Reliability

This entry focuses on reliability, which is a desired property of the scores obtained from measurement instruments. Therefore, reliability is relevant in a variety of contexts, such as scores on multiple-choice questions on an achievement test and Likert-type scale responses on a survey questionnaire. The concept of reliability is premised on the idea that observed responses are imperfect representations of an unobserved, hypothesized latent variable. In the social sciences, it is typically this unobserved characteristic, rather than the observed responses themselves, that is of interest to researchers. Specifically, reliability serves to quantify the precision of measurement instruments over numerous consistent administration conditions or replications and, thus, the trustworthiness of the scores produced with the instrument. This entry discusses several frameworks for estimating reliability.