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

Academic production has increased in the area of health, increasingly demanding high quality in publications of great impact. One of the ways to consider quality is through methods that increase the consistency of data analysis, such as reliability which, depending on the type of data, can be evaluated by different coefficients, especially the alpha coefficient. Based on this, the present review systematically gathers scientific articles produced in the last five years, which in a methodological manner gave the α coefficient psychometric use as an estimator of internal consistency and reliability in the processes of construction, adaptation and validation of instruments. The identification of the studies was conducted systematically in the databases BioMed Central Journals, Web of Science, Wiley Online Library, Medline, SciELO, Scopus, Journals@Ovid, BMJ and Springer, using inclusion and exclusion criteria. Data analyses were performed by means of triangulation, content analysis and descriptive analysis. It was found that most studies were conducted in Iran (f=3), Spain (f=2) and Brazil (f=2). These studies aimed to test the psychometric properties of instruments, with eight studies using the α coefficient to assess reliability and nine for assessing internal consistency. All studies were classified as methodological research when their objectives were analyzed. In addition, four studies were also classified as correlational and one as descriptive-correlational. It can be concluded that though the α coefficient is widely used as one of the main parameters for assessing internal consistency of questionnaires in health sciences, its use as an estimator of trust of the methodology used and internal consistency has some critiques that should be considered.