Assessment and Diagnosis

Reliability is the consistency, stability, or repeatability of test scores. A reliable test yields similar results across administrations, items, or raters. It is a necessary but not sufficient condition for validity.

Test-retest reliability is assessed by administering the same test to the same people on two occasions and correlating the scores. It measures temporal stability of the construct over time.

Carryover or practice effects: examinees remember items or improve from prior exposure, artificially inflating the correlation. Long intervals introduce true change in the construct, lowering it.

It is the correlation between scores on two equivalent versions of a test built from the same content domain. It controls for memory/practice effects that affect test-retest reliability.

Internal consistency reflects the degree to which items on a test measure the same construct, estimated from a single administration. Common indices are Cronbach's alpha, KR-20, and split-half reliability.

Cronbach's alpha estimates internal consistency as the mean of all possible split-half correlations; it is appropriate for items scored on a continuous or multi-point scale. Values increase with more items and higher inter-item correlations.