reliability, alternate forms

n. The level of a test's measurement error determined by examining the consistency of the scores of two different tests constructed from the same content domain. Alternate forms reliability (also called parallel forms or equivalent forms reliability) is concerned with error caused by item sampling. Specifically, alternate forms reliability is designed to examine the error variance that can be attributed to the selection of a specific item or subset of items for inclusion in a measure. To calculate alternate forms reliability, two equivalent forms of a test are developed to measure the same construct or attribute (the specific items on the tests are different, but the tests are designed to be equivalent in target, difficulty, structure, etc.); the two different versions of the test are then administered to the same group of people on the same day (counterbalancing the order of administration). The scores of the two tests are then correlated to determine the reliability coefficient. Correlation coefficients closer to 0 represent low reliability while coefficients closer to 1 represent high reliability. While alternate forms reliability is a rigorous and valuable method of estimating a test's reliability, it is often shunned by test developers because of the constraints associated with developing an alternate version of the target test. – BJM