Which statement about test reliability would Cronbach's alpha assess?

Which statement about test reliability would Cronbach's alpha assess?

• A. It measures the correlation of test items with the external criterion.
• B. It compares the test with another form.
• C. It is a measure of internal consistency reliability.
• D. It assesses inter-rater reliability.

Answer: (C)

Cronbach's alpha measures internal consistency reliability, meaning how well the items on a test work together to measure the same underlying construct. It’s derived from the average inter-item correlations and the number of items: when items correlate well with each other, alpha increases, indicating the set of items hangs together as a coherent scale. This is not about how a test relates to an external criterion, nor about stability over time or agreement between scorers. Those are different reliability or validity concepts: correlations with a criterion reflect criterion-related validity, test-retest reliability looks at stability over time, and inter-rater reliability concerns scorer agreement. Interpreting alpha, values closer to 1 indicate better internal consistency, with common guidelines roughly describing .70 as acceptable, .80 as good, and very high values (.95+) potentially suggesting item redundancy. Keep in mind that alpha assumes the scale is measuring one dimension; for multidimensional tests, alphas for subscales or other approaches may be more appropriate.

Cronbach's alpha measures internal consistency reliability, meaning how well the items on a test work together to measure the same underlying construct. It’s derived from the average inter-item correlations and the number of items: when items correlate well with each other, alpha increases, indicating the set of items hangs together as a coherent scale.

This is not about how a test relates to an external criterion, nor about stability over time or agreement between scorers. Those are different reliability or validity concepts: correlations with a criterion reflect criterion-related validity, test-retest reliability looks at stability over time, and inter-rater reliability concerns scorer agreement.

Interpreting alpha, values closer to 1 indicate better internal consistency, with common guidelines roughly describing .70 as acceptable, .80 as good, and very high values (.95+) potentially suggesting item redundancy. Keep in mind that alpha assumes the scale is measuring one dimension; for multidimensional tests, alphas for subscales or other approaches may be more appropriate.