Fiabilidad de los instrumentos de medición en ciencias de la salud

Abstract

This article presents the most frequent types of reliability that can be evaluated when using measurement instruments or procedures to perform research in the health sciences and provides examples for each case. The most frequently used statistical tests according to the type of reliability under evaluation and the type of variables measured are also discussed. Reliability provides information on the reproducibility of the results obtained by the procedure used for measurement; it is the degree of stability achieved in the results when a measurement is repeated under identical conditions. The four elements that must be considered when evaluating reliability are: a) interobserver reliability, which refers to the consistency between two distinct observers when they evaluate the same measurement in a single individual, b) intraobserver reliability, which evaluates the degree of consistency in the same observer when performing a measurement, c) test-retest reliability, which indicates the extent to which an instrument provides similar results when applied in a single individual on more than one occasion but under identical conditions, and d) internal consistency, which is the property that defines the level of agreement or conformity of a set of measurements within themselves. The most appropriate statistical tests described in the article according to the type of data to be measured are kappa index, the weighted kappa index, the intraclass correlation coefficient and Crohnbach's alpha.