Retest reliability

from Wikipedia, the free encyclopedia

Retest reliability (from English retest and reliability "repeat test reliability") generally designates the degree of accuracy with which a certain characteristic is repeatedly measured by a certain test procedure in psychology . In a narrower sense, it describes the degree of agreement between the test results for the same test subjects and with the same test for several repeat tests. The retest reliability simply means how reliable psychometric tests are.


In quality assurance, the majority of the measuring procedures in psychology ( psychometric tests) consistently show a low level of reliability because psychological tests are subject to falsifications that are not based on the measured characteristic. An exception are quality-assured, standardized IQ tests and some Big Five personality tests based on the OCEAN model .

Reliability measurements

In general, reliability ("reliability") is a generic term for a number of concepts that only concern certain aspects of measurement accuracy. The value for the measured reliability is expressed by the correlation coefficient of the two tests (reliabilities greater than 0.80 are desired).

A reliability measurement with the aid of retest reliability is problematic because the stability of the measured feature is assumed. This assumption has not yet been adequately proven empirically; many psychological characteristics seem to change over time, i.e. H. to be variable.

Another problem is the fact that a person can remember the test on the new run. This can lead to learning effects in intelligence tests, which influences the reliability measurement and can lead to pseudo-reliability.

Other methods of measuring reliability are the parallel test reliability, the split-half method and the internal consistency.