Reliability, Validity and Uni-Dimensionality: A Primer

Though measurement theory has traditionally got full attention and appreciation in scientific researches, its service in social sciences is young in years. Most of the popular definitions of measurement refer only to measuring objects or events in science; however, the phenomena to be measured in social sciences are mostly abstract ones, unobservable, latent, variables. Measurement in social sciences is hence to seek to construct these latent variables from aggregation of relevant observable items of questions or rating scale items in an interview questionnaire. These constructs are interpreted as manifestations of abstract traits. A number of criteria have come up to test the feasibility of this interpretation such as reliability, validity, uni-dimensionality, etc. The present paper seeks to present a primer on these three criteria.