A large number of people were asked to answer only the odd-numbered questions on a test. Later, they were asked to answer only the even-numbered questions. By comparing each person's two scores, a determination of the test's __________ reliability could be made.

Split-half reliability divides a test into two equivalent halves (odd vs even items) and correlates the scores. This estimates internal consistency reliability - whether all items on the test measure the same construct. It's different from test-retest (same test twice) or equivalent forms (two different versions of the same test).