Resumo

LIVACIC-ROJAS, PABLO; VALLEJO, GUILLERMO  e  FERNANDEZ, PAULA. PROCEDIMIENTOS ESTADÍSTICOS ALTERNATIVOS PARA EVALUAR LA ROBUSTEZ MEDIANTE DISEÑOS DE MEDIDAS REPETIDASALTERNATIVE STATISTICAL PROCEDURES TO ASSESS THE ROBUSTNESS USING REPEATED MEASURES DESIGNS. rev.latinoam.psicol. [online]. 2006, vol.38, n.3, pp.579-598. ISSN 0120-0534.

The purpose of this investigation was assessing the robustness of a series of new procedures under the traditional ANOVA approach to analyze some data corresponding to repeated measures by means of Monte Carlo simulation. In order to study the efficacy of these procedures for the control of Type I rate of errors, four variables were handled: sample size; structure of dispersion matrixes; relationship between the group sizes and the dispersion matrices, and population distribution pattern. Results suggest that the Improved General Approach, Proc Mixed program with the Kenward-Roger solution, Brown-Forsythe and Welch James procedures offer a very good behavior to control the Type I rate of errors both for the main effect of the measurement opportunities as well as the effect of interaction, under all the assessed conditions.

Palavras-chave : Robustness; ANOVA; Type I rate of error; alternative procedures; SAS Proc; Mixed; Monte Carlo simulation..

        · resumo em Espanhol     · texto em Espanhol     · Espanhol ( pdf )