Research design: measurement, reliability, and validity

Abstract

The concept of measuring constructs is discussed. An explanation of reliability and validity of measures is presented. Reliability is consistency in measurement over repeated measures. Reliable measures are those with low random (chance) errors. Reliability is assessed by one of four methods: retest, alternative-form test, split-halves test, or internal consistency test. Validity is measuring what is intended to be measured. Valid measures are those with low nonrandom (systematic) errors. Validity is assessed by one of three methods: content validation, criterion-related validation, and construct validation. It is particularly important for the researcher in the behavioral or social sciences to assure himself that measures used are both reliable and valid.