2013, Number 1
El problema de la comparación de métodos. El caso de los errores proporcionales e iguales
Full text
Language: Spanish
References: 10
Page: 179-192
PDF size: 531.72 Kb.
ABSTRACT
The performance of the ordinary least-squares regression, the Passing-Bablok regression, and the Deming regression with the Α coefficient adjusted to the unity, or estimated from the within-replicates variances, as solutions of the methods comparison problem, was assessed by means of statistical-mathematical simulation techniques for the case of proportional but equal analytical errors. Ordinary least-squares regression returned biased estimates of the slope of the methods comparison straight line, along with an elevated rejection factor of the null hypothesis Ho: β = 1. It is likely Deming regression to be sensitive to the dependence of analytical imprecision regarding the level of concentration of the analyte. Use of Passing-Bablok regression should be promoted as an acceptable solution (although not an adequate one) of the methods comparison problem through its inclusion into the available statistical management packages. Santana Porbén S. The problem of methods comparison. Proportional and equal errors.REFERENCES
Delgado Ramos A, Ramos Salazar R, Martínez Canalejo H, Santana Porbén S. Procederes de regresión lineal como soluciones al problema de la comparación de métodos. I. Errores analíticos constantes e iguales. RCAN Rev Cubana Aliment Nutr 2010;20:191-67. Reimpreso de: Contacto Químico [Michoacán] 2007;2(6):21-3.
Ramos Salazar R, Delgado Ramos A, Martínez Canalejo H, Santana Porbén S. Procederes de regresión lineal como soluciones al problema de la comparación de métodos. II. Errores analíticos constantes pero diferentes. Rev Cubana Aliment Nutr 2010;20:338-50. Reimpreso de: Rev Mex Patol Clín 2001;48:223-32.