ABSTRACT

To be a valid measurement, it MUST be reliable and objective, though the reverse is not true. Reliability and objectivity can be statistically examined and when there is no criterion measure, they may be taken as evidence (not proof) of validity. Test–retest is the strongest assessment of reliability, but since that requires two independent administrations of the test, additional split-half methods have been devised, such as Cronbach’s alpha. Other factors such as fatigue or learning may hurt reliability and must be considered.