Abstract

Psychometrics refers to the measurement of hypothetical constructs, such as cognitive abilities or attitudes, that are not directly observed but instead are inferred. Psychometrics also refers to special statistics about the properties of scores from psychological tests such as score reliability coefficients and validity coefficients, among other kinds. Described in this entry are the origins and characteristics of classical measurement theory, which was developed from about 1900–1960 and is still relevant to this day. Classical measurement theory is the basis for introducing students to psychological measurement. The same principles should also guide the selection of measures for use in research projects that involve psychological constructs, which are pertinent in many human studies. Also briefly considered are two approaches to psychometrics in modern measurement theory. These newer methods build on classical measurement theory or depend on statistical techniques that were impractical before the availability of relatively affordable fast computers with large memory capacities. One is generalizability theory, which extends the classical perspective on score reliability. The other modern approach is item response theory, also known as latent trait theory, which consists of families of mathematical models that relate observed responses on test items to underlying continuous variables, or latent traits, that represent target hypothetical constructs. Journal article reporting standards for psychometrics are also discussed.