test-retest reliability

n. The level of a test's measurement error, determined by examining the consistency of test scores from multiple administrations of the same measure. Test-retest reliability utilizes time sampling to evaluate the error associated with administering a test on two or more different occasions. To calculate test-retest reliability, a test must be administered to the same group of people on at least two separate, specified occasions; then the correlation of the scores from the two testing sessions is calculated to determine the reliability coefficient. Correlation coefficients closer to 0 represent low reliability while coefficients closer to 1 represent high reliability. Test-retest reliability estimates are only meaningful when the target of the measurement is stable. For example, constructs such as intelligence are assumed to be stable and thus should not change much from one testing session to another, while other characteristics, such as reading ability, are not stable and are likely to change dramatically from age 5 to age 7. When interpreting test-retest reliability data, one must ensure that changes in test scores are not due to random variations in testing conditions or the impact of carryover effects, practice effects, and reactivity. – BJM