click below
click below
Normal Size Small Size show me how
test validity & reli
Question | Answer |
---|---|
test validity | test must be valid - it must actually measure what it is supposed/designed to measure |
content validity | the content of the test, including all its subtests and items, adequately measures what it is designed to measure |
Criterion-related validity | the test can adequately predict performance on other tasks |
construct validity | the test provides a good reflection of the theory on which it is based and that there is empirical evidence supporting the theory |
internal validity | design of the research and the procedure used to conduct the study if a study has gaps or flaws, then the study may be considered lacking internal validity |
internal validity example | researchers need to be confident that the specific method used to conduct a study actually tests the hypothesis and that the hypothesis has been tested in a convincing way design of research - matched, independent group, repeated measure |
external validity | the conclusions can be generalised/applied to the population from which the sample used in the study was drawn |
test reliability | refers to the ability of a test to consistently measure what it is supposed to measure each time it is given |
test-retest reliability | involves giving the test to the same group of people on two different occasions and then comparing the two sets of scores. if the test is reliable then each person should achieve similar scores on the subtests and the overall test each time they do it |
test-retest reliability problem | test takers may benefit from 'test practice effects' and perform better on the test when re-tested because of their prior experience with the test |
eliminating test-retest reliability problem | parallel-forms reliability split-half reliability |
parallel-forms reliability | alternate forms reliability giving another version of the same test instead of using exactly the same test twice. if scores on the two tests are similar it suggests that they measure the same thing |
split-half reliability (internal consistency) | involves dividing the original test into halves and examining the correlation between scores on each half. if the halves of the test are composed of items with similar levels of difficulty each individual should have similar scores on the halves |
inter-rating reliability | involves checking that different administrators get similar results from it. i.e. two admins diagnose 2 diff IQ |