Quant #08: Reliability and Validity

When a variable consists of multiple items (questions), it helps to ensure that all the items reflect the underlying concept. Reliability refers to the property of measurement that assesses the extent to which various items reflect the underlying concept. One of the most commonly used tests of reliability is Cronbach's alpha.

In addition to being reliable, a measure must be valid. The validity of a measure concerns its degree of accuracy, i.e., a valid measure assesses what it is intended to assess, assesses only what it is supposed to assess, and assesses all aspects of what it is supposed to assess. One of the most common validity tests is content validity, which is concerned with the degree to which various items in a questionnaire collectively cover the material that the questionnaire is intended to cover.

Although it is important to use content validity in a study, it is often desirable to use more than one validity test. Thus, other validity tests are conducted; these can include tests such as construct validity. Construct validity can be defined as the evaluation of the extent to which a measure assesses the construct it is intended to measure. It is regarded as one of the most important validity tests as it deals with what the instrument is measuring. Factor analysis is often used to conduct construct validity (i.e., examining whether a certain group of items together measure a pre-specified abstract construct.)