of research

a study’s methods and design should

have a rationale/justification for why they were selection