Una prueba de rangos para la alternativa de escala en dos muestras provenientes de la distribución lambda generalizada
Cargando...
Autores
Másmela Caita, Luis Alejandro
Director
Tipo de contenido
Document language:
Español
Fecha
Título de la revista
ISSN de la revista
Título del volumen
Documentos PDF
Resumen
Se construye y se estudian las bondades de una Prueba de Rangos Localmente Más Potente (PRLMP) para el problema de escala en dos muestras, construyendo la estadística de puntajes a partir de densidades de la DLG. Se compara en términos de su eficiencia relativa asintótica con las pruebas de Mood, Ansari-Bradley y Klotz. Se encontró que para muestras de poblaciones normales, algunas de las pruebas superan a las pruebas de Mood y Ansari-Bradley. Con muestras de poblaciones logística y exponencial doble un conjunto de pruebas propuestas supero las tres pruebas con que se comparó.
Abstract: The advantages a Locally Most Powerfull Ranks Test (LMPRT) are built and studied for the scale problem in two samples by building the score function using as function of density the density of the Generalized Lambda Distribution. It is compared in terms of their asymptotic relative efficiency with the tests of Mood, Ansari-Bradley and Klotz. It was found that for normal samples, some of the tests overcome to the tests of Mood and Ansari-Bradley. With logistics samples and double exponential samples a group of proposed tests defeated the three tests with which it was compared.
Abstract: The advantages a Locally Most Powerfull Ranks Test (LMPRT) are built and studied for the scale problem in two samples by building the score function using as function of density the density of the Generalized Lambda Distribution. It is compared in terms of their asymptotic relative efficiency with the tests of Mood, Ansari-Bradley and Klotz. It was found that for normal samples, some of the tests overcome to the tests of Mood and Ansari-Bradley. With logistics samples and double exponential samples a group of proposed tests defeated the three tests with which it was compared.