What is test-retest reliability?

What is test-retest reliability?

• A. Consistency of reports across different observers.
• B. Consistency of a measure when repeated at two points in time.
• C. The measure's relation to other measures as theory predicts.
• D. Reliability is required for validity.

Answer: (B)

Test-retest reliability measures temporal stability: if you administer the same measure to the same people on two different occasions, the scores should be similar if the underlying trait hasn’t changed and the measure is stable. The degree of similarity (often shown by a high correlation) indicates how reliably the instrument reproduces results over time. This is exactly what the option describing consistency of a measure when repeated at two points in time captures. Inter-rater reliability, about consistency across different observers, isn’t the same issue as stability over time. Validity-related concepts describe how a measure relates to other measures or constructs, not how consistent it is across repeated administrations. And while reliability is necessary for validity, the statement that reliability is required for validity isn’t the definition of test-retest reliability.

Test-retest reliability measures temporal stability: if you administer the same measure to the same people on two different occasions, the scores should be similar if the underlying trait hasn’t changed and the measure is stable. The degree of similarity (often shown by a high correlation) indicates how reliably the instrument reproduces results over time. This is exactly what the option describing consistency of a measure when repeated at two points in time captures.

Inter-rater reliability, about consistency across different observers, isn’t the same issue as stability over time. Validity-related concepts describe how a measure relates to other measures or constructs, not how consistent it is across repeated administrations. And while reliability is necessary for validity, the statement that reliability is required for validity isn’t the definition of test-retest reliability.