Reliability and Validity A Quick Review

Abstract

The purpose of this article is to provide a brief review of reliability and validity testing. These concepts are important to researchers who are choosing techniques and/or developing tools that will be applied and evaluated in diabetes education practice. Several types of reliability and validity testing are defined, and an easy-to-use check sheet is provided for research purposes. Following testing for the basic aspects of reliability and validity such as face and construct validity, a tool may be appropriate for use in practice settings. Those conducting comprehensive outcomes evaluations, however, may desire additional validation such as testing for external validity. Diabetes educators can and should incorporate rigorous testing for these important aspects when conducting assessments of techniques and tools relating to diabetes self-management training.

Desenvolvimento e validação de instrumento para avaliar o impacto de um programa de intervenção em comércios de alimentos em área de deserto alimentar/Development and validity of an instrument to evaluate the impact of an intervention in food stores at an urban food desert

“…(FITZNER, 2007) e considera-se que este tipo de validade é garantido também pelo processo de elaboração do instrumento (PASQUALI et al, 2004). …”

Psychometric properties of the Persian version of the Time to Relapse Questionnaire (TRQ) in substance use disorder

“…Values equal to or above 0.70 are often considered to be satisfactory. However, Fitzner (2007) considered a respectable reliability to be between 0.70 and 0.80 and a very good reliability to be between 0.80 and 0.90 (23). Furthermore, to evaluate the questionnaire's stability, test-retest reliability was measured using the intraclass correlation coefficient (ICC).…”

Estimadores de consistencia interna en las investigaciones en salud: el uso del coeficiente alfa

“…Sin embargo, por ser un método que correlaciona los ítems del test, el valor alfa puede ser modificado, donde un alto coeficiente α no siempre significará un alto grado de consistencia interna (10) . Es decir, el valor de coeficiente α aumenta cuando aumentan las intercorrelaciones entre los ítems, cuando se mide el mismo constructo del elemento.…”

“…Para el análisis de confiabilidad de los datos longitudinales es utilizado un test específico denominado coeficiente de correlación intraclase (CCI) para medidas continuas y el coeficiente de Kappa para las medidas dicotómicas y ordinales (2,3) . Para las investigaciones transversales, el test que evalúa la equivalencia puede ser verificado básicamente por dos métodos: 1) consistencia de formas paralelas, que consiste en un mismo instrumento con diferentes presentaciones y que producen los mismos resultados (4,5) ; 2) evaluación de consistencia interevaluadores o interobservadores, que es la capacidad de un test aplicado por múltiples evaluadores para conseguir los mismos resultados (5)(6)(7)(8)(9)(10)(11)(12) . Por otro lado, la consistencia interna puede ser evaluada por medio de cuatro métodos: 1) mitad dividida (5,(7)(8)(9)13,14) ; 2) correlaciones totales totales (13) ; 3) coeficiente de Kuder-Richardson (13) ; y 4) coeficiente alfa (α) (13) .…”