scholarly journals Type I Error Rates of the Kenward-Roger Adjusted Degree of Freedom F-test for a Split-Plot Design with Missing Values

2007 ◽  
Vol 6 (1) ◽  
pp. 66-80 ◽  
Author(s):  
Miguel A. Padilla ◽  
James Algina
Methodology ◽  
2013 ◽  
Vol 9 (4) ◽  
pp. 129-136 ◽  
Author(s):  
Pablo Livacic-Rojas ◽  
Guillermo Vallejo ◽  
Paula Fernández ◽  
Ellián Tuero-Herrero

We examined the selection of covariance structures and the Type I error rates of the Criterion Selector Akaike’s (Akaike’s Information Criteria, AIC) and the Correctly Identified Model (CIM). Data were analyzed with a split-plot design through the Monte Carlo simulation method and SAS 9.1 statistical software. We manipulated the following variables: sample size, relation between group size and dispersion matrix size, type of dispersion matrix, and form of the distribution. Our findings suggest that AIC selects heterogeneous covariance structure more frequently than original covariance structure. Specifically, AIC mostly selected heterogeneous covariance structures and displayed slightly higher Type I error rates than the CIM. These were mostly associated with main and interaction effects for the ARH and RC structures and a marked tendency toward liberality. Future research needs to assess the power levels exhibited by covariance structure selectors.


2005 ◽  
Vol 65 (1) ◽  
pp. 42-50 ◽  
Author(s):  
Christine E. Demars

Sign in / Sign up

Export Citation Format

Share Document