RELIABILITY AND VALIDITY

Reliability and Validity

Reliability and Validity

RELIABILITY

Reliability refers to the consistency of a measure's results for each application of the measure. Reliable measures demonstrate variation that is strictly attributable to manipulation of the variables under investigation.

VALIDITY

Validity is the degree to which a test measures what it claims to measure. If the test is not valid, then results cannot be accurately interpreted and applied.

Types of Reliability

Test-Retest Reliability

The most common form of reliability measure, test-retest reliability refers to correlating multiple applications of a single measure, conducted at different times.

Generally, r = 0.80 is indicative of a reliable measure.

Multiple Forms

Similar to test-retest reliability, multiple forms reliability refers to the application of equivalent forms of a measure (e.g. the same measure, with differently worded questions), and their correlation.

As before, r = 0.80 is indicative of a reliable measure.

Multiple forms reliability has an advantage over test-retest reliability, in that it can be applied in a single session of testing. In this circumstance, subjects are given both forms of the measure concurrently.

Types of Validity

Face Validity

A subjective form of validity measure, which associates the variable of interest with the proposed study variable, by relying heavily on logic and common sense.

Given its nature, face validity should only be considered as a starting point.

Content Validity

A less subjective form of validity measure than face validity, although it does extend from face validity, which relies on an assessment of whether the proposed measure incorporates all content of a particular construct.

Criterion Validity

A much more objective measure than both face validity and content validity, criterion validity relies on comparison between the proposed measure and a measure already established to be valid, as concerns the variable of interest.

Concurrent Validity

In applying your proposed measure and a measure with established validity to the same data set, concurrent validity refers to strong correlation between the measures' results.

A correlation of r = 0.50 is the minimum acceptable for declaring concurrent validity.

Predictive Validity

A valuation of your proposed measure's ability to predict future events.

Unlike concurrent validity, predictive validity first requires a collection of scores on the measure, followed by a collection of scores on the criterion measure at a later time.

Construct Validity

A complex validity measure, first identified by Lee J. Cronbach and Paul E. Meehl in 1995, construct validity evaluates whether your proposed measure correlates with all concepts related to the theory under investigation.

Example of validity in human service research

In Appendix

Example of reliability in human service research

Can be considered ...